ArticlePDF Available

Online and free! Ten years of online machine translation: origins, developments, current use and future prospects

  • Unversity of Bologna at Forlì, Italy


Marking the ten-year anniversary of the launch of Babel Fish, the first ever free online machine translation (MT) service that went live on the Internet in late 1997, this paper sketches the background that led to its development, giving an account of its origins and of the early stages of its evolution. Several competitors have entered the field of web-based MT over the last decade, and the paper offers a review of the most significant contributions in the literature with a particular focus on two key issues: firstly, the role that these online MT tools have played in meeting the translation needs of the users, and secondly the impact that they have had on the MT-related industry and business. Information coming from a variety of sources, including data on current usage supplied by the online MT providers themselves for the purposes of this study, testifies to the massive increase in the use of the leading multilingual online MT services over the last ten years. On this basis, the conclusion assesses the future prospects of Internet-based MT.
Online and Free! Ten Years of Online Machine Translation:
Origins, Developments, Current Use and Future Prospects
Federico Gaspari John Hutchins
School of Informatics, University of Manchester
PO Box 88, Manchester M60 1QD
United Kingdom
89 Christchurch Road
Norwich NR2 3NG
United Kingdom
Marking the ten-year anniversary of the launch of Babel Fish, the first ever free online machine translation (MT) service that went live
on the Internet in late 1997, this paper sketches the background that led to its development, giving an account of its origins and of the
early stages of its evolution. Several competitors have entered the field of web-based MT over the last decade, and the paper offers a
review of the most significant contributions in the literature with a particular focus on two key issues: firstly, the role that these online
MT tools have played in meeting the translation needs of the users, and secondly the impact that they have had on the MT-related
industry and business. Information coming from a variety of sources, including data on current usage supplied by the online MT
providers themselves for the purposes of this study, testifies to the massive increase in the use of the leading multilingual online MT
services over the last ten years. On this basis, the conclusion assesses the future prospects of Internet-based MT.
Keywords: online MT, Internet, Web, history
1. Origins and Early Developments
Fulfilling predictions since the late 1980s that MT
services would become available on the Internet, the first
online MT service was provided from 1988 by the Systran
Centre in Paris to subscribers of the French postal
service’s Minitel network (restricted to France). Users
could send texts for translation from their PC or
Macintosh and receive results (22 lines a minute, at a
charge of about $1.20 per page) – the language pairs
offered were French to English, German to English, and
English to French (Gachot, 1989). According to Ryan
(1987: 100), MT software provided by Systran was
potentially accessible to 4.5 million users of Minitel in
The next concrete proposal was for an online service
that would be available more widely. In September 1992,
it was announced that CompuServe was investigating the
possibility of offering MT to its subscribers (Harrison,
1992). This project involved a six-month evaluation
period to test the output quality and the overall
performance of existing MT systems for the language pair
English-German (in both directions): “CompuServe’s
basic goal for MT is to provide draft-quality translation
directly to end users. […] We suspect that there is a
market for low-cost translations, even if the quality is less
than ideal” (Harrison, 1992: 11). Although not yet
offering online MT as it is commonly understood today,
i.e. in the form of services that users can directly log on to
in order to have input texts or webpages of their choice
translated, the pioneering experiences at CompuServe laid
the foundations for further crucial developments in the
following 15 years which are the focus of this paper.
Mary Flanagan, a computational linguist based in the
USA, led the Advanced Technologies Group at
CompuServe from 1992 until 1998, and regularly reported
the groundbreaking developments at CompuServe. For
example, Flanagan & Jensen (1994) describe the early
implementation of an entirely automated MT process,
whereby the messages posted in English on selected
CompuServe forums were periodically collected, fed
through Intergraph’s Transcend system, and the output in
French and German displayed in parallel versions that
could be read online by people unfamiliar with English.
The original forums (in English) and the machine-
translated versions (in French and German) presented the
same contents and the same structure, i.e. the threading
and sequence of the multilingual postings were identical.
The usage statistics reported in Flanagan (1995) are quite
impressive: MT was used for translation between English
and three other languages, namely French, German and
Spanish, and during the first month of operation, on just
one of the more than 600 specialist interest forums
available on CompuServe, more than 900,000 words were
translated at a speed of over 3,000 words per minute.
Details were also given of further plans and services (e.g.
a low-cost post-editing service for email translation) to be
launched for CompuServe subscribers, with an assessment
of the commercial opportunities offered by the
deployment of MT in the online environment, as well as
of the associated challenges.
In a later report, Flanagan focuses on the usage
patterns and on the reactions of the users who are exposed
to MT output for the first time in the online environment:
25% of them abandon the service after receiving the first
translations, possibly because they are surprised by the
poor quality of the raw output and find it impossible to
understand or use effectively. Interestingly, following the
launch of their online MT facility, CompuServe received
“hundreds of angry e-mail messages, as well as hundreds
of resumes from translators” offering their services
(Flanagan, 1996a: 193). The report also says that “users
were overwhelmingly satisfied with the quality of the
translations […] and several large users routinely submit
jobs totalling more than 10,000 words per week” (ibid.:
194). Flanagan (1996b: 244) summarises the philosophy
of these initial attempts to offer online MT to registered
users as follows:
CompuServe has taken a pragmatic approach to
MT technology, focusing on finding a market
niche for what it can do – generate rapid, very
rough draft, information-scanning quality
translations in an environment where quick
scanning for content is more important than high
However, retaining some quality was still a concern.
Consequently, CompuServe’s Document Translation
Service offered its subscribers the possibility of uploading
documents for MT and requesting an optional post-editing
service, which was charged at a rate per-word that was ten
times higher than the rate for raw MT output (Flanagan,
1996b: 245). Flanagan (1997a: 25) states that 85% of the
requests were submitted for unedited translation (i.e. raw
MT output), and that additional professional post-editing
at the higher rate was requested in only 15% of the cases.
Commenting on CompuServe’s MT-related services,
Bonthrone (1996: 4) reports that post-edited jobs tended
to be larger than those for which only raw MT was
requested. As a result, in terms of word count there was a
more balanced ratio of roughly 60% raw MT output vs.
40% post-edited content. Flanagan perceptively referred
to online translations as “MT’s new frontier” (Flanagan
1997b) – as the following years were to demonstrate.
Whilst the importance of the role played by
CompuServe in introducing MT to a large population of
new users (more than two million, according to MT News
International, 1999: 15) cannot be overestimated, the
access to MT provided by CompuServe via the Web was
still restricted, in that it was limited only to registered
subscribers. The online MT scene changed radically on
December 9, 1997 with the launch of Babel Fish, which
made MT available free of charge to any Internet user. It
was the result of a partnership between Systran Software
Inc. and the well-known search engine AltaVista.1 After
its experience with Minitel in France, Systran had started
to offer online translations of webpages from its own
website since 1996 (Yang & Lange, 1998: 276). Now, it
made the service more widely available.
2. Online MT and Its Use
It was Babel Fish that ensured the unprecedented global
visibility and accessibility of MT on the Internet.
However, it came with a few surprises. Yang & Lange
(1998: 282), reporting on users’ feedback and usage
behaviour in the first few months of its operation,
observed significant usage in areas that were either not
anticipated by the developers and providers of the service
(e.g. as a tool for language learning, cf. McCarthy, 2004
and Somers et al., 2006), or deprecated by them (i.e. as an
entertainment tool, getting it to perform ‘back-and-forth’
translations or to translate idioms). Initially Babel Fish
covered ten language combinations involving five major
European languages, and Yang & Lange (2003) provide
an update on its usage by a growing group of Internet
users (more information is available in the Appendix).
They report that between 1998 and 1999 the most
popular language combination was for translations from
English into Spanish, followed by English-French,
1 In 2006 the service was acquired by Yahoo!, becoming
officially known as Yahoo! Babel Fish (Flournoy, 2006).
German-English, Spanish-English, English-German and
French-German. They reported a wide variety of motives
by users (assimilation, dissemination, communication,
language learning, and of course entertainment), and an
amazing variety of texts were fed to the system, from
chatroom jargon to X-rated material, adult content, taboo
words and risqué terms. They revealed that over half the
‘texts’ submitted were less than five words, and only a
quarter were longer than 20 words (Yang & Lange, 2003).
Within a short time after its launch, the Babel Fish
service was no longer unique. Other MT vendors joined in
offering free online services. By 2000 there were more
than ten companies involved; apart from Babel Fish
(AltaVista) and CompuServe, these included
FreeTranslation (Transparent Language Inc.), Gist-in-
Time (Alis Technologies), iTranslator (Lernout &
Hauspie), MT Ave (Toshiba), My Translator (Apex
Network), PARS (Lingvistica), ProMT, and Reverso
(Softissimo). In many cases, these free services were
augmented by charged post-editing and/or human
translation services. Currently there are over 30 free
online MT services (see current issue of Compendium of
translation software).2
A major factor in the rapid popularity of web-based
MT services has undoubtedly been that they are available
free to users,3 and that results are (almost) instantaneous.
Limitation of the amount of text has been no impediment
for most users. Offering an MT service without charge
was clearly seen as a means of promoting sales of full MT
systems, both to the general public and to companies –
particularly since purchased systems would not have the
limitations (in length of texts, functions and facilities) of
the online service. To what extent the MT vendors have
benefitted in this way from their free online services is
unknown. However, it may well be that they have profited
from the leasing of their software to other web-based
services (e.g. news and current affairs providers).4
Information on current use of online MT from a
Japanese perspective comes in a report by Yamada et al.
(2005) on the MT market in Japan. A questionnaire-based
online survey elicited information from 4,000 respondents
between February 2003 and February 2005. The data
show a slight but steady increase in the use of online MT
services in this period. Not surprisingly, the
overwhelming majority of MT activity in Japan involves
English and Japanese. Those with limited knowledge of
English are particularly likely to use online MT to
translate content of websites available only in English.
However, no specific information is given on MT usage
by those with no knowledge of English at all. In this
period, on the other hand, there was a 5% rise in the
number of Japan-based professional translators using
online MT as part of their work.
The availability of free translation services has had
some less attractive consequences. It has given the
opportunity for users to exploit the known inadequacies of
automatic translation. One particularly unfortunate
example of the questionable use of Babel Fish and, for
2 Available at
3 But not necessarily free to providers; services such as
AltaVista and Yahoo! pay a fee to Systran.
4 Economic aspects are not covered here; for obvious reasons,
providers are unwilling to disclose commercially sensitive data.
that matter, of any MT software – is provided by Watters
& Patel (1999), who attempted to evaluate Babel Fish by
translating proverbs from English into other languages.
They argue that:
using a set of the most commonly known proverbs
in English, it should be possible to evaluate how
well direct translation systems are able to process
semantic information, and whether they correctly
select the appropriate sense of a word, where
multiple senses exist. (ibid.: 155)
Their evaluation method consisted in translating four
proverbs into each of the five target languages initially
supported by Babel Fish in combination with English, and
then back again into the original source language, and
trying to account for the mistranslations that occur in the
process. On this fairly impressionistic basis, the authors
analysed the data in detail, and, not surprisingly, they
found that the results in general were rather disappointing.
However, the authors do concede that their conclusions
“are limited to the extent that the translation performance
of expert human translators was not tested against” the
online MT service (ibid.: 159).
In subsequent years, numerous commentators have
enjoyed finding fault with online MT and, by
implication, with MT systems in general (e.g. Budiansky,
1998). The principal method is to input sentences which
contain one or more ambiguous words or ambiguous
syntactic structures. Naturally, the results are garbage and
often amusing.5 A major problem with the use of online
MT – one recognised by the MT community – is that most
users are not aware of the limitations of MT. Church &
Hovy (1993: 246) emphasised that “it should be clear to
the users what the system can and cannot do”, whatever
the type of MT service – but particularly, for a service
intended for large-scale use by the general public.
3. Use and Misuse
Before the appearance of free online MT, the acceptability
and usefulness of ‘less-than-perfect’ MT was a matter of
discussion. Church & Hovy (1993) argued that the MT
community should seek ‘niche applications’ where poor
quality (‘crummy’) MT would be acceptable, where
expectations were not too high, and the service “should be
attractive to the intended users” (ibid.). At the time, their
suggested applications involved variations on existing
practice, such as rapid post-editing and draft versions for
human translators. They did refer, however, to previous
experience of users with ‘raw’ unedited output (going
back to the 1960s and the Georgetown system – cf.
Henisz-Dostert, 1979), who were often happy with
unedited MT – particularly if otherwise no translation was
available. There were clearly many situations in which
‘crummy’ MT would be acceptable, and online MT
represents just such a ‘niche application’.
Such has been the uptake of free online MT that it is
now arguable that this form of MT is more widely known
to the general public than the corporate use of edited and
unedited MT in the production of technical material and
5 There is even a website devoted to ‘exposing’ the foolishness
of online MT at
reception/19/index.html [accessed 5 July 2007].
in the circulation of administrative documentation. Even
though there are a few exceptions, it is therefore all the
more strange that the MT community has largely ignored
discussion of these services and their impact on the image
of MT in the wider world. (The bibliography of this paper
lists a high proportion of the total English-language
literature on the topic.)
McLaughlin & Schwall (1998) outline the MT
products and services with the greatest potential on the
Internet. They present a case study focusing on Lernout &
Hauspie, at that time a leading provider of MT solutions
on/for the Web, reporting on its customer base, tools and
products. Ananiadou (1998), reviewing trends in MT in
Europe and Japan, provides details on language coverage,
modes of use and conditions of service of four leading
commercial providers of online MT (two based in Japan
and two based in Europe), and illustrates the dynamism
and growing demand in this area.
Prompted by the unanticipated ways and unreasonable
expectations of quality which some users have of online
MT, Bennett (2000) examines its merits and drawbacks,
and explains some of the major challenges involved in
processing unpredictable input. A technique often used
intuitively by some users to evaluate online MT is the so-
called “round-trip translation” (or RTT, also sometimes
referred to as “back-and-forth translation”). It is, however,
a technique without solid theoretical or empirical
foundations. Somers (2005) demonstrates that RTT is not
as useful as some lay-users of MT on the Web may think.
He conducted two separate experiments, based on a
variety of texts and involving five different free online
MT services (i.e. Babel Fish, FreeTranslation, Systran,
ProMT and Worldlingo). His conclusion is that although
non-experts in MT might see some value in it, the RTT
technique does neither help to reveal the quality of a
particular MT system nor to indicate the “machine
translatability” of a specific text.
4. Translators and Online MT
Fulford (2002) reports on an exploratory study conducted
between 2001 and 2002 to investigate the uptake of MT
among freelance translators based in the United Kingdom.
Of the 30 individuals who were interviewed, only two
(i.e. less than 7%) “were actively using MT in their work”
(Fulford, 2002: 119). Interestingly, however, eight of the
professional translators who were interviewed (26%)
“stated that they had ‘occasionally’ made use of web-
based MT systems to produce an initial rough draft of a
translation, or to ‘get ideas for producing a translation,
before polishing the output manually ready for
presentation to a client” (ibid.). This suggests that
although, in general, professional translators are reluctant
to invest in MT software and to integrate it into their
working practice, online MT is seen by some as a
potentially valuable translation aid.
Fulford & Granell-Zafra (2004a) give a progress
report with 390 responses received from freelancers. The
authors conclude that “[b]eyond terminology and
document consultation / look-up, there was little or no
actual use being made of online systems, such as MT”
(Fulford & Granell-Zafra, 2004a: 59). In another paper,
providing additional data from the same survey, Fulford
& Granell-Zafra (2004b) report that only 3% of the 390
freelance translators who took part in the survey made use
of online MT systems as part of their work (ibid.: 41),
compared with much higher percentages for the use of
other Internet-based tools such as e-mail (93%), search
engines (85%), online dictionaries and glossaries (78%)
and multilingual terminology databases (59%).
5. Online MT and the MT Community
The ready availability of online MT systems on the
Internet over the last ten years has had direct and indirect
effects on all those involved with MT, namely not only on
the users, but also on the wider MT research community
and especially on the MT industry and the MT vendors.
The earliest indication dates back more than two years
before the launch of Babel Fish, recognising the pioneer
developments by CompuServe at that time. In a survey of
the users and usage of MT in Europe and the Americas,
Brace et al. (1995) predict an “upsurge in the use of MT
on-line”, which they call an “impressive development”.
The paper also refers to the experience of Internet-based
providers who offered added-value services to customers
requiring post-edited polished versions of online MT
output, demonstrating that in the mid-1990s the MT
support services were attracting growing commercial
At the AMTA conference in 1996, two of the speakers
on a panel “MT Online: The Future is Now!” (AMTA,
1996: 220 ff.) were David Clements of Globalink and
Patricia O’Neill-Brown of the US Department of
Commerce. Clements (1996) presents a number of
scenarios in which the availability of online MT is the
ideal solution to real communication problems. However,
the language used in Internet-based exchanges is often
stylistically and grammatically sloppy, and thus presents
unprecedented and largely unpredictable challenges. On
the other hand, he emphasises that translation technology
is essential to enable communication and interactions on
the Internet, and concludes that the “convergence of the
Internet, e-mail and various online services, as well as the
increasing popularity of the personal computer over the
whole world, presents the greatest opportunity yet to bring
MT “to the masses”” (Clements, 1996: 221). O’Neill-
Brown (1996) points out that trivial but still important
technical aspects can prevent the large-scale or effective
deployment of online MT tools, that large volumes of
documents are available only in paper form, and that
issues of encoding can have adverse effects on MT
processing of online texts. In a number of respects, the
problems raised by O’Neill-Brown have been resolved in
subsequent years.
Westfall (1996) raises a series of interesting questions
regarding the legal implications of online MT, with
particular reference to the potential risk of litigation and
lawsuit against MT companies and providers. They may
be liable if web-based MT services are used to translate,
disseminate and distribute (on the Internet or otherwise)
text protected by copyright, information of a
commercially sensitive nature, or content that is illegal for
whatever reason. Another potential liability might arise if
an incorrect translation provided by an online system
leads to safety violations. She concludes that “[t]he legal
issues surrounding machine translation on-line will need
to be defined within the next year” (Westfall, 1996: 231).
Unfortunately, this has not yet happened, as a few years
later Yang & Lange (1998: 282-283; 2003: 205-206)
comment on a number of outstanding legal issues with
regard to Babel Fish where users are threatening to take
legal action for specific incidents resulting from
mistranslations, or are demanding that the service should
provide a clear disclaimer that it cannot accept any
liability in connection with its use (cf. Gaspari, 2004: 69).
In an overview of MT research and development in
Canada, Macklovitch (1997: 204) claims that “the Internet
is transforming classic MT”. In considering the impact of
this evolving scenario on the MT industry, he describes
the commercial strategy of one major Canadian-based
software company. The approach of Alis Technologies
was to specialise at this time in the customisation and re-
selling of MT products manufactured by other vendors,
and to provide a one-stop solution to corporate clients
needing Internet-based translation and multilingual online
content management. The same author returns to this
topic in a later paper, reinforcing these arguments on the
basis of the developments and trends detected in the
intervening few years (Macklovitch, 2001: 27), describing
the attempts of the MT industry to capitalise on the
demand for automated solutions to translation needs in the
online environment. Similar discussions are also found in
Smith (2001a) and Baron (2003: 118).
Allen (2000) focuses on the impact of online MT
services on the MT market in the late 1990s, and offers a
perspective on their value for the users as well as for MT
companies. This contribution sheds light on the
connections between the several free online MT services
available on the Internet and the commercial strategies of
a relatively small number of companies specialising in
MT system development and implementation. When
Internet users try out free MT services that are based on
related commercial products, companies see an
opportunity to make them paying customers who for a fee
receive higher-quality translation services. They offer, for
example, the possibility of translating unlimited amounts
of text (whereas length limits and connection timeout
constraints apply to non-paying occasional users) and the
option of activating domain-specific dictionaries to
improve the quality of translations of specialised texts
(whilst free web-based MT services typically have only
general-language dictionaries that cannot be customised
or augmented).
6. Customising Online MT
Limitations on text lengths and on dictionary coverage are
obvious impediments to corporate users. However, many
of them are reluctant to invest time and money in full-
scale in-house MT systems. For this reason, we are seeing
increasingly the use of Internet-based MT engines as the
basis for enhanced translation services. Smith (2001b)
describes a pilot project to test the feasibility of offering
an online MT application powered by Systran to the
multilingual employees of PricewaterhouseCoopers
(PwC) over the firm’s intranet. The proposed MT facility
covers 20 language pairs and users can select specialised
dictionaries to improve the quality of the MT output. The
quality of the results is variable, and additional fine-
tuning was needed. Further developments are reported in
Smith (2003): some customisation at the terminological
level to tailor the MT dictionaries to the needs of the
company; the number of language pairs covered by
Systran nearly doubled, rising to 37; and, the most
popular language combination among PwC employees is
for translations from English into Spanish, nearly double
the next most popular, from Spanish to English – which is
in line with usage for the free online service provided by
Babel Fish. The paper also reports that approximately
130,000 translation requests coming from 7,300
individuals around the world have been processed by the
MT engine since it was first introduced. User feedback
indicates a prevalence of positive reactions, although
there is also evidence of negative responses, and some
possible future enhancements to the online MT facility are
Kübler (2002) focuses on similar efforts to enhance
the performance of online MT services, but outside the
corporate environment. This study addresses the
challenges of combining available tools and resources to
customise dictionaries used by online MT systems, in an
attempt to enable technical translators to become more
time-efficient in their work. The experiments make use of
Systranet, an online MT service powered by Systran,
which gives users access to a dictionary management
facility to create and augment personalised dictionaries, to
improve the quality of translations of specialised and
domain-specific documentation. The paper focuses on the
translation of technical material from English into French.
After initial investment of time to identify relevant
terminology and feed it into Systranet’s customisable
dictionaries, noticeable improvements were achieved in
the quality of the raw output provided by the web-based
MT system. The output produced with personalised
dictionaries provides a target-language draft good enough
for post-editing and polishing into a final text of
professional standard, saving time and guaranteeing
productivity gains.
7. Current Use of Online MT
The users of online MT are probably the largest group of
MT users, and yet we know very little about them. How
satisfied are they with the quality of online MT? How
often do they use online MT? How well do they know the
language(s) they want to translate into or from? What
types of translation ‘errors’ are most irritating? Do they
want more specialised services (e.g. for medical texts, for
technical documentation)? How much do they use online
MT for emails and chatrooms? Do they use MT for
business purposes? What are their social, occupational
and age groups? etc., etc. Much of the discussion about
online MT is based on speculation. As far as we know,
there has been no large-scale survey of users and of what
they expect from online MT now or in the future. It could
be instructive to hear the views of users themselves,
particularly as this area of MT provision is now what the
majority of the general public consider to be the most (or
even ‘best’) that MT can do.
Very little data regarding the use of online MT
services is publicly available, and most companies
associated with web-based MT systems seem reluctant to
publicise information regarding, for example, the nature
of the texts submitted for translation or to reveal feedback
received from the users (Zervaki, 2002), although usage
patterns are constantly monitored. In addition, the
providers of online MT regard information on a wide
range of topics, particularly having to do with the inner
workings of their web-based MT systems, as proprietary
and confidential, given its sensitive nature in a highly
competitive field. To date, Yang & Lange (1998) and
Yang & Lange (2003) are the key sources of information
in this respect, although the perspective that they offer is
rather limited and outdated, being focused exclusively on
the usage of Babel Fish on two census days in June 1998
and November 1999.
In an attempt to find more up-to-date and
representative information about the overall current usage
of online MT services, and with a view to comparing it
with the scant reports that are already available, the
authors have approached a number of people directly
associated with some of the leading online MT services.
Eventually, we received usage data regarding three of the
major providers of web-based MT systems, namely
Yahoo! Babel Fish, FreeTranslation and Systran. The
survey was centred on the following four core questions:
1) most popular and frequently used language pairs;
2) number of translation requests per unit of time (day,
week or month);
3) ratio of translation types, i.e. for webpages vs.
passages of plain text;
4) length of plain-text translations (number of words per
translation job).
We also emphasised that we would welcome any other
insight or detail that the providers of web-based MT were
willing to supply to us for the purposes of this paper. The
Appendix at the end of the paper shows the data on
current usage that our contacts have kindly shared with us
(alongside the information about Babel Fish for 1998 and
1999). As a result, the Appendix allows for some
comparison across the leading online MT services, giving
at the same time an overall picture of the scale of the
traffic and of the volume of translation requests handled at
present by the web-based MT systems involved in the
Responses from MT providers about the use of their
online services indicate that the most popular language
pairs remain English to and from Spanish and English to
and from French, and that (as expected) for each
particular country, the most popular pair is English to and
from the native language. Overall, the volume of online
MT continues to grow Flournoy (personal
communication) reports that the advent of Yahoo’s Babel
Fish has not affected in any way the volume of traffic on
AltaVista’s MT service, which is still in existence and
operating as usual.
What is perhaps surprising is that translation from
webpages is much less common than plain text (only 2%
of Yahoo! Babel Fish, less than 10% of FreeTranslation
and no more than 17% on AltaVista is webpage
translation). No less surprising is that most users are using
the online services to look up or check translations of
single words or very short phrases. This may perhaps
mean that most users have some knowledge of the
original language and require only occasional assistance,
i.e. they use the services as electronic dictionaries. Some
of these users may in fact be translators (as described in
section 4) or readers seeking only to extract small pieces
of information from texts. And a few may be ‘testing’ MT
out of curiosity or for entertainment (as mentioned in
section 2). This low use of online MT for translating more
substantial texts (whether plain text or webpages) may
suggest also that the number of users with poor or no
knowledge of the source languages is relatively small.
However, as emphasised above, we really know
almost nothing about who the users of online MT services
are, what their language knowledge is, what they are
looking for when using online MT services, and how
much awareness they possess of the limitations and
potential of MT software. We hope that this gap will be
filled by future research efforts, possibly to be undertaken
in close collaboration with the MT providers, who would
certainly benefit from a deeper understanding of the
reasons and circumstances leading users to access online
translation services.
8. Conclusion: Future Prospects of Online MT
The comparison between the current usage of the major
online MT services and the volume of translations
handled by Babel Fish in the late 1990s, when web-based
MT was still a newcomer to the Internet, indicates that
over the years the growth in demand on the part of the
users has been constant and substantial. To match this
demand and offer a reliable and efficient service, the
providers invest significant resources to support the
growing amount of traffic. In addition, the developers are
continuously working to improve the facilities and
capabilities offered by the online MT systems (Flournoy,
2006), with a special emphasis on adding new language
pairs likely to attract interest and enhancing translation
quality, by creating more lexical entries in the dictionaries
and including more powerful rules in the systems all the
The surprisingly low current use of online MT for
translating texts longer than a few words or phrases (as
reported by service providers) may mean that these free
services do not yet represent as great a revolution in
overall MT usage as might have been anticipated when
they were first introduced. In other words, online MT is
meeting a demand for occasional ‘translation’ that
traditional translation services could not meet – a demand
primarily for translation assistance from users already
with some knowledge of the ‘source’ language. Although
in volume terms this may currently be the dominant use,
there is still a substantial demand for online translation of
longer texts and for translation from unknown languages
even if it is only a small fraction of usage, out of the
millions of translation requests per day it represents a
large demand and one not in any way to be dismissed.
This is a demand that the providers have to meet and one
which is likely to grow over time.
If the trends observed in the past decade continue, the
interest of Internet users for online MT services is set to
remain high for the foreseeable future, and it can be safely
assumed that the providers will try hard to benefit from
this. It is not clear, however, whether over time the users
of online MT services, particularly casual or occasional
ones, will become more aware of the limitations of these
systems, and will have more reasonable expectations
regarding the level of quality that they can realistically
offer. There is the likelihood that new and extended
support services will be offered (e.g. easier options for
post-edited or human translation, advice on how to
implement MT-based multilingual communication
strategies, etc.). A vital challenge consists in the education
of general users on how to gain the maximum benefit
from using online MT services, as advocated by Somers
(2003: 523), with a view to strengthening the confidence
of average web surfers in the use of web-based MT
(Gaspari, 2006).
As yet, online MT (and, indeed, MT in general) gives
poor results for the kind of colloquial and ill-formed
language found in electronic mail, chatrooms and blogs.
In future, there may be several online MT services
devoted to such text types;6 and it is to be hoped that more
research effort will be directed to this neglected area if
only for the sake of the reputation of machine translation
itself. Many MT service providers have access to vast
caches of emails, webpages and blogs, which could be
analysed and utilised by current statistical MT techniques
(cf. Yang & Lange, 1998: 280). Perhaps it is being done
already – we do not know.
More speculatively, the coming of Internet telephony
suggests the ultimate possibility of online MT of spoken
language – initially perhaps in constrained domains for
some very specific tasks. More immediately, there must
be the expectation that free online MT services will
provide an ever wider range of language pairs. Translation
from English is available into a large number of
languages, although there is always room for more –
particularly, as always, for languages of Africa and India.
Translation into English is less well served – even some
European languages are not provided for (Czech,
Lithuanian, Polish). Free online MT services are available
for some non-English pairs, such as Chinese-Japanese,
Korean-Japanese, French-German, Spanish-Portuguese;
but there is clearly a need for many more. What can be in
no doubt is that online MT will continue to grow, and that
it will become the principal focus of MT activity and
research in the not too distant future.
The authors would like to thank the following people for
their invaluable help in the preparation of this paper, and
in particular for their assistance in providing the data on
the current usage of the leading online MT services
included in the Appendix: Raymond Flournoy (Yahoo!
Babel Fish), Jay Marciano (FreeTranslation), Cris Fitch
(Systran). Many thanks also to three anonymous
reviewers for their helpful comments on an earlier draft of
this paper.
Allen, J. (2000) “The value of Internet translation
portals”. International Journal for Language and
Documentation 6(August-September): 45-46
AMTA (1996) Proc. 2nd AMTA Conference
Ananiadou, S. (1998) “Machine translation trends in
Europe and Japan”. Proc. 20th Translating and the
Baron, N.S. (2003) “Language of the Internet”. Farghaly,
A. (ed.), Stanford Handbook for Language Engineers.
Stanford: CSLI Publications. 59-127
Bennett, S. (2000) “Taking the babble out of Babel Fish”.
Language International 12(3): 20-21
6 The company Translution has launched an online service for
translating electronic mail, with limited functionality (see
7 Papers from proceedings of AMTA, EAMT and MT Summit
conferences are available at
Bonthrone, R. (1996) “Translation and the Internet Part
2”. LISA Newsletter 5(2). Available at
Translation_and_Net-0396.pdf [accessed 5 July
Brace, C., M. Vasconcellos & L.C. Miller (1995) “MT
users and usage: Europe and the Americas”. Proc. MT
Summit V
Budiansky, S. (1998) “Lost in translation” Atlantic
Monthly 282(6): 80-84
Church, K.W. & E. Hovy (1993) “Good applications for
crummy machine translation”. Machine Translation
8(4): 239-258
Clements, D. (1996) “The value of online MT in the age
of the ‘cyber society’”. Proc. 2nd AMTA Conference.
Flanagan, M. & P. Jensen (1994) “Machine translation on
CompuServe Forums”. Proc. 1st AMTA Conference.
Flanagan, M. (1995) “MT in the online environment:
challenges and opportunities”. Proc. MT Summit V
Flanagan, M. (1996a) “Two years online: experiences,
challenges and trends”. Proc. 2nd AMTA Conference.
Flanagan, M. (1996b) “System demonstration: online MT
at CompuServe”. Proc. 2nd AMTA Conference. 244-246
Flanagan, M. (1997a) “MT today: emerging roles, new
successes”. Machine Translation 12(1-2): 25-27
Flanagan, M. (1997b) “Online translation: MT’s new
frontier”. Proc. 19th Translating and the Computer
Flournoy, R. (2006) “Say “hello” to my little friend”.
Posting dated April 27, 2006. Available at
000295.html [accessed 5 July 2007]
Fulford, H. (2002) “Freelance translators and machine
translation: an investigation of perceptions, uptake,
experience and training needs”. Proc. 6th EAMT
Workshop. 117-122
Fulford, H. & J. Granell-Zafra (2004a) “The freelance
translator’s workstation: an empirical investigation”.
Proc. 9th EAMT Workshop. 53-61
Fulford, H. & J. Granell-Zafra (2004b) “The uptake of
online tools and web-based language resources by
freelance translators: implications for translator
training, professional development, and research”.
Proc. 2nd International Workshop on Language
Resources for Translation Work, Research and
Training. 37-44
Gachot, J. (1989) “Machine translation in everyday use”.
Proc. MT Summit II. 72-73
Gaspari, F. (2004) “Integrating on-line MT services into
monolingual web-sites for dissemination purposes: an
evaluation perspective”. Proc. 9th EAMT Workshop. 62-
Gaspari, F. (2006) “The added value of free online MT
services: confidence boosters for linguistically-
challenged Internet users, a case study for the language
pair Italian-English”. Proc. 7th AMTA Conference. 46-
Harrison, C. (1992) “CompuServe investigates machine
translation”. MT News International, issue 3,
September 1992. 11
Henisz-Dostert, B. (1979) “Users’ evaluation of machine
translation”. Henisz-Dostert, B., R.R. Macdonald & M.
Zarechnak (eds.), Machine translation. The Hague:
Mouton. 147-244
Kübler, N. (2002) “Teaching commercial MT to
translators: bridging the gap between human and
machine”. Proc. 6th EAMT Workshop. 155-162
Macklovitch, E. (1997) “MT R&D in Canada”. Teller, V.
& B. Sundheim (eds.), Proc. MT Summit IV. 200-210
Macklovitch, E. (2001) “Recent trends in translation
technology”. Proc. 2nd International Conference The
Translation Industry Today. 23-47
McCarthy, B. (2004) “Does online machine translation
spell the end of take-home translation assignments?”
CALL-EJ Online 6(1). Available at http://www.
al/6-1/mccarthy.html [accessed 5 July 2007]
McLaughlin, S. & U. Schwall (1998) “Spicing up the
information soup: machine translation and the
Internet”. Farwell, D., L. Gerber & E. Hovy (eds.),
Proc. 3rd AMTA Conference. Berlin: Springer. 384-397
MT News International. Newsletter of the IAMT (1999)
8(1), issue no. 22, July 1999
O’Neill-Brown, P. (1996) “Online machine translation”.
Proc. 2nd AMTA Conference. 222-223
Ryan, J.P. (1987) “Systran: a machine translation system
to meet user needs”. Proc. MT Summit. 99-103
Smith, R. (2001a) “Machine translation: potential for
progress”. English Today 17(4): 38-44
Smith, R. (2001b) “Using information technology to
optimise translation processes at
PricewaterhouseCoopers Madrid”. Proc. MT Summit
VIII. 341-344
Smith, R. (2003) “Overview of PwC/Systranet on-line
MT facility”. Proc. 25th Translating and the Computer
Somers, H. (2003) “Machine translation: latest
developments”. Mitkov, R. (ed.), Oxford Handbook of
Computational Linguistics. Oxford: OUP. 512-528
Somers, H. (2005) “Round-trip translation: what is it good
for?” Proc. Australasian Language Technology
Workshop 2005. 127-133. Available at
~harold/RoundTrip.doc [accessed 5 July 2007]
Somers, H., F. Gaspari & A. Niño (2006) “Detecting
inappropriate use of free online machine translation by
language students a special case of plagiarism
detection”. Proc. 11th EAMT Conference. 41-48
Watters, A. & M. Patel (1999) “Semantic processing
performance of Internet machine translation systems”.
Internet Research: Electronic Networking Applications
and Policy 9(2): 153-160
Westfall, E. (1996) “Legal implications of MT on-line”.
Proc. 2nd AMTA Conference. 231-232
Yamada, S., S. Kodama, T. Matsuoka, H. Araki, Y.
Murakami, O. Takano & Y. Sakamoto (2005) “A
Report on the Machine Translation Market in Japan”.
Proc. MT Summit X. 55-62
Yang, J. & E.D. Lange (1998) “SYSTRAN on AltaVista:
a user study on real-time machine translation on the
Internet”. Farwell, D., L. Gerber & E. Hovy (eds.),
Proc. 3rd AMTA Conference. Berlin: Springer. 275-285
Yang, J. & E. Lange (2003) “Going live on the Internet”.
Somers, H. (ed.), Computers and translation: a
translator’s guide. Amsterdam: John Benjamins. 191-
Zervaki, T. (2002) “Online free translation services”.
Proc. 24th Translating and the Computer
AltaVista Babel Fish (from December 1997 until early 2006)
Date of launch 9 December 1997
Primary URL
Initial language pairs 5 bidirectional language pairs (i.e. 10 separate combinations) including major
European languages, i.e. EN<>FR, EN<>DE, EN<>ES, EN<>IT, EN<>PT
Most popular language pairs In decreasing order of frequency: EN>ES, EN>FR, DE>EN, ES>EN, EN>DE,
FR>DE, EN>IT, EN>PT, IT>EN, PT>EN (Nov. 1999)
Volume of translations 500,000 per day (May 1998); 740,218 (10 Nov. 1999); 1.3 million per day (Oct. 2000)
Types and details of translations
42.3% webpages vs. 57.6% plain text (June 1998); 17.6% webpages vs. 82.4% plain
text (Nov. 1999); 50%+ of translations are one- or two-word phrases (May 1998);
average length of texts submitted approx. 20 words (Nov. 1999)
User feedback Between January and May 1998 more than 5,000 emails sent by users with linguistic
comments or feedback on translations – estimated 95% positive
Table 1: Information on the use of AltaVista Babel Fish extracted from Yang & Lange (1998) and Yang & Lange (2003)
Yahoo! Babel Fish (since April 2006)
Date of launch 27 April 2006 (effectively a re-launch under the Yahoo! brand)
Primary URL
Initial language pairs 38 language pair-directions, including EN<>KO, DU<>FR, GR<>EN, etc.
Most popular language pairs EN<>ES (main/US sites); EN<>FR (UK site); in every non-English-speaking region,
the most popular pair is always the local native language<>EN
Types and details of translations
- 2% URLs vs. 98% text for requests submitted directly via the Yahoo! Babel Fish
portal (not including requests from other sources, such as search results or the toolbar)
- The launch of Yahoo! Babel Fish in April 2006 has not affected the volume of traffic
on AltaVista Babel Fish (which is still in existence and operating as usual)
- The language pair traditional Chinese<>simplified Chinese was developed in-house
by Yahoo! Babel Fish and is the only one which is not also on AltaVista Babel Fish
Table 2: Information on the use of Yahoo! Babel Fish as of April 2007
Date of launch 21 June 1999 (official announcement in press release), but the service had been running
in “stealth mode” for several weeks before this official launch
Primary URL
Most popular language pairs
Based on a sample of 1 million translation requests to a single server: EN>ES (34.83%
of total); ES>EN (22.17%); EN>FR (11.85%); FR>EN (7.31%); DE>EN (5.09%);
EN>DE (4.91%)
Volume of translations
3,500 on first day of stealth operation; 50,000 per day (Dec. 1999); 100,000 per day
(Dec. 2000); 1 million per day (late 2002); nearly 3 million per day (Jan. 2006); 3.4
million per day, corresponding to 50 million source words (Sep. 2006)
Types and details of translations
- More than 90% of the requests are for plain text translations; average number of
source words per translation request is 15
- Average of 3 translations per minute done in 6-week beta period (mid-1999)
- Peak usage is usually between 8:00 and 10:00 PM Eastern Standard Time, with up to
4,000 translation requests per minute received during this time
Table 3: Information on the use of FreeTranslation as of September 20068
Services powered by Systran
Most popular language pairs EN<>FR; EN<>ES; EN<>DE; RU>EN; EN<>ZH; EN<>JA; more than 40 language
pairs in total
Volume of translations More than 30 million pages translated per day on all services powered by Systran
Types and details of translations
On average plain-text translations are becoming increasingly shorter, and it is very
common for people to use the service to look up and translate single words
Table 4: Information on the use of Systran (all services) as of June 2007
8 Updates about this service are regularly posted by its developers at [accessed 5 July 2007].
... The MT applications have been continuously developed for decades and have received the attention from scholars and computational linguists. The history, background, development, and evolution of OMT has been studied by Gaspari and Hutchins (2007). According to their paper, the MT have been available on the Internet (restricted to France) since the late 1980s. ...
... According to their paper, the MT have been available on the Internet (restricted to France) since the late 1980s. The Systran Centre was the first company to provide OMT service translating texts from French to English, German to English, and English to French (Gachot, 1989 as cited in Gaspari & Hutchins, 2007). It was reported by Ryan (1987) in Gaspari and Hutchins' study that their online software has been used by up to 4.5 million users in France. ...
Full-text available
The present study aims to investigate how Thai EFL university students use Google Translate (GT) in English writing, how they post-edit (PE) its outputs, and how they view GT use in English writing. The participants were 15 third-year non-English major students from three universities in Thailand. The data collection tools were an interview and two writing assignments. After the data analysis, the findings revealed the students’ behavior of GT use and their output PE as well as their attitudes toward GT use in English writing. The results reported the students always used GT in completing writing tasks at both sentence and paragraph levels, and most students did PE the outputs before applying them. However, a few students used the outputs with no PE because they trusted in GT more than they did in themselves. Regarding the PE level, the students intended to address lexical and syntax errors, so their correcting covered the light level. The results also revealed mixed messages in their attitudes toward GT use in English writing. Most students viewed GT as a helpful, reliable assistant enhancing their writing quality, but some raw GT outputs of phrases, idioms, long sentences, and paragraphs were found incomprehensible. Also, the students acquired some bad habits from using GT. However, most students disagreed with not being allowed to use GT in English writing. The study recommended language teachers to provide Thai EFL students adequate instructions for the effective use of GT and its output PE.
... Research in [3] shows that 1980 is considered a flagship moment of MT with new developments emerging whilst "more dramatic development took place in MT in the 1990 since computers became more powerful with much higher storage capabilities". This crowned with the invention of the internet, a source of translation for the general public [4]. Automatic translation tools available on the Internet translate billions of documents daily, which would take human translators months. ...
... MT systems Evaluation (MTE) is crucial to the development of MT systems. The author in [4] shows that the evaluation of MT is central to determine the effectiveness and performance of MT systems. Machine Translation Evaluation is essential to all endusers: researchers, designers and users to select the best system to use [16]. ...
Full-text available
The study aims to verify the capacity of Facebook Translation Service in translating English Facebook posts into Arabic in terms of two criteria: adequacy and fluency in line with the Translation Automation User Society (TAUS) scales. To ensure consistency and objectivity as recommended by TAUS, six evaluators, native speakers of Arabic and near-native speakers of English, rated the same data on each scale. The evaluators were acquainted with fluency and adequacy scales along with MT limitations and potentials. Once the corpus was uploaded and sent to the evaluators using TAUS tools, they had to assign scores online on 1-4 rating scales. Then, each report was displayed online on the TAUS reports tool. Evaluators’ responses were combined in thematic categories and were calculated to obtain frequencies and percentages. The study found that FTS provided fluent output with highest percentage of the scale good equal to 3 on a scale from 1 to 4, where the output is assessed as flowing smoothly with minor linguistic errors. Moreover, FTS succeeded in generating an adequate output with the highest percentage of responses as ‘most’ equal to 3 on a scale from 1 to 4, where almost the full meaning of the source is deemed to be transferred in the target language. This study is useful since it highlights the role of Facebook Translation service in translating, educating the public and fighting COVID-19. Consequently, such research would encourage the use and research on the potentiality of MT and FTS in dealing with abrupt crises, such as COVID-19.
... Despite the criticism involving the use of translation as a means for language teaching and learning and the widespread adoption of the direct method, teachers and language instructors of the 21st century were challenged by the emergency and popularization of translation technologies such as Machine Translation systems (MT), especially with the free availability of popular tools such as Google Translate system (GT). Although the quality of the output of these systems was considered poor in their earlier stages of development [4], advances in the field of Artificial Intelligence have brought about significant improvements to the MT output leading to an increase in MT users especially in the past five years. Way [5] notes with respect to MT usage on Facebook that "with 2 billion translations being provided on a daily basis, and almost 1 billion users seeing these translations each month, the numbers are truly staggering". ...
Full-text available
In this article, we address the question of whether exposure to the translated output of MT systems could result in changes in the cognitive processing of English as a second language (L2 English). To answer this question, we first conducted a survey with 90 Brazilian Portuguese L2 English speakers with the aim of understanding how and for what purposes they use web-based MT systems. To investigate whether MT systems are capable of influencing L2 English cognitive processing, we carried out a syntactic priming experiment with 32 Brazilian Portuguese speakers. We wanted to test whether speakers re-use in their subsequent speech in English the same syntactic alternative previously seen in the MT output, when using the popular Google Translate system to translate sentences from Portuguese into English. The results of the survey show that Brazilian Portuguese L2 English speakers use Google Translate as a tool supporting their speech in English as well as a source of English vocabulary learning. The results of the syntactic priming experiment show that exposure to an English syntactic alternative through GT can lead to the re-use of the same syntactic alternative in subsequent speech even if it is not the speaker’s preferred syntactic alternative in English. These findings suggest that GT is being used as a tool for language learning purposes and so is indeed capable of rewiring the processing of L2 English syntax.
... aids thousands of daily users with its free online translation service. However, after its inception in the World Wide Web, users started to report translation errors, which the founder and providers of the service did not expect [9][10][11][12]. To date, although ...
... In a broad historical overview, Gaspari & Hutchins (2007) stated that Babel Fish, which was launched on December 9, 1997, was the first free online MT. Since then, many free MT online services became widely available and are regarded as a fast tool available for all internet users. ...
Full-text available
The purpose of this study was to determine whether significant differences exist in Chinese-to-English translation accuracy between moderate to higher-level human translators and commonly employed freely available machine translation tools. A Chinese-to-English language proficiency structure test and a Chinese-to-English phrase and sentence translation test were given to a large sample of machine (n=10) and human translators (n=133) who are native Chinese speakers with at least 15 years of familiarity with the English language. Results demonstrated that native Chinese speakers with this minimum level of English proficiency were significantly better at translating sentences and phrases from Chinese to English, compared to the ten freely available online machine translation applications, which unexpectedly showed a considerable degree of variation in translation accuracy among them. These results indicate that humans with at least a moderate level of exposure to a non-native language make far fewer translation errors compared to machine translation tools. This outcome is understandable, given the unique human ability to take into account subtle linguistic variants, context, and capricious meaning associated with the language and culture of different groups.
... Qualitative improvements have tended to be incremental rather than exponential. Statistical machine translation (SMT) became the dominant MT paradigm for comprehension and production purposes following its adoption by Internet MT providers and the appearance of the first commercial SMT systems in the 2000s (Gaspari and Hutchins, 2007). Since then, MT for production has become mainstream, to the extent that by 2016, 56% of language service S. Castilho ADAPT Centre -Dublin City University Tel.: +353 17006719 E-mail: providers offered an integrated, client-or project-specific MT system as part of their workflows, using in-house or third-party engines trained by one of the many dedicated MT providers (Lommel and DePalma, 2016). ...
Full-text available
This article reports a multifaceted comparison between statistical and neural machine translation (MT) systems that were developed for translation of data from massive open online courses (MOOCs). The study uses four language pairs: English to German, Greek, Portuguese, and Russian. Translation quality is evaluated using automatic metrics and human evaluation, carried out by professional translators. Results show that neural MT is preferred in side-by-side ranking, and is found to contain fewer overall errors. Results are less clear-cut for some error categories, and for temporal and technical post-editing effort. In addition, results are reported based on sentence length, showing advantages and disadvantages depending on the particular language pair and MT paradigm.
Machine Translation (MT) has become widely used throughout the world as a medium of communication between those who live in different countries and speak different languages. However, translation between distant languages constitutes a challenge for machines. Therefore, translation evaluation is poised to play a significant role in the process of designing and developing effective MT systems. This book evaluates three prominent MT systems, including Google Translate, Microsoft Translator, and Sakhr, each of which provides translation between English and Arabic. In the book Almahasees scrutinizes the capacity of the three systems in dealing with translation between English and Arabic in a large corpus taken from various domains, including the United Nation (UN), the World Health Organization (WHO), the Arab League, Petra News Agency reports, and two literary texts: The Old Man and the Sea and The Prophet. The evaluation covers holistic analysis to assess the output of the three systems in terms of Translation Automation User Society (TAUS) adequacy and fluency scales. The text also looks at error analysis to evaluate the systems’ output in terms of orthography, lexis, grammar, and semantics at the entire-text level and in terms of lexis, grammar, and semantics at the collocation level. The research findings contained within this volume provide important feedback about the capabilities of the three MT systems with respect to English<>Arabic translation and paves the way for further research on such an important topic. This book will be of interest to scholars and students of translation studies and translation technology. Available for pre-order. Item will ship after November 30, 2021
In this chapter, we argue that education and training in translation quality assessment (TQA)is being neglected for most, if not all, stakeholders of the translation process, from translators, post-editors, and reviewers to buyers and end-users of translation products and services. Within academia, there is a lack of education and training opportunities to equip translation students, even at postgraduate level, with the knowledge and skills required to understand and use TQA. This has immediate effects on their employability and long-term effects on professional practice. In discussing and building upon previous initiatives to tackle this issue, we provide a range of viewpoints and resources for the provision of such opportunities in collaborative and independent contexts across all modes and academic settings, focusing not just on TQA and machine translation training, but also on the use of assessment strategies in educational contexts that are directly relevant to those used in industry. In closing, we reiterate our argument for the importance of education and training in TQA, on the basis of all the contributions and perspectives presented in the volume.
Graduates of translation programmes increasingly encounter machine translation in the language industry. In response to this identified market need, translation education programmes have begun to incorporate machine translation into the curriculum. In order for translation graduates to serve as professional post-editors in the language industry, this content must be embedded in multiple courses across the curriculum, rather than concentrating the material in a stand-alone course or module. This article argues for the incorporation of terminology management, controlled authoring, post-editing, and engine tuning into translation practice courses to prepare students for the evolving market.
The share of the economy related to translation activities is growing steadily under the influence of the globalization of exchanges. Today it numbers dozens of billions of which an increasing share belongs to machine translation. Various factors, such as migratory flows or the propagation of mobile telephony, prompt new translation practices in a variety of languages with simultaneous coverage enabled by networks. Nevertheless, is it true as we intuitively believe that translation promotes linguistic and cultural diversity? This article originates from a study conducted for UNESCO’s world report on cultural diversity (2009). This study notably reveals that 75% of all books are translated from three languages with 55% being from English. On a planetary scale, translation is dominated by some twenty languages, primarily European. In the new world economic order, the urgent and paradoxical task is to “rebabelize” the world.
Conference Paper
Full-text available
This paper presents the experiment that was led at the University Paris 7 to teach machine translation to translation students. Linking machine translation with translation-training in order to help students acquire the linguistic and technical skills increasingly required on the translation market has become part of a number of translation syllabuses in the last few years. Although, MT knowledge is more and more required for professional translators, most of them are still skeptical, not to say hostile to MT. Our objective here is then twofold: showing translation trainees why and how MT can be used for professional purposes, from the text to be translated to the final result.
Full-text available
This paper reports on an experiment inves-tigating how effective free online machine translation (MT) is in helping Internet us-ers to access the contents of websites writ-ten only in languages they do not know. This study explores the extent to which us-ing Internet-based MT tools affects the confidence of web-surfers in the reliability of the information they find on websites available only in languages unfamiliar to them. The results of a case study for the language pair Italian-English involving 101 participants show that the chances of identifying correctly basic information (i.e. understanding the nature of websites and finding contact telephone numbers from their web-pages) are consistently en-hanced to varying degrees (up to nearly 20%) by translating online content into a familiar language. In addition, confidence ratings given by users to the reliability and accuracy of the information they find are significantly higher (with increases be-tween 5 and 11%) when they translate websites into their preferred language with free online MT services.
Full-text available
On-line machine translation (MT) services are becoming increasingly popular among Internet users. In particular, over the last few years there has been a dramatic increase in the number of monolingual web-sites that rely on Internet-based MT systems to disseminate their contents in a variety of languages, which seems to be one of the most interesting areas in the current use of MT technology. This paper is based on preliminary observations of these recent developments and reports on how on-line MT services are actually integrated into a sample of monolingual web-sites only available in English, attempting to evaluate the success of the strategies used to incorporate web-based MT technology for dissemination purposes. The discussion suggests in conclusion that the overall lack of a user-oriented approach and the limited consideration of issues of user-friendliness make the integration of on-line MT into the sample of monolingual web-sites largely ineffective.
This paper considers the popular but ques- tionable technique of 'round-trip transla- tion' (RTT) as a means of evaluating free on-line Machine Translation systems. Two experiments are reported, both relat- ing to common requirements of lay-users of MT on the web. In the first we see whether RTT can accurately predict the overall quality of the MT system. In the second, we ask whether RTT can predict the translatability of a given text. In both cases, we find RTT to be a poor predictor of quality, with high BLEU and F-scores for RTTs when the forward translation was poor. We discuss why this is the case, and conclude that, even if it seemed obvi- ous that RTT was good for nothing, at least we now have some tangible evi- dence.
The notion of a translator's workstation has been widely discussed at various points in the history of translation and computers, and a number of tools and language resources have been proposed for inclusion in it, ranging from general purpose text-editing facilities, to tools designed specifically for translators, such as translation memory and terminology management software. This paper reports on the progress of a project that has been initiated to investigate which of the many available tools and language resources translators today are actually incorporating into their workstations, and which they deem to be useful in supporting their work. Specifically, in this paper, the findings of a survey of UK translators are presented, focussing specifically on the levels of uptake of a wide range of tools and language resources. To date, some 400 responses to this survey have been received, logged and analysed.
A status report on a controversial and much-maligned area of technological research and development. Machine Translation does not often make the headlines, though it can happen. In the non-English-speaking world, at least, the last time the spotlight briefly fell on this normally unexciting activity was in September 1998, at the height of the Clinton-Lewinsky scandal. The report on the case by the Independent Council, Kenneth Starr, containing an account of the President's intimate dealings with Miss Lewinsky, was placed on the Internet and to satisfy the morbid curiosity of surfers everywhere was immediately translated into the world's major languages using the free translation software available on a number of WWW sites. The results, not surprisingly, were laughable. A capable human being would have had a hard enough time translating such a potent combination of technical and colloquial English: the MT applications were quite out of their depth. The silliest gaffes appeared in newspapers and everyone agreed that the on-line MT programs were useless; the subject was soon forgotten.
An exploratory study has been initiated to determine the uptake of machine translation (MT) among UK freelance translators, to assess their perceptions and experience of it, and to identify their perceived MT training needs. A preliminary analysis of findings suggests that, although the uptake of MT has, to date, been rather low in the freelance translation community, there was among the respondents a keen interest, to learn more about MT and to explore the possibility of using it more in their translation assignments. The respondents expressed a desire for MT training, with a particular preference for self-directed training material allowing experimentation with MT on 'real' assignments for their clients.
This paper describes how information technology is used by the Translation Department of PricewaterhouseCoopers in Madrid to optimise translation processes. It commences by describing a mechanism for handling workflow via the corporate network, designed to maximise speed and efficiency in translation requests and also to function as an automated record for administration purposes. This is followed by an appraisal of the CAT system used in the Translation Department, namely the Trados Workbench and related applications. Finally, an ongoing project for making MT (Systran) available to PwC employees around the world over the Firm's intranet is outlined.
Freelance translators today have at their disposal an ever-increasing array of online tools and language resources, including search engines, electronic mail, dictionaries, document archives, and terminology databases. A survey of working practices has been conducted among UK translators, with responses received to date from around 400 freelancers. Included in the survey was an investigation of the uptake of online services by translators, the findings of which are reported in this paper. These findings indicate that freelance translators have adopted, and are making extensive use of, both general online tools, and more specialised online terminology resources. Less use is being made, however, of mailing lists, document archives, and online machine translation. The paper concludes with a discussion of the findings, together with a number of research questions arising from those findings.