TT-Assig #4 - see attached
31
THE DEVELOPMENT OF
TRANSLATION TECHNOLOGY
1967–2013
Chan Sin-wai
the chinese university \ff h\fng k\fng, h\fng k\fng, china
Introduction
The history of translation technology, or more specifically computer-aided translation, is short,
but its development is fast. It is generally recognized that the failure of machine translation in
the 1960s led to the emergence of computer-aided translation. The development of computer-
aided translation from its beginning in 1967 as a result of the infamous ALPAC report (1966)
to 2013, totalling 46 years, can be divided into four periods. The first period, which goes from
1967 to 1983, is a period of germination. The second period, covering the years between 1984
and 1993, is a period of steady growth. The third period, which is from 1993 to 2003, is a
decade of rapid growth. The last period, which includes the years from 2004 to 2013, is a
period of global development.
1967–1983: A period of germination
Computer-aided translation, as mentioned above, came from machine translation, while
machine translation resulted from the invention of computers. Machine translation had made
considerable progress in a number of countries from the time the first computer, ENIAC, was
invented in 1946. Several events before the ALPAC report in 1966 are worth noting. In 1947,
one year after the invention of the computer, Warren Weaver, President of the Rockefeller
Foundation and Andrew D. Booth of Birkbeck College, London University, were the first
two scholars who proposed to make use of the newly invented computer to translate natural
languages (Chan 2004: 290−291). In 1949, Warren Weaver wrote a memorandum for peer
review outlining the prospects of machine translation, known in history as ‘Weaver’s
Memorandum’. In 1952, Yehoshua Bar-Hillel held the first conference on machine translation
at the Massachusetts Institute of Technology, and some of the papers were compiled by
William N. Locke and Andrew D. Booth into an anthology entitled Machine Translation of
Languages: Fourteen Essays, the first book on machine translation (Locke and Booth 1955). In
1954, Leon Dostert of Georgetown University and Peter Sheridan of IBM used the IBM701
machine to make a public demonstration of the translation of Russian sentences into English,
which marked a milestone in machine translation (Hutchins 1999: 1−16; Chan 2004: 4
125−226). In the same year, the inaugural issue of Mechanical Translation, the first journal in the
field of machine translation, was published by the Massachusetts Institute of Technology
(Yngve 2000: 50−51). In 1962, the Association for Computational Linguistics was founded in
the United States, and the journal of the association, Computational Linguistics, was also
published. It was roughly estimated that by 1965, there were eighteen countries or research
institutions engaged in the studies on machine translation, including the United States, former
Soviet Union, the United Kingdom, Japan, France, West Germany, Italy, former
Czechoslovakia, former Yugoslavia, East Germany, Mexico, Hungary, Canada, Holland,
Romania, and Belgium (Zhang 2006: 30−34).The development of machine translation in the United States since the late 1940s, however,
fell short of expectations. In 1963, the Georgetown machine translation project was terminated,
which signifies the end of the largest machine translation project in the United States (Chan
2004: 303). In 1964, the government of the United States set up the Automatic Language
Processing Advisory Committee (ALPAC) comprising seven experts to enquire into the state
of machine translation (ALPAC 1966; Warwick 1987: 22–37). In 1966, the report of the
Committee, entitled Languages and Machines: Computers in Translation and Linguistics, pointed
out that ‘there is no immediate or predictable prospect of useful machine translation’ (ALPAC
1966: 32). As machine translation was twice as expensive as human translation, it was unable
to meet people’s expectations, and the Committee recommended that resources to support
machine translation be terminated. Its report also mentioned that ‘as it becomes increasingly
evident that fully automatic high-quality machine translation was not going to be realized for
a long time, interest began to be shown in machine-aided translation’ (ibid.: 25). It added that
machine translation should shift to machine-aided translation, which was ‘aimed at improved
human translation, with an appropriate use of machine aids’ (ibid.: iii), and that ‘machine-aided
translation may be an important avenue toward better, quicker, and cheaper translation’ (ibid.:
32). The ALPAC report dealt a serious blow to machine translation in the United States,
which was to remain stagnant for more than a decade, and it also made a negative impact on
the research on machine translation in Europe and Russia. But this gave an opportunity to
machine-aided translation to come into being. All these events show that the birth of machine-
aided translation is closely related to the development of machine translation. Computer-aided translation, nevertheless, would not be possible without the support of
related concepts and software. It was no mere coincidence that translation memory, which is
one of the major concepts and functions of computer-aided translation, came out during this
period. According to W. John Hutchins, the concept of translation memory can be traced to
the period from the 1960s to the 1980s (Hutchins 1998: 287−307). In 1978, when Alan Melby
of the Translation Research Group of Brigham Young University conducted research on
machine translation and developed an interactive translation system, ALPS (Automated
Language Processing Systems), he incorporated the idea of translation memory into a tool
called ‘Repetitions Processing’, which aimed at finding matched strings (Melby 1978; Melby
and Warner 1995: 187). In the following year, Peter Arthern, in his paper on the issue of
whether machine translation should be used in a conference organized by the European
Commission, proposed the method of ‘translation by text-retrieval’ (Arthern 1979: 93).
According to Arthern,
This information would have to be stored in such a way that any given portion of
text in any of the languages involved can be located immediately … together with its
translation into any or all of the other languages which the organization employs. (Arthern 1979: 95)
S. Chan Development of translation technology5
In October 1980, Martin Kay published an article, entitled ‘The Proper Place of Men and
Machines in Language Translation’, at the Palo Alto Research Center of Xerox. He proposed
to create a machine translation system in which the display on the screen is divided into two
windows. The text to be translated appears in the upper window and the translation would be
composed in the bottom one to allow the translator to edit the translation with the help of
simple facilities peculiar to translation, such as aids for word selection and dictionary
consultation, which are labelled by Kay as a translator amanuensis (Kay 1980: 9−18). In view of
the level of word-processing capacities at that time, his proposal was inspiring to the
development of computer-aided translation and exerted a huge impact on its research later on.
Kay is generally considered a forerunner in proposing an interactive translation system. It can be seen that the idea of translation memory was established in the late 1970s and the
1980s. Hutchins believed that the first person to propose the concept of translation memory
was Arthern. As Melby and Arthern proposed the idea almost at the same time, both could be
considered as forerunners. And it should be acknowledged that Arthern, Melby, and Kay made
a great contribution to the growth of computer-aided translation in its early days. The first attempt to deploy the idea of translation memory in a machine translation system
was made by Alan Melby and his co-researchers at Brigham Young University, who jointly
developed the Automated Language Processing Systems, or ALPS for short. This system
provided access to previously translated segments which were identical (Hutchins 1998: 291).
Some scholars classify this type of full match a function of the first generation translation
memory systems (Gotti et al. 2005: 26−30; Kavak 2009; Elita and Gavrila 2006: 24−26). One
of the major shortcomings of this generation of computer-aided translation systems is that
sentences with full matching were very small in number, minimizing the reusability of
translation memory and the role of translation memory database (Wang 2011: 141). Some researchers around 1980 began to collect and store translation samples with the
intention of redeploying and sharing their translation resources. Constrained by the limitations
of computer hardware (such as its limited storage space), the cost of building a bilingual database
was high, and with the immaturity in the algorithms for bilingual data alignment, translation
memory technology had been in a stage of exploration. As a result, a truly commercial
computer-aided translation system did not emerge during the sixteen years of this period and
translation technology failed to make an impact on translation practice and the translation
industry.
1984–1992: A period of steady growth
The eight years between 1984 and 1992 are a period of steady growth for computer-aided
translation and for some developments to take place. Corporate operation began in 1984,
system commercialization, in 1988, and regional expansion, in 1992.
Company operation
It was during this period that the first computer-aided translation companies, Trados in
Germany and Star Group in Switzerland, were founded in 1984. These two companies later
had a great impact on the development of computer-aided translation. The German company was founded by Jochen Hummel and Iko Knyphausen in Stuttgart,
Germany, in 1984. Trados GmbH came from TRAnslation and DOcumentation Software.
This company was set up initially as a language service provider (LSP) to work on a translation
project they received from IBM in the same year. As the company later developed computer- 6
aided translation software to help complete the project, the establishment of Trados GmbH is
regarded as the starting point of the period of steady growth in computer-aided translation
(Garcia and Stevenson 2005: 18–31; http://www.lspzone.com).Of equal significance was the founding of the Swiss company STAR AG in the same year.
STAR, an acronym of Software, Translation, Artwork, and Recording, provided manual
technical editing and translation with information technology and automation. Two years
later, STAR opened its first foreign office in Germany in order to serve the increasingly
important software localization market and later developed STAR software products, GRIPS
and Transit for information management and translation memory respectfully. At the same
time, client demand and growing export markets led to the establishment of additional overseas
locations in Japan and China. The STAR Group still plays an important role in the translation
technology industry (http://www.star-group.net). It can be observed that during this early period of computer-aided translation, all companies
in the field were either established or operated in Europe. This Eurocentric phenomenon was
going to change in the next period.
System commercialization
The commercialization of computer-aided translation systems began in 1988, when Eiichiro
Sumita and Yutaka Tsutsumi of the Japanese branch of IBM released the ETOC (Easy to
Consult) tool, which was actually an upgraded electronic dictionary. Consultation of a
traditional electronic dictionary was by individual words. It could not search phrases or
sentences with more than two words. ETOC offered a flexible solution. When inputting a
sentence to be searched into ETOC, the system would try to extract it from its dictionary. If
no matches were found, the system would make a grammatical analysis of the sentence, taking
away some substantive words but keeping the form words and adjectives which formed the
sentence pattern. The sentence pattern would be compared with bilingual sentences in the
dictionary database to find sentences with a similar pattern, which would be displayed for the
translator to select. The translator could then copy and paste the sentence onto the Editor and
revise the sentence to complete the translation. Though the system did not use the term
translation memory and the translation database was still called a ‘dictionary’, it nevertheless
had essentially the basic features of translation memory of today. The main shortcoming of this
system is that as it needs to make grammatical analyses, its programming would be difficult and
its scalability would be limited. If a new language were to be added, a grammatical analysis
module would have to be programmed for the language. Furthermore, as the system could
only work on perfect matching but not fuzzy matching, it drastically cut down on the reusability
of translations (Sumita and Tsutsumi 1988: 2). In 1988, Trados developed TED, a plug-in for text processor tool that was later to become,
in expanded form, the first Translator’s Workbench editor, developed by two people and their
secretary (Garcia and Stevenson 2005: 18–31). It was around this time that Trados made the
decision to split the company, passing the translation services part of the business to INK in the
Netherlands, so that they could concentrate on developing translation software (http://www.
translationzone.com). Two years later, the company also released the first version of MultiTerm as a memory-
resident multilingual terminology management tool for DOS, taking the innovative approach
of storing all data in a single, freely structured database with entries classified by user-defined
attributes (Eurolux Computers 1992: 8; http://www.translationzone.com; Wassmer 2011).
S. Chan Development of translation technology7
In 1991 STAR AG also released worldwide the Transit 1.0 (‘Transit’ was derived from the
phrase ‘translate it’) 32-bit DOS version, which had been under development since 1987 and
used exclusively for in-house production. Transit featured the modules that are standard
features of today’s CAT systems, such as a proprietary translation editor with separate but
synchronized windows for source and target language and tag protection, a translation memory
engine, a terminology management component and project management features. In the
context of system development, the ideas of terminology management and project management
began with Transit 1.0. Additional products were later developed for the implementation and
automation of corporate product communications: TermStar, WebTerm, GRIPS,
MindReader, SPIDER and STAR James (http://www.star-group.net). One of the most important events in this period is obviously the release of the first
commercial system, Trados, in 1992, which marks the beginning of commercial computer-
aided translation systems.
Regional expansion
The year 1992 also marks the beginning of the regional expansion of computer-aided
translation. This year witnessed some significant advances in translation software made in
different countries. First, in Germany, Translator’s Workbench I and Translator’s Workbench
II (DOS version of Trados) were launched within the year, with Workbench II being a
standalone package with an integrated editor. Translator’s Workbench II comprises the TW II
Editor (formally TED) and MultiTerm 2. Translator’s Workbench II was the first system to
incorporate a ‘translation memory’ and alignment facilities into its workstation. Also of
considerable significance was the creation by Matthias Heyn of Trados’s T Align, later known
as WinAlign, the first alignment tool on the market. In addition, Trados began to open a
network of global offices, including Brussels, Virginia, the United Kingdom and Switzerland
(Brace 1994; Eurolux Computers 1992; http://www.translationzone.com; Hutchins 1998:
287–307). Second, in the United States, IBM launched its IBM Translation Manager / 2 (TM/2), with
an Operating System/2 (OS/2) package that integrated a variety of translation aids within a
Presentation Manager interface. TM/2 had its own editor and a translation memory feature
which used fuzzy search algorithms to retrieve existing material from its translation database.
TM/2 could analyse texts to extract terms. TM/2 came with lemmatizers, spelling lists, and
other linguistic resources for nineteen languages, including Catalan, Flemish, Norwegian,
Portuguese, Greek, and Icelandic. External dictionaries could also be integrated into TM/2,
provided they were formatted in Standard Generalized Markup Language (SGML). TM/2
could be linked to logic-based machine translation (Brace 1992a). This system is perhaps the
first hybrid computer-aided translation system that was integrated with a machine translation
system (Brace 1993; Wassmer 2011). Third, in Russia, the PROMT Ltd was founded by two doctorates in computational
linguistics, Svetlana Sokolova and Alexander Serebryakov, in St. Petersburg in 1991. At the
beginning, the company mainly developed machine translation (MT) technology, which has
been at the heart of the @promt products. Later, it began to provide a full range of translation
solutions: machine translation systems and services, dictionaries, translation memory systems,
data mining systems (http://www.promt.com). Fourth, in the United Kingdom, two companies specializing in translation software
production were founded. First, Mark Lancaster established the SDL International, which
served as a service provider for the globalization of software (http://www.sdl.com). Second, 8
ATA Software Technology Ltd, a London-based software house specializing in Arabic
translation software, was established in 1992 by some programmers and Arabic software
specialists. The company later developed a series of machine translation products (Arabic and
English) and MT and TM hybrid system, Xpro7 and online translation engine (http://www.
atasoft.com).
1993–2003: A period of rapid growth
This period, covering the years from 1993 to 2003, is a period of rapid growth, due largely to
(1) the emergence of more commercial systems; (2) the development of more built-in functions;
(3) the dominance of Windows operation systems; (4) the support of more document formats;
(5) the support of more languages for translation; and (6) the dominance of Trados as a market
leader.
(1) The emergence of more commercial systems
Before 1993, there were only three systems available on the market, including Translator’s
Workbench II of Trados, IBM Translation Manager / 2, and STAR Transit 1.0. During this
ten-year period between 1993 and 2003, about twenty systems were developed for sale,
including the following better-known systems such as Déjà Vu, Eurolang Optimizer (Brace
1994), Wordfisher, SDLX, ForeignDesk, Trans Suite 2000, Yaxin CAT, Wordfast, Across,
OmegaT, MultiTrans, Huajian, Heartsome, and Transwhiz. This means that there was a six-
fold increase in commercial computer-aided translation systems during this period. Déjà Vu is the name of a computer-aided translation system developed by Atril in Spain after
1993. A preliminary version of Déjà Vu, a customizable computer-aided translation system that
combined translation memory technology with example-based machine translation techniques,
was initially developed by ATRIL in June to fulfil their own need for a professional translation
tool. At first, they worked with machine translation systems, but the experiments with machine
translation were extremely disappointing, and subsequent experiences with translation memory
tools exposed two main shortcomings: all systems ran under MS-DOS and were capable of
processing only plain text files. Then, ATRIL began considering the idea of writing its own
translation memory software. Déjà Vu 1.0 was released to the public in November 1993. It was with an interface for
Microsoft Word for Windows 2.0, which was defined as the first of its kind. Version 1.1
followed soon afterwards, incorporating several performance improvements and an integrated
alignment tool (at a time when alignment tools were sold as expensive individual products),
and setting a new standard for the translation tool market (http://www.atril.com). Déjà Vu, designed to be a professional translation tool, produced acceptable results at an
affordable price. Déjà Vu was a first in many areas: the first TM tool for Windows; the first
TM tool to directly integrate into Microsoft Word; the first 32-bit TM tool (Déjà Vu version
2.0); and the first affordable professional translation tool. In the following year, Eurolang Optimizer, a computer-aided translation system, was
developed by Eurolang in France. Its components included the translator’s workstation, pre-
translation server with translation memory and terminology database, and project management
tool for multiple languages and users (Brace 1994). In Germany, Trados GmbH announced the release of the new Windows version of
Translator’s Workbench, which could be used with standard Windows word processing
packages via the Windows DDE interface (Brace 1994). In June 1994 Trados released
S. Chan Development of translation technology9
MultiTerm Professional 1.5 which was included in Translator’s Workbench, which had fuzzy
search to deliver successful searches even when words were incorrectly spelt, a dictionary-style
interface, faster searches through use of new highly compressed data algorithms, drag and drop
content into word processor and integrated programming language to create powerful layouts
(http://www.translationzone.com). In Hungary, Tibor Környei developed the WordFisher for Microsoft Word macro set. The
programme was written in the WordBasic language. For translators, it resembled a translation
memory programme, but provided a simpler interface in Word (Környei 2000). In 1995, Nero AG was founded in Germany as a manufacturer of CD and DVD application
software. Later, the company set up Across Systems GmbH as a division, which developed and
marketed a tool of the same name for corporate translation management (CTM) that supported
the project and workflow manage
ment of translations (Schmidt 2006; German 2009: 9–10).
During the first half of 1996, when Windows 95 was in its final stages of beta testing, Atril
Development S.L. in Spain began writing a new version of Déjà Vu − not just porting the
original code to 32 bits, but adding a large number of important functionalities that had been
suggested by the users. In October, Atril released Déjà Vu beta v2.0. It consisted of the universal
editor, Déjà Vu Interactive (DVI), the Database Maintenance module with an alignment tool,
and a full-featured Terminology Maintenance module (Wassmer 2007: 37–38). In the same year, Déjà Vu again was the first TM tool available for 32-bit Windows and
shipped with a number of filters for DTP packages − including FrameMaker, Interleaf, and
QuarkXPress − and provided extensive project management facilities to enable project
managers to handle large, multi-file, multilingual projects. In 1997, developments in France and Germany deserve mentioning. In France, CIMOS
released Arabic to English translation software An-Nakel El-Arabi, with features like machine
translation, customized dictionary and translation memory. Because of its deep sentence
analysis and semantic connections, An-Nakel Al-Arabi could learn new rules and knowledge.
CIMOS had previously released English to Arabic translation software (MultiLingual 1997). In
Germany, Trados GmbH released WinAlign as a visual text alignment tool as the first fully-
fledged 32-bit application in Trados. Mircosoft decided to base its internal localization memory
store on Trados and consequently acquired a share of 20 per cent in Trados (http://www.
translationzone.com). The year 1998 marks a milestone in the development of translation technology in China and
Taiwan. In Beijing, Beijing Yaxincheng Software Technology Co. Ltd. 北京雅信誠公司 was
set up as a developer of translation software. It was the first computer-aided translation software
company in China. In Taipei, the Inventec Corporation released Dr Eye 98 (譯典通) with
instant machine translation, dictionaries and termbases in Chinese and English (http://www.
dreye.com.tw). In the same year, the activities of SDL and International Communications deserve special
mention. In the United Kingdom, SDL began to acquire and develop translation and
localization software and hardware − both for its own use in client-specific solutions, and to
be sold as free-standing commercial products. At the end of the year, SDL also released SDLX,
a suite of translation memory database tools. SDLX was developed and used in-house at SDL,
and therefore was a mature product at its first offering (Hall 2000; MultiLingual 1998). Another
British company, International Communications, a provider of localization, translation and
multilingual communications services, released ForeignDesk v5.0 with the full support of
Trados Translator’s Workbench 2.0 and WinAlign, S-Tagger. Then, Lionbridge Technologies
Inc. acquired it (known as Massachusetts-based INT’L.com at the transaction) and later in
November 2001 decided to open-source the ForeignDesk suite free of charge under BSD 10
licence. ForeignDesk was originally developed by International Communications around 1995
(MultiLingual 2000).In June 1999, Beijing YaxinCheng Software Technology Co. Ltd. established Shida CAT
Research Centre (實達 CAT 研究中心), which later developed Yaxin CAT Bidirectional
v2.5 (Chan 2004: 338). In June, SJTU Sunway Software Industry Ltd. acquired one of the
most famous CAT products in China at the moment − Yaxin CAT from Beijing YaxinCheng
Software Technology Co. Ltd., and it released the Yaxin CAT v1.0 in August. The release of
this software signified, in a small way, that the development of computer-aided systems was no
longer a European monopoly. In France, the first version of Wordfast PlusTools suite of CAT (Computer-Assisted
Translation) tools was developed. One of the developers was Yves A. Champollion, who
incorporated Wordfast LLC later. There were only a few TM software packages available in
the first version. It could be downloaded freely before 2002, although registration was required
(http://www.wordfast.net/champollion.net). In the United States, MultiCorpora R&D Inc. was incorporated, which was exclusively
dedicated to providing language technology solutions to enterprises, governments, and
language service providers (http://www.multicorpora.com). In the United Kingdom, following the launch of SDL International’s translation database
tool, SDLX, SDL announced SDL Workbench. Packaged with SDLX, SDL Workbench
memorized a user’s translations and automatically offered other possible translations and
terminology from a user’s translation database within the Microsoft Word environment. In line
with its ‘open’ design, it was able to work with a variety of file formats, including Trados and
pre-translated RTF files (MultiLingual 1999). The year 2000 was a year of activities in the industry. In China, Yaxin CAT v2.5 Bidirectional
(English and Chinese) was released with new features like seventy-four topic-specific lexicons
with six million terms free of charge, project analysis, project management, share translation
memory online and simultaneous editing of machine output (Chen 2001). In Germany, OmegaT, a free (GPL) translation memory tool, was publicly released. The
key features of OmegaT were basic (the functionality was very limited), free, open-source,
cross-operation systems as it was programmed in Java (http://www.omegat.org; Prior 2003). In Ireland, Alchemy Software Development Limited announced the acquisition of Corel
CATALYST™, which was designed to boost the efficiency and quality of globalizing software
products and was used by over 200 software development and globalization companies
worldwide (http://www.alchemysoftware.ie). In the United Kingdom, SDL International announced in April the release of SDLX 2.0,
which was a new and improved version of SDLX 1.03 (http://www.sdl.com). It also released
SDL Webflow for managing multilingual website content (http://www.sdlintl.com). In Germany, Trados relocated its headquarters to the United States in March and became a
Delaware corporation. In France, Wordfast v3.0 was released in September. The on-the-fly tagging and un-tagging
of HTML (HyperText Markup Language) files was a major breakthrough in the industry.
Freelance translators could translate HTML pages without worrying about the technical hurdles. Not much happened in 2001. In Taiwan, Inventec Corporation released Dr Eye 2001, with
new functions like online search engine, full-text machine translation from English to Chinese,
machine translation from Japanese to Chinese and localization plug-in (Xu 2001). In the
United Kingdom, SDL International released SDLX 4.0 with real-time translation, a flexible
software licence and enhanced capabilities. In the United States, Trados announced the launch
of Trados 5 in two versions, Freelance and Team (http://www.translationzone.com).
S. Chan Development of translation technology11
In contrast, the year 2002 was full of activities in the industry.
In North America, MultiCorpora R&D Inc. in Canada released MultiTrans 3, providing
corpus-based translation support and language management solution. It also introduced a new
translation technology called Advanced Leveraging Translation Memory (ALTM). This model
provided past translations in their original context and required virtually no alignment
maintenance to obtain superior alignment results. In the United States, Trados 5.5 (Trados
Corporate Translation Solution™) was released. MultiCorpora released MultiTrans 3.0, which
introduced an optional client-server add-on, so it could be used in a web-based, multi-user
environment or as a standalone workstation. Version 3 supported TMX and was also fully
Unicode compliant (Locke and Giguère 2002: 51). In Europe and the United Kingdom, SDL International released its new SDLX Translation
Suite 4, and then later that year released the elite version of the suite. The SDLX Translation
Suite features a modular architecture consisting of five to eight components: SDL Project
Wizard, SDL Align, SDL Maintain, SDL Edit and SDL TermBase in all versions, and SDL
Analyse, SDL Apply and SDLX AutoTrans in the Professional and Elite versions (Wassmer
2003). In Germany, MetaTexis Software and Services released in April the first official version
1.00 of MetaTexis (http://www.metatexis.com). In Asia, Huajian Corporation in China released Huajian IAT, a computer-aided translation
system (http://www.hjtek.com). In Taiwan, Otek launched in July Transwhiz Power version
(client/server structure), which aimed at enterprise customers (http://www.otek.com.tw). In
Singapore, Heartsome Holdings Pte. Ltd. was founded to develop language translation
technology (Garcia and Stevenson 2006: 77). North America and Europe were active in translation technology in 2003.
In 2003, MultiCorpora R&D Inc. in Canada released MultiTrans 3.5 which had new and
improved capabilities, including increased processing speed of automated searches, increased
network communications speed, improved automatic text alignment for all languages, and
optional corpus-based pre-translation. Version 3.5 also offered several new terminology
management features, such as support for additional data types, additional filters, batch updates
and added import and export flexibility, as well as full Microsoft Office 2003 compatibility,
enhanced Web security and document analysis capabilities for a wider variety of document
formats (MultiLingual 2003). In the United States, Trados 6 was launched in April and Trados
6.5 was launched in October with new features like auto concordance search, Word 2003
support and access to internet TM server (Wassmer 2004: 61). In Germany, MetaTexis version 2.0 was released in October with a new database engine.
And MetaTexis version ‘Net/Office’ was released with new features that supported Microsoft
PowerPoint and Excel files, Trados Workbench, and could be connected with Logoport
servers (http://www.metatexis.com). In Russia, PROMT, a developer of machine translation products and services, released a
new version @promt XT with new functions like processing PDF file formats, which made
PROMT the first among translation software that supported PDF. Also, one of the editions,
@promt Expert integrated translation memory solutions (Trados) and a proprietary terminology
extraction system (http://www.promt.com). In France, Atril, which was originally founded in Spain but which relocated its group
business to France in the late 1990s, released Déjà Vu X (Standard, Professional, Workgroup
and Term Sever) (Harmsen 2008). Wordfast 4, which could import and translate PDF contents,
was also released (http://www.wordfast.net). Some developers of machine translation systems also launched new versions with a translation
memory component, such as LogoVista, An-Nabel El-Arabi and PROMT (http://www. 12
promt.com). Each of these systems was created with distinct philosophies in its design, offering
its own solutions to problems and issues in the work of translation. This was aptly pointed out
by Brace (1994):Eurolang Optimizer is based on an ambitious client / server architecture designed
primarily for the management of large translation jobs. Trados Workbench, on the
other hand, offers more refined linguistic analysis and has been carefully engineered
to increase the productivity of single translators and small workgroups.
(2) The development of more built-in functions
Computer-aided translation systems of the first and second periods were usually equipped with
basic components, such as translation memory, terminology management, and translation
editor. In this period, more functions were developed and more components were gradually
integrated into computer-aided translation systems. Of all the new functions developed, tools
for alignment, machine translation, and project management were most significant. Trados
Translator’s Workbench II, for example, incorporated T Align, later known as WinAlign, into
its workstation, followed by other systems such as Déjà Vu, SDLX, Wordfisher, and MultiTrans.
Machine translation was also integrated into computer-aided translation systems to handle
segments not found in translation memories. IBM’s Translation Manager, for example,
introduced its Logic-Based Machine Translation (LMT) to run on IBM mainframes and
RS/6000 Unix systems (Brace 1993). The function of project management was also introduced
by Eurolang Optimizer in 1994 to better manage translation memory and terminology
databases for multiple languages and users (Brace 1992a).
(3) The dominance of Windows Operating System
Computer-aided translation systems created before 1993 were run either in the DOS system
or OS/2 system. In 1993, the Windows versions of these systems were first introduced and
they later became the dominant stream. For example, IBM and Trados GmbH released a
Windows version of TM/2 and of Translator’s Workbench respectively in mid-1993. More
Windows versions came onto the market, such as the preliminary version of ATRIL’s Déjà Vu
1.0 in June in Spain. Other newly released systems running on Windows include SDLX,
ForeignDesk, Trans Suite 2000, Yaxin CAT, Across, MultiTrans, Huajian, and TransWhiz.
(4) The support of more document formats
Computer-aided translation systems of this period could handle more document formats
directly or with filters, including Adobe InDesign, FrameMaker, HTML, Microsoft
PowerPoint, Excel, Word, QuarkXPress, even PDF by 2003. Trados 6.5, for example,
supported all the widely used file formats in the translation community, which allowed
translators and translation companies to translate documents in Microsoft Office 2003 Word,
Excel and PowerPoint, Adobe InDesign 2.0, FrameMaker 7.0, QuarkXPress 5, and PageMaker.
(5) The support of translation of more languages
Translation memory is supposed to be language-independent, but computer-aided translation
systems developed in the early 1990s did not support all languages. In 1992, Translator
S. Chan Development of translation technology13
Workbench Editor, for example, supported only five European languages, namely, German,
English, French, Italian and Spanish, while IBM Translation Manager / 2 supported 19
languages, including Chinese, Korean and other OS/2 compatible character code sets. This
was due largely to the contribution of Unicode, which provided the basis for the processing,
storage, and interchange of text data in any language in all modern software, thereby allowing
developers of computer-aided translation systems to gradually resolve obstacles in language
processing, especially after the release of Microsoft Office 2000. Systems with Unicode support
mushroomed, including Transit 3.0 in 1999, MultiTerm and WordFisher 4.2.0 in 2000,
Wordfast Classic 3.34 in 2001, and Tr-AID 2.0 and MultiTrans 3 in 2002.
(6) The dominance of Trados as a market leader
As a forerunner in the field, Trados became a market leader in this period. As observed by
Colin Brace, ‘Trados has built up a solid technological base and a good market position’ in its
first decade. By 1994, the company had a range of translation software, including Trados
Translator’s Workbench (Windows and DOS versions), MultiTerm Pro, MultiTerm Lite, and
MultiTerm Dictionary. Its technology in translation memory and file format was then widely
used in other computer-aided translation systems and its products were most popular in the
industry. From the late 1990s, a few systems began to integrate Trados’s translation memory
into their systems. In 1997, ProMemoria, for example, was launched with its translation
memory component provided by Trados. In 1998, International Communications released
ForeignDesk 5.0 with the full support of Trados Translator’s Workbench 2.0, WinAlign, and
S-Tagger. In 1999, SDLX supported import and export formats such as Trados and tab-
delimited and CSV files. In 2000, Trans Suite 2000 was released with the capacity to process
Trados RTF file. In 2001, Wordfast 3.22 could directly open Trados TMW translation
memories (Translator’s Workbench versions 2 and 3). In 2003, PROMT XT Export integrated
Trados’s translation memory. In October 2003, MetaTexis ‘Net/Office’ 2.0 was released and
was able to work with Trados Workbench.
2004–2013: A period of global development
Advances in technology have given added capabilities to computer-aided translation systems.
During the last nine years, while most old systems have been upgraded on a regular basis, close
to thirty new systems have been released to the market. This situation has offered a wider range
of choices for buyers to acquire systems with different packages, functions, operation systems,
and prices. One of the most significant changes in this period is the addition of new computer-aided
translation companies in countries other than those mentioned above. Hungary is a typical
example. In 2004, Kilgray Translation Technologies was established by three Hungarian
language technologists. The name of the company was made up of the founders’ surnames: Kis
Balázs (KI), Lengyel István (L), and Ugray Gábor (GRAY). Later, the company launched the
first version of MemoQ, an integrated Localization Environment (ILE), in 2005. MemoQ’s first
version had a server component that enabled the creation of server projects. Products of Kilgray
included MemoQ, MemoQ server, QTerm, and TM Repository (http://www.kilgray.com). Another example is Japan. In Japan, Rozetta Corporation released TraTool, a computer-
aided translation system with translation memory, an integrated alignment tool, an integrated
terminology tool and a user dictionary. The product is still commercially available but no
major improvement has been made since its first version (http://www.tratool.com). 14
Yet another example is Poland, where AidTrans Soft launched its AidTrans Studio 1.00, a
translation memory tool. But the company was discontinued in 2010 (http://www.
thelanguagedirectory.com/translation/translation_software). New versions of computer-aided translation systems with new features are worth noting. In
the United Kingdom, ATA launched a new Arabic Memory Translation system, Xpro7 which
had the function of machine translation (http://www.atasoft.com). SDL Desktop Products, a
division of SDL International, announced the launch of SDLX 2004. Its new features included
TMX Certification, seamlessly integrating with Enterprise systems such as online terminology
and multilingual workflow management, adaptation of new file formats, synchronized web-
enabled TM, and Knowledge-based Translation (http://www.sdl.com). In the United States,
Systran released Systran Professional Premium 5.0, which contained integrated tools such as
integrated translation memory with TMX support, a Translator’s Workbench for post-editing
and ongoing quality analysis (http://www.systransoft.com). Multilizer Inc., a developer of
globalization technologies in the United States, released a new version of Multilizer, which
included multi-user translation memory with Translation Memory Manager (TMM), a
standalone tool for maintaining Multilizer Translation Memory contents. TMM allowed
editing, adding and deleting translations, and also included a briefcase model for working with
translations off-line (http://www.multilizer.com). In Ukraine, Advanced International Translations (AIT) started work on user-friendly
translation memory software, later known as AnyMen, which was released in December 2008. In 2005, translation technology moved further ahead with new versions and new functions.
In North America, MultiCorpora in Canada released MultiTrans 4, which built on the
foundation of MultiTrans 3.7 and had a new alignment process that was completely automated
(MultiLingual 2005d). Trados, incorporated in the United States, produced Trados 7 Freelance,
which supported twenty additional languages, including Hindi. At an operating system level,
Microsoft Windows 2000, Windows XP Home, Windows XP Professional, and Windows
2003 Server were supported. More file formats were now directly supported by TagEditor.
MultiCorpora also introduced MultiTrans 4, which was designed to meet the needs of large
organizations by providing the newest efficiencies for translators in the areas of text alignment
quality, user-friendliness, flexibility and web access (http://www.multicorpora.com). In Europe, Lingua et Machina, a memory translation tool developer, released SIMILIS v1.4,
its second-generation translation tool. SIMILIS uses linguistic parsers in conjunction with the
translation memory paradigm. This function allowed for the automatic extraction of bilingual
terminology from translated documents. Version 1.4 brought compatibility with the Trados
translation memory format (Text and TMX) and a new language, German (MultiLingual
2005b). In Switzerland, STAR Group released Transit XV Service Pack 14. This version
extended its capabilities with a number of new features and support of 160 languages and
language versions, including Urdu (India) and Urdu (Pakistan). It supported Microsoft Word
2003 files and had MySpell dictionaries (MultiLingual 2005a). PROMT released @promt 7.0
translation software, which supported the integrated translation memory, the first of its kind
among PROMT’s products (http://www.promt.com). In the United Kingdom, SDL Desktop Products released the latest version of its translation
memory tool SDLX 2005, which expanded the Terminology QA Check and automatically
checked source and translations for inconsistent, incomplete, partial or empty translations,
corrupt characters, and consistent regular expressions, punctuation, and formatting. Language
support had been added for Maltese, Armenian and Georgian, and the system could handle
more than 150 languages (MultiLingual 2005c). In June, SDL International acquired Trados
for £35 million. The acquisition provided extensive end-to-end technology and service
S. Chan Development of translation technology15
solutions for global information assets (http://www.translationzone.com). In October, SDL
Synergy was released as a new project management tool on the market.In Asia, Huajian Corporation in China released in June Huajian Multilingual IAT network
version (華建多語 IAT 網絡版) and in October Huajian IAT (Russian to Chinese) standalone
version (http://www.hjtrans.com). In July, Beijing Orient Yaxin Software Technology Co.
Ltd. released Yaxin CAT 2.0, which was a suite including Yaxin CAT 3.5, CAM 3.5, Server,
Lexicons, Translation Memory Maintenance and Example Base. In Singapore, Heartsome
Holdings Pte. Ltd. released Heartsome Translation Suite, which was composed of three
programs: an XLIFF Editor in which source files were converted to XLIFF format and
translated; a TMX Editor that dealt with TMX files; and a Dictionary Editor that dealt with
TBX files (Garcia and Stevenson 2006: 77). In Taiwan, Otek released Transwhiz 9.0 for
English, Chinese and Japanese languages (http://www.otek.com.tw). Significant advances in translation technology were made in 2006 particularly in Europe, the
United Kingdom, and the United States. In Europe, Across Systems GmbH in Germany released in September its Corporate
Translation Management 3.5, which marked the start of the worldwide rollout of Across
software (MultiLingual 2006a). In the United Kingdom, SDL International released in February
SDL Trados 2006, which integrated with Translators Workbench, TagEditor, SDLX editing
environments and SDL MultiTerm. It included new support for Quark, InDesign CS2 and
Java (http://www.sdl.com). In the United States, MultiCorpora launched TextBase TM
concept (http://www.multicorpora.com). Apple Inc. released in August AppleTrans, a text
editor specially designed for translators, featuring online corpora which represented ‘translation
memory’ accessible through documents. AppleTrans helped users localize web pages (http://
developer.apple.com). Lingotek, a language search engine developer in the United States,
launched a beta version of a collaborative language translation service that enhanced a
translator’s efficiency by quickly finding meaning-based translated material for re-use.
Lingotek’s language search engine indexed linguistic knowledge from a growing repository of
multilingual content and language translations, instead of web pages. Users could then access
its database of previously translated material to find more specific combinations of words for
re-use. Such meaning-based searching maintained better style, tone, and terminology. Lingotek
ran completely within most popular web browsers, including initial support for Internet
Explorer and Firefox. Lingotek supported Word, Rich Text Format (RTF), Open Office,
HTML, XHTML and Excel formats, thereby allowing users to upload such documents directly
into Lingotek. Lingotek also supported existing translation memory files that were TMX-
compliant memories, thus allowing users to import TMX files into both private and public
indices (MultiLingual 2006b). In 2007, Wordfast 5.5 was released in France. It was an upgrade from Wordfast 4, in which
Mac support was completely overhauled. This version continued to offer translators
collaboration community via a LAN. Each Wordfast licence granted users the ability to search
Wordfast’s web-based TM and knowledge base, VLTM (http://www.wordfast.net). In
Germany, a group of independent translators and programmers under the GNU GPL licence
developed in October Anaphraseus, a computer-aided translation tool for creating, managing
and using bilingual translation memories. Originally, Anaphraseus was developed to work with
the Wordfast TM format, but it could also export and import files in TMX format (http://
anaphraseus.sourceforge.net). In Hungary, Kilgray Translation Technologies released in
January MemoQ 2.0. The main theme for the new version was networking, featuring a new
resource server. This server not only stored translation memory and term bases, but also offered
the possibility of creating server projects that allowed for the easy distribution of work among 16
several translators and ensured productivity at an early stage of the learning curve. Improvements
on the client side included support for XML and Adobe FrameMaker MIF file formats;
improvements to all other supported file formats; and support for the Segmentation Rule
eXchange standard, auto-propagation of translated segments, better navigation and over a
hundred more minor enhancements (Multilingual 2007). In Russia, MT2007, a freeware, was
developed by a professional programmer Andrew Manson. The main idea was to develop easy-
to-use software with extensive features. This software lacked many features that leading systems
had. In the United Kingdom, SDL International released in March SDL Trados 2007, which
had features such as a new concept of project delivery and supply chain, new one-central-view
dashboard for new project wizard, PerfectMatch, automated quality assurance checker and full
support for Microsoft Office 2007 and Windows Vista.In the United States, MultiCorpora’s Advanced Leveraging launched WordAlign which
boasted the ability to align text at the individual term and expression level (http://www.
multicorpora.com). MadCap Software Inc., a multi-channel content authoring company,
developed in May MadCap Lingo, an XML-based, fully-integrated Help authoring tool and
translation environment. MadCap Lingo offered an easy-to-use interface, complete Unicode
support for all left-to-right languages for assisting localization tasks. Across Systems GmbH and
MadCap Software announced a partnership to combine technical content creation with
advanced translation and localization. In June, Alchemy Software Development Ltd. and
MadCap Software, Inc. announced a joint technology partnership that combined technical
content creation with visual TM technology. In 2008, Europe again figured prominently in computer-aided translation software
production. In Germany, Across Systems GmbH released in April Across Language Server 4.0
Service Pack 1, which contained various extensions in addition to authoring, such as
FrameMaker 8 and SGML support, context matching, and improvements for web-based
translations via crossWeb (MultiLingual 2008a). It also introduced in July its new Language
Portal Solution (later known as Across Language Portal) for large-scale organizations and
multinational corporations, which allowed customers operating on an international scale to
implement Web portals for all language-related issues and for all staff levels that need to make
use of language resources. At the same time Across released the latest update to the Across
Language Server, offering many new functions for the localization of software user interfaces
(http://www.across.net). In Luxembourg, Wordbee S.A. was founded as a translation software
company focusing on web-based integrated CAT and management solutions (http://www.
wordbee.com). In Eastern Europe, Kilgray Translation Technologies in Hungary released in September
MemoQ 3.0, which included a new termbase and provided new terminology features. It
introduced full support for XLIFF as a bilingual format and offered the visual localization of
RESX files. MemoQ 3.0 was available in English, German, Japanese and Hungarian (http://
kilgray.com). In Russia, Promt released in March 8.0 version with major improvement in its
translation engine, translation memory system with TMX files import support, and dialect
support in English (UK and American), Spanish (Castilian and Latin American), Portuguese
(Portuguese and Brazilian), German (German and Swiss) and French (French, Swiss, Belgian,
Canadian) (http://www.promt.com). In Ukraine, Advanced International Translations (AIT)
released in December AnyMen, a translation memory system compatible with Microsoft
Word. In Uruguay, Maxprograms launched in April Swordfish version 1.0-0, a cross-platform
computer-aided translation tool based on the XLIFF 1.2 open standard published by OASIS
(http://www.maxprograms.com). In November, this company released Stingray version
1.0-0, a cross-platform document aligner. The translation memories in TMX, CSV or Trados
S. Chan Development of translation technology17
TXT format generated by Stingray could be used in most modern computer-aided translation
systems (http://www.maxprograms.com).In Ireland, Alchemy Software Development, a company in visual localization solutions,
released in July Alchemy PUBLISHER 2.0, which combined visual localization technology
with translation memory for documentation. It supported standard documentation formats,
such as MS Word, XML, application platforms such as Windows 16/22/64x binaries, web-
contents formats such as HTML, ASP, and all derivative content formats (http://www.
alchemysoftware.ie). In North America, JiveFusion Technologies, Inc. in Canada officially launched Fusion One
and Fusion Collaborate 3.0. The launches introduced a new method of managing translation
memories. New features included complete contextual referencing. JiveFusion also integrated
Fusion Collaborate 3.0 with TransFlow, a project and workflow management solution by
Logosoft (MultiLingual 2008b). In the United States, MadCap Software, Inc. released in
February MadCap Lingo 2.0, which included the Darwin Information Typing Architecture
standard, Microsoft Word and a range of standard text and language formats. In September, it
released MadCap Lingo 3.0, which included a new project packager function designed to
bridge the gap between authors and translators who used other translation memory system
software and a new TermBase Editor for creating databases of reusable translated terms. In Asia, Yaxin CAT 4.0 was released in China in August with some new features including
a computer-aided project platform for project management and huge databases for handling
large translation projects. In Taiwan, Otek released Transwhiz 10 for translating English,
Chinese and Japanese languages, with fuzzy search engine and Microsoft Word workstation
(http://www.otek.com.tw). The year 2009 witnessed the development of Autshumato Integrated Translation
Environment (ITE) version 1.0, a project funded by the Department of Arts and Culture of
the Republic of South Africa. It was released by The Centre for Text Technology (CTexT®)
at the Potchefstroom Campus of the North-West University and University of Pretoria after
two years of research and development. Although Autshumato ITE was specifically developed
for the eleven official South African languages, it was in essence language independent, and
could be adapted for translating between any language pair. In Europe, Wordfast released in January Wordfast Translation Studio, a bundled product
with Wordfast Classic (for Microsoft Word) and Wordfast Pro (a standalone CAT platform).
With over 15,000 licences in active use, Wordfast claimed itself the second most widely used
translation memory tool (http://www.wordfast.net). In Germany, Across Systems GmbH
released in May Across Language Server 5.0, which offered several options for process
automation as well as for workflow management and analysis. Approximately fifty connections
were available for interacting with other systems (MultiLingual 2009b). In September, STAR
Group in Switzerland released Transit
NXT (Professional, Freelance Pro, Workstation, and
Freelance). Service pack 1 for Transit NXT/TermStar NXT contained additional user interface
languages for Chinese, Spanish, Japanese, and Khmer, enhanced alignment usability, support
for QuarkXpress 7, and proofreading for internal repetitions. In the United Kingdom, SDL announced in June the launch of SDL Trados® Studio 2009
in the same month, which included the latest versions of SDL MultiTerm, SDL Passolo
Essential, SDL Trados WinAlign, and SDL Trados 2007 Suite. New features included Context
Match, AutoSuggest, QuickPlace (http://www.sdl.com). In October, SDL released its
enterprise platform SDL TM Server™ 2009, a new solution to centralize, share, and control
translation memories (http://www.sdl.com). 18
In North America, JiveFusion Technologies Inc. in Canada released in March Fusion 3.1 to
enhance current TMX compatibility and the capability to import and export to TMX while
preserving the complete segment context (MultiLingual 2009a). In the United States, Lingotek
introduced software-as-a-service collaborative translation technology which combined the
workflow and computer-aided translation capabilities of human and machine translation into
one application. Organizations could upload new projects, assign translators (paid or unpaid),
check the status of current projects in real time and download completed documents from any
computer with web access (MultiLingual 2009c). In Asia, Beijing Zhongke LongRay Software and Technology Ltd. Co. in China released in
September LongRay CAT 3.0 (standalone edition), a CAT system with translation memory,
alignment, dictionary and terminology management and other functions (http://www.zklr.
com). In November, Foshan Snowman Computer Co. Ltd. released Snowman version 1.0 in
China (http://www.gcys.cn). Snowman deserves some mentioning because (1) it was new; (2)
the green trial version of Snowman could be downloaded free of charge; (3) it was easy to use as
its interface was user-friendly and the system was easy to operate; and (4) it had the language pair
of Chinese and English, which caters to the huge domestic market as well as the market abroad. Most of the activities relating to computer-aided translation in 2010 took place in Europe
and North America. In Germany, Across Systems GmbH released in August Across Language Server v. 5 Service
Pack 1, which introduced a series of new functionalities and modes of operation relating to the
areas of project management, machine translation, crowdsourcing and authoring assistance
(http://new.multilingual.com). In October, MetaTexis version 3.0 was released, which
imported filter for Wordfast Pro and Trados Studio translation memories and documents
(http://www.metatexis.com). In France, Wordfast LLC released in July Wordfast Pro 2.4
(WFP) with over sixty enhancements. This system was a standalone environment that featured
a highly customizable interface, enhanced batch processing functionality, and increased file
format support (http://www.wordfast.net). In October, Wordfast LLC created an application
to support translation on the iPhone and iPad in the Wordfast Anywhere environment (http://
www.wordfast.net). In Hungary, Kilgray Translation Technologies released in February
MemoQ 4.0, which was integrated with project management functions for project managers
who wanted to have more control and enable translators to work in any translation tool. In
October, the company released MemoQ 4.5, which had a rewritten translation memory
engine and improvements to the alignment algorithm (http://www.kilgray.com). In France,
Atril released in March TeaM Server, which allowed translators with Déjà Vu Workgroup to
work on multinational and multisite translation projects on a LAN or over the Internet, sharing
their translations in real-time, ensuring superior quality and consistency. TeaM Server also
provided scalable centralized storage for translation memories and terminology databases. The
size of translation repositories and the number of concurrent users were only limited by the
server hardware and bandwidth (http://www.atril.com). In October, Atril released Déjà Vu
X2 in four editions: Editor, Standard, Professional, and Workgroup. Its new features included
DeepMiner data extraction engine, new StartView interface, and AutoWrite word prediction.
In Switzerland, STAR Group released in October Transit NXT Service Pack 3 and TermStar
NXT. Transit NXT Service Pack 3 contained the following improvements: support of
Microsoft Office 2007, InDesign CS5, QuarkXpress 8 and QuarkXpress 8.1, and PDF
synchronization for MS Word files. In the United Kingdom, SDL released in March a new subscription level of its SDL Trados
Studio, which included additional productivity tools for translators such as Service Pack 2,
enabling translators to plug in to multiple automatic translation tools. The company also did a
S. Chan Development of translation technology19
beta launch of SDL OpenExchange, inviting the developer community to make use of standard
open application programming interfaces to increase the functionality of SDL Trados Studio
(Multilingual 2010a). In September, XTM International released XTM Cloud, which was a
totally online Software-as-a-Service (SaaS) computer-assisted translation tool set, combining
translation workflow with translation memory, terminology management and a fully featured
translator workbench. The launch of XTM Cloud means independent freelance translators
have access to XTM for the first time (http://www.xtm-intl.com). In Ireland, Alchemy
Software Development Limited released in May Alchemy PUBLISHER 3.0, which supports
all aspects of the localization workflow, including form translation, engineering, testing, and
project management. It also provided a Machine Translation connector which was jointly
developed by PROMT, so that documentation formats could be machine translated (http://
www.alchemysoftware.ie; http://www.promt.com). In North America, IBM in the United States released in June the open source version of
OpenTM/2, which originated from the IBM Translation Manager. OpenTM/2 integrated
with several aspects of the end-to-end translation workflow (http://www.opentm2.org).
Partnering with LISA (Localization Industry Standards Association), Welocalize, Cisco, and
Linux Solution Group e.V. (LiSoG), IBM aimed to create an open source project that provided
a full-featured, enterprise-level translation workbench environment for professional translators
on OpenTM/2 project. According to LISA, OpenTM/2 not only provided a public and open
implementation of translation workbench environment that served as the reference
implementation of existing localization industry standards, such as TMX, it also aimed to
provide standardized access to globalization process management software (http://www.lisa.
org; LISA 2010). Lingotek upgraded in July its Collaborative Translation Platform (CTP) to a
software-as-a-service product which combined machine translation, real-time community
translation, and management tools (MultiLingual 2010b). MadCap Software, Inc. released in
September MadCap Lingo v4.0, which had a new utility for easier translation alignment and a
redesigned translation editor. Systran introduced in December Desktop 7 Product Suite, which
included the Premium Translator, Business Translator, Office Translator, and Home Translator.
Among them, Premium Translator and Business Translator were equipped with translation
memory and project management features. In South America, Maxprograms in Uruguay released in April Swordfish II, which
incorporated Anchovy version 1.0-0 as glossary manager and term extraction tool, and added
support for SLD XLIFF files from Trados Studio 2009 and Microsoft Visio XML Drawings,
etc. (http://www.maxprograms.com). In 2011, computer-aided translation was active in Europe and America.
In Europe, ATRIL / PowerLing in France released in May Déjà Vu X2, a new version of
its computer-assisted translation system, which had new features such as DeepMiner data
mining and translation engine, SmartView Interface and a multi-file and multi-format
alignment tool (MultiLingual 2011). In June, Wordfast Classic v6.0 was released with features
such as the ability to share TMs and glossaries with an unlimited number of users, improved
quality assurance, AutoComplete, and improved support for Microsoft Word 2007/2010 and
Mac Word 2011 (http://www.wordfast.net). In Luxembourg, the Directorate-General for
Translation of the European Commission released in January its one million segments of
multilingual Translation Memory in TMX format in 231 language pairs. Translation units
were extracted from one of its large shared translation memories in Euramis (European
Advanced Multilingual Information System). This memory contained most, but not all, of the
documents of the Acquis Communautaire, the entire body of European legislation, plus some
other documents which were not part of the Acquis. In Switzerland, the STAR Group released 20
in February Service Pack 4 for Transit NXT and TermStar NXT. Transit NXT Service Pack
4 contained the following improvements: support of MS Office 2010, support of Quicksilver
3.5l, and preview for MS Office formats. In Eastern Europe, Kilgray Translation Technologies
in Hungary released in June TM Repository, the world’s first tool-independent Translation
Memory management system (http://kilgray.com). Kilgray Translation Technologies later
released MemoQ v 5.0 with the AuditTrail concept to the workflow, which added new
improvements like versioning, tracking changes (to show the difference of two versions),
X-translate (to show changes on source texts), the Post Translation Analysis on formatting tags
(Kilgray Translation Technologies 2011). In the United Kingdom, XTM International released in March XTM 5.5, providing both
Cloud and On-Premise versions, which contained customizable workflows, a new search and
replace feature in Translation Memory Manager and the redesign of XTM Workbench (http://
www.xtm-intl.com). In North America, MultiCorpora R&D Inc. in Canada released in May MutliTrans Prism,
a translation management system (TMS) for project management, translation memory and
terminology management (MultiCorpora 2011). In 2012, the development of computer-aided translation in various places was considerable
and translation technology continued its march to globalization. In North America, the development of computer-aided translation was fast. In Canada,
MultiCorpora, a provider of multilingual asset management solutions, released in June
MultiTrans Prism version 5.5. The new version features a web editing server that extends
control of the management of translation processes, and it can be fully integrated with content
management systems. In September, Terminotix launched LogiTerm 5.2. Its upgrades and
new features, including indexing TMX files directly in Bitext database, reinforced the fuzzy
match window, and adjusted buttons (http://terminotix.com/news/newsletter). In December,
MultiCorpora added new machine translation integrations to its MultiTrans Prism. The
integration options include Systran, Google and Microsoft (http://www.multicorpora.com).
In Asia, there was considerable progress in computer-aided translation in China. Transn
Information Technology Co., Ltd. released TCAT 2.0 as freeware early in the year. New
features of this software include the Translation Assistant (翻譯助理) placed at the sidebar of
Microsoft Office, pre-translation with TM and termbase, source segment selection by
highlighting (自動取句) (http://www.transn.com). In May, Foshan Snowman Computer Co.
Ltd. released Snowman 1.27 and Snowman Collaborative Translation Platform (雪人 CAT 協
同翻譯平臺) free version. The platform offers a server for a central translation memory and
termbase so that all the users can share their translations and terms, and the reviewers can view
the translations simultaneously with translators. It also supports online instant communication,
document management and online forum (BBS) (http://www.gcys.cn). In July, Chengdu
Urelite Tech Co. Ltd. (成都優譯信息技術有限公司), which was founded in 2009, released
Transmate, including the standalone edition (beta), internet edition and project management
system. The standalone edition was freely available for download from the company’s website.
The standalone edition of Transmate is targeted at freelancers and this beta release offers basic
CAT functions, such as using TM and terminology during translation. It has features such as
pre-translation, creating file-based translation memory, bilingual text export and links to an
online dictionary website and Google MT (http://www.urelitetech.com.cn). Heartsome Translation Studio 8.0 was released by the Shenzhen Office of Heartsome in
China. Its new features include pre-saving MT results and external proofreading file export in
RTF format. The new and integrated interface also allows the user to work in a single unified
environment in the translation process (http://www.heartsome.net).
S. Chan Development of translation technology21
In Japan, Ryan Ginstrom developed and released Align Assist 1.5, which is freeware to align
source and translation files to create translation memory. The main improvement of this
version is the ability to set the format of a cell text (http://felix-cat.com). In October, LogoVista
Corporation released LogoVista PRO 2013. It can support Microsoft Office 2010 64-bit and
Windows 8. More Japanese and English words are included and the total number of words in
dictionaries is 6.47 million (http://www.logovista.co.jp). In Europe, the developments of computer-aided translation systems are noteworthy.
In the Czech Republic, the MemSource Technologies released in January MemSource
Editor for translators as a free tool to work with MemSource Cloud and MemSource Server.
The Editor is multiplatform and can be currently installed on Windows and Macintosh (http://
www.memsource.com). In April, this company released MemSource Cloud 2.0. MemSource
Plugin, the former CAT component for Microsoft Word, is replaced by the new MemSource
Editor, a standalone translation editor. Other new features include adding comments to
segments, version control, translation workflow (only in the Team edition), better quality
assurance and segmentation (http://blog.memsource.com). In December, MemSource
Technologies released MemSource Cloud 2.8. It now encrypts all communication by default.
This release also includes redesigned menu and tools. Based on the data about previous jobs,
MemSource can suggest relevant linguistics for translation jobs (http://www.memsource.com). In France, Wordfast LLC released Wordfast Pro 3.0 in April. Its new features include
bilingual review, batch TransCheck, filter 100 per cent matches, split and merge TXML files,
reverse source/target and pseudo-translation (http://www.wordfast.com). In June Atril and
PowerLing updated Déjà Vu X2. Its new features include an incorporated PDF converter and
a CodeZapper Macro (http://www.atril.com). In Germany, Across Language Server v 5.5 was released in November. New features such as
linguistic supply chain management are designed to make project and resources planning more
transparent. The new version also supports the translation of display texts in various formats, and
allows the protection of the translation units to ensure uniform use (http://www.across.net). In Hungary, Kilgray Translation Technologies released in July MemoQ 6.0 with new
features like predictive typing and several new online workflow concepts such as FirstAccept
(assign job to the first translator who accepted it on the online workflow), GroupSourcing,
Slicing, and Subvendor group (http://kilgray.com). In December, the company released
MemoQ 6.2. Its new features include SDL package support, InDesign support with preview,
new quality assurance checks and the ability to work with multiple machine translation engines
at the same time (http://kilgray.com). In Luxembourg, Wordbee in October designed a new business analysis module for its Wordbee
translation management system, which provides a new dashboard where over 100 real-time
reports are generated for every aspect of the localization process (http://www.wordbee.com). In Switzerland, STAR Group released Service Pack 6 for Transit
NXT and TermStar NXT .
The improvements of Service Pack 6 of Transit NXT contain the support of Windows 8 and
Windows Server 2012, QuarkXPress 9.0-9.2, InDesign CS6, integrated OpenOffice spell
check dictionaries, 10 additional Indian languages (http://www.star-group.net). In the United Kingdom, XTM International, a developer of XML authoring and translation
tools, released in April XTM Suite 6.2. Its updates include a full integration with machine
translation system, Asia Online Language Studio and the content management system XTRF.
In October, the company released XTM Suite 7.0 and a new XTM Xchange module in XTM
Cloud intended to increase the supply chain. Version 7.0 includes project management
enhancements, allowing users to group files, assign translators to specific groups or languages,
and create different workflows for different languages (http://www.xtm-intl.com). 22
During this period, the following trends are of note.
1
The systematic compatibility with Windows and Microsoft Office
Of the sixty-seven currently available systems on the market, only one does not run on
the Windows operation systems. Computer-aided translation systems have to keep up
with the advances in Windows and Microsoft Office for the sake of compatibility.
Wordfast 5.51j, for example, was released in April 2007, three months after the release of
Windows Vista, and Wordfast 5.90v was released in July 2010 to support Microsoft Office
Word 2007 and 2010.
2
The integration of workflow control into CAT systems
Besides re-using or recycling translations of repetitive texts and text-based terminology,
systems developed during this period added functions such as project management, spell
check, quality assurance, and content control. Take SDL Trados Studio 2011 as an
example. This version, which was released in September 2011, has a spell checking
function for a larger number of languages and PerfectMatch 2.0 to track changes of the
source documents. Most of the systems on the market can also perform ‘context match’,
which is the identical match with identical surrounding segments in the translation
document and in the translation memory.
3
The availability of networked or online systems
Because of the fast development of new information technologies, most CAT systems
during this period were server-based, web-based and even cloud-based CAT systems,
which had a huge storage of data. By the end of 2012, there were fifteen cloud-based
CAT systems available on the market for individuals or enterprises, such as Lingotek
Collaborative Translation Platform, SDL World Server, and XTM Cloud.
4
The adoption of new formats in the industry
Data exchange between different CAT systems has always been a difficult issue to handle
as different systems have different formats, such as dvmdb for Déjà Vu X, and tmw for SDL
Trados Translator’s Workbench 8.0. These program-specific formats cannot be mutually
recognizable, which makes it impossible to share data in the industry. In the past, the
Localization Industry Standards Association (LISA) played a significant role in developing
and promoting data exchange standards, such as SRX (Segmentation Rules eXchange),
TMX (Translation Memory eXchange), TBX (Term-Bese eXchange) and XLIFF (XML
Localisation Interchange File Format). (http://en.wikipedia.org/wiki/XLIFF). It can be
estimated that the compliance of industry standards is also one of the future directions for
better data exchange.
Translation technology on a fast track: a comparison of the developments of
computer-aided translation with human translation and machine translation
The speed of the development of translation technology in recent decades can be illustrated
through a comparison of computer-aided translation with human translation and machine
translation.
The development of human translation
Human translation, in comparison with machine translation and computer-aided translation,
has taken a considerably longer time and slower pace to develop. The history of human
translation can be traced to 1122 \bc when during the Zhou dynasty (1122–255 \bc), a foreign
S. Chan Development of translation technology23
affairs bureau known as Da xing ren 大行人 was established to provide interpreting services for
government officials to communicate with the twelve non-Han minorities along the borders
of the Zhou empire (Chan 2009: 29−30). This is probably the first piece of documentary
evidence of official interpreting in the world. Since then a number of major events have taken place in the world of translation. In 285 \bc,
there was the first partial translation of the Bible from Hebrew into Greek in the form of the
Septuagint (Worth 1992: 5−19). In 250 \bc, the contribution of Andronicus Livius to translation
made him the ‘father of translation’ (Kelly 1998: 495−504). In 67, Zhu Falan made the first
translation of a Buddhist sutra in China (Editorial Committee 1988: 103). In 1141, Robert de
Retines produced the first translation of the Koran in Latin (Chan 2009: 47). In 1382, John
Wycliffe made the first complete translation of the Bible in English (Worth 1992: 66−70). In
1494, William Tyndale was the first scholar to translate the Bible from the original Hebrew
and Greek into English (Delisle and Woodsworth 1995: 33−35). In 1611, the King James
Version of the Bible was published (Allen 1969). In 1814, Robert Morrison made the first
translation of the Bible into Chinese (Chan 2009: 73). In 1945, simultaneous interpreting was
invented at the Nuremberg Trials held in Germany (Gaiba 1998). In 1946, the United Bible
Societies was founded in New York (Chan 2009: 117). In 1952, the first conference on
machine translation was held at the Massachusetts Institute of Technology (Hutchins 2000: 6,
34−35). In 1953, the Fédération Internationale des Traducteurs (FIT), or International
Association of Translators, and the Association Internationale des Interprètes de Conference
(AIIC), or the International Association of Conference Interpreters, were both founded in
Paris (Haeseryn 1989: 379−84; Phelan 2001). In 1964, with the publication of Toward a Science
of Translating in which the concept of dynamic equivalent translation was proposed, Eugene A.
Nida was referred to as the ‘Father of Translation Theory’ (Nida 1964). In 1972, James S.
Holmes proposed the first framework for translation studies (Holmes 1972/1987: 9−24, 1988:
93−98). In 1978, Even-Zohar proposed the Polysystem Theory (Even-Zohar 1978: 21−27). A total of some seventeen major events took place during the history of human translation,
which may be 3,135 years old. This shows that in terms of the mode of production, human
translation has remained unchanged for a very long time.
The development of machine translation
In comparison with human translation, machine translation has advanced enormously since its
inception in the 1940s. This can be clearly seen from an analysis of the countries with research
and development in machine translation during the last seventy years. Available information shows that an increasing number of countries have been involved in
the research and development of machine translation. This is very much in evidence since the
beginning of machine translation in 1947. Actually, long before the Second World War was
over and the computer was invented, Georges Artsrouni, a French-Armenian engineer, created
a translation machine known as ‘Mechanical Brain’. Later in the year, Petr Petrovi
č Smirnov-
Troyanskij (1894−1950), a Russian scholar, was issued a patent in Moscow on 5 September
for his construction of a machine which could select and print words while translating from
one language into another or into several others at the same time (Chan 2004: 289). But it was not until the years after the Second World War that the climate was ripe for the
development of machine translation. The invention of computers, the rise of information
theory, and the advances in cryptology all indicated that machine translation could be a reality.
In 1947, the idea of using machines in translating was proposed in March by Warren Weaver
(1894−1978), who was at that time the vice president of the Rockefeller Foundation, and 24
Andrew D. Booth of Birkbeck College of the University of London. They wanted to make
use of the newly invented computer to translate natural languages. Historically speaking, their
idea was significant in several ways. It was the first application of the newly invented computers
to non-numerical tasks, i.e. translation. It was the first application of the computer to natural
languages, which was later to be known as computational linguistics. It was also one of the first
areas of research in the field of artificial intelligence. The following year witnessed the rise of information theory and its application to translation
studies. The role of this theory has been to help translators recognize the function of concepts
such as information load, implicit and explicit information, and redundancy (Shannon and
Weaver 1949; Wiener 1954). On 15 July 1948, Warren Weaver, director of the Rockefeller
Foundation’s natural sciences division, wrote a memorandum for peer review outlining the
prospects of machine translation, known in history as ‘Weaver’s Memorandum’, in which he
made four proposals to produce translations better than word-for-word translations (Hutchins
2000: 18−20). The first machine translation system, the Georgetown-IBM system for Russian−English
translation, was developed in the United States in June 1952. The system was developed by
Leon Dostert and Paul Garvin of Georgetown University and Cuthbert Hurd and Peter
Sheridan of IBM Corporation. This system could translate from Russian into English (Hutchins
1986: 70−78). Russia was the second country to develop machine translation. At the end of 1954, the
Steklov Mathematical Institute of the Academy of Sciences began work on machine translation
under the directorship of Aleksej Andreevi
č Ljapunov (1911−1973), a mathematician and
computer expert. The first system developed was known as FR-I, which was a direct translation
system and was also considered one of the first generation of machine translation systems. The
system ran on STRELA, one of the first generation of computers (Hutchins 2000: 197−204). In the same year, the United Kingdom became the third country to engage in machine
translation. A research group on machine translation, Cambridge Language Research Group,
led by Margaret Masterman, was set up at Cambridge University, where an experimental
system was tried on English-French translation (Wilks 2000: 279−298). In 1955, Japan was the fourth country to develop machine translation. Kyushu University was
the first university in Japan to begin research on machine translation (Nagao 1993: 203−208).
This was followed by China, which began research on machine translation with a Russian−
Chinese translation algorithm jointly developed by the Institute of Linguistics and the Institute
of Computing Technology (Dong 1988: 85−91; Feng 1999: 335−340; Liu 1984: 1−14). Two years later, Charles University in Czechoslovakia began to work on English–Czech
machine translation (http://www.cuni.cz). These six countries were the forerunners in machine translation. Other countries followed
suit. In 1959, France set up the Centre d’Études de la Traduction Automatique (CETA) for
machine translation (Chan 2009: 300). In 1960, East Germany had its Working Group for
Mathematical and Applied Linguistics and Automatic Translation, while in Mexico, research
on machine translation was conducted at the National Autonomous University of Mexico
(Universidad Nacional Autonoma de Mexico) (http://www.unam.mx). In 1962, Hungary’s
Hungarian Academy of Sciences conducted research on machine translation. In 1964 in
Bulgaria, the Mathematical Institute of the Bulgarian Academy of Sciences in Sofia set up the
section of ‘Automatic Translation and Mathematical Linguistics’ to conduct work on machine
translation (http://www.bas.bg; Hutchins 1986: 205−06). In 1965, the Canadian Research
Council set up CETADOL (Centre de Traitement Automatique des Données Linguistiques)
to work on an English−French translation system (Hutchins 1986: 224).
S. Chan Development of translation technology25
But with the publication of the ALPAC Report prepared by the Automatic Language
Processing Advisory Committee of the National Academy of Sciences, which concluded with
the comment that there was ‘no immediate or predictable prospect of useful machine
translation’, funding for machine translation in the United States was drastically cut and interest
in machine translation waned considerably (ALPAC 1966; Warwick 1987: 22−37). Still,
sporadic efforts were made in machine translation. An important system was developed in the
United States by Peter Toma, previously of Georgetown University, known as Systran, an
acronym for System Translation. To this day, this system is still one of the most established and
popular systems on the market. In Hong Kong, The Chinese University of Hong Kong set up
the Hung On-To Research Laboratory for Machine Translation to conduct research into
machine translation and developed a practical machine translation system known as ‘The
Chinese University Language Translator’, abbreviated as CULT (Loh 1975: 143−155, 1976a:
46−50, 1976b: 104−05; Loh, Kong and Hung 1978: 111−120; Loh and Kong 1979: 135−148).
In Canada, the TAUM group at Montreal developed a system for translating public weather
forecasts known as TAUM-METEO, which became operative in 1977. In the 1980s, the most important translation system developed was the EUROTRA system,
which could translate all the official languages of the European Economic Community (Arnold
and Tombe 1987: 1143−1145; Johnson, King and Tombe 1985: 155−169; King 1982; King
1987: 373−391; Lau 1988: 186−191; Maegaard 1988: 61−65; Maegaard and Perschke 1991:
73−82; Somers 1986: 129−177; Way, Crookston and Shelton 1997: 323−374). In 1983, Allen
Tucker, Sergei Nirenburg, and others developed at Colgate University an AI-based multilingual
machine translation system known as TRANSLATOR to translate four languages, namely
English, Japanese, Russian, and Spanish. This was the beginning of knowledge-based machine
translation in the United States (http://www.colgate.edu). The following year, Fujitsu
produced ATLAS/I and ATLAS/II translation systems for translation between Japanese and
English in Japan, while Hitachi and Market Intelligence Centre (QUICK) developed the
ATHENE English−Japanese machine translation system (Chan 2009: 223). In 1985, the
ArchTran machine translation system for translation between Chinese and English was
launched in Taiwan and was one of the first commercialized English−Chinese machine
translation systems in the world (Chen, Chang, Wang and Su 1993: 87−98). In the United
States, the METAL (Mechanical Translation and Analysis of Languages) system for translation
between English and German, supported by the Siemens Company in Munich since 1978 and
developed at the University of Texas, Austin, became operative (Deprez, Adriaens, Depoortere
and de Braekeleer 1994: 206−212; Lehmann, Bennett and Slocum et al. 1981; Lehrberger
1981; Little 1990: 94−107; Liu and Liro 1987: 205−218; Schneider 1992: 583−594; Slocum,
Bennett, Bear, Morgan and Root 1987: 319−350; White 1987: 225−240). In China, the
TranStar English−Chinese Machine Translation System, the first machine-translation product
in China, developed by China National Computer Software and Technology Service
Corporation, was commercially available in 1988 (http://www.transtar.com.cn). In Taiwan,
the BehaviorTran, an English−Chinese machine translation system, was also launched in the
same year. In the 1990s, Saarbrucken in Germany formed the largest and the most established machine
translation group in 1996. The SUSY (Saarbrücker Ubersetzungssystem/The Saarbrücken
Machine Translation System) project for German to English and Russian to German machine
translation was developed between 1972 and 1986 (rz.uni-sb.de). In 1997, Dong Fang Kuai Che
東方快車 (Orient Express), a machine translation system developed by the China Electronic
Information Technology Ltd. in China, was commercially available (Chan 2004: 336) while in
Taiwan, TransBridge was developed for internet translation from English into Chinese (http:// 26
www.transbridge.com.tw). The first year of the twenty-first century witnessed the development
of BULTRA (BULgarian TRAnslator), the first English−Bulgarian machine translation tool,
by Pro Langs in Bulgaria (Chan 2004: 339).What has been presented above shows very clearly that from the beginning of machine
translation in 1947 until 1957, six countries were involved in the research and development of
machine translation, which included Massachusetts Institute of Technology and Georgetown
University in the United States in 1952, Academy of Sciences in Russia and Cambridge
University in the United Kingdom in 1954, Kyushu University in Japan in 1955, the Institute
of Linguistics in China in 1956, and Charles University in Czechoslovakia in 1957. By 2007,
it was found that of the 193 countries in the world, 30 have conducted research on computer
or computer-aided translation, 9 actively. This means that around 16 per cent of all the
countries in the world have been engaged in machine translation, 30 per cent of which are
active in research and development. The 31 countries which have been engaged in the research
and development of machine translation are: Belgium, Brazil, Bulgaria, Canada, China,
Czechoslovakia, Denmark, Finland, France, Germany, Hungary, India, Italy, Japan, Korea,
Macau, Malaysia, Mexico, the Netherlands, Luxemburg, Poland, Russia, Singapore, Slovenia,
Spain, Sweden, Switzerland, Taiwan, Tunisia, the United Kingdom, and the United States. Of
these, the most active countries are China and Japan in Asia, France, Germany, the Netherlands,
the United Kingdom, and Russia in Europe, and Canada and the United States in North
America. The huge increase in the number of countries engaged in machine translation and
the fast development of systems for different languages and language pairs show that machine
translation has advanced by leaps and bounds in the last 65 years.
Conclusion
It should be noted that computer-aided translation has been growing rapidly in all parts of the
world in the last 47 years since its inception in 1967. Drastic changes have taken place in the
field of translation since the emergence of commercial computer-aided translation systems in
the 1980s. In 1988, as mentioned above, we only had the Trados system that was produced in
Europe. Now we have more than 100 systems developed in different countries, including
Asian countries such as China, Japan, and India, and the northern American countries, Canada
and the United States. In the 1980s, very few people had any ideas about computer-aided
translation, let alone translation technology. Now, it is estimated that there are around 200,000
computer-aided translators in Europe, and more than 6,000 large corporations in the world
handle their language problems with the use of corporate or global management computer-
aided translation systems. At the beginning, computer-aided translation systems only had
standalone editions. Now, there are over seventeen different types of systems on the market. According to my research, the number of commercially available computer-aided translation
systems from 1984 to 2012 is 86. Several observations on these systems can be made. First,
about three computer-aided translation systems have been produced every year during the last
28 years. Second, because of the rapid changes in the market, nineteen computer-aided
translation systems failed to survive in the keen competition, and the total number of current
commercial systems stands at 67. Third, almost half of the computer-aided translation systems
have been developed in Europe, accounting for 49.38 per cent, while 27.16 per cent of them
have been produced in America. All these figures show that translation technology has been on the fast track in the last five
decades. It will certainly maintain its momentum for many years to come.
S. Chan Development of translation technology27
References
Allen, Ward (ed.) (1969) Translating for King James, Nashville, TN: Vanderbilt University Press.
ALPAC (Automatic Language Processing Advisory Committee) (1966) Languages and Machines: Computers in Translation and Linguistics, A Report by the Automatic Language Processing Advisory Committee,
Division of Behavioral Sciences, National Academy of Sciences, National Research Council,
Washington, DC: National Academy of Sciences, National Research Council, 1966.
Arnold, Doug J. and Louis des Tombe (1987) ‘Basic Theory and Methodology in EUROTRA’, in Sergei Nirenburg (ed.) Machine Translation: Theoretical and Methodological Issues, Cambridge: Cambridge
University Press, 114−135.
Arthern, Peter J. (1979) ‘Machine Translation and Computerized Terminology Systems: A Translator’s viewpoint’, in Barbara M. Snell (ed.) Translating and the Computer: Proceedings of a Seminar, London:
North-Holland Publishing Company, 77−108.
Brace, Colin (1992) ‘Trados: Smarter Translation Software’, Language Industry Monitor Issue September– October. Available at: http://www.lim.nl/monitor/trados-1.html.
Brace, Colin (1993) ‘TM/2: Tips of the Iceberg’, Language Industry Monitor Issue May−June. Available at: Retrieved from http://www.mt-archive.
Brace, Colin (1994) ‘Bonjour, Eurolang Optimizer’, Language Industry Monitor Issue March-April. Available at: http://www.lim.nl/monitor/optimizer.html.
Chan, Sin-wai (2004) A Dictionary of Translation Technology, Hong Kong: The Chinese University Press.
Chan, Sin-wai (2009) A Chronology of Translation in China and the West, Hong Kong: The Chinese University Press.
Chen, Gang (2001) ‘A Review on Yaxin CAT2.5’, Chinese Science and Technology Translators Journal 14(2).
Chen, Shuchuan, Chang Jing-shin, Wang Jong-nae, and Su Keh-yih (1993) ‘ArchTran: A Corpus-based Statistics-oriented English-Chinese Machine Translation System’, in Sergei Nirenburg (ed.) Progress in
Machine Translation, Amsterdam: IOP Press, 87−98.
Deprez, F., Greert Adriaens, Bart Depoortere, and Gert de Braekeleer (1994) ‘Experiences with METAL at the Belgian Ministry of the Interior’, Meta 39(1): 206−212.
Delisle, Jean and Judith Woodsworth (eds) (1995) Translators through History, Amsterdam and Philadelphia: John Benjamins Publishing Company and UNESCO Publishing.
Dong, Zhendong (1988) ‘MT Research in China’, in Dan Maxwell, Klaus Schubert, and Toon Witkam (eds) New Directions in Machine Translation, Dordrecht, Holland: Foris Publications, 85−91.
Editorial Committee, A Dictionary of Translators in China 《中國翻譯家詞典》編寫組 (ed.) (1988) 《中 國翻譯家詞典》 (A Dictionary of Translators in China), Beijing: China Translation and Publishing
Corporation 中國對外翻譯出版公司.
Elita, Natalia and Monica Gavrila (2006) ‘Enhancing Translation Memories with Semantic Knowledge’, Proceedings of the 1
st Central European Student Conference in Linguistics, 29-31 May 2006, Budapest,
Hungary: 24−26.
Eurolux Computers (1992) ‘Trados: Smarter Translation Software’, Language Industry Monitor 11, September−October. Available at: http://www.lim.nl.
Even-Zohar, Itamar (1978) Papers in Historical Poetics, Tel Aviv: The Porter Institute for Poetics and Semiotics, Tel Aviv University.
Feng, Zhiwei 馮志偉 (1999)〈中國的翻譯技術:過去、現在和將來〉(Translation Technology
in China: Past, Present, and Future), in Huang Changning 黃昌寧 and Dong Zhendong 董振東 (eds)
《計算器語言學文集》(Essays on Computational Linguistics), Beijing: Tsinghua University Press, 335–
440.
Gaiba, Francesca (1998) The Origins of Simultaneous Interpretation: The Nuremberg Trial, Ottawa: University of Ottawa Press.
Garcia, Ignacio and Vivian Stevenson (2005) ‘Trados and the Evolution of Language Tools: The Rise of the De Facto TM Standard – And Its Future with SDL’, Multilingual Computing and Technology 16(7).
Garcia, Ignacio and Vivian Stevenson (2006) ‘Heartsome Translation Suite’, Multingual 17(1): 77. Available at: http://www.multilingual.com.
German, Kathryn (2009) ‘Across: An Exciting New Computer Assisted Translation Tool’, The Northwest Linguist 9−10.
Gotti, Fabrizio, Philippe Langlais, Elliott Macklovitch, Didier Bourigault, Benoit Robichaud, and Claude Coulombe (2005) ‘3GTM: A Third-generation Translation Memory’, Proceedings of the 3
rd Computational
Linguistics in the North-East (CLiNE) Workshop, Gatineau, Québec, Canada, 26–30. 28
Haeseryn, Rene (1989) ‘The International Federation of Translators (FIT) and its Leading Role in the Translation Movement in the World’, in Rene Haeseryn (ed.) Roundtable Conference FIT-UNESCO:
Problems of Translator in Africa, Belgium: FIT, 379−384.
Hall, Amy (2000) ‘SDL Announces Release of SDLX Version 2.0’, SDL International. Available at: http://www.sdl.com/en/about-us/press/1999/SDL_Announces_Release_of_SDLX_Version_2_0.
asp.
Harmsen, R. (2008) ‘Evaluation of DVX’. Available at: http://rudhar.com.
Holmes, James S. (1972, 1987) ‘The Name and Nature of Translation Studies’, in Gideon Toury (ed.) Translation across Cultures, New Delhi: Bahri Publications: Pvt. Ltd., 9−24.
Holmes, James S. (1988) ‘The Name and Nature of Translation Studies’, in James S. Holmes (ed.) Translated! Papers on Literary Translation and Translation Studies, Amsterdam: University of Amsterdam,
93−98.
http://anaphraseus.sourceforge.net.
http://blog.memsource.com.
http://developer.apple.com.
http://en.wikipedia.org/wiki/XLIFF
http://felix-cat.com.
http://new.multilingual.com.
http://terminotix.com/news/newsletter.
http://www.across.net.
http://www.alchemysoftware.ie.
http://www.atasoft.com.
http://www.atril.com.
http://www.bas.bg.
http://www.colgate.edu.
http://www.cuni.cz.
http://www.dreye.com.tw.
http://www.gcys.cn.
http://www.heartsome.net.
http://www.hjtek.com.
http://www.hjtrans.com.
http://www.kilgray.com.
http://www.lisa.org.
http://www.logovista.co.jp.
http://www.maxprograms.com.
http://www.memsource.com.
http://www.metatexis.com.
http://www.multicorpora.com.
http://www.multilizer.com.
http://www.omegat.org.
http://www.opentm2.org.
http://www.otek.com.tw.
http://www.promt.com.
http://www.sdl.com.
http://www.sdlintl.com.
http://www.star-group.net.
http://www.systransoft.com.
http://www.thelanguagedirectory.com/translation/translation_software.
http://www.transbridge.com.tw.
http://www.translationzone.com.
http://www.transn.com.
http://www.transtar.com.cn.
http://www.tratool.com.
http://www.unam.mx.
http://www.urelitetech.com.cn.
http://www.wordbee.com.
http://www.wordfast.com.
http://wordfast.net/champollion.net.
S. Chan Development of translation technology29
http://www.xtm-intl.com.
http://www.zklr.com.
Hutchins, W. John (1986) Machine Translation: Past, Present and Future, Chichester: Ellis Horwood.
Hutchins, W. John (1998) ‘The Origins of the Translator’s Workstation’, Machine Translation 13(4): 287−307.
Hutchins, W. John (1999) ‘The Development and Use of Machine Translation System and Computer- based Translation Tools’, in Chen Zhaoxiong (ed.) Proceedings of the International Conference on MT and
Computer Language Information Processing, Beijing: Research Center of Computer and Language
Engineering, Chinese Academy of Sciences, 1−16.
Hutchins, W. John (2000) Early Years in Machine Translation, Amsterdam and Philadelphia: John Benjamins.
Johnson, R.I., Margaret King, and Louis des Tombe (1985) ‘Eurotra: A Multilingual System under Development’, Computational Linguistics 11(2−3): 155−169.
Kavak, Pinar (2009) ‘Development of a Translation Memory System for Turkish to English’, Unpublished MA dissertation, Bo
ğaziçi University, Turkey.
Kay, Martin (1980) ‘The Proper Place of Men and Machines in Language Translation’, Research Report CSL-80-11, Xerox Palo Alto Research Center, Palo Alto, CA.
Kelly, Louis G. (1998) ‘Latin Tradition’, in Mona Baker (ed.) Routledge Encyclopedia of Translation Studies, London and New York: Routledge, 495−504.
Kilgrary Translation Technologies (2011) ‘What’s New in MemoQ’. Available at: http://kilgray.com/ products/memoq/whatsnew.
King, Margaret (1982) EUROTRA: An Attempt to Achieve Multilingual MT, Amsterdam: North-Holland.
King, Margaret (ed.) (1987) Machine Translation Today: The State of the Art, Edinburgh: Edinburgh University Press.
Környei, Tibor (2000) ‘WordFisher for MS Word: An Alternative to Translation Memory Programs for Freelance Translators?’ Translation Journal 4(1). Available at: http://accurapid.com/journal/11wf.htm.
Lau, Peter Behrendt (1988) ‘Eurotra: Past, Present and Future’, in Catriona Picken (ed.) Translating and the Computer 9: Potential and Practice, London: The Association for Information Management, 186−91.
Lehmann, Winfred P., Winfield S. Bennett, Jonathan Slocum et al. (1981) The METAL System, New York: Griffiss Air Force Base.
Lehrberger, John (1981) The Linguistic Model: General Aspects, Montreal: TAUM Group, University of Montreal.
LISA (2010) ‘IBM and the Localization Industry Standards Association Partner to Deliver Open-Source Enterprise-level Translation Tools’. Available at: http://www.lisa.org/OpenTM2.1557.0.html.
Little, Patrick (1990) ‘METAL – Machine Translation in Practice’, in Catriona Picken (ed.) Translation
and the Computer 11: Preparing for the Next Decade, London: The Association for Information
Management, 94−107.
Liu, Jocelyn and Joseph Liro (1987) ‘The METAL English-to-German System: First Progress Report’, Computers and Translation 2(4): 205−218.
Liu, Yongquan et al. 劉湧泉等 (1984) 《中國的機器翻譯》(Machine Translation in China), Shanghai:
Knowledge Press.
Locke, Nancy A and Marc-Olivier Giguère (2002) ‘MultiTrans 3.0’, MultiLingual Computing and Technology 13(7): 51.
Locke, William Nash and Andrew Donald Booth (eds) (1955) Machine Translation of Languages: Fourteen Essays, Cambridge, MA: MIT Press.
Loh, Shiu-chang (1975) ‘Machine-aided Translation from Chinese to English’, United College Journal 12(13): 143−155.
Loh, Shiu-chang (1976a) ‘CULT: Chinese University Language Translator’, American Journal of Computational Linguistics Microfiche, 46, 46−50.
Loh, Shiu-chang (1976b) ‘Translation of Three Chinese Scientific Texts into English by Computer’, ALLC Bulletin 4(2): 104−05.
Loh, Shiu-chang, Kong Luan, and Hung Hing-sum (1978) ‘Machine Translation of Chinese Mathematical Articles’, ALLC Bulltein 6(2): 111−120.
Loh, Shiu-chang and Kong Luan (1979) ‘An Interactive On-line Machine Translation System (Chinese into English)’, in Barbara M. Snell (ed.) Translating and the Computer, Amsterdam: North-Holland,
135−148.
Maegaard, Bente (1988) ‘EUROTRA: The Machine Translation Project of the European Communities’, Literary and Linguistic Computing 3(2): 61−65. 30
Maegaard, Bente and Sergei Perschke (1991) ‘Eurotra: General Systems Design’, Machine Translation 6(2): 73−82.
Melby, Alan K. (1978) ‘Design and Implementation of a Machine-assisted Translation System’, Proceedings of the 7
th International Conference on Computational Linguistics, 14−18 August 1978, Bergen, Norway.
Melby, Alan K. and Terry C. Warner (1995) The Possibility of Language: A Discussion of the Nature of Language, with Implications for Human and Machine Translation, Amsterdam and Philadelphia: John
Benjamins.
MultiCorpora Inc. (2011) ‘MultiCorpora Launches New Translation Management System’. Available at: http://www.multicorpora.com/news/multicorpora-launches-new-translation-management-system.
MultiLingual (1997) ‘CIMOS Releases Arabic to English Translation Software’, MultiLingual 20 December 1997. Available at: http://multilingual.com/newsDetail.php?id=422
MultiLingual (1998) ‘SDL Announces Translation Tools’, MultiLingual 23 September 1998. Available at: http://multilingual.com/newsDetail.php?id=154.
MultiLingual (1999) ‘SDL Announces SDL Workbench and Product Marketing Executive’, MultiLingual
22 Feburary 1999. Available at: http://multilingual.com/newsDetail.php?id=12.
MultiLingual (2003) ‘MultiCorpora R&D Releases MultiTrans 3.5’, MultiLingal 17 October 2003.
Available at: http://multilingual.com/newsDetail.php?id=3219.
MultiLingual (2005a). ‘STAR Releases Transit Service Pack 14’, MultiLingual 15 April 2005. Available
at: http://multilingual.com/newsDetail.php?id=4169.
MultiLingual (2005b) ‘SIMILIS Version 1.4 Released’, MultiLingual 27 April 2005. Available at: http://
multilingual.com/newsDetail.php?id=4187.
MultiLingual (2005c) ‘SDL Releases SDLX 2005’, MultiLingual 5 May 2005. Available at: http:// multilingual.com/newsDetail.php?id=4216.
MultiLingual (2005d) ‘MultiCorpora Announces the Release of MultiTrans 4’, MultiLingual 31 August
2005. Available at: http://multilingual.com/newsDetail.php?id=4425.
MultiLingual (2006a) ‘Across Rolls out New Version 3.5’, MultiLingual 20 November 2006. Available at: http://multilingual.com/newsDetail.php?id=5372.
MultiLingual (2006b) ‘Lingotek Announces Beta Launch of Language Search Engine’, MultiLingual 15 August 2006. Available at: http://multilingual.com/newsDetail.php?id=5168.
MultiLingual (2007) ‘Kilgray Releases Version 2.0 of MemoQ’, MultiLingual 25 January 2007. Available
at: http://multilingual.com/newsDetail.php?id=5467.
MultiLingual (2008a) ‘Across Language Server 4.0 SP1’, MultiLingual 21 April 2008. Available at: http:// multilingual.com/newsDetail.php?id=6228.
MultiLingual (2008b) ‘Fusion One and Fusion Collaborate 3.0’, MultiLingual 28 November 2008. Available at: http://multilingual.com/newsDetail.php?id=6568.
MultiLingual (2009a) ‘Fusion 3.1’, MultiLingual 19 March 2009. Available at: http://multilingual.com/ newsDetail.php?id=6734.
MultiLingual (2009b) ‘Across Language Server V.5’, MultiLingual 13 May 2009. Available at: http://
multilingual.com/newsDetail.php?id=6834.
MultiLingual (2009c) ‘Lingotek Launches Crowdsourcing Translation Platform’, MultiLingual 22 October 2009. Available at: http://multilingual.com/newsDetail.php?id=7103.
MultiLingual (2010a) ‘SDL Trados Studio’, MultiLingual 3 March 2010. Available at: http://multilingual. com/newsDetail.php?id=7298.
MultiLingual (2010b) ‘Collaborative Translation Platform 5.0’, MultiLingual 27 July 2010. Available at: http://multilingual.com/newsDetail.php?id=7544.
MultiLingual (2011) ‘Déjà Vu X2’, MultiLingual 24 May 2011. Available at: http://multilingual.com/
newsDetail.php?id=933.
Nagao, Makoto (1993) ‘Machine Translation: The Japanese Experience’, in Sergei Nirenburg (ed.) Progress in Machine Translation, Amsterdam: IOS Press, 203−208.
Nida, Eugene A. (1964) Toward a Science of Translating, Leiden: E.J. Brill.
Phelan, Mary (2001) The Interpreter’s Resource, Clevedon: Multilingual Matters Ltd.
Prior, Marc (2003) ‘Close Windows, Open Doors’, Translation Journal 7(1). Available at: http://accurapid. com/journal/23linux.htm.
Schmidt, Axel (2006) ‘Integrating Localization into the Software Development Process’, TC World March 2006.
Schneider, Thomas (1992) ‘User Driven Development: METAL as an Integrated Multilingual System’, Meta 37(4): 583−594.
S. Chan Development of translation technology31
Shannon, Claude L. and Warren Weaver (1949) The Mathematical Theory of Communication, Urbana, IL: University of Illinois Press.
Slocum, Jonathan, Winfield S. Bennett, J. Bear, M. Morgan, and Rebecca Root (1987) ‘METAL: The LRC Machine Translation System’, in Margaret King (ed.) Machine Translation Today: The State of the
Art, Edinburgh: Edinburgh University Press, 319−350.
Somers, Harold L. (1986) ‘Eurotra Special Issue’, Multilingual 5(3): 129−177.
Sumita, Eiichiro and Yutaka Tsutsumi (1988) ‘A Translation Aid System Using Flexible Text Retrieval Based on Syntax-matching’, in Proceedings of the 2
nd International Conference on Theoretical and Methodological
Issues in Machine Translation of Natural Languages, Pittsburgh, Pennsylvania: Carnegie Mellon University.
Available online at: http:// www.mt-archive. info/TMI-1988-Sumita.pdf.
Wang, Zheng 王正 (2011) 〈翻譯記憶系統的發展歷程與未來趨勢〉(Translation Memory Systems: A Historical Sketch and Future Trends),《編譯論叢》(Compilation and Translation Review) 4(1):
133−160.
Warwick, Susan (1987) ‘An Overview of Post-ALPAC Developments’, in Margaret King (ed.) Machine Translation Today: The State of the Art, Edinburgh: Edinburgh University Press, 22−37.
Wassmer, Thomas (2003) ‘SDLX TM Translation Suite 2003’, Translation Journal 7(3).
Wassmer, Thomas (2004) ‘Trados 6.5’, MultiLingual Computing and Technology 15(1): 61.
Wassmer, Thomas (2007) ‘Comparative Review of Four Localization Tools: Déjà Vu, MULTILIZER, MultiTrans and TRANS Suite 2000, and Their Various Capabilities’, MultiLingual Computing and
Technology 14(3): 37−42.
Wassmer, Thomas (2011) ‘Dr Tom’s Independent Software Reviews’. Available at: http://www. localizationworks.com/DRTOM/Trados/TRADOS.
Way, Andrew, Ian Crookston, and Jane Shelton (1997) ‘A Typology of Translation Problems for Eurotra Translation Machines’, Machine Translation 12(4): 323−374.
White, John S. (1987) ‘The Research Environment in the METAL Project’, in Sergei Nirenburg (ed.) Machine Translation: Theoretical and Methodological Issues, Cambridge: Cambridge University Press,
225−240.
Wiener, Norbert (1954) The Human Use of Human Beings: Cybernetics and Society, New York: Houghton Mifflin.
Wilks, Yorick (2000) ‘Magaret Masterman’, in W. John Hutchins (ed.) Early Years in Machine Translation, Amsterdam and Philadelphia: John Benjamins Publishing Company, 279−298.
Worth, Roland H. (1992) Bible Translations: A History through Source Documents, Jefferson, NC, and London: McFarland and Company, Inc., Publishers.
Xu, Jie (2001) ‘Five Amazing Functions of Dr Eye 2001’ (Dr Eye 2001 譯典通 5 大非凡功能), 《廣東
電腦與電訊》(Computer and Telecom) (3).
Yngve, Victor H. (2000) ‘Early Research at M.I.T. in Search of Adequate Theory’, in W. John Hutchins (ed.) Early Years in Machine Translation, Amsterdam and Philadelphia: John Benjamins, 39−72.
Zhang, Zheng 張政 (2006)《計算機翻譯研究》(Studies on Machine Translation), Beijing: Tsinghua University Press 清華大學出版社. 32
2
COMPUTER-AIDED TRANSLATION
Major concepts
Chan Sin-wai
the chinese university \ff h\fng k\fng, h\fng k\fng, china
Introduction
When the term computer-aided translation is mentioned, we often associate it with the
functions a computer-aided translation system offers, such as toolbars, icons, and hotkeys, the
built-in tools we can use, such as online dictionaries, browsers, and the computational hitches
we often encounter when working on a computer-aided translation project, such as chaotic
codes. What is more important is to see beyond the surface of computer-aided translation to
find out the major concepts that shape the development of functions in translation technology.
Concepts, which are relatively stable, govern or affect the way functions are designed and
developed, while functions, which are fast-changing, realize the concepts through the tasks
they perform. As a major goal of machine translation is to help human translators, a number of
functions in computer-aided translation systems have been created to enable machine processing
of the source with minimum human intervention. Concepts, moreover, are related to what
translators want to achieve in translating. Simply put, translators want to have a controllable
(controllability) and customizable (customizability) system, which is compatible with file formats
(compatibility) and language requirements, and behaves as well as (simulativity) or even better
than (emulativity) a human translator, to allow them to work together (collaborativity) to produce
quality translations (productivity). We have therefore identified seven major concepts which are
important in computer-aided translation: simulativity, emulativity, productivity, compatibility,
controllability, customizability, and collaborativity. The order in which concepts are arranged
can be memorized more easily by their acronym SEPCCCC.
Simulativity
The first concept of computer-aided translation is simulativity, which is about the way in
which a computer-aided translation system models the behaviour of a human translator by
means of its functions, such as the use of concordancers in text analysis to model after
comprehension on the part of the human translator and the creation of a number of quality
assurance tools to follow the way checking is done by a human translator. There are a number of ways to illustrate man–machine simulativity. Computer-aided translation33
(1) Goal of translation
The first is about the ultimate goal of translation technology. All forms of translation (machine
translation, computer-aided translation and human translation) aim at obtaining high-quality
translations. In the case of machine translation, the goal of a fully automatic high-quality
translation (FAHQT) is to be achieved through the use of a machine translation system without
human intervention. In the case of computer-aided translation, the same goal is to be achieved
through a computer-aided translation system that simulates the behaviour of a human translator
through man−machine interaction.
(2) Translation procedure
A comparison of the procedures of human translation with those of computer-aided translation
shows that the latter simulates the former in a number of ways. In manual translation, various
translation procedures have been proposed by translation scholars and practitioners, ranging
from the two-stage models to eight-stage ones, depending on the text type and purposes of
translation. In machine translation and computer-aided translation, the process is known as
technology-oriented translation procedure.
(a) Two-stage model
In human translation, the first type of translation procedure is a two-stage one, which includes
the stage of source text comprehension and the stage of target text formulation, as shown
below:
Figure 2.1 A two-stage model for human translation
Figure 2.1 is a model for human translators with the ability of comprehension. As a computer-
aided translation system does not have the ability of comprehension, it cannot model after
human translation with this two-stage model. It can, however, work on a two-stage translation
with the use of its system dictionary, particularly in the case of a language-pair-specific system,
as in Figure 2.2:
Figure 2.2 A two-stage dictionary-based language-pair-specific model 34
Another two-stage model of computer-aided translation is a terminology-based system, as
shown in Figure 2.3:
Figure 2.3 A two-stage terminology-based CAT system
(b) Three-stage models
The second type of translation procedure is a three-stage model. This section covers five
variations of this model proposed by Eugene Nida and Charles Taber (1969), Wolfram Wilss
(1982), Roger Bell (1991), Basil Hatim and Ian Mason, and Jean Delisle (1988) respectively. A
three-stage example-based computer-aided translation system is shown to illustrate the
simulation of human translation by computer-aided translation.
(i) model by eugene nida and charles taber
The first model of a three-stage translation procedure involving the three phases of analysis,
transfer, and restructuring was proposed by Eugene Nida and Charles Taber ([1969] 1982:
104). They intended to apply elements of Chomsky’s transformational grammar to provide
Bible translators with some guidelines when they translate ancient source texts into modern
target texts, which are drastically different in languages and structures. Nida and Taber describe
this three-stage model as a translation procedure in which
the translator first analyses the message of the source language into its simplest and
structurally clearest forms, transfers it at this level, and then restructures it to the level
in the receptor language which is most appropriate for the audience which he intends
to reach. (Nida and Taber [1969] 1982: 484)
Analysis is described by these two scholars as ‘the set of procedures, including back transformation
and componential analysis, which aim at discovering the kernels underlying the source text and
the clearest understanding of the meaning, in preparation for the transfer’ (Nida and Taber [1969]
1982: 197). Transfer, on the other hand, is described as the second stage ‘in which the analysed
material is transferred in the mind of the translator from language A to language B’ (ibid.: 104).
Restructuring is the final stage in which the results of the transfer process are transformed into a
‘stylistic form appropriate to the receptor language and to the intended receptors’.
S. Chan Computer-aided translation35
In short, analysis, the first stage, is to analyse the source text, transfer, the second stage, is to
transfer the meaning, and restructuring, the final stage, is to produce the target text. Their
model is shown in Figure 2.4.
Figure 2.4 Three-stage model by Nida and Taber (1964)
(ii) model by wolfram wilss
The second three-stage model was proposed by Wolfram Wilss (1982) who regards translation
procedure as a linguistic process of decoding, transfer and encoding. His model is shown in
Figure 2.5.
Figure 2.5 Three-stage model by Wolfram Wilss (1982)
(iii) model by roger bell
Another three-stage model of note is by Roger Bell whose translation procedure framework is
divided into three phases: the first phase is source text interpretation and analysis, the second,
translation process, and the third, text reformulation (see Figure 2.6). The last phase takes into
consideration three factors: writer’s intention, reader’s expectation, and the target language
norms (Bell 1991). 36
Figure 2.6 Model of Roger Bell
(iv) model by basil hatim and ian mason
This model, proposed by Basil Hatim and Ian Mason, is a more sophisticated three-stage
model, which involves the three steps of source text comprehension, transfer of meaning, and
target text assessment (see Figure 2.7). At the source text comprehension level, text parsing,
specialized knowledge, and intended meaning are examined. At the meaning transfer stage,
consideration is given to the lexical meaning, grammatical meaning, and rhetorical meaning.
At the target text assessment level, attention is paid to text readability, target language
conventions, and the adequacy of purpose.
Figure 2.7 A three-stage model by Basil Hatim and Ian Mason
(v) model of jean delisle
The fourth model of a three-stage translation procedure was proposed by Jean Delisle (1988:
53−69) (see Figure 2.8). Deslisle believes that there are three stages in the development of a
translation equivalence: comprehension, reformulation, and verification: ‘comprehension is
S. Chan Computer-aided translation37
based on decoding linguistic signs and grasping meaning, reformulation is a matter of reasoning
by analogy and re-wording concepts, and verification involves back-interpreting and choosing
a solution’ (1988: 53).Parallel to human translation, a three-stage model in computer-aided translation is the
example-based system. The input text goes through the translation memory database and
glossary database to generate fuzzy matches and translations of terms before getting the target
text. The procedure of an example-based computer-aided translation system is shown in
Figure 2.9.
Figure 2.8 A three-stage model of Jean Delisle
Figure 2.9 Three-stage example-based computer-aided translation model
(c) Four-stage model
The third type of translation procedure is a four-stage one. A typical example is given by
George Steiner ([1975] 1992) who believes that the four stages of translation procedure are:
knowledge of the author’s times, familiarization with author’s sphere of sensibility, original
text decoding, and target text encoding (see Figure 2.10). 38
Figure 2.10 Model of George Steiner (1975)
For computer-aided translation, a four-stage model is exemplified by webpage translation
provided by Yaxin. The first stage is to input the Chinese webpage, the second stage is to
process the webpage with the multilingual maintenance platform, the third stage is to process
it with the terminology database, and the final stage is to generate a bilingual webpage. The
Yaxin translation procedure is shown in Figure 2.11.
Figure 2.11 Yaxin’s four-stage procedure
(d) Five-stage model
The fourth type of translation procedure is a five-stage one, as proposed by Omar Sheikh Al-
Shabab (1996: 52) (see Figure 2.12). The first stage is to edit the source text, the second,
interpret the source text, the third, interpret it in a new language, the fourth, formulate the
translated text, and the fifth, edit the formulation.In computer-aided translation, a five-stage model is normally practised. At the first stage, the
Initiating Stage, tasks such as setting computer specifications, logging in a system, creating a
profile, and creating a project file are performed. At the second stage, the Data Preparation
Stage, the tasks involve data collection, data creation, and the creation of terminology and
translation memory databases. At the third stage, the Data Processing Stage, the tasks include
data analysis, the use of system and non-system dictionaries, the use of concordancers, doing
S. Chan Computer-aided translation39
Figure 2.12 Model of Omar Sheikh Al-Shabab
pre-translation, data processing by translation by computer-aided translation systems with
human intervention, or by machine translation systems without human intervention, or data
processing by localization systems. At the fourth stage, the Data Editing Stage, the work is
divided into two types. One type is data editing for computer-aided translation systems, which
is about interactive editing, the editing environments, matching, and methods used in
computer-aided translation. Another type is data editing for computer translation systems,
which is about post-editing and the methods used in human translation. At the last or fifth
stage, the Finalizing Stage, the work is mainly updating databases. The fives stages in computer-aided translation are illustrated in Figure 2.13.
Figure 2.13 Five-stage technology-oriented translation procedure model
It can be seen that though there are both five-stage models in human translation and computer-
aided translation and the tasks involved are different, the concept of simulativity is at work at
almost all stages.
(e) Eight-stage model
The fifth type of translation procedure is an eight-stage one, as proposed by Robert Bly (1983). 40
Robert Bly, who is a poet, suggests an eight-stage procedure for the translation of poetry: (a)
set down a literal version; (b) find out the meaning of the poem; (c) make it sound like English;
(d) make it sound like American; (e) catch the mood of the poem; (f) pay attention to sound;
(g) ask a native speaker to go over the version; and (h) make a final draft with some adjustments
(see Figure 2.14).
Figure 2.14 Model by Robert Bly (1983)
In computer-aided translation, there is no eight-stage model. But other than the five-stage
model, there is also a seven-stage model, which is shown in Figure 2.15.
Figure 2.15 Seven-stage computer-aided translation procedure
The seven stages of computer-aided translation go from sample text collection to termbase
creation, translation memory database creation, source text selection, data retrieval, source text
translation and finally data updating.
S. Chan Computer-aided translation41
All in all, we can say that when compared to human translation, computer-aided translation
is simulative, following some of the stages in human translation.
Emulativity
There are obviously some functions which are performable by a computer-aided translation
system, but not by a human translation. This is how technology can emulate human
translation. Computer-aided translation, with the help of machine translation, simulates
human translation, and it also emulates human translation in a number of areas of computer-
aided translation, some of which are mentioned below.
Alt-tag translation
This function of machine translation allows the user to understand the meaning of text
embedded within images (Joy 2002). The images on a web site are created by IMG tag (inline
image graphic tag), and the text that provides an alternative message to viewers who cannot see
the graphics is known as ALT-tag, which stands for ‘alternative text’. Adding an appropriate
ALT-tag to every image within one’s web site will make a huge difference to its accessibility.
As translators, our concern is the translation of the alternative text, as images are not to be
translated anyway.
Chatroom translation
Machine translation has the function to translate the contents of a chatroom, known as ‘chat
translation’ or ‘chatroom translation’. Chat translation systems are commercially available for
the translation of the contents of the Chatroom on the computer. As a chat is part of
conversational discourse, all the theoretical and practical issues relating to conversational
discourse can be applied to the study of chat translation. It should be noted that this kind of
online jargon and addressivity are drastically different from what we have in other modes of
communication. The function of Chatroom is available in some systems, such as Fluency, as one of the
resources. This function has to be purchased and enabled in the Fluency Chat Server to allow
clients to be connected to this closed system for internal communications. For standalone
version users, the function of Chat will be provided by Fluency Chat Server provided by its
company, the Western Standard (Western Standard 2011: 39).
Clipboard translation
This is to copy a text to the clipboard from any Windows application for a machine translation
system to translate the clipboard text, and the translated text can then be pasted in the original
or any other location. One of the systems that translate clipboards is Atlas.
Conversion between metric and British systems
A function that can be easily handled by machine translation but not so easily by human
translation is the conversion of weight, volume, length, or temperature from metric to British
or vice versa. Fluency, for example, can do the metric/British conversion the target text box
with the converted units. 42
Currency conversion
Some computer-aided translation systems can do currency conversion. With the use of
Currency Converter, a function in Fluency, and access to the Internet to get the currency
conversion rates, systems can convert a currency in a country into the local country currency.
The number of currencies that can be handled by a system is relatively large. Fluency, for
example, supports the conversion of currencies of around 220 countries. The conversion of
multiple currencies is also supported.
Email translation
This refers to the translation of emails by a machine translation system (Matsuda and Kumai
1999; Rooke 1985: 105−115). The first online and real-time email translation was made in
1994 by the CompuServe service which provided translation service of emails from and to
English and French, German or Spanish. Email translation has since become a very important
part of daily communication and most web translation tools have email translators to translate
emails. As emails are usually conversational and often written in an informal or even
ungrammatical way, they are difficult for mechanical processing (Fais and Ogura 2001; Han,
Gates and Levin 2006). One of the systems that translates emails is Atlas.
Foreign language translation
One of the most important purposes of using translation software is to translate a source text
the language of which is unfamiliar to the user so as to explain its contents in a language
familiar to the user. It is found that a majority of the commercial machine translation systems
are for translation among Indo-European languages or major languages with a large number of
speakers or users. Software for translation between major languages and minor languages are
relatively small in number.
Gist translation
Another area where machine translation differs fundamentally from human translation is gist
translation, which refers to a translation output which expresses only a condensed version of
the source text message. This type of rough translation is to get some essential information
about what is in the text for a user to decide whether to translate it in full or not to serve some
specific purposes.
Highlight and translate
This function allows the user to highlight a part of the text and translate it into the designated
language. The highlighted text is translated on its own without affecting the rest of the text.
Instant transliteration
This refers to a function of machine translation which can transliterate the words of a text with
a certain romanization system. In the case of Chinese, the Hanyu Pinyin Romanization system
for simplified characters is used in mainland China, while the Wade-Giles Romanization
system for traditional characters is used in Taiwan.
S. Chan Computer-aided translation43
Mouse translation
This is to translate sentences on a web page or on applications by simply clicking the mouse.
Systems that provide mouse translation include Atlas.
Online translation
This is the translation of a text by an online machine translation system which is available at all
times on demand from users. With the use of online translation service, the functions of
information assimilation, message dissemination, language communication, translation
entertainment, and language learning can be achieved.
Pre-translation
Machine translation is taken to be pre-translation in two respects. The first is as a kind of
preparatory work on the texts to be translated, including the checking of spelling, the
compilation of dictionaries, and the adjustment of text format. The second is taken to be a draft
translation of the source text which can be further revised by a human translator.
Sentence translation
Unlike human translation which works at the textual level, machine translation is sentential
translation. In other words, machine translation is a sentence-by-sentence translation. This
type of translation facilitates the work of post-editing and methods which are frequently used
in translating sentences in translation practice to produce effective translations can be used to
produce good translations from machine translation systems.
Web translation
This refers to the translation of information on a web page from one language into another.
Web-translation tools are a type of translation tools which translate information on a web page
from one language into another. They serve three functions: (1) as an assimilation tool to
transmit information to the user; (2) as a dissemination tool to make messages comprehensible;
and (3) as a communication tool to enable communication between people with different
language backgrounds.
Productivity
As translation technology is a field of entrepreneurial humanities, productivity is of great
importance. Productivity in computer-aided translation is achieved through the use of
technology, collective translation, recycling translations, reusing translations, professional
competence, profit-seeking, labour-saving, and cost-saving.
Using technology to increase productivity
The use of technology to increase productivity needs no explanation. As early as 1980, when
Martin Kay discussed the proper place of men and machines in language translation, he said: 44
Translation is a fine and exacting art, but there is much about it that is mechanical
and routine and, if this were given over to a machine, the productivity of the translator
would not only be magnified but his work would become more rewarding, more
exciting, more human.(Kay 1980: 1)
All computer-aided translation systems aim to increase translation productivity. In terms of the
means of production, all translation nowadays is computer-aided translation as virtually no one
could translate without using a computer.
Collective translation to increase productivity
Gone are the days when bilingual competence, pen and paper, and printed dictionaries made
a translator. Gone are the days when a single translator did a long translation project all by
himself. It is true that in the past, translation was mainly done singly and individually.
Translation was also done in a leisurely manner. At present, translation is done largely through
team work linked by a server-based computer-aided translation system. In other words,
translation is done in a collective manner.
Recycling translations to increase productivity
To recycle a translation in computer-aided translation is to use exact matches automatically
extracted from a translation memory database. To increase productivity, the practice of
recycling translations is followed in computer-aided translation. Networked computer-aided
translation systems are used to store centralized translation data, which are created by and
distributed among translators. As this is the case, translators do not have to produce their own
translations. They can simply draw from and make use of the translations stored in the bilingual
database to form their translation of the source text. Translation is therefore produced by
selection.
Reusing translations to increase productivity
To reuse a translation in computer-aided translation is to appropriate terms and expressions
stored in a term database and translation memory database. It should be noted that while in
literary translation, translators produce translations in a creative manner, translators in practical
translation reuse and recycle translations as the original texts are often repetitive. In the present
age, over 90 per cent of translation work is in the area of practical translation. Computer-aided
translation is ideal for the translation of repetitive practical writings. Translators do not have to
translate the sentences they have translated before. The more they translate, the less they have
to translate. Computer-aided translation therefore reduces the amount a translator needs to
translate by eliminating duplicate work. Some systems, such as Across, allow the user to
automatically reuse existing translations from the trans
lation memory. It can be seen that
‘reduce, reuse, recycle’ are the three effective ways of increasing profitability (de Ilarraza,
Mayor and Sarasola 2000).
S. Chan Computer-aided translation45
Professional competence to increase productivity
Translators have to work with the help of translation technology. The use of computer-aided
translation tools has actually been extended to almost every type of translation work. Computer-
aided translation tools are aimed at supporting translators and not at replacing them. They
make sure that translation quality is maintained as ‘all output is human input’. As far as the use
of tools is concerned, professional translation is technological. In the past, translators used only
printed dictionaries and references. Nowadays, translators use electronic concordancers, speech
technology, online terminology systems, and automatic checkers. Translation is about the use
of a workbench or workstation in translation work. Translation competence or knowledge and skills in languages are not enough today. It is
more realistic to talk about professional competence, which includes linguistic competence,
cultural competence, translation competence, translator competence, and technological
competence. Professional competence is important for translators as it affects their career
development. A remark made by Timothy Hunt is worth noting: ‘Computers will never
replace translators, but translators who use computers will replace translators who don’t’ (Sofer
2009: 88). What has happened in the field of translation technology shows that Hunt’s remark
may not be far off the mark. In the 1980s, very few people had any ideas about translation
technology or computer-aided translation. Now, SDL alone has more than 180,000 computer-
aided translators. The total number of computer-aided translators in the world is likely to be
several times higher than the SDL translators.
Profit-seeking to increase productivity
Translation is in part vocational, in part academic. In the training of translators, there are
courses on translation skills to foster their professionalism, and there are courses on translation
theories to enhance their academic knowledge. But there are very few courses on translation
as a business or as an industry. It should be noted that translation in recent decades has
increasingly become a field of entrepreneurial humanities as a result of the creation of the
project management function in computer-aided translation systems. This means translation is
now a field of humanities which is entrepreneurial in nature. Translation as a commercial
activity has to increase productivity to make more profits.
Labour-saving to increase productivity
Computer-aided translation systems help to increase productivity and profits through labour-
saving, eliminating repetitive translation tasks. Through reusing past translations, an enormous
amount of labour is saved. Computer-aided translation tools support translators by freeing
them from boring work and letting them concentrate on what they can do best over machines,
i.e. handling semantics and pragmatics. Generally, this leads to a broader acceptance by
translators. The role of a translator, therefore, has changed drastically in the modern age of
digital communication. Rather than simply translating the document, a computer-aided
translator has to engage in other types of work, such as authoring, pre-editing, interactive
editing, post-editing, term database management, translation memory database management,
text alignment and manual alignment verification. It is estimated that with the use of translation
technology, the work that was originally borne by six translators can be taken up by just one. 46
Cost-saving to increase productivity
Computer-aided translation is also cost-saving. It helps to keep the overhead cost down as
what has been translated needs not to be translated again. It helps to improve budget planning. Other issues relating to cost should also be taken into account. First, the actual cost of the
tool and its periodic upgrades. Second, the licensing policy of the system, which is about the
ease of transferring licences between computers or servers, the incurring of extra charges for
client licences, the lending of licences to one’s vendors, freelances, and the eligibility for free
upgrades. Third, the cost that is required for support, maintenance, or training. Fourth, the
affordability of the system for one’s translators. Fifth, the user-friendliness of the system to
one’s computer technicians and translators, which affects the cost of production.
Compatibility
The concept of compatibility in translation technology must be considered in terms of file
formats, operating systems, intersystem formats, translation memory databases, terminology
databases, and the languages supported by different systems.
Compatibility of file formats
One of the most important concepts in translation technology is the type of data that needs to
be processed, which is indicated by its format, being shown by one or several letters at the end
of a filename. Filename extensions usually follow a period (dot) and indicate the type of
information stored in the file. A look at some of the common file types and their file extensions
shows that in translation technology, text translation is but one type of data processing, though
it is the most popular one. There are two major types of formats: general documentation types and software development
types.
(I) General documentation types
(1) Text files
All computer-aided translation systems which use Microsoft Word as text editor can process all
formats recognized by Microsoft Word. Throughout the development of translation
technology, most computer-aided translation systems process text files (.txt). For Microsoft
Word 2000−2003, text files were saved and stored as .doc (document text file/word processing
file); for Microsoft Word 2007−2011, documents were saved and stored as .docx (Document
text file (Microsoft Office 2007)), .dotx (Microsoft Word 2007 Document Template). Other
types of text files include .txt (Text files), .txml (WordFast files), and .rtf (Rich Text files). All automatic and interactive translation systems can process text files, provided the text
processing system has been installed in the computer before processing begins. Some of the
computer-aided translation systems which can only translate text files include: Across,
AidTransStudio, Anaphraseus (formerly known as OpenWordfast), AnyMem (.docx or higher),
Araya, Autshumato Integrated Translation Environment (ITE), CafeTran, Déjà Vu, Esperantilo,
Fluency, Fusion, OmegaT, Wordfast, and WordFisher. Computer-aided translation systems
which can translate text files as well as other formats include CafeTran, Esperantilo, Felix,
Fortis, GlobalSight, Google Translator Toolkit, Heartsome Translation Suite, Huajian IAT,
Lingo, Lingotek, MadCap Lingo, MemoQ, MemOrg, MemSource, MetaTexis, MultiTrans,
S. Chan Computer-aided translation47
OmegaT+, Pootle, SDL-Trados, Similis, Snowman, Swordfish, TM-database, Transit,
Wordfast, XTM, and Yaxin.
(2) Web-page files
HyperText Markup Language (HTML) is a markup language that web browsers use to
interpret and compose text, images and other material into visual or audible web pages. HTML
defines the structure and layout of a web page or document by using a variety of tags and
attributes. HTML documents are stored as .asp (Active Server Pages), .aspx (Active Server Page
Extended), .htm (Hypertext Markup Language), .html (Hypertext Markup Language Files),
.php (originally: Personal Home Page; now: Hypertext Preprocessor), .jsp (JavaServer Pages),
.sgml (Standard Generalized Markup Language File), .xml (Extensible Markup Language file),
.xsl (Extensible Stylesheet Language file) files format, which were available since late 1991.
Due to the popularity of web pages, web translation has been an important part of automatic
and interactive translation systems. Many systems provide comprehensive support for the
localization of HTML-based document types. Web page localization is interchangeable with
web translation or web localization. Systems that handle HTML include Across, AidTransStudio, Alchemy Publisher, Araya,
Atlas, CafeTran, CatsCradle, Déjà Vu, Felix, Fluency, Fortis, GlobalSight, Google Translator
Toolkit, Heartsome Translation Suite, Huajian IAT, Lingo, Lingotek, LogiTerm, MemoQ,
MemOrg, MetaTexis, MultiTrans, Okapi Framework, OmegaT, OmegaT+, Open Language
Tools, Pootle, SDL-Trados, Similis, Snowman, Swordfish, TM-database, TransSearch, Transit,
Transolution, and XTM.
(3) PDF files
Portable Document Format (PDF) (.pdf ) is a universally accepted file interchange format
developed by Adobe in the 1990s. The software that allows document files to be transferred
between different types of computers is Adobe Acrobat. A PDF file can be opened by the
document format, which might require editing to make the file look more like the original, or
can be converted to an rtf file for data processing by a computer-aided translation system. Systems that can translate Adobe PDF files and save them as Microsoft Word documents
include Alchemy Publisher, CafeTran, Lingo, Similis, and Snowman.
(4) Microsoft Office PowerPoint files
Microsoft PowerPoint is a presentation program developed to enable users to create anything
from basic slide shows to complex presentations, which are comprised of slides that may
contain text, images, and other media. Versions of Microsoft Office PowerPoint include
Microsoft PowerPoint 2000–2003, .ppt (General file extension), .pps (PowerPoint Slideshow), .pot
(PowerPoint template); Microsoft PowerPoint 2007/2011, which are saved as .pptx (Microsoft
PowerPoint Open XML Document), .ppsx (PowerPoint Open XML Slide Show), .potx
(PowerPoint Open XML Presentation Template), and .ppsm (PowerPoint 2007 Macro-
enabled Slide Show). Systems that can handle Powerpoint files include Across, AidTransStudio, Alchemy
Publisher, CafeTran, Déjà Vu, Felix, Fluency, Fusion, GlobalSight, Lingotek, LogiTerm,
MadCap Lingo, MemoQ, MemSource, MetaTexis, SDL-Trados, Swordfish, TM-database,
Transit, Wordfast, XTM, and Yaxin. 48
(5) Microsoft Excel files
Different versions of Microsoft Excel include Microsoft Excel 2000–2003 .xls (spreadsheet),
.xlt (template); Microsoft Excel 2007: .xlsx (Microsoft Excel Open XML Document), .xltx
(Excel 2007 spreadsheet template), .xlsm (Excel 2007 macro-enabled spreadsheet) The computer-aided translation systems that can translate Excel files include Across,
AidTransStudio, Déjà Vu, Felix, GlobalSight, Lingotek, LogiTerm, and MemoQ, MemOrg,
MetaTexis, MultiTrans, Snowman, Wordfast, and Yaxin.
(6) Microsoft Access files
One of the computer-aided translation systems which can handle Access with .accdb (Access
2007–2010) file extension is Déjà Vu.
(7) Image files
The processing of image data, mainly graphics and pictures, is important in computer-aided
translation. The data is stored as .bmp (bitmap image file), .jpg (Joint Photographic Experts
Group), and .gif (Graphics Interchange Format). One of the computer-aided translation systems
that is capable of handling images is CafeTran.
(8) Subtitle files
One of the most popular subtitle files on the market is .srt (SubRip Text). OmegaT is one of
the computer-aided systems that supports subtitle files.
(9) Adobe InDesign files
Adobe InDesign is desktop publishing software. It can be translated without the need of any
third party software by Alchemy Publisher and AnyMem. For Alchemy Publisher, the .indd file
must be exported to an .inx format before it can be processed. Other computer-aided translation
systems that support Adobe Indesign files include Across, Déjà Vu, Fortis, GlobalSight,
Heartsome Translation Suite, Okapi Framework, MemoQ, MultiTrans, SDL-Trados,
Swordfish, Transit, and XTM.
(10) Adobe FrameMaker Files
Adobe FrameMaker is an authoring and publishing solution for XML. FrameMaker files, .fm,
.mif and .book, can be opened directly by a system if it is installed with Adobe FrameMaker. Computer-aided translation systems that can translate Adobe FrameMaker files include
Across, Alchemy Publisher (which requires a PPF created by Adobe FrameMaker before
translating it. Alchemy Publisher supports FrameMaker 5.0, 6.0, 7.0, 8.0, 9.0, FrameBuilder
4.0, and FrameMaker + sgml), CafeTran, Déjà Vu, Fortis, GlobalSight, Heartsome Translation
Suite, Lingo, Lingotek, MadCap Lingo, MemoQ, MetaTexis, MultiTrans, SDL-Trados,
Swordfish, Transit, Wordfast, and XTM.
(11) Adobe PageMaker files
Systems that support Adobe PageMaker 6.5 and 7 files include Déjà Vu, GlobalSight,
MetaTexis, and Transit.
(12) AutoCAD files
AutoCAD, developed and first released by Autodesk, Inc. in December 1982, is a software
application for computer-aided design (CAD) and drafting which supports both 2D and 3D
S. Chan Computer-aided translation49
formats. This software is now used internationally as the most popular drafting tool for a range
of industries, most commonly in architecture and engineering. Computer-aided translation systems that support AutoCad are CafeTran, Transit, and
TranslateCAD.
(13) DTP tagged text files
DTP stands for Desktop Publishing. A popular desktop publishing system is QuarkXPress. Systems that support desktop publishing include Across, CafeTran, Déjà Vu, Fortis,
GlobalSight, MetaTexis, MultiTrans, SDL-Trados, and Transit.
(14) Localization files
Localization files include files with the standardized format for localization .xliff (XML
Localization Interchange File Format) files, .ttx (XML font file format) files, and .po (Portable
Object). Computer-aided translation systems which process XLIFF files include Across Language
Server, Araya, CafeTran, Esperantilo, Fluency, Fortis, GTranslator, Heartsome Translation
Suite, MadCap Lingo, Lingotek, MemoQ, Okapi Framework, Open Language Tools, Poedit,
Pootle, Swordfish, Transolution, Virtaal, and XTM.
(II) Software development types
(1) Java Properties files
Java Properties files are simple text files that are used in Java applications. The file extension of
Java Properties file is .properties. Computer-aided translation systems that support Java Properties File include Déjà Vu,
Fortis, Heartsome Translation Suite, Lingotek, Okapi Framework, OmegaT+, Open Language
Tools, Pootle, Swordfish, and XTM.
(2) OpenOffice.org/StarOffice
StarOffice of the Star Division was a German company that ran from 1984 to 1999. It was
succeeded by OpenOffice.org, an open-sourced version of StarOffice owned by Sun
Microsystems (1999–2009) and by Oracle Corporation (2010–2011), which ran from
1999−2011.Currently it is Apache OpenOffice. The format of OpenOffice is .odf (Open
Document Format). Computer-aided translation systems which process this type of file include AidTransStudio,
Anaphraseus, CafeTran, Déjà Vu, Heartsome Translation Suite, Lingotek, OmegaT,
OmegaT+, Open Language Tools, Pootle, Similis, Swordfish, Transolution, and XTM.
(3) Windows resource files
These are simple script files containing startup instructions for an application program, usually
a text file containing commands that are compiled into binary files such as .exe and .dll. File
extensions include .rc (Record Columnar File), .resx (NET XML Resource Template).
Computer-aided translation systems that process this type of files include Across, Déjà Vu,
Fortis, Lingotek, MetaTexis, and Okapi Framework. 50
Compatibility of operating systems
One of the most important factors which determined the course of development of computer-
aided translation systems is their compatibility with the current operating systems on the
market. It is therefore essential to examine the major operating systems running from the
beginning of computer-aided translation in 1988 to the present, which include, among others,
the Windows of Microsoft and the OS of Macintosh.
Microsoft Operating Systems
In the world of computing, Microsoft Windows has been the dominant operating system.
From the 1981 to the 1995, the x86-based MS-DOS (Microsoft Disk Operating System) was
the most commonly used system, especially for IBM PC compatible personal computers.
Trados’s Translator’s Workbench II, developed in 1992, is a typical example of a computer-
aided translation system working on DOS.DOS was supplemented by Microsoft Windows 1.0, a 16-bit graphical operating
environment, released on 20 November 1985 (Windows 2012). In November 1987, Windows
1.0 was succeeded by Windows 2.0, which existed till 2001. Déjà Vu 1.0, released in 1993,
was one of the systems compatible with Windows 2.0. Windows 2.0 was supplemented by
Windows 286 and Windows 386. Then came Windows 3.0, succeeding Windows 2.1x. Windows 3.0, with a graphical
environment, is the third major release of Microsoft Windows, and was released on 22 May
1990. With a significantly revamped user interface and technical improvements, Windows 3
became the first widely successful version of Windows and a rival to Apple Macintosh and the
Commodore Amiga on the GUI front. It was followed by Windows 3.1x. During its lifespan
from 1992−2001, Windows 3.1x introduced various enhancements to the still MS-DOS-
based platform, including improved system stability, expanded support for multimedia,
TrueType fonts, and workgroup networking. Trados’s Translator’s Workbench, released in
1994, was a system that was adaptable to Windows 3.1x. Except for Windows and DOS, OS/2 is also one of the operation systems that support
computer-aided translation systems, especially in late 1980s and early 1990s.
Apple Operating Systems
Mac OS (1984−2000) and OS X (2001−) are two series of graphical user interface-based
operating systems developed by Apple Inc. for their Macintosh line of computer systems. Mac
OS was first introduced in 1984 with the original Macintosh and this series was ended in 2000.
OS X, first released in March 2001, is a series of Unix-based graphical interface operating
systems. Both series share a general interface design, but have very different internal architectures. Only one computer-aided translation system, AppleTrans, is designed for OS X. Its initial
released was announced in February 2004 and the latest updated version was version 1.2(v38)
released in September 2006, which runs on Mac OS X 10.3 or later. Another computer-aided translation system, Wordfast Classic was released to upgrade its
support of the latest text processor running on Mac OS X, such as Wordfast Classic 6.0, which
is compatible for MS Word 2011 for Mac. Other computer-aided translation systems that can run on Mac OS or OS X are cross-
platform software, rather than software developed particularly for Mac. Examples are Java-
based applications, such as Autshumato, Heartsome, OmegaT, Open Language Tools and
S. Chan Computer-aided translation51
Swordfish. Besides, all cloud-based systems can support Mac OS and OS X, including
Wordbee, XTM Cloud, Google Translator’s Toolkit, Lingotek Collaborative Translation
Platform, MemSource Cloud, and WebWordSystem.OS/2 is a series of computer operating systems, initially created by Microsoft and IBM, then
later developed by IBM exclusively. The name stands for ‘Operating System/2’. Until 1992, the early computer-aided translation systems ran either on MS-DOS or OS/2.
For example, IBM Translation Manager/2 (TM/2) was released in 1992 and run on OS/2.
ALPS’s translation tool also ran on OS/2. But OS/2 had a much smaller market share compared
with Windows in early 1990s. Computer-aided translation system developers therefore
gradually shifted from OS/2 and MS-DOS to Windows or discontinued the development of
OS/2 and MS-DOS compatible computer-aided translation systems. By the end of the 1990s,
most computer-aided translation systems mainly ran on Windows, although some developers
offered operating-system customization services upon request. OS/2 4.52 was released in
December 2001. IBM ended its support to OS/2 on 31 December 2006.
Compatibility of databases
Compatibility of translation memory databases
TMX (Translation Memory eXchange), created in 1998, is widely used as an interchange
format between different translation memory formats. TMX files are XML (eXtensible Markup
Language) files whose format was originally developed and maintained by OSCAR (Open
Standards for Container/Content Allowing Re-use) of the Localization Industry Standards
Association. The latest official version of the TMX specification, version 1.4b, was released in
2005. In March 2011 LISA was declared insolvent; as a result its standards were moved under
the Creative Commons licence and the standards specification relocated. The technical
specification and a sample document of TMX can be found on the website of The Globalization
and Localization Association. TMX has been widely adopted and is supported by more than half of the current computer-
aided translation systems on the market. The total number of computer-aided translation
systems that can import and export translation memories in TMX format is 54, including
Across, Alchemy Publisher, Anaphraseus, AnyMem, Araya, ATLAS, Autshumato, CafeTran,
Crowdin, Déjà Vu, EsperantiloTM, Felix, Fluency, Fortis, Fusion, GE-CCT, GlobalSight,
Google Translator Toolkit, Heartsome, Huajian IAT, Lingotek, LogiTerm, LongRay CAT,
MadCap Lingo, MemoQ, MemSource, MetaTexis, MT2007, MultiTrans, OmegaT,
OmegaT+, Open Language Tools, OpenTM2, OpenTMS, PROMT, SDL Trados, Snowball,
Snowman, Swordfish, Systran, Text United, The Hongyahu, TM Database, Transit,
Translation Workspace, Transwhiz, TraTool, Webwordsystem, Wordbee Translator, Wordfast
Classic and Wordfast Pro, XTM, Yaxin CAT, and 翻訳ブレイン (Translation Brain).
Compatibility of terminology databases
Compatibility of terminology databases is best illustrated by TermBase eXchange (TBX),
which covers a family of formats for representing the information in a high-end termbase in a
neutral intermediate format in a manner compliant with the Terminological Markup
Framework (TMF) (Melby 2012: 19−21). Termbase Exchange is an international standard as well as an industry standard. The industry
standard version differs from the ISO standard only by having different title pages. Localization 52
Industry Standards Association, the host organization for OSCAR that developed Termbase
Exchange, was dissolved in February 2011. In September 2011, the European
Telecommunications Standards Institute (ETSI) took over maintenance of the OSCAR
standards. ETSI has established an interest group for translation/localization standards and a
liaison relationship with the International Organization for Standardization (ISO) so that TBX
can continue to be published as both an ISO standard and an industry standard. There are many types of termbases in use, ranging from huge termbases operated by
governments, to medium-size termbases maintained by corporations and non-governmental
organizations, to smaller termbases maintained by translation service providers and individual
translators. The problem addressed by the designers of term exchange was that existing
termbases are generally not interoperable. They are based on different data models that use a
variety of data categories. And even if the same data category is used for a particular piece of
information, the name of the data category and the values allowed for the data category may
be different.
Compatibility of rules
segmentation rules exchange
Segmentation Rules eXchange (SRX) is an XML-based standard that was maintained by
Localization Industry Standards Association, until it became insolvent in 2011 and then this
standard is now maintained by the Globalization and Localization Association (GALA). Segmentation Rules eXchange provides a common way to describe how to segment text for
translation and other language-related processes. It was created when it was realized that
translation memory exchange leverage is lower than expected in certain instances due to
differences in how tools segment text. Segmentation Rules eXchange is intended to enhance
the translation memory exchange so that translation memory data that is exchanged between
applications can be used more effectively. Having the segmentation rules that were used when
a translation memory was created will increase the leverage that can be achieved when
deploying the translation memory data.
Compatibility with the languages supported
As computer-aided translation systems cannot identify languages, language compatibility is
therefore an important concept in translation technology. There are a large number of
languages and sub-languages in the world, totalling 6,912. But the number of major languages
computers can process is relatively small. It is therefore important to know whether the
languages that require machine processing are supported by a system or not. With the aid of unicode, most of the languages in the world are supported in popular
computer-aided translation systems. Unicode is a computing industry standard for the consistent
encoding, representation and handling of text expressed in most of the world’s writing systems. There are basically two major types of language and sub-language codes. Some systems, such
as OmegaT and XTM, use letters for language codes (2 or 3 letters) and language-and-region
codes (2+2 letters), which can be selected from a drop-down list. OmegaT follows the ISO
639 Code Tables in preparing its code list. French for example, is coded fr with the language-
and region code for French (Canada) as fr-CA. The following is a list of languages supported by Wordfast Classics and XTM, two of the
nine computer-aided translation systems chosen for analysis in this chapter.
S. Chan Computer-aided translation53
wordfast classic
Wordfast can be used to translate any of the languages supported by Microsoft Word. The
number of languages supported by Microsoft is 91, with a number of sub-languages for some
major languages.
[Afro-Asiatic] Arabic (Algeria), Arabic (Bahrain), Arabic (Egypt), Arabic (Iraq), Arabic (Jordan),
Arabic (Kuwait), Arabic (Lebanon), Arabic (Libya), Arabic (Morocco), Arabic (Oman), Arabic
(Qatar), Arabic (Saudi Arabia), Arabic (Syria), Arabic (Tunisia), Arabic (U.A.E.), Arabic
(Yemen), Hebrew, Maltese
[Altaic] Azeri (Cyrillic), Azeri (Latin), Japanese, Korean, Turkish
[Austro-Asiatic] Vietnamese
[Austronesian] Indonesian, Malay (Brunei Darussalam), Malaysian
[Basque] Basque
[Dravidian] Kannada, Malayalam, Tamil, Telugu
[Indo-European] Afrikaans, Albanian, Armenian, Assamese, Belarusian, Bengali, Bulgarian,
Byelorussian, Catalan, Croatian, Czech, Danish, Dutch, Dutch (Belgian), English (Australia),
English (Belize), English (Canadian), English (Caribbean), English (Ireland), English (Jamaica),
English (New Zealand), English (Philippines), English (South Africa), English (Trinidad),
English (U.K.), English (U.S.), English (Zimbabwe), Faroese, Farsi, French (Belgian), French
(Cameroon), French (Canadian), French (Congo), French (Cote d’Ivoire), French
(Luxembourg), French (Mali), French (Monaco), French (Reunion), French (Senegal), French
(West Indies), Frisian (Netherlands), Gaelic (Ireland), Gaelic (Scotland), Galician, German,
German (Austria), German (Liechtenstein), German (Luxembourg), Greek, Gujarati, Hindi,
Icelandic, Italian, Kashmiri, Konkani, Latvian, Lithuanian, Macedonian (FYRO), Marathi,
Nepali, Norwegian (Bokmol), Norwegian (Nynorsk), Oriya, Polish, Portuguese, Portuguese
(Brazil), Punjabi, Rhaeto-Romance, Romanian, Romanian (Moldova), Russian, Russian
(Moldova), Sanskrit, Serbian (Cyrillic), Serbian (Latin), Sindhi, Slovak, Slovenian, Sorbian,
Spanish (Argentina), Spanish (Bolivia), Spanish (Chile), Spanish (Colombia), Spanish (Costa
Rica), Spanish (Dominican Republic), Spanish (Ecuador), Spanish (El Salvador), Spanish
(Guatemala), Spanish (Honduras), Spanish (Nicaragua), Spanish (Panama), Spanish (Paraguay),
Spanish (Peru), Spanish (Puerto Rico), Spanish (Spain), Spanish (Traditional), Spanish
(Uruguay), Spanish (Venezuela), Swedish, Swedish (Finland), Swiss (French), Swiss (German),
Swiss (Italian), Tajik, Ukrainian, Urdu, Welsh
[Kartvelian] Georgian
[Niger-Congo] Sesotho, Swahili, Tsonga, Tswana, Venda, Xhosa, Zulu
[Sino-Tibetan] Burmese, Chinese, Chinese (Hong Kong SAR), Chinese (Macau SAR), Chinese
(Simplified), Chinese (Singapore), Chinese (Traditional), Manipuri, Tibetan 54
[Tai-Kadai] Laothian, Thai
[Turkic] Tatar, Turkmen, Uzbek (Cyrillic), Uzbek (Latin)
[Uralic] Estonian, Finnish, Hungarian, Sami Lappish
xtm
The languages available in XTM are 157, not including varieties within a single language.
These languages are as follows:
[Afro-Asiatic] Afar, Amharic, Arabic, Hausa, Hebrew, Maltese, Oromo, Somali, Sudanese
Arabic, Syriac, Tigrinya,
[Altaic] Azeri, Japanese, Kazakh, Korean, Mongolian, Turkish
[Austro-Asiatic] Khmer, Vietnamese
[Austronesian] Fijian, Indonesian, Javanese, Malagasy, Malay, Maori, Nauru, Samoan, Tagalog,
Tetum, Tonga
[Aymaran] Aymara
[Bantu] Kikongo
[Basque] Basque
[Constructed Language] Esperanto, Interlingua, Volapk
[Dravidian] Kannada, Malayalam, Tamil, Telugu
[English Creole] Bislama
[Eskimo-Aleut] Greenlandic, Inuktitut, Inupiak
[French Creole] Haitian Creole
[Hmong-Mien] Hmong
[Indo-European] Afrikaans, Armenian, Assamese, Asturian, Bengali, Bihari, Bosnian, Breton,
Bulgarian, Byelorussian, Catalan, Corsican, Croatian, Czech, Danish, Dari, Dhivehi, Dutch,
English, Faroese, Flemish, French, Frisian, Galician, German, Greek, Gujarati, Hindi, Icelandic,
Irish, Italian, Kashmiri, Konkani, Kurdish, Latin, Latvian, Lithuanian, Macedonian, Marathi,
Montenegrin, Nepali, Norwegian, Occitan, Oriya, Pashto, Persian, Polish, Portuguese,
Punjabi, Rhaeto-Romance, Romanian, Russian, Sanskrit, Sardinian, Scottish Gaelic, Serbian,
Sindhi, Singhalese, Slovak, Slovenian, Sorbian, Spanish, Swedish, Tajik, Ukrainian, Urdu,
Welsh, Yiddish
S. Chan Computer-aided translation55
[Kartvelian] Georgian
[Ngbandi-based Creole] Sango
[Niger-Congo] Chichewa, Fula, Igbo, Kinyarwanda, Kirundi, Kiswahili, Lingala, Ndebele,
Northern Sotho, Sesotho, Setswana, Shona, Siswati, Tsonga, Tswana, Twi, Wolof, Xhosa,
Yoruba, Zulu
[Northwest Caucasian] Abkhazian
[Quechuan] Quechua
[Romanian] Moldavian
[Sino-Tibetan] Bhutani, Burmese; Chinese, Tibetan
[Tai-Kadai] Laothian, Thai
[Tupi] Guarani
[Turkic] Bashkir, Kirghiz, Tarar, Turkmen, Uyghur, Uzbek
[Uralic] Estonian, Finnish, Hungarian
Several observations can be made from the languages supported by the current eleven systems.(1) The number of languages supported by language-specific systems is small as they need to
be supplied with language-specific dictionaries to function well. Yaxin is best for English−
Chinese translation, covering two languages, while most non-language-specific systems support
around or above 100 languages. (2) For the seven systems developed in Europe, the United Kingdom, and the United States,
which include Across, Déjà Vu, MemoQ, OmegaT, SDL Trados, Wordfast, and XTM, the
Indo-European languages take up around 51.89 per cent, while the proportion of the non-
Indo-European languages is 48.11 per cent. Table 2.1 shows the details:
Table 2.1 Statistics of languages supported by 7 CAT systems
Name of the
system Number of
languages
supported Number of
language families
supported Number and percentage of
Indo-European languages
Number and percentage of
non-Indo-European
languages
Across 121 1861 (50.41%) 60 (49.59%)
Déjà Vu 132 2166 (50%) 66 (50%)
MemoQ 102 1654 (52.94%) 48 (47.06%)
OmegaT 90 1448 (53.33%) 42 (46.67%)
SDL Trados 115 1862 (53.91%) 53 (46.09%)
Wordfast 91 1354 (59.34%) 37 (40.66%)
XTM 157 2668 (43.31%) 89 (56.69%) 56
Controllability
One of the main differences between human translation and computer-aided translation lies in
the degree of control over the source text. In human translation, there is no need, or rather it
is not the common practice, to control how and what the author should write. But in
computer-aided translation, control over the input text may not be inappropriate as the output
of an unedited or uncontrolled source language text is far from satisfactory (Adriaens and
Macken 1995: 123−141; Allen and Hogan 2000: 62−71; Arnold et al. 1994;
Hurst 1997:
59−70; Lehtola, Tenni and Bounsaythip 1998: 16−29; Mitamura 1999: 46−52; Murphy et al.
1998; Nyberg et al. 2003: 245−281; Ruffino 1985: 157−162). The concept of controllability is realized in computer-aided translation by the use of
controlled language and the method of pre-editing.
Controllability by the use of controlled language
An effective means of achieving controllability in translation technology is controlled language
(see Figure 2.16). The idea of controlled language was created, partly at least, as a result of the
problems with natural languages which are full of complexities, ambiguities, and robustness
(Nyberg et al. 2003: 245−281). A strong rationale for controlled language is that a varied
source text generates a poor target text, while a controlled source text produces a quality target
text. (Bernth 1999). Controlled language is therefore considered necessary (Caeyers 1997:
91−103; Hu 2005: 364−372). Controlled language, in brief, refers to a type of natural language developed for specific
domains with a clearly defined restriction on controlled lexicons, simplified grammars, and
style rules to reduce the ambiguity and complexity of a text so as to make it easier to be
understood by users and non-native speakers and processed by machine translation systems
(Chan 2004: 44; Lux and Dauphin 1996: 193−204). Control over the three stages of a translation procedure, which include the stage of inputting
a source text, the stage of transfer, and the stage of text generation, is generally regarded as a
safe guarantee of quality translation. Control of the source text is in the form of controlled
authoring, which makes the source text easier for computer processing (Allen 1999; Chan
2004: 44; van der Eijk and van Wees 1998: 65−70; Zydron 2003). The text produced is a
‘controlled language text’ (Melby 1995: 1). There is also control over the transfer stage. And
the output of a machine translation system is known as ‘controlled translation’ (Carl 2003:
16−24; Gough and Way 2004: 73−81; Rico and Torrejon 2004; Roturier 2004; Torrejón
2002: 107−116), which is alternatively known as a ‘controlled target language text’ (Chan
2004: 44). In short, a controlled text is easier to be processed by machine translation systems to
produce a quality output.
Goals and means of controlled language
Controlled language is used by both humans and computers. The goals of controlled language
are to make the source text easier to read and understand. These goals are to be achieved at the
lexical and sentential levels. At the lexical level, controlled language is about the removal of lexical ambiguity and the
reduction in homonymy, synonymy, and complexity. This is to be achieved by one-to-one
correspondence in the use and translation of words, known as one-word one-meaning. An
example is to use only the word ‘start’ but not similar words such as ‘begin’, ‘commence’,
S. Chan Computer-aided translation57
Figure 2.16 Controlled language
‘initiate’, and ‘originate’. The second method is to use the preferred language, such as American
English but not British English. The third method is to have a limited basic vocabulary
(Bjarnestam 2003; Chen and Wu 1999; Probst and Levin 2002: 157−167; Wasson 2000:
276−281), which can be illustrated by the use of a controlled vocabulary of 3,100 words in
aircraft-maintenance documentation at the European Association of Aerospace Industries
(AECMA) in 1980 (AECMA 1995).At the sentential level, controlled language is about the removal of syntactical ambiguity, the
simplification of sentence structures, limitations on sentence length, and constraints on voice,
tense, and other grammatical units. To do all these, there are a limited number of strictly
stipulated writing rules to follow. The European Association of Aerospace Industries had 57
writing rules. Short sentences are preferred over long and complex sentences. And there is also
a limit on the number of words in a sentence. For procedural text, there should be no more
than twenty words. For descriptive texts, the number is twenty-five. There are also rules
governing grammatical well-formedness (Loong 1989: 281−297), restricted syntax, and the use
of passive construction in procedural texts. At the suprasentential level, there is a limit of six
sentences in a paragraph, the maximum number of clauses in a sentence, and the use of separate
sentences for sequential steps in procedural texts. This means setting limits on the length of a sentence, such as setting the number of words
at twenty, using only the active voice, and expressing one instruction or idea by one sentence.
Controlled language checkers
Controlled language cannot be maintained manually; it relies on the use of different kinds of
checkers, which are systems to ensure that a text conforms to the rules of a particular controlled
language (Fouvry and Balkan 1996: 179−192). There is the automatic rewriting system, which
is specially developed for controlled language, rewriting texts automatically into controlled
language without changing the meaning of the sentences in the original in order to produce a
high-quality machine translation. There is the controlled language checker, which is software
that helps an author to determine whether a text conforms to the approved words and writing
rules of a particular controlled language. Checkers can also be divided into two types: in-house controlled language checker and
commercial controlled language checker. In-house controlled language checkers include the 58
PACE (Perkins Approved Clear English) of Perkins Engines Ltd, the Controlled English of
Alcatel Telecom, and the Boeing Simplified English Checker of the Boeing Company (Wojcik
and Holmback 1996: 22−31). For commercial controlled language checkers, there are a number
of popular systems. The LANTmaster Controlled Checker, for example, is a controlled
language checker developed by LANT in Belgium. It is based on work done for the METAL
(Mechanical Translation and Analysis of Languages) machine translation project. It is also based
on the experience of the Simplified English Grammar and Style Checker (SECC) project
(Adriaens 1994: 78–88; Adriaens and Macken 1995: 123−141). The MAXit Checker is another
controlled language software developed by Smart Communications Incorporation to analyse
technical texts written in controlled or simplified English with the use of more than 8,500
grammar rules and artificial intelligence to check the clarity, consistency, simplicity, and global
acceptance of the texts. The Carnegie Group also produced the ClearCheck, which performs
syntactic parsing to detect such grammatical problems as ambiguity (Andersen 1994: 227).
Advantages and disadvantages of controlled language
The advantages of controlled language translation are numerous, including high readability,
better comprehensibility, greater standardization, easier computer processing, greater
reusability, increased translatability, improved consistency, improved customer satisfaction,
improved competitiveness, greater cost reduction in global product support, and enhanced
communication in global management.There are a number of disadvantages in using controlled language, such as expensive system
construction, high maintenance cost, time-consuming authoring, and restrictive checking
process.
Controlled language in use
As the advantages of using controlled language outweigh its disadvantages, companies started
to use controlled language as early as the 1970s. Examples of business corporations which used
controlled languages include Caterpillar Fundamental English (CFE) of the Caterpillar
Incorporation in 1975 (Kamprath et al. 1998: 51−61; Lockwood 2000: 187−202), Smart
Controlled English of the Smart Communications Ltd in 1975, Douglas Aircraft Company in
1979, the European Association of Aerospace Industries (AECMA) in 1980, the KANT
Project at the Center for Machine Translation, Carnegie Mellon University in 1989 (Allen
1995; Carbonell et al. 1992: 225−235; Mitamura et al. 1994: 232−233; Mitamura and Nyberg
1995: 158−172; Mitamura et al. 2002: 244−247; Nyberg and Mitamura 1992: 1069−1073;
Nyberg et al. 1997; Nyberg et al. 1998: 1−7; Nyberg and Mitamura 2000: 192−195), the
PACE of Perkins Engines Ltd. in 1989, ScaniaSwedish in Sweden in 1995 (Almqvist and Hein
1996: 159−164; Hein 1997), General Motors in 1996, Ericsson English in Sweden in 2000,
Nortel Standard English in the United Kingdom in 2002, and Oce Technologies English in
Holland in 2002.
Controlled language in computer-aided translation systems
The concept of controlled language is realized in controlled authoring in computer-aided
translation systems. Authoring checking tools are used to check and improve the quality of the
source text. There is an automatic rewriting system which is usually used as a tool to realize
controlled authoring. One of the computer-aided translation systems that performs controlled
S. Chan Computer-aided translation59
authoring is Star Transit. This system provides automatic translation suggestions from the
translation memory database from a speedy search engine and it is an open system that can
integrate with many authoring systems.
Customizability
Customizability, etymologically speaking, is the ability to be customized. More specifically, it
refers to the ability of a computer or computer-aided translation system to adapt itself to the
needs of the user. Customizing a general-purpose machine translation system is an effective
way to improve MT quality.
Editorial customization
Pre-editing is in essence a process of customization. The customization of machine translation
systems, which is a much neglected area, is necessary and essential as most software on the
market are for general uses and not for specific domains. Practically, system customization can
be taken as part of the work of pre-editing as we pre-edit the words and expressions to facilitate
the production of quality translation.The degree of customization depends on the goals of translation, and the circumstances and
the type of text to be translated.
Language customization
It is true that there are many language combinations in computer-aided translation systems to
allow the user to choose any pair of source and target languages when creating a project, yet
many users only work with a limited set of source and target languages. XTM, a cloud-based
system, allows the user to set language combinations through the Data section. In the language
combinations section, the project administrator or user can reduce and customize the available
languages to be used, set the language combinations for the entire system and set specific
language combinations for individual customers (XTM International 2012: 15). Language customization in XTM, for example, can be conducted on the Customize tab
where there are three options for the user to modify and use language combinations. The first
option is ‘system default language combinations’, which is the full set of unmodified language
combinations. The second option is ‘system defaults with customized language combinations’,
which is the full set of language combinations in which the user may have customized some
parameters. The third option is ‘customized language combinations only’, which include only
the language combinations that the user has customized. It is possible to add or delete the
source and target languages in the selected customized option.
Lexicographical customization
Lexicographical customization is best shown in the creation of custom dictionaries for each
customer, other than the dictionaries for spell checking. This means that multiple translators
working on projects for the same customer will use the same custom dictionary. 60
Linguistic customization
As far as linguistic customization is concerned, there are basically two levels of customization:
lexical customization and syntactical customization.
Lexical customization
Lexical customization is to customize a machine translation system by preparing a customized
dictionary, in addition to the system dictionary, before translating. This removes the
uncertainties in translating ambiguous words or word combinations. It must be pointed out,
however, that the preparation of a customized dictionary is an enormous task, involving a lot
of work in database creation, database maintenance, and database management.
Syntactical customization
Syntactical customization, on the other hand, is to add sentences or phrases to the database to
translate texts with many repetitions. Syntactical customization is particularly important when
there is a change of location for translation consumption. The translation memory databases
built up in Hong Kong for the translation of local materials, for example, may not be suitable
for the production of translations targeted at non-Hong Kong readers, such as those in mainland
China.
Resource customizationWebsite customization
Some computer-aided translation systems allow the user to create resource profile settings.
Each profile in Fluency, for example, has four customized uniform resource locators (URLs)
associated with it. URLs are the Internet addresses of information. Each document or file on
the Internet has a unique address for its location. Fluency allows the user to have four URLs
of one’s preference, two perhaps for specialized sites and two general sites.
Machine translation system customization
Some systems are connected to installed machine translation systems the terminology databases
of which can be customized for the generation of output, thus achieving terminological
consistency in the target text.
Collaborativity
Collaborativity is about continuously working and communicating with all parties relating to
a translation project, from the client to the reviewer, in a shared work environment to generate
the best benefits of team work. Computer-aided translation is a modern mode of translation
production that works best in team translation. In the past and decreasingly at present, individual
translation has been the norm of practice. At present and increasingly in the future, team
translation is the standard practice. A number of systems, such as Across and Wordfast, can allow users to interact with each
other through the translation memory server and share translation memory assets in real time.
S. Chan Computer-aided translation61
Translation is about management. Translation business operates on projects. Translation
technology is about project management, about how work is to be completed by translation
teams. With the use of translation technology, the progress of translation work is under control
and completed with higher efficiency. The best way to illustrate this point is project
collaboration, which allows translators and project managers to easily access and distribute
projects and easily monitor their progress. The work of translation in the present digital era is done almost entirely online with the help
of a machine translation or computer-aided translation system. This can be illustrated with
SDL-Trados 2014, which is a computer-aided translation system developed by SDL
International and generally considered to be the most popular translation memory system on
the market. Figure 2.17 shows the dashboard of SDL-Trados 2014.
Figure 2.17 Dashboard of SDL-Trados 2014
Figure 2.18 List of current projects 62
Figure 2.19 Project details
Workflow of a translation project
To start a project, the first stage of the workflow is the creation of a termbase and a translation
memory database, as shown in Figure 2.20.
Figure 2.20 Workflow of a translation project: the first stage
S. Chan Computer-aided translation63
In other words, when the Project Manager has any publications, files or web pages to
translate, he will send them to the translators of a department or unit, or freelancers for
processing. They will create translation units and term databases from these pre-translated
documents and save these databases in the SDL-Trados 2014 Server. This is the first stage of
the workflow. After the creation of translation memory and term databases, as shown in Figure 2.21, the
Project Manager can then initiate a translation project and monitor its progress with the use of
SDL-Trados 2014 (as indicated by ). He can assign and distribute source files to in-house and
/ or freelance translators by emails (as indicated by ). Translators can then do the translation
by (i) reusing the translation memories and terms stored in the databases; (ii) adding new words
or expressions to the translation memory and term databases (as indicated by ). When the
translation is done, translators send their translated files back to the Project Manager on or
before the due date (as indicated by ). When the Project Manager receives the translated
files, he updates the project status, finalizes the project and marks it as ‘complete’ (as indicated
by ). To make sure that SDL-Trados 2014 has a smooth run, a technical support unit to maintain
the SDL-Trados server may be necessary (as indicated by ).
Figure 2.21 Workflow of a translation project: the second stage
A translation team usually consists of the following members.
Project manager
A project manager is a professional in the field of project management. The responsibilities of
a project manager include the following:
1
plan, execute, and close projects
(When planning a project, the project manager works on the overall resources and budget
of the project. When executing a project, the project manager can add or import customers
and subcontract projects.)
2
create clear and attainable project objectives;
3
build the project requirements; and
4
manage cost, time, and scope of projects. 64
Terminologist
A terminologist is one who manages terms in the terminology database. There are two types
of terminologists: (1) customer-specific terminologists who can only access the terminology of
one customer; and (2) global experts who can access all the terms in the systems for all
customers.
Conclusion
This chapter is possibly the first attempt to analyse the concepts that have governed the growth
of functionalities in computer-aided translation systems. As computing science and related
disciplines advance, more concepts will be introduced and more functions will be developed
accordingly. However, it is believed that most of the concepts discussed in this chapter will last
for a long time.
References
Adriaens, Geert (1994) ‘Simplified English Grammar and Style Correction in an MT Framework: The LRE SECC Project’, in Translating and the Computer 16, London: The Association for Information
Management, 78−88.
Adriaens, Geert and Lieve Macken (1995) ‘Technological Evaluation of a Controlled Language Application: Precision, Recall and Convergence Tests for SECC’, in Proceedings of the 6th International
Conference on Theoretical and Methodological Issues in Machine Translation (TMI-95), 5−7 July 1995,
University of Leuven, Leuven, Belgium, 123−141.
AECMA (1995) ‘A Guide for the Preparation of Aircraft Maintenance Documentation in the International Aerospace Maintenance Language−Issue 1’, Brussels, Belgium.
Al-Shabab, Omar Sheikh (1996) Interpretation and the Language of Translation: Creativity and Convention in Translation, London: Janus Publishing Company.
Allen, Jeffrey (1995) Review of the Caterpillar KANT English-French MT System, Internal Technical Report, Peoria, IL: Technical Information Department, Caterpillar Inc.
Allen, Jeffrey (1999) ‘Adapting the Concept of “Translation Memory” to “Authoring Memory” for a Controlled Language Writing Environment’, in Translating and the Computer 20, London: The
Association for Information Management.
Allen, Jeffrey and Christopher Hogan (2000) ‘Toward the Development of a Post-editing Module for Raw Machine Translation Output: A Controlled Language Perspective’, in Proceedings of the 3
rd
International Workshop on Controlled Language Applications (CLAW 2000), 29−30 April 2000, Seattle,
WA, 62−71.
Almqvist, Ingrid and Anna Sågvall Hein (1996) ‘Defining ScaniaSwedish − Controlled Language for Truck Maintenance’, in Proceedings of the 1st International Workshop on Controlled Language Applications
(CLAW-96), Leuven, Belgium, 159–164.
Andersen, Peggy (1994) ‘ClearCheck Demonstration’, in Proceedings of the 1st Conference of the Association for Machine Translation in the Americas: Technology Partnerships for Crossing the
Language Barrier (AMTA-1), 5−8 October 1994, Columbia, MD, 227.
Arnold, Doug J., Lorna Balkan, R. Lee Humphreys, Seity Meijer, and Louisa Sadler (1994) Machine Translation: An Introductory Guide, Manchester and Oxford: NCC Blackwell.
Bell, Roger T. (1991) Translation and Translating: Theory and Practice, London and New York: Longman.
Bernth, Arendse (1999) Tools for Improving E-G MT Quality, Yorktown Heights, NY: IBM T.J. Watson Research Center.
Bjarnestam, Anna (2003) ‘Internationalizing a Controlled Vocabulary Based Search Engine for Japanese’, in Proceedings of the Localization World Conference 2003, 14−16 October 2003, Seattle, WA.
Bly, Robert (1983) The Eight Stages of Translation, Boston, MA: Rowan Tree Press.
Caeyers, Herman (1997) ‘Machine Translation and Controlled English’, in Proceedings of the 2nd Workshop of the European Association for Machine Translation: Language Technology in Your Organization? 21−22 May
1997, University of Copenhagen, Copenhagen, Denmark, 91−103.
S. Chan Computer-aided translation65
Carbonell, Jaime G., Teruko Mitamura, and Eric H. Nyberg (1992) ‘The KANT Perspective: A Critique of Pure Transfer (and Pure Interlingua, Pure Statistics,...)’, in Proceedings of the 4 th International Conference
on Theoretical and Methodological Issues in Machine Translation of Natural Languages: Empiricist vs Rationalist
Methods in MT (TMI-92), Montreal, Quebec, Canada, 225−235.
Carl, Michael (2003) ‘Data-assisted Controlled Translation’, in Proceedings of the Joint Conference Combining the 8
th International Workshop of the European Association for Machine Translation and
the 4th Controlled Language Applications Workshop: Controlled Language Translation (EAMT-
CLAW-2003), Dublin City University, Ireland, 16−24.
Chan, Sin-wai (2004) A Dictionary of Translation Technology, Hong Kong: The Chinese University Press.
Chen, Kuang-Hua and Chien-Tin Wu (1999) ‘Automatically Controlled-vocabulary Indexing for Text Retrieval’, in Proceedings of the International Conference on Research in Computational Linguistics
(ROCLING-XII), Taipei, Taiwan.
de Ilarraza, Arantxa Diaz, Aingeru Mayor, and Kepa Sarasola (2000) ‘Reusability of Wide-coverage Linguistic Resources in the Construction of Multilingual Technical Documentation’, in Proceedings of
the International Conference on Machine Translation and Multilingual Applications in the New Millenium (MT-
2000), University of Exeter, England.
Delisle, Jean (1988) Translation: An Interpretive Approach, Patricia Logan and Monica Creery (trans.), Ottawa and London: University of Ottawa Press.
Fais, Laurel and Kentaro Ogura (2001) ‘Discourse Issues in the Translation of Japanese Email’, in Proceedings of the 5th Pacific Association for Computational Linguistics Conference (PACLING-2001),
Fukuoka, Japan.
Fouvry, Frederik and Lorna Balkan (1996) ‘Test Suites for Controlled Language Checkers’, in Proceedings of the 1st International Workshop on Controlled Language Applications, Katholieke Universiteit, Leuven,
Belgium, 179−192.
Gough, Nano and Andy Way (2004) ‘Example-based Controlled Translation’, in Proceedings of the 9th Workshop of the European Association for Machine Translation: Broadening Horizons of Machine
Translation and Its Applications, Foundation for International Studies, Malta, 73−81.
Han, Benjamin, Donna Gates, and Lori S. Levin (2006) ‘Understanding Temporal Expressions in Emails’, in Proceedings of the Human Language Technology Conference − Annual Meeting of the North
American Chapter of the Association for Computational Linguistics (HLT-NAACL-2006), New
York.
Hein, Anna Sagvall (1997) ‘Scania Swedish-A Basis for Multilingual Translation’, in Translating and the Computer 19, London: The Association for Information Management.
http://www.alchemysoftware.ie/index.html.
http://www.helicon.co.at/aboutus.html.
http://www.internetworldstats.com/stats.htm.
http://www.lisa.org/Glossary.
http://www2.multilizer.com/company.
http://www.passolo.com.
http://www.schaudin.com.
Hu, Qingping 胡清平 (2005)〈受控語言及其在漢英機器翻譯裏的應用前景〉(Controlled Language and Its Prospective Application in Chinese-English Machine Translation), in Luo Xuanmin 羅選民
(ed.) 《語言認識與翻譯研究》(Language, Cognition and Translation Studies), Beijing: Foreign
Language Press 外文出版社, 364−372.
Hurst, Matthew F. (1997) ‘Parsing for Targeted Errors in Controlled Languages’, in Ruslan Mitkov and Nicolas Nicolov (eds) Recent Advances in Natural Language Processing, Amsterdam and Philadelphia: John
Benjamins, 59−70.
Joy, Lorna (2002) ‘Translating Tagged Text − Imperfect Matches and a Good Finished Job’, Translating and the Computer 24, London: The Association for Information Management.
Kamprath, Christine, Eric Adolphson, Teruko Mitamura, and Eric H. Nyberg (1998) ‘Controlled Language Multilingual Document Production: Experience with Caterpillar Technical English’, in
Proceedings of the 2nd International Workshop on Controlled Language Applications (CLAW-98), Language
Technologies Institute, Carnegie Mellon University, Pittsburgh, PA, 51−61.
Kay, Martin (1980) ‘The Proper Place of Men and Machines in Language Translation’, Research Report CSL-80-11, Xerox Palo Alto Research Center, Palo Alto, CA.
Lehtola, Aarno, Jarno Tenni, and Catherine Bounsaythip (1998) ‘Controlled Language—An Introduction’, in Proceedings of the 2nd International Workshop on Controlled Language Applications, Carnegie Mellon
University, Pittsburgh, PA, 16−29. 66
Lockwood, Rose (2000) ‘Machine Translation and Controlled Authoring at Caterpillar’, in Robert C. Sprung (ed.) Translating into Success: Cutting-edge Strategies for Going Multilingual in a Global Age,
Amsterdam and Philadelphia: John Benjamins, 187–202.
Loong, Cheong Tong (1989) ‘A Data-driven Control Strategy for Grammar Writing Systems’, Machine Translation 4(4): 281–297.
Lockwood, Rose (2000) ‘Machine Translation and Controlled Authoring at Caterpillar’, in Robert C. Sprung (ed.) Translating into Success: Cutting-edge Strategies for Going Multilingual in a Global Age,
Amsterdam and Philadelphia: John Benjamins, 187−202.
Lux, Veronika and Eva Dauphin (1996) ‘Corpus Studies: A Contribution to the Definition of a Controlled Language’, in Proceedings of the 1st International Workshop on Controlled Language Applications (CLAW-
96), Leuven, Belgium, 193−204.
Matsuda, Junichi and Hiroyuki Kumai (1999) ‘Transfer-based Japanese-Chinese Translation Implemented on an E-mail System’, in Proceedings of MT Summit VII: MT in the Great Translation Era, 13−17
September 1999, Kent Ridge Digital Labs, Singapore.
Melby, Alan K. (1995) The Possibility of Language: A Discussion of the Nature of Language, with Implications for Human and Machine Translation, Amsterdam and Philadelphia: John Benjamins.
Melby, Alan K. (2012) ‘Terminology in the Age of Multilingual Corpora’, The Journal of Specialized
Translation 18: 7−29.
Mitamura, Teruko, Eric H. Nyberg, and Jaime G. Carbonell (1994) ‘KANT: Knowledge-based, Accurate Natural Language Translation’, in Proceedings of the 1st Conference of the Association for Machine Translation
in the Americas: Technology Partnerships for Crossing the Language Barrier (AMTA-1), 5−8 October 1994,
Columbia, MD, 232−233.
Mitamura, Teruko and Eric H. Nyberg (1995) ‘Controlled English for Knowledge Based MT: Experience with the KANT System’, in Proceedings of the 6th International Conference on Theoretical and Methodological
Issues in Machine Translation (TMI-95), Leuven, Belgium, 158−172.
Mitamura, Teruko (1999) ‘Controlled Language for Multilingual Machine Translation’, in Proceedings of MT Summit VII: MT in the Great Translation Era, Singapore, 46−52.
Mitamura, Teruko, Eric H. Nyberg, Kathy Baker, Peter Cramer, Jeongwoo Ko, David Svoboda, and Michael Duggan (2002) ‘The KANTOO MT System: Controlled Language Checker and Lexical
Maintenance Tool’, in Stephen D. Richardson (ed.) AMTA-02: Proceedings of the 5
th Conference of the
Association for Machine Translation in the Americas, Machine Translation: From Research to Real Users, 6−12
October 2002, Tiburon, CA, 244−247.
Murphy, Dawn, Jane Mason, and Stuart Sklair (1998) ‘Improving Translation at the Source’, in Translating and the Computer 20, London: The Association for Information Management.
Nida, Eugene A. (1969) ‘Science of Translation’, Language 45(3): 483−498.
Nida, Eugene A. and Charles R. Taber ([1969] 1982) The Theory and Practice of Translation, Leiden: E.J. Brill.
Nyberg, Eric H. and Teruko Mitamura (1992) ‘The KANT System: Fast, Accurate, High-quality Translation in Practical Domains’, in Proceedings of the 14th International Conference of Computational
Linguistics (COLING-92), 23−28 August 1992, Nantes, France, 1069−1073.
Nyberg, Eric H., Teruko Mitamura, and Jaime G. Carbonell (1997) ‘The KANT Machine System: From R&D to Initial Deployment’, in LISA Workshop on Intergrating Advanced Translation Technology, Seattle,
WA.
Nyberg, Eric H., Christine Kamprath, and Teruko Mitamura (1998) ‘The KANT Translation System: from R&D to Large-Scale Deployment’, LISA Newsletter 2(1): 1−7.
Nyberg, Eric H. and Teruko Mitamura (2000) ‘The KANTOO Machine Translation Environment’, in John S. White (ed.) Envisioning Machine Translation in the Information Future, Berlin: Springer Verlag,
192−195.
Nyberg, Eric H., Teruko Mitamura, and Willem-Olaf Huijsen (2003) ‘Controlled Language for Authoring and Translation’, in Harold L. Somers (ed.) Computers and Translation: A Translator’s Guide,
Amsterdam and Philadelphia: John Benjamins, 245−281.
Probst, Katharina and Lori S. Levin (2002) ‘Challenges in Automated Elicitation of a Controlled Bilingual Corpus’, in Proceedings of the 9th International Conference on Theoretical and Methodological Issues in Machine
Translation (TMI-2002), Keihanna, Japan, 157−167.
Rico, Celia and Enrique Torrejon (2004) ‘Controlled Translation as a New Translation Scenario − Training the Future User’, in Translating and the Computer 26, London: The Association for Information
Management.
S. Chan Computer-aided translation67
Rooke, Robert (1985) ‘Electronic Mail’, in Catriona Picken (ed.) Translation and Communication: Translating and the Computer 6, London: The Association for Information Management, 105−115.
Roturier, Johann (2004) ‘Assessing Controlled Language Rules: Can They Improve Performance of Commercial Machine Translation Systems?’ Translating and the Computer 26, London: The Association
for Information Management.
Ruffino, J. Richard (1985) ‘The Impact of Controlled English on Machine Translation’, in Patricia E. Newman (ed.) American Translators Association Conference − 1985, Medford, NJ: Learned Information,
Inc., 157−162.
Sofer, Morry (2009) The Translator’s Handbook, Rockville, MD: Schreiber Publishing.
Steiner, George ([1975] 1992) After Babel: Aspect of Language and Translation, 2nd edition, Oxford: Oxford University Press.
Torrejón, Enrique (2002) ‘Controlled Translation: A New Teaching Scenario Tailor-made for the Translation Industry’, in Proceedings of the 6th Workshop of the European Association for Machine Translation:
Teaching Machine Translation, Manchester, England, 107−116.
van der Eijk, Pim and Jacqueline van Wees (1998) ‘Supporting Controlled Language Authoring’, in Proceedings of the 3rd Workshop of the European Association for Machine Translation: Translation
Technology: Integration in the Workflow Environment, Geneva, Switzerland, 65−70.
Wasson, Mark (2000) ‘Large-scale Controlled Vocabulary Indexing for Named Entities’, in Proceedings of the 6th Applied Natural Language Processing Conference, Seattle, Washington, DC, USA, 276−281.
Western Standard (2011) Fluency Translation Suite 2011Fluency User Manual V2.5.04. Utah: © 2009–2011 Western Standard Translation. Available at: http://www.westernstandard.com/FluencyInstalls/
FluencyDocumentation.pdf.
Wilss, Wolfram (1982) The Science of Translation: Problems and Methods, Tubingen, Germany: Gunter Narr Verlag.
Windows (2012) “A History of Windows,” Windows.
Wojcik, Richard H. and Heather Holmback (1996) ‘Getting a Controlled Language Off the Ground at Boeing’, in Proceedings of the 1st International Workshop on Controlled Language Applications (CLAW-96),
Katholieke Universiteit Leuven, Belgium, 22−31.
XTM International Ltd. (2012) XTM for CMS Explained, Bucks: XTM International Ltd. Available at: http://www.xtm-intl.com/files/content/xtm/resources/XTM%20for%20CMS%20Explained%20
2012-01.pdf.
Zydron, Andrzej (2003) ‘Xml: tm – Using XML Technology to Reduce the Cost of Authoring and Translation’, in Translating and the Computer 25, London: The Association for Information Management.