This paper is published only on my web site, www.nbrigham.org.
It was originally written for a graduate class at the University of Michigan
School of Information.
If you prefer to see the pdf
version, click here.
If you would like to reproduce the paper or parts of it for a class or
discussion, you are free to do so, as long as you give the author full
credit.
I would also appreciate your letting me know it is being used by sending
a communication to
. If you would like to make more
extensive use of the paper, please ask permission at the above email address.
And
finally, if you have new information or suggestions on how to update it,
I would be most appreciative. The basic points are still valid, but I'd
be interested in seeing more recent analyses.
Abstract
This paper probes the history, potential, and reality of
Machine Translation in terms of the language needs of global communications.
In examining the nature of language, the literature reflects both a respect
for the need of peoples to define themselves through a local culture and
their urgency to function in a global world. This dilemma is echoed in
international online discussions.
It’s now widely recognized that global communications
must be available in as many languages as feasible, and MT could offer
a means to communicate both locally and globally. MT has long been associated
with an idealistic drive to find a universal foundation for all language,
but this has proved unfeasible because the linguistic task of reducing
languages into a universal set of symbols denies the very nature and cultural
value of language. What makes language irreducible is also what makes
it important.
As the need for MT and commercial applications grow, major
institutions and developers are designing niche products to fulfill specific
needs. The search for more effective general translation also continues,
and the programs now available to a broad public appear to offer tremendous
potential. However, their limitations must be recognized, and to use them
effectively, people have to make some accommodation to the machine. How
to do that without perverting the languages that translation is supposed
to capture is a challenge that must be met. While some ideas are presented
for how to tailor messages, norms, and expectations to the real abilities
of MT without destroying the ability to freely communicate, it’s
acknowledged that norms for communications will develop through experience
and experimentation.
Introduction
In today’s global economy, as more people increasingly need or
want to communicate with counterparts abroad, we are confronted with daunting
cultural and language divisions. The fact that most of the world’s
people can’t communicate in American English, the predominate language
of international commerce and communications, is reinforcing a global
digital divide.
Globalism has intensified the demand for translation. To reach customers
in Mexico and Canada, for example, businesses must offer packaging and
technical instructions in three languages. In 1994 Mexico closed Wal-Mart’s
doors for a day because the company failed to make sure all products were
available in Spanish.1 If the world’s
non-English speakers want to be more than passive receivers of information,
dynamic translation is needed. This paper investigates the global language
divide, particularly from the point of view of the average citizen impacted
by globalization, and the potential of Machine Translation (MT) to close
that gap. Today people who access the Internet now have at their fingertips
free programs they can use online or download, that claim to translate
growing numbers of languages. Can these programs revolutionize international
communications?
Machine Translation programs appear to be much better than
the programs available a few years ago, and there should be a way they
can be useful to people who can’t afford professional translation.
Can they help the global have-nots express themselves in their own language
without losing crucial meaning? Or must we change language itself for
international communications? This paper investigates the uses people
are making of MT, how much more progress can be expected, and the implications
for society and culture.
Language and Culture in a Global Economy
Nine out of ten global citizens don’t speak English.
The world’s people speak 3,000 to 8,000 different languages (depending
on what is defined as a dialect.2 But most of
the communication through new communications technology is in English,
even as that technology facilitates a vast new level of international
commerce and communications. Of the world’s 50 largest enterprises
in global information services (press, publishing, TV radio and cinema),
41 are based in the U.S. or United Kingdom.3
Language, says Manuel Castells, is often experienced as
the repository of community, thought structures, and values that people
refuse to relinquish. Castells points to Catalunya as a culture that maintained
a strong sense of community around a language that was suppressed over
centuries by ruling powers. Language “represents a system of codes,”
explains Castells, “crystallizing historically a cultural configuration”
and is “a fundamental attribute to self-recognition… a linkage
between the public and the private sphere and between the past and the
present.” In response to “homogenization by the ideology of
modernization [and materialism] and the power of global media,”
language is also “the trench of cultural resistance…”4
that also leads to growing “social fragmentation.”5
Ian Clark views the disintegration of nations “into a narrow
ethnic and linguistic form” as “pockets of resistance to globalism”
that both resist the “powerful homogenizing influence” of
today’s world and also carry “the potential for instability
and conflict.”6
Languages on the Internet
When Alis Technologies and the Internet Society studied
the languages used on the Internet in 1997, they found that English dominated
84% of all sites.7 In 1998, Colin Haynes wrote
that 80 percent of Web sites were created in English-speaking countries.8
Yet by 1999, 217 countries were connected to the Internet, ten times the
number that were online in the beginning of the decade.9
A June 1999 report from Computer Economics said that just 54 percent of
Internet users are English-speaking and predicted that that the proportion
will fall to 43 percent of users by 2005.10
People are beginning to use the Internet to communicate
in other languages. In late 1997 The Economist reported that in three
years Web sites posted in English had fallen from 98% to 82%.11
By 1998, Yoshi Mikami of Asia Info Network claimed that "99
% of the sites created in Japan are in Japanese."12
Major sites such as CNN and Yahoo are now translated into several languages.13
The Appel du Comité Européen Pour le Respect des Cultures
et des Langues en Europe (C.E.R.C.L.E.) has a web site translated into
11 languages, with the express intention to fight “the triumph of
a sole language and of a unique thinking.” Europe, it says, must
“claim its identity and affirm its personality” and “build
a radiant community of equal, united yet diverse peoples.”14
The Human Languages Page hosts links to dozens of Web pages in languages
like Wolof, Yiklamu, and Papiementu.15 Yet proponents
of globalism like Mansell and When see the emergence of “netizens
who form common language communities” as potential “barriers
to communications and knowledge sharing.” 16
The Search for Viable Machine Translation
The review of online international discussions presented
in Appendix A reveals personal struggles of people who want to enjoy cross-culture
communication. One of the themes coming out of these discussions is the
yearning for an automatic way to translate messages that, says Frenchman
Michel Elie, would “help preserve cultural diversity while making
the world a little smaller" 17
This yearning has deep roots. In the 19th century, mathematician
Carles Babbage convinced the British government to finance research on
a “computing machine” by promising that it would lead to automated
translation of spoken languages.18 In 1946 American
mathematician Warren Weaver was inspired by the British Colossus computer
success in breaking German encryption codes to launch the first Machine
Translation project.19 In 1948 Claude Shannon
of Bell Telephone Laboratories collaborated with Weaver on transforming
natural language to mathematically processable bits and bytes.20
To translate a text from Russian to English, Warren explained in a 1949
memo to the Rockefeller Foundation, “all I need to do is strip off
the code in order to retrieve the information.”21
The first MT conference was held in the U.S. in 1952. 22
When the Georgetown Automatic Translating System (GAT) was
unveiled in 1954, it was promoted as a breakthrough because it translated
several simple sentences into Russian on an IBM mainframe computer.23
Its vocabulary was limited to 250 words, however, and it handled only
six rules of grammar and contained no random data. This, say Church and
Hovy was “a canned demo of the worst kind,” because it set
unrealistic expectations. 24
Work on artificial intelligence was spurred in part by the
cold war. But in 1959 Israeli linguist Yehoshua Bar-Hillel reported that
machines produced mediocre translations that could be useful only in special
contexts or in collaboration with humans.25
He illustrated the difficulty of, for example, translating the pair of
sentences: “The box was in the pen” and “The pen was
in the box.” The METAL translation system was developed at the University
of Texas in 1959, and later picked up by Siemens of Germany. The Georgetown
GAT system continued to be developed, largely through military funding,
and in 1964 the Atomic Energy Commission at Oak Ridge took it up. GAT
became the SYSTRAN system when it was commercialized in California in
1964 by Peter Toma, and over the decades it was attached to huge databases
of technical and military terms and idiomatic expressions. NASA used it
for the Apollo-Suyoz mission. 26
Research was set back by a scathing April, 1964-66 report
from the Automatic Language Processing Advisory Committee (ALPAC) of the
U.S. National Academy of Science. MT wasn’t netting enough results
to make it worth the money, concluded ALPAC, and when compared with human
capabilities, it might never be a success. The real problem, Colin Haynes
pointed out, was that although MT was developed for military use and had
business applications, it was condemned because it couldn’t reach
the unrealistic “ideal objective” of translation without post-editing
that was claimed by the Georgetown project. 27
Faced with mounting needs for translation, the Commission
of European Communities purchased an English-French version of SYSTRAN
in 1975. A Russian-French system was developed in 1966 by the Group of
Grenoble CETA for translating mathematical and physics texts. The Logos
system was used in the U.S. to translate English into Vietnamese during
Vietnam War, and in 1976 the Pan-American Health Organization created
the SPANAM system that helped specially trained translators get results,
doubling or tripling their productivity.28
As accelerated global commerce forced multinational companies
to develop and market products in several languages, the Japanese Ministry
of Industry and big private groups got involved in commercial research
and development of MT in the 1970s. Fujitsu’s Atlas system was the
first in the world to translate between Japanese and English at more than
60,000 words per hour. 29 The Air Force’s
work on the METAL system was commercialized in 1989. 30
But the systems were used mostly as workstation tools for professional
translators. 31
Some systems could stand alone for specialized uses. In
1977 the Centre Canadien de l’Environment installed the TAUM group’s
Meteo system for translating weather reports. Xerox adopted Systran for
internal translations by creating a Multinational Customized English that’s
easier to translate, and Toshiba’s AS-TRANSAC translated complex
technical manuals. In 1983 the ALPS (Automated Language Processing System)
appeared with an interactive feature that allows users to enrich the system’s
dictionaries.32
In 1987 British Telecom launched the world’s first
telephone interpreter to handle hotel bookings. 33
Work on speech translation continues, notably at Carnegie-Mellon University
and through a government-funded team of engineers in Kyoto, Japan.34
In 1991 the Unicode consortium of software publishers created
a standard for text that allows computers to communicate in all the world’s
major languages, by including accents & special characters that standard
ASCII 8-bit code can’t handle. That same year the International
Association for Machine Translation was founded in Washington D.C. 35
But for business, language can still be what Haynes calls
“the Ultimate Non-Tariff Barrier to Commerce.” 36
Whenever Lucent sells a telephone switching system overseas, it can require
translation of some 15,000 pages of documentation. 37
Global pressures to reduce time-to-market are also spurring the use of
MT: A one-day delay in launching new drug costs pharmaceutical industry
between $60,000 to $600,000 in non-sales, say Hartley and Paris. Aircraft
maintenance manuals need to be updated – and retranslated -- four
times a year.38 A Wards Automotive News reader
survey pinpointed communication as “one of the biggest roadblocks
facing truly global engineering.” To facilitate communication, Japanese
and American engineers began using pictographs, diagrams, hieroglyphics
and sketches. 39
In 1994 CompuServe began offering free translation on online
discussion forums, and in December, 1997, Alta Vista became the first
major web search site to offer free MT online. 40
Failure to Find a Universal Language
Much of the research has centered on finding or creating
a universal foundation onto which any language could be translated. Otherwise,
the standard “transfer” method of translation means that every
time you add a new language to a system, you have to write new transfer
components between that language and every other language. So if you extend
a system of three languages into five, you need to write 14 new transfer
components, one for each new pair. It appeared easier to translate into
an “intermediate language only the MT system understands”
and then reconstruct meaning into any target language from that base.41
MT pioneer Warren Weaver assumed he could find a “great
open basement, common to all the towers” of language, 42
and Bar-Hillel’s 1951 report stated that developing MT required
the discover of a “stock of concepts held in common by humanity,”
The DLT (Distributa Lingvo-Tradukado) system was developed around Esperanto,
an “ideal” universal language created in 1887. 43
Progress in linguistic theory culminated in 1965 with the publication
of Aspects of the Theory and Syntax by Noam Chomsky.44
Chomsky’s approach relied on the notions of deep structure and surface
structure linked by transformations, but the transformations worked only
in one direction, from depth to surface; reversibility would be crucial
to any Interlingua translation system. 45
The search for an Interlingua, or Linguistic Knowledge System,
was pursued by the European Community’s EUROTRA project 46
and others, including Carbonnnel, Colgate, and Carnegie Mellon Universities.
But the Grenoble group reoriented itself in 1971 from interlingual toward
transfer models, and in 1986 the METAL system also rejected the interlingual
approach.47 The DLT Project ended in 1990.48
The failure of the interlingua model provides insight into
the difficulties of machine translation. Alan Melby was part of a team
that, beginning in 1970, spent five years reducing language to sememes
(language-independent concepts) that could be translated into several
target languages.49 By 1978 he hit a “Wall,”
and concluded that “the language-independent universal sememes we
were looking for do not exist!” Melby says that the interlingua
researchers failed “because of a fundamental misunderstanding of.
. .languages.”50 “I had a degree
in mathematics,” explained Melby. “The world was supposed
to be a nice, tidy place.” But that approach “leave[s] true
creativity behind. All meaning becomes mechanical combinations of atomic
word senses.” 51 He developed a new respect
for the Whorf-Sapir hypothesis that languages are not interchangeable
and one’s world view is influenced by one’s language,52
and his continuing work focuses on productivity tools for human translators.53
New Approaches to Machine Translation
Yorick Wilks of Sheffield University is among those developing
programs that comb through already-translated text to empirically derive
grammatical rules. “It doesn’t matter how brain-dead the approach,”
says Eduard Hovy, president of Association for Machine Translation in
the Americas. As computers become more powerful, “doggedness”
and computing power matter more than linguistic breakthroughs.”
54 In what Budiansky calls “brilliant
stupidity,” 55 the computer tallies coincidences
in translation and creates probability tables for combinations of up to
three words. Another experiment searches the Internet itself, sometimes
using HTML tags for guidance, for translated word groups it then stores
in a personalized dictionary.56
Other researchers are trying to program “ meaning”
into the computer with “nested belief spaces representing differing
points of view of the participants in the translation process as well
as the actors referred to by source-language text.” They try to
include “default reasoning” about events, so they can make
machine’s translation seem coherent within the beliefs of the target-language
audience. For example, terms like “el tercer piso” (literally
“third floor” in Spanish) could mean either the third or fourth
floor, depending on cultural floor-naming conventions. 57
Evaluations of Publicly Available Machine Translation Systems
The number of web sites offering free translation was growing
when this paper was written; JETServ 58 even
offered an online forum in which “Americans and Japanese can communicate--without
worries of a language barrier” In 1998 Globalink’s marketing
consultant optimistically predicted “accuracy rates of “98%
plus.”59
But available translation programs have serious short-comings.
An attempt to translate a CNN article about a street protest into Spanish
using AltaVista (See Appendix B) disintegrated into nonsense, but French
translations of the same passage were better. Since hopes for an Interlingua
are largely dead, translations must be evaluated not only in terms of
the system, but of each language pair. When Amber Lewis evaluated how
four systems treated a badly-written business letter, she concluded that
the system “is not cost-effective because it will still require
extensive human editing.”60 However, many
such samples show errors that machine technology should be capable of
fixing, such as pairing a singular article with a plural noun or creating
non-words like “adddestr”.61
Many MT translations are generally understandable, but there
are still basic problems with words like “beam” that have
multiple meanings, and machines can’t guess as well as people do.
MT is “now recognized as a tool rather than a panacea,” reports
Frievalds. One factor is that “the input must be precise to produce
useful output.” For example, so that a machine won’t interpret
“inspect for wear and damage” as “inspect it and then
damage it” you would have to write: “inspect for wear and
also for damage.” 62 Conceding that available
systems aren’t good at prose, Mansel and When conclude that they’re
nonetheless “reasonably proficient” at translating web sites.63
Adaptations to Machine Technology
MT is still largely used for translators, offering tools
like memory software that memorizes a translator’s past work and
suggests a translation when similar sentences appear. Medtronic, a medical
device manufacturer that translates its documents into 12 languages, has
glossaries that increase translator productivity by 25 percent a year.64
Haynes says that in 1995 experts concluded that: “on
complex technical terminology, machines can do better” because there
are fewer nuances to technical terms. MT is also “especially valuable”
where literary values aren’t important, and you just need to pull
information out of the source text. To make that easier, organizations
are standardizing their use of terms, and programs are being developed
for specific domains, such as medicine or mining; even some programs available
on the Web allow users to choose from specialized databases. The effectiveness
of MT depends on the attitudes and competence of the humans using them,
concludes Haynes.65
Modern controlled-language tools like Clearcheck (Hayes ’94) are
designed to adapt human language to the machine rather than vice versa,
by constraining user input. Caterpillar Corporation’s Controlled
English program, for example, limits the number of words used to 8,000
and establishes syntax rules, leading to 40 percent faster translations.
66 Interactive programs like Carnegie-Mellon’s
DIPLOMAT prompt the user to intervene during the parsing process to resolve
ambiguity and/or add words and phrases they commonly use. That way, computers
“learn” with each translation to expand their capability.67
The Potential for Popular Machine Translation
With MT for the first time available for free to people who could never
before communicate across languages, can we have it both ways? Will MT
help promote world understanding, while also allowing people to use the
unique forms of language expression that are so important to our cultural
identity? Can the average person translate ideas to counterparts throughout
the world, just as the rich and powerful have been doing?
While powerful institutions can hire translators for pre- and post-editing,
train employees, and invest in customized programs, these options aren’t
available to the average person. Moreover, in personal communications
people tend to use slang, local expressions and words like “surfing”
that change over time. Many languages will never be machine translated,
and some others can only be translated to and from English.
The basic problem is that every language holds within it attitudes that
cannot be fully translated by human or machine. The French adjective “sympatique”
describes a quality that the English language doesn’t recognize,
for example. And although the Spanish “se me cayó el libro”
may be translated to “I dropped the book,” it reflects a different
attitude, stressing the accidental nature of such events by treating the
speaker as an observer; in the English version, there’s no distinction
between dropping the book by accident or on purpose.
Making Machine Translation Work
While expecting MT to fully translate the complexities of language remains
an unrealistic standard, MT can help people get access to a vast amount
of information and extract the essence of the meaning. In going beyond
that to create their own original messages, people have three alternatives
if they want to get acceptable results without help from a human translator:
1. Make adjustments before sending a message.
While avoiding extreme “controlled language” approaches, people
can learn to speak carefully and add visual hints such as graphics, if
the desire to communicate is strong. See Appendices C and D for tips.
It’s also a good idea to translate results back into the original
language and if something’s completely off-base, reword the original.68
2. Check
the translation in progress. Some Web programs allow users to list words,
such as proper names, that they do NOT want translated. Communication
with the person receiving the translation can also become interactive;
but people have to be willing to send back translated sentences that are
unclear for clarification, and ask questions. That involves delays that
interfere with direct communication and can also means getting over conventions
where people want to avoid anything that might imply criticism of the
sender. Programs like Translator that allow people to add their own expressions
to standard dictionaries can help, and will likely become more widely
available.69
3. All
parties to the communication adjust their expectations and tolerance.
“There are many millions of people around the world, particularly
younger people under 30, using the new technology. . . who have no problem
at all in accepting the raw English output of the better MT systems as
being acceptable as the fractured Americanized English that they use as
a common language when they get together with foreign contemporaries on
line, or face-to-face in our increasingly global society,” asserts
Haynes. He even expects a computerized equivalent of pidgin English to
develop.70
Since the goal is comprehension, people should be prepared to “take
in stride the mis-translation or even non-translation of text” in
cases where they can “figure out fairly easily” what the other
person is trying to say, advise Church and Hovy. 71
They point out that “workers at the EURATOM Research center in Italy
and the Atomic Energy Commission’s Oak Ridge lab managed to communicate
using the primitive Georgetown system available from 1963 through 1973,
with no pre- or post-editing.” Moreover, an evaluation found that
96 percent would recommend MT “even though the texts were said to
require almost twice as much time to read as original English texts”
and 21 percent were intelligible. Some said “the machine is more
honest” and predictable – they got used to MT style, and mistakes
were easily discarded. 72 Haynes finds that
the hardest thing for people to accept is imperfect grammar or vocabulary
in written texts in their mother tongue. “This is an important point
to bear in mind as we examine the validity of machine translations in
real world situations.”73
The question remains: If you have to modify what’s familiar and
expressive in language for universal translation, how valuable is it?
Technical manuals are fine, but what about communication between people?
Does this adaptation pervert the cultural expression that translation
should potentially protect? The answer relies in part on the fact that
dynamic cultures always adapt to new circumstances. Hopefully the interaction
between MT and the people who use it will create new norms that are still
impossible to predict. In the global telecommunications age, the experiment
seems worth the trouble.
Acceptance of imperfection also depends on a level of trust and familiarity
over time. It means giving someone who appears to have said something
terrible the benefit of the doubt, filling in blanks for concepts that
are incomplete, and developing commonly understood shortcuts. The more
shared experiences people gain by communicating with MT tools over time,
the more successful they’ll be.
Appendix A:
Review of Online Discussions of Language
Those who join international discussions online hotly debate the use
of English and minority languages.. “A language is the bedrock of
most civilisations,” claims Etse Ladzekpo in an African discussion
group conducted in English. “Would it not be great if members of
this group were able to communicate via this medium in a typical African
language?. . . For the present I'm prepared to use the colonial but universal
language,” she adds. ”I'm willing to adapt and apply the technology
which is predominantly Anglo-centric and Anglo-phonic in ways which would
benefit my community.” Another participant thinks “one of
the fundamental ways to democratise web access is by exploring technologies
that can make use of African languages in conveying web content.”74
At the 1997 Virtual Conference on the Right to Communicate*: hosted by
IDRC/PanAsia Networking, an online discussion that was itself was translated
into three languages discussed the language dilemma. One participant bemoaned
the fact that young people in his native Belgium, a trilingual country,
aren’t using French. In his bilingual work environment, he claims
“You get more respect from other people if you talk multiple languages.”
Writing in Spanish, Hawaiian Edward Sills said reading other languages
is "a little like listening to a conversation through a brick wall.
I catch a word here and there and I understand the context a little.”
Writing in English, Spaniard Luistxo Fernandex said that in any bilingual
forum “The logical choice of language will always go toward the
dominant language.” Another participant complained that “
the vast majority of [Americans] act as though … EVERYBODY in the
world should understand and speak English."75
“I can think of nothing more horrible than one world, one language,
one culture,” continued Sills from Hawaii. “Is that the only
alternative to divisive, mistrustful, warring, insularity?. . . I'm an
amateur naturalist. Observing nature I see that there is great diversity.
Our survival is dependent on that diversity, and I suspect that the survival
of our individual cultures rests on diversity as well. Use the technology
to cement diversity, minorities, languages into the culture.." To
this, Fernandez responded: " Well, If I posted this message in Basque,
maybe my action gets praised as 'politically correct', but it is nonsense
from a communicational point of view." Adding that there aren’t
any monolingual Basques, he felt that attempts at establishing Basque
language sites on the Internet would result in “marginalization
of Basque." Cvi Solt suggested that a universal, easy-to-learn language
like Esperanto be used as a second or third language in international
communications.76
In another international discussion, a participant said “I very
much enjoy postings in multiple languages. When I have the time, I can
compare the translations and pick up a word or phase or two. But I would
hate to see this become a requirement for participation. Funding really
should be made to construct inexpensive and reliable language translators.”
Michel Elie of Montpellier France praised efforts to comprehend other
languages, adding that dual language postings “would improve mutual
understanding.” He suggested that such translations “in the
future might be assisted by an automatic translator." The goal, he
added, is to “help preserve cultural diversity while making the
world a little smaller."77
The Ayacucho chat group originating at a Venezuelan University disintegrated
into bitter debate over one participant’s untranslated posting in
English. A respondent complained in Spanish that “the people of
Venezuela .. can’t even listen to music in Spanish.” Another
confessed that since he didn’t speak English, for a long time he
suffered in silence about messages in English, not realizing it bothered
other participants as well. Another praised the “magnificent essay”
that had been submitted in English and in Spanish suggested that “your
minds are walled off.” Eventually someone translated the essay.78
Caoimhín P. Ó Donnaíle who created the Web site
“European minority (or minoritized!) language” 79
warned in a French email message to Marie Lebert that the Internet “has
contributed and will contribute to the continuing development of English
as the world language.” The Internet can also be “a tremendous
aid to minority languages,” he believes, but “that won’t
happen all by itself; people must choose to defend their language.”
He sees the popularity of the many web sites that offer language courses
as helpful. Welcoming the UNICODE system that facilitates multilingualism
on the Web, he sees machine translation as a way to help cultures revive
repressed languages.80
Henk Slettenhaar, a trilingual professor at the Webster University, Geneva,
Switzerland, wrote in December, 1998 that “Local communities. ..
on the Web should use the local language first and foremost. . I much
prefer to read the original with difficulty than to get a bad translation.
. I see a real need for bilingual websites.” 81
Bruce Girard director of Comunica.org. wrote in the Indian language Chasqui
about the Agencia Informativa Púlsar, of which he’s a founding
director, which uses the Internet to communicate for Latin Americans in
both Spanish and in Quechua.82
The French language is vigorously promoted online by La Délégation
Générale à la Langue Française (DGLF). It
includes the creation of three lists moderated by the DGLF to facilitate
exchanges of ideas and information in French and also discuss linguistic
diversity, politics and dynamics, along with technical guides to help
people communicate digitally in French.83
Windows a été cassé, signes déchirés
vers le bas et
plusieurs Suisse les agents de police ont été blessés
samedi comme en gros 1,300 les démonstrateurs ont essayé
d'interrompre le Monde Économique Forum ici, mais un mur
serré de police a gardé les contestataires bien loin du
centre de conférence où Président Bill Clinton a
parlé à
la conférence quelques heures plus tôt. Peu après
4 p.le
m. le temps local, plusieurs contestataires dans le la
foule bruyante a commencé déchirer et mettre feu à
plusieurs panneaux d'affichage en bois qui courent le
long du principal promenade dans ceci ville normalement
paisible haut dans le Suisse Alpes. Président Clinton
n'était pas dans le centre de conférence quand les
protestations ont commencé. Police dans équipement de
l'émeute plein et les boucliers en osier ont tiré un
petit nombre de caoutchouc plus tard balles et aérosol du
poivre usagé subjuguer trois contestataires qui a essayé
d'avancer mbeyond un ensemble de la barrière du grillage
au-dessus par police le long de la principale route qui
mène à la conférence le centre. Contrôlez porte-parole
qu'Alois Hafner a dit à deux agents de police les
blessures de la tête souffertes. Il a dit que quelques
démonstrateurs avaient aussi été le blessé,
et que deux
ont été arrêtés.
French back to English:
Windows has been broken, torn downwards signs and
several Switzerland police's agents have been wounded
Saturday as on the whole 1,300 the demonstrators tried
to interrupt the World Economic Forum here, but a wall tight police kept
the protesters well far from the center of conference where President
Bill Clinton spoke to the conference a few hours earlier. Shortly after
4 p.m. the local time, several protesters in the the loud crowd began
to tear and to put fire to several noticeboards made of wood that runs
the long of the principal walk in this city normally restful high in the
Switzerland Alps. President Clinton was not in the center of conference
when the protests began. Police in equipment of the full riot and the
wicker shields pulled one small number of rubber later bullets and spray
of the used pepper to captivate three protesters who tried to advance
mbeyond a whole of the gate of the wire fencing above by police along
the main road who lead to the conference centers it. Control spokesman
that Hafner standards told two agents of police the injuries of the head
suffered. He/it said that some demonstrators had also been the injured,
and that two have been stopped.
English into Spanish:
Windows esté borracho, colgajo de las señales abajo y
varios suizo se dañaron policía sábado como aproximadamente
1,300 demostradores intentaron romper el Mundo Económico
El foro aquí, pero una pared firme de policía guardó
a las
protestadoras bien fuera del centro de la conferencia
dónde Presidente Bill Clinton habló antes a la conferencia
unas horas. Poco después 4 postmeridiano el tiempo local,
varios protestadores en el la muchedumbre bulliciosa
empezó rasgando abajo y poniendo el fuego a varias
carteleras de madera a lo largo de que corren el principal
pasee en este pueblo normalmente-pacífico alto en el suizo
Alpes. Presidente Clinton no estaba en el centro de la
conferencia cuando las protestas empezaron. Vigile el
vestido del alboroto por completo y los escudos tejidos de
mimbres dispararon un número pequeño de caucho después
las balas y usó el rocío de pimienta para dominar a tres
protestadores quién intentó al mbeyond de antemano un
juego de barrera de cerco a por policía a lo largo del
camino principal que lleva a la conferencia el centro.
Portavoz policíaco que Alois Hafner les dijo a dos
policía las heridas de cabeza sufridas. Él dijo que
algunos demostradores también habían sido el herido, y
que se han arrestado dos.
Spanish back into English:
Windows is drunk, torn piece of the signs below and
several Swiss was damaged police Saturday like approximately
1,300 demonstrators tried to break the Economic World
The forum here, but police's firm wall kept to those
protesting well outside of the center of the conference
where President Bill Clinton spoke before to the conference
some hours. Soon after 4 postmeridiano the local time,
several protesters in the the bustling crowd
it began ripping below and putting the fire to several
wooden billboards along that you/they run the main one
go for a walk in this high usually-peaceful town in the Swiss
Alps. President Clinton was not in the center of the
he/she confers when the protests began. Watch over the one
dress of the din completely and the knitted shields of
wickers shot a small number of rubber later
the bullets and it used the dew of pepper to dominate at three
protesting who attempted to the mbeyond ahead of time a
game of fence barrier to for police along the one
I walk main that takes to the conference the center.
Police spokesman that Alois Hafner told at two
police the suffering head wounds. He said that
some demonstrators had also been the wounded, and
that two have been arrested.
English original:
Windows were smashed, signs torn down and several Swiss
policemen were injured Saturday as roughly 1,300
demonstrators attempted to disrupt the World Economic
Forum here, but a tight wall of police kept the protesters
well away from the conference center where President Bill
Clinton spoke to the conference a few hours earlier.
Shortly after 4 p.m. local time, several protesters in the
boisterous crowd began ripping down and setting fire to
several wooden billboards that run along the main
promenade in this normally-peaceful town high in the Swiss
Alps. President Clinton was not in the conference center
when the protests began. Police in full riot gear and
wicker shields later fired a small number of rubber
bullets and used pepper spray to subdue three protesters
who tried to advance mbeyond a fence barrier set up by
police along the main road leading to the conference
center. Police spokesman Alois Hafner said two policemen
suffered head wounds. He said some demonstrators had also been injured,
and that two have been arrested.
Appendix C:
Practical Tips for Pre-Editing
Abridged from: Colin Haynes, Breaking Down the Language Barriers
1. Always run the draft for translation through
grammar-checking software, which can catch overly complex construction,
compound verbs and obscure phrasing (which they often flag as being in
the passive voice).
2. Use a word processor or use that function
in a MT program.
3. Use a thesaurus to simplify uncommon usages.
4. Stick to a logical sequences of events,
without flashbacks.
5. Spell out abbreviations when they’re
first used, with the abbreviations put in all-caps in brackets.
6. Avoid idiomatic, slang and regional or national expressions.
7. Don’t use complex compound structures.
8. Be precise. Avoid fuzzy language.
9. Don’t make the comprehension of the text dependent on
formatting like italics or indents.
10. Try to use the ISO Format for dates.
11. Be careful with contracts, where language may have a precise
but obscure legal meaning.
12. Translate back and forth (back to the original language) to
see where the translation goes astray, and reword.
Appendix D:
Tips for Preparing Your Document for Translation
From: http://www.multilingual.com/ ADAM JONES
Translating English materials into other languages has its share of pitfalls,
many of which can be avoided. At SimulTrans, we look for the following
primary difficulties at the beginning of a translation to prevent problems
and ensure consistency and clarity in the target language:
• Maintain consistency of terminology.
• Strive for clarity and use simple, direct sentences with basic
grammatical construction.
• International users generally prefer straightforward, factual
wording.
• Provide a list of all terms which should remain in English (for
example, proper names, product names and titles) to alert the translator
.
• Provide a definition or explanation for terms taken out of context
(for example, disjointed phrases or ad headlines)
• Avoid words that can have multiple meanings
• Avoid abbreviations, acronyms and contractions (for example, Esc.,
config., FDA)
• Avoid long noun strings or modifier chains (for example, magnetic
storage media, move program item, compatible software and hardware), as
they can cause confusion about what modifies what.
• Inserting a hyphen can often clarify meaning.
• Avoid present participles (for example, “selecting icons
and menu items”). Instead use the infinitive form of the verb (“to
select icons and menu items”), as present participles do not have
counterparts in many languages.
• Avoid singular words that can be multiple parts of speech (for
example, program, display, feature, control)
• Avoid plays on words, puns, slang and idiomatic expressions (for
example, “take a byte out of your budget”).
• Avoid Latinisms such as e.g. and i.e., which do not have equivalents
in some language families.
Adam Jones is the Vice President of Customer Programs at SimulTrans and
can be reached at AdamJ@simultrans.com. This article reprinted from #13
Volume 8 Issue 2 of MultiLingual Computing & Technology published
by MultiLingual Computing, Inc. 319 North First Ave., Sandpoint, Idaho,
USA, 208-263-8178, Fax: 208-263-6310.
Bibliography
Anis, Jacques. “Ordinateurs et Traduction: Survol d’un Demi-Siecle.”
Languages v. 28 n. 116 (Dec., 1994) 111-112.
Arnold, D. Balkan, L., Humphreys, R,.Lee, Meijer, S., and Sadler, L. Machine
Translation, An Introductory Guide. NCC Blackwell, Ltd. UK, 1994.
Bethoney, Herb. “Machine Translation: Better than Nothing.”
PC Week v. 15 n. 50 (Dec. 14, 1998) 82.
Budiansky, Stephen. “Lost in Translation.” The Atlantic Monthly
v. 282 n. 6 (Dec. 1998) 80-84.
Castells, Manuel. The Power of Identity: The Information Age: Economy,
Society and Culture v. I & II. Blackwell Publishers, 1997.
Church, Kenneth W. and Hovy, Eduard H. “Good Applications for Crummy
Machine Translation.” Machine Translation v. 8 n. 4 (1993), 239-258.
Clark, Ian. Globalization and Fragmentation: International Relations in
the Twentieth Century. Oxford University Press, 1997.
Dawson, Chester “Gift of Tongues: Translation Software Being Developed
in Japan Makes Users Understandable in Several Languages.” Far Eastern
Economic Review (Nov. 18, 1999) 48.
Economist, The. “A Gift of Tongues (Language Translations Done by
Computers).” The Economist. v. 345 n. 8039 (18 Oct., 1997) 81.
Ganzel, Rebecca. “Tech Trends: Universal Translator? Not quite.”
Training v. 36 n. 4 (April 1999) 22.
Gantz, John. “Coming Soon: Language Barriers.” Computerworld.
v. 32 n. 5 (2 Feb. 1998) 33.
Gunkel, David J. Lingua ex Machina: Computer-mediated Communication and
the Tower of Babel. Johns Hopkins University Press and Society for Literature
and Science, 1999. http://muse.jhu.edu/j...configurations/v007/7.1gunkel.html
Lewis, Amber L. The Practical Implications of a Minimum Machine Translation
Unit, Babel: Fédération Internationale de traducteurs Revue,
43: 2 p. 138-150
Loffler-Laurian, Anne-Marie, La Traductin Automatique: Son utilization
par le ‘Grand Public
Freivalds, John. “The Technology of Translation.” Management
Review. (July-August, 1999) 48.
Friedman, Thomas. The Lexus and the Olive Tree. 1999.
Hartley, Anthony and Paris, Cecile. “Multilingual Document Production
from Support for Translating to Support for Authoring.” Machine
Translation. v. 12, no.1-2 (1997).
Haynes, Colin. Breaking Down the Language Barriers. Aslib, the Association
for Information Management, 1998.
Helmreigh, Stephen, and Farwell, David. “Translation Differences
and Pragmatics-Based MT.” Machine Translation. v. 1, n. 3l (1998).
Ikeno, Atsushi; Murata, Toshiki; Shimohata, Sayori, and Yamamoto, Hideki.
“Human Dimension Beyond Point and Click: Machine Translation Using
Internet Natural Language Resources.” Telecom Interactive 99; International
Telecommunication Union Infrastructure Summit (1999). http://www2.itu.int/itudocs/telecom/inter@99/list/2812.pdf
Mansell, Robin and When, Uta, eds. Knowledge Societies: Information Technology
for Sustainable Development. Published for the United Nations by Oxford
University Press, 1998.
McEnery, Tony; Wilson, Andres; Sanchez-Leon, Fernando; Nieto-Serrano,
Amalio. “Multilingual Resources for European Languages: Contributions
of the CRATER Project.” Literary and Linguistic Computing v. 12
n. 4 (1997)
Mealing, Stuart, and Yazdani, Masoud. Multilingual Meaning: Bridging the
Language Barrier with Intelligent Systems. Intellect Books, 1993.
Melby, Alan K. The Possibility of Language,: A Discussion of the Nature
of of Language, with Implications for Human and Machine Translation. John
Benjamin B.V., 1995.
Moad, Jeff. “Machine Translation – The Next Generation.”
PC Week v. 15 n. 4 (Jan 26, 1998) 76.
Richardson, Robert. “Sell Globally, Speak Locally (Using Foreign
Languages as a Web Marketing Tool).” Home Office Computing v. 17
n. 2 (Feb. 1999) 92
Winter, Drew. “You’re global. Now how the heck do you communicate?”
Ward’s Auto World v. 35 n. 3 (March 1999) 45.
Technology Guide. “Word for Word: Translation Software Delivers
the Gist of the Web.” Fortune special issue n. 32 (Winter 2000)
32
Webliography
Africa discussion group.http://www.egroups.com/group/africa_web_content_owner/info.html
Alis Technologies and Internet Society. “Web Languages Hit Parade.”
(June 1997) http://babel.alis.com:8080/palmares.html
Americans and foreign languages chat group. (Nov. 5 1997) http://bluedog.cc.emory.edu/archives/cml/Nov_97/0526.html.
AltaVista Babel Fish. http://babelfish.altavista.com
Appel du Comité européen pour le respect des cultures et
des langues en Europe ( C.E.R.C.L.E. ), http://PersoWeb.francenet.fr/~mbonnaud/cerclefr.html
Babylonet http://www.babylonet.net/
Bills, Bonnie, “Web World Wide: The Net Revoluton Goes Global.”
CNET Special Report. http://home.cnet.com/specialreports/0-6014-7-1538058.html
(Feb. 9, 2000)
Center for Machine Translation (CMT) at the School of Computer Science
at Carnegie Mellon University. http://www.lti.cs.cmu.edu/Research/cmt-projects.html
Computer Economics, http://www.computereconomics.com/
Edit. http://www.edit.fr/
Edit Online. http://www.edit.fr/news/index.html
Enteleky. http://www.enteleky.com/trans.html
European Association for Machine Translation. http://www.lim.nl/eamt/.
European minority (or minoritized!) languages, http://www.smo.uhi.ac.uk/saoghal/mion-chanain/Failte_en.html
Girard, Bruce, “Minority Languages and Information Ethics: Pluralism,
Radio and the Internet article,” originally published in Chasqui,
The Latin-American Journal of Communication, http://commposite.uqam.ca/videaz/bio/brgien.html
Go.com. http://translator.go.com/
Institute of Information Science, Sinica, Taiwan.. http://www.iis.sinica.edu.tw/engtext.htm
InterTran. http://www.tranexp.com:2000/
ISPO Information Society Promotion Office Electronic Democracy discussion
list, http://bscw2.ispo.cec.be/ispo/lists/el-democracy/msg00036.html
Lebert, Marie. “Le multilinguisme sur le Web.” Le Centre d'Expertise
et de Veille Inforoutes et Langues (Ceveil). Gouvernement du Québec
(Fév. 1999) http://www.ceveil.qc.ca/multi0.htm
Lernout & Hauspie. http://www.lhs.com/default2.htm
Logos. http://www.logos.it/
Jones, Matt. “Geek Page: Machine Translation Using Interlinguas
to Map between Languages.” WIRED Online v. 4 n. 10 (Oct. 1 1966)
http://www.wired.com/wired/archive/4.10/geek.html
Machine Translation, Geneva Switzerland. http://www.lim.nl/eamt/
NUA Internet Surveys. http://www.nua.ie/surveys/how_many_online/index.html
QwikTrans http://www.qwiktrans.com/
Transparent Language web site. http://www.freetranslation.com/
Systran Translation Software. http://www.systransoft.com/ and http://www.systran.fr/
Venezuela discussion group, http://venezuela.mit.edu/listas/ayacucho-anteriores/ayacucho940601/0017.html
Virtual Conference on the Right to Communicate, http://commposite.uqam.ca/videaz
hosted by IDRC/PanAsia Networking, http://www.PanAsia.org.sg/conferen.htm
Footnotes
1 John Freivalds, “The Technology of Translation,”
Management Review (July-August 1999) 48. 2 Colin Haynes, Breaking Down the Language Barriers.
(Aslib, the Association for Information Management, 1998) 40. 3 Robin Mansell and Uta When, eds. Knowledge
Societies: Information Technology for Sustainable Development. (Published
for the United Nations by Oxford University Press, 1998) 145. 4 Manuel Castells,. The Power of Identity: The
Information Age: Economy, Society and Culture v. I. (Blackwell Publishers,
1997) 42. 5 Ibid., 3. 6 Ian Clark, Globalization and Fragmentation:
International Relations in the Twentieth Century (Oxford University Press,
1997) 27. 7 Alis Technologies and Internet Society, “Web
Languages Hit Parade.” (June 1997) http://babel.alis.com:8080/palmares.html. 8 Haynes, Breaking Down the Language Barriers,
16. 9 Bonnie Bills, “Web World Wide: The Net
Revoluton Goes Global.” (CNET Special Report. Feb. 9, 2000), http://home.cnet.com/specialreports/0-6014-7-1538058.html
10 Computer Economics, http://www.computereconomics.com/ 11 Mansell and When, eds. Knowledge Societies:
Information Technology for Sustainable Development, 145. 12 Marie Lebert, “Le multilinguisme sur
le Web.” (Le Centre d'Expertise et de Veille Inforoutes et Langues
(Ceveil). Gouvernement du Québec, Fév. 1999) http://www.ceveil.qc.ca/multi0.htm
13 http://cnn.com/INDEX/languages/; http://www.yahoo.com.
Appel du Comité européen pour le respect des cultures et
des langues en Europe ( C.E.R.C.L.E. ), 14 http://PersoWeb.francenet.fr/~mbonnaud/cerclefr.html. 15 http://cgi.hardlink.com/~chambers/cgi/HLPsearch.cgi?stype=DIV&DIVS=Languages+and+Literature 16 Mansell and When, eds. Knowledge Societies:
Information Technology for Sustainable Development, 145 p. 91. 17 ISPO Information Society Promotion Office
Electronic Democracy discussion list, http://bscw2.ispo.cec.be/ispo/lists/el-democracy/msg00036.html 18 David J. Gunkel, Lingua ex Machina: Computer-mediated
Communication and the Tower of Babel (Johns Hopkins University Press and
Society for Literature and Science, 1999),
http://muse.jhu.edu/j...configurations/v007/7.1gunkel.html 19 Jeff Moad, “Machine Translation –
The Next Generation.” (PC Week Jan 26, 1998), 76. 20 Haynes, Breaking Down the Language Barriers,
16. 21 Arnold, Balkan, Humphreys, Lee, Meijer and
Sadler, Machine Translation, An Introductory Guide (NCC Blackwell, Ltd.,
1994) 13. 22 Haynes, Breaking Down the Language Barriers,
10. 23 Jacques Anis, “Ordinateurs et Traduction:
Survol d’un Demi-Siecle.”(Languages Dec., 1994), 111-112 24 Kenneth W. Church and Eduard H. Hovy, “Good
Applications for Crummy Machine Translation,” (Machine Translation,
1993), 239-258 25 Anis, “Ordinateurs et Traduction: Survol
d’un Demi-Siecle,” 111 26 Freivalds, “The Technology of Translation,”
48 27 Haynes, Breaking Down the Language Barriers,
12. 28 Arnold, Balkan, Humphreys, Lee, Meijer and
Sadler, Machine Translation, An Introductory Guide 29 Haynes, Breaking Down the Language Barriers,
12. 30 Anis, “Ordinateurs et Traduction: Survol
d’un Demi-Siecle,” 114 31 M. Kay, The Proper Place of Men and Machines
in Language Translation,” (Unpublished manuscript, Xerox, Palo.Alto,
1980), 2 32 Anis, “Ordinateurs et Traduction: Survol
d’un Demi-Siecle,” 118 33 Ibid. 34 Chester Dawson, “Gift of Tongues: Translation
Software Being Developed in Japan Makes Users Understandable in Several
Languages.” (Far Eastern Economic Review, Nov. 18, 1999), 48 35 Anis, “Ordinateurs et Traduction: Survol
d’un Demi-Siecle.” 36 Haynes, Breaking Down the Language Barriers. 37 Freivalds, “The Technology of Translation,”
48 38 Anthony Hartley and Cecile Paris, “Multilingual
Document Production from Support for Translating to Support for Authoring.”
(Machine Translation. v. 12, no.1-2 1997), 112 39 Drew Winter, “You’re global.
Now how the heck do you communicate?” Ward’s Auto World, March
1999), 45 40 Moad, “Machine Translation –
The Next Generation,” 76 41 Haynes, Breaking Down the Language Barriers. 42 Gunkel, Lingua ex Machina: Computer-mediated
Communication and the Tower of Babel, 72 43 Ibid. 44 Lebert, “Le multilinguisme sur le Web,”
http://www.ceveil.qc.ca/multi0.htm 45 Alan K. Melby, The Possibility of Language:
A Discussion of the Nature of Language, with Implications for Human and
Machine Translation (John Benjamin B.V., 1995), 179-180 46 Haynes, Breaking Down the Language Barriers. 47 Gunkel, Lingua ex Machina: Computer-mediated
Communication and the Tower of Babel 48 Anis, “Ordinateurs et Traduction: Survol
d’un Demi-Siecle.” 49 Melby, The Possibility of Language: A Discussion
of the Nature of Language, with Implications for Human and Machine Translation
(John Benjamin B.V., 1995) 50 Ibid.,74 51 Ibid., 180-182. 52 Ibid., 49-50. 53 According to Gunkel, (Lingua ex Machina:
Computer-mediated Communication and the Tower of Babel, p. 85) the pre-
Babel ideal of universal language has always been a controversial subject,
in any case, with Jacques Derrida claiming a “universal tongue would
have been a particular language imposed by violence, by force, by violent
hegemony over the rest of the world.” Critics link the universal-language
movement to colonial expansion of Europe . .The Ars magna universal-language
projects of Ramon Lull (1232-1316), “were initially devised for
the purpose of converting the heathen to Christianity,” Gunkel says,
pointing out that MT techniques were not developed at first for intercultural
communication, but by the U.S. Defense Department for national security.
They began as an extension of wartime cryptanalysis by computer. p.85 54 Stephen Budiansky, “Lost in Translation”
(The Atlantic Monthly, Dec. 1998) 83. 55 Melby, The Possibility of Language: A Discussion
of the Nature of Language, with Implications for Human and Machine Translation. 56 Ikeno, Murata, Shimohata, and Yamamoto, “Human
Dimension Beyond Point and Click: Machine Translation Using Internet Natural
Language Resources.” http://www2.itu.int/itudocs/telecom/inter@99/list/2812.pdf
57 Stephen Helmreigh and David Farwell, “Translation
Differences and Pragmatics-Based MT.” (Machine Translation, 1998). 58 http://www.jetserv.com/cbusa.htm 59 Lebert, “Le multilinguisme sur le Web,”
http://www.ceveil.qc.ca/multieng4.htm 60 Amber Lewis, “ The Practical Implications
of a Minimum Machine Translation Unit,” (Babel: Fédération
Internationale de traducteurs Revue, v 43 n 2 ) 138-150 61 Rebecca Ganzel, “Tech Trends: Universal
Translator? Not quite.” (Training, April 1999), 22 62 Freivalds, “The Technology of Translation,
” 48. 63 Mansell and When, eds. Knowledge Societies:
Information Technology for Sustainable Development, 145. 64 Ibid. 65 Haynes, Breaking Down the Language Barriers. 66 Freivalds, “The Technology of Translation,”
48. 67 Center for Machine Translation (CMT) at the
School of Computer Science at Carnegie Mellon University. http://www.lti.cs.cmu.edu/Research/cmt-projects.html 68 Stuart Mealing, , and Masoud Yazdani, Multilingual
Meaning: Bridging the Language Barrier with Intelligent Systems. Intellect
Books, 1993 69 Haynes, Breaking Down the Language Barriers. 70 Ibid. 71 Church and Hovy, “Good Applications
for Crummy Machine Translation,” 239-258 72 Ibid., 253 73 Haynes, Breaking Down the Language Barriers. 74 Africa discussion group, http://www.egroups.com/group/africa_web_content_owner/info.html
75 Virtual Conference on the Right to Communicate,
http://commposite.uqam.ca/videaz
Ibid. 76 ISPO Information Society Promotion Office
Electronic Democracy discussion list, 77 http://bscw2.ispo.cec.be/ispo/lists/el-democracy/msg00036.html 78 Venezuela discussion group, http://venezuela.mit.edu/listas/ayacucho-anteriores/ayacucho940601/0017.html 79 European minority (or minoritized!) languages,
http://www.smo.uhi.ac.uk/saoghal/mion-chanain/Failte_en.html 80 Lebert, “Le multilinguisme sur le Web,”
http://www.ceveil.qc.ca/multi0.htm 81 Ibid. 82 Bruce Girard, “Minority Languages and
Information Ethics: Pluralism, Radio and the Internet article,”
originally published in Chasqui, The Latin-American Journal of Communication,
http://commposite.uqam.ca/videaz/bio/brgien.html 83 La Délégation Générale
à la Langue Française, http://www.culture.fr/culture/dglf/garde.htm