Professional Documents
Culture Documents
Translation and Technology - Chapitre1. Quah PDF
Translation and Technology - Chapitre1. Quah PDF
Deflnition of Terms 7
1 MT
translation
In translation technology, terms commonly used to describe The initial goal of machine translation was to build a fully automatic
tools are as follows: high-quality machine translation that did not require any human
interuention. At a 1952 conference, however, Bar-Hillel reported that
. machine translation (MT); building a fully automatic translation system was unrealistic and years
. (MAHT);
machine-aided/assisted human translation later still remained convinced that a fully automatic high-quality
. (HAMT);
human-aided/assisted machine translation machine translation system was essentially unattainable (Bar-Hillel
. computer-aided/assisted translation (CAT);
1960120O3: 45). Instead, what has emerged in its place is machine
. -u.hirr"-aided/assisted
translation (MAT); translation, placed between FAHQT and HAMT on the continuum of
.fuIlyautomatichigh-quality(machine)tlanslation(FAHQT/FAHQMT). Figure 1.1. The main aim of machine translation is still to generate
ganslation automatically, but it is no longer required that the output
Distinctions between some of theJe terms are
not always clear' For quality is high, rather that it is fit-for-purpose (see Chapters 2 and 3).
(CAT) is often the term used in
example, computer-aided translation As for human-aided machine translation and machine-aided human
the Iocalization industry (see the second
Transiation Studies (TS) and translation, the boundary between these two areas is especially unclear.
the community which develops
put, of this chapter), while software Eoth classes are considered to be computer-aided translation as indicated
it'machine-aided translation' (MAT)' As
itt,,yp" of tool prefers to call
and in the field
ln Figure 1.1 (Tong 1994: 4,730; see also Slocum 1988; Hutchins and
the more familiar term among professional translators 8ôrners L992). However, in Schadek and Moses (ZOOI), a different
is used throughout
of Translation Studies, 'computer-aided translation' dlassification has been proposed where only machine-aided human
thebooktolepresentbothcclmputer-aidedtranslationandmachine-aided
,aided' is chosen instead of 'assisted" as , dnnslation is viewed as synonymous with computer-aided translation.
translation tobls, and the term tfiuman-aided machine translation 1S considered AS a separate category
lDhe reasoning
alsoin,human.aidedmachinetranslation,and,machine.aidedhuman behind the vlew offe red by Schadek and M oses not
translation ffifficult to understand. At least theoreti cally th e difference between
distinguishes four types of translation relating
human and two is obvious. For human-aided machine translation, the
Figure 1 1
a linear continuum
machine involvement 1n a classification along is the principal translator, while in machine-aided human
992: 1 48) This classification, it is a human. In practice, however, it may be less easy today
ln troduced by Hutchins and Somers 1
to sustain as
now more than a decade old, will become harder distinguishable boundary between them. The blurring of
we sha II see n Chapters 3, 4 and is further complicated when human-aided machine translation
tools become multifunctional, a
remalns useful a point
Nevertheless, the concePt ln Figure 1 1 AS as a subclass of machine translation, an approach
chosen
ln relation to technologY
reference for classifying transla tion (2002). Since human-aided machine translation has
6
r
Definition of Terms 9
8 Translation andTechnologY
texts from one natural
a feature that is closer to application of computers to the task of translating
the machine as the principal translator -. it i;"guug" to anothêr' (http://www'eamt'org/mt'html)
while the Interna-
machine-aided human translation -
machine translation it'u"'to iioria Àro.iution of Machine Translation
(IAMT) defines machine transla-
it under the catesory of computer-ar-ded
at a time [sic] and
makes little sense to inciuOe
book adopts the àistinctlon suggested
iiàn ut taking 'input in the form of tull sentences
translation. Consequently, this oe-neratins coirespànding full sentences
(not necessarily of good quality)'
i" *ttich human translation is
machine-aided are essentially variants of the same
by Schadek and Moses BOOfj
translation and human-aided iH"i.t r"i z000ai. These definitions texts and target or
synonymous with i-oïiutt'-uiO"a ào.r."p, focusing on source or 'input-lariguage'
machine translation is a class
on its own' texts'
have been olaced in these individual "ilËrrn",
'output-language'
above includes human intervention.
A decade ago most toots couta àr in. definitions
ho*"""'' is no longer suitable for describinS some form of human
classes. A linear continîum'
as they contai; features
ftom more than Oit"rr, such as Arnold etal' (Ig94t L), mention
part of the process of
many integrated systems of today i'i.ru"ntion: 'the attempt to automate all or
or," .tur, free ct apter ul' irunstating fiom one human language
to another' (my italics)' When
'i"ll1'-1"^î,i:.:i:',î"J[Ti;ii:Tll::l
1" is now requlr(
is mentioned in a definition, it often
state of translation technol<lgy and ,.o,,,. ro'- of human intervention
types are not entirely separate '1992: 4O8). This view is echoed by Archer
continuum, these four translation
*u"y *oi" similarities than when they
were ;;;;;., ,murky, (Balkan
distinct, and they ,inozt tOOl, according to whom scholars and researchers still disagree
"";;;;t involvement
first presented by Hutchins and
Somers'
àïin. definition of machine translation with respect to theforthcoming, it
for describing àf hu.nunr. Howevet, since no other term has been
Table 1'1 An example of a table automated as well
translation tYPes Io.rit"""t to be used to refer to systems that are fully
u''t'o'"withhumaninvolvement(Somers2003b:1).Figure1.2shows
MT HAMT CAT HT
howasourceJanguagetextcanbeprocessedbyamachinetranslation
ryrr... If the taïget text is produced automatically there is no before'
n
human
1,,
Topic A human intervention may be employed
Topic B
intervention; however,
Arnold etal. L994).
àuring and/or after machine translation (see also
translation; HAMT=human-aided
machine
Furitrer distinctions are also made' A machine translation system'
MT=machine
',ârf HT human
uiio.t, CAT = computer-aided translation
=
(2000a), can be classified as operating on one of
according to Hutchins
translation;
tables in order
changes is to use a series of
One way of presenting these with the topics
translation types
to show the relationsd ;il;;t" fàur 1.1', where the
discussed in this book, Is
litustrateo in principle in Table TL text
groups of topics'
in relation to selected SL text
four translation types are reviewed Machine
in detail in Chapter 7 when all
The series of tables *iff Ut described translation
perspec-
Each table adopts a different system
relevant topics have U""" ài"u""d' with respect to transla'
kinds of relationship
tive in ordeï to srrow oiiiererrt
tion and technologY'
systems are MaTra Pro and Lite developed at the National Centre for
Unedited SL text [H]
Software Technology based in Mumbai, India, that translate from
Pre-edited SL English into Hindi. Human-aided machine translation systems have
text [H] been implemented at Schreiber Translations, Inc., Foreign Language
Services, Inc. and Ralph McEIroy Translation Company, all companies
that are employed to translate patents for the United States (US) Patent
Machine TL text and Trademark Office.
Controlled
translation
system
language (SL)
Post-editing [H] Machine-aided human translation
Text [H]
for this is that the difference between this category and full
The reason
transla-
machine translation is blurred since some definitions of machine SL = sourcê language; TL = target language; H = human
for human translators to carry out pre- and post-editing,
tion also allow
of human-aided machine translation Ftgure 1.5 Machine-aided human translation model
as indicated earlier. Examples
14 Translation and TechnologY Definition of Terms 15
Human translation
Box 1.1 (Continued)
Nowadays,itiScommontofindprofessionaltlanslatolsusingsome
A description of
xi"a cornputer_aided translation tool in their work. words through the Google search engine that comes with experience.
"r at work is presented he[e, written
,rr.t, u modern-day human translator Sometimes some ingenuity is required. Just typing in the Japanese
for this book by a professional translator based in Singapore word and the word'English' into the search fieid [where a query
specifically
urrd sydn.y, Stephen Moore, who translates
between English andJapanese. word or phrase is entered for Google to perform the search] is
been
wlth-voot.,s permission, his piece was edited; explanations have unlikely to produce any useful results. A good first attempt for a
clarification. His description of a difficult technical term is to guess one of the component words
added in square brackets for further
mo<lern-day translator at work is as shown
in Box 1'1' found in the term. As an example, I would type in the Japanese
The account is from a translator who is highly computel literate and term ær'{tEîS7 i )vA in the search field and then add the
willing to learn to use new tools in order to improve his translation English word 'film'. This may lead me to the English equivalent,
servicà. His attitude guarantees that he is not
being left behind in the which is 'anisotropic conductive film' (a type of material used in
competitiveenvironmentolsecuringtranslation|obs.Thismaynot the electronics industrY).
necessarily be the case for every tlanslator.
Researchers such as Gaspari One point I might add is that even with transliterations from
(2001), Yuste Rodrigo (2001) and Bowker (2002) have
discussed resistance English to Japanese I have to be careful. One example is 'shiito'
[based on 'sheet'] as in 'polarization shiito', which can appear in
English technical material but the proper English term is really
,polarization film'. Sometimes there is no English equivalent but
Box 1'1 A translator at work
enough examples of its use in the source language can be found to
pair of
I have registered with a number of agencies for the language pinpoint the meaning. However, I would never use this method to
and English specializing in the chemical-related field. Various
Japanese 'hazard a guess'. If unsure, I have to seek help either in a mailing
also know
compa.rier and organizations in the chemical industry list I subscribe to or go back to my client. It may turn out to be a
my iapabilities due to my rvork as a journalist in this area and
have
term that only my client's company uses.
ueenwort<ingwithmeonaregularbasisforsometime.Partofmy With the growth of the chemical industry's interest in China, it
regular work involves translation of newspaper articles
on a daily basis.
is advantageous to have the knowledge of the Chinese language and
and generally
I r"eceive Word documents via email in the late aftemoon, capability on my computer to assist me with tracking down names of
I am committed to returning the translations within two days' places and organizations. Although the Japanese pronunciation
The first thing I do is to ascertain whether the material is based on a and character style are slightly different when compared to the
press release. If so, there is a likelihood that an English
language
Chinese characters, I sometimes have to revert to a Chinese dictionary
version of the press release can be found on the Internet. The next
to look up the relevant characters and pronunciation, and then
press release
best scenario would be that the Japanese language
is
input the Chinese characters into a search engine such as Google.
posted on the Intemet. This can help as background
material' As Japanese characters are derived from old Chinese characters,
Whilstworkinglalwaysmaintainabroadbandconnectiontoan there are many similarities, even with the simplified characters
online dictionary to look up technical words and the search engine that Mainland China [the People's Republic of China] now uses,
and the bilingual
Google for words that cannot be found in the online [Simplified Chinese script has fewer strokes per character than
(fapanese-English) dictionaries. other dictionaries that I use include Traditional Chinese, for example compared to ,ffi for ,horse,
àne for ctremical-specific terms and one for the names
of organ2ations. Ê and compared to ll for ,treasure,].-i For example, I tried to look
For words that cannot be sourced using online oI conventional up the name of an obscure Chinese company (written in
dictionaries, I may employ a variety of search engines' There is a Simplified Chinese: ËtfijgâtÊ,ÂrênËËmâ\E) found
in a Japanese
technique, howevet, to finding English equivalents of Japanese language press release. Hence
the search was unsuccessful. So I took out
my Chinese dictionary, and systematically looked up each character
a-
16 Translation andTechnology Definition of Terms 77
from its radical. [A radical is a semantic classifier. It categorizes precluding extensive use of the Web. For some clients who are Micro-
words based on their meanings rather than spelling since Chinese is soft Office sawy, I use tags [hidden comments or remarks which are
logographic in nature - written from left to right or top to bottom - inserted in the translated text by the translator] when I am unsure of
where each character stands for a single word or a single syllabic my translation or want to clarify it. The client or the client,s proof
morpheme (Boltz 1996: 191,199), for example, the character F 'moon' reader will similarly use tags for questions and clarifications.
is the radical of the word E , which means 'shares'in a company.] I have also started to experiment with a translation memory tool
The Pinyin romanization of the obscure Chinese company name called WordFast. The shareware program is a Microsoft macro that can
is 'Ningbo Lianhe Touzhi Konggu Youxian Gongsi'. [Pinyin be used on any version of Office and on any operating system. [A
romanization is a form of spelling based on sound in Mandarin. It macro in Microsoft is a saved sequence of commands or keyboard
is also known as Hanyu Pinyin, which literally means 'Han strokes that can be stored and recalled with a single command or
language pinyin'. Romanization refers to the phonetic notation and keyboard strokel. Initial tests with my laptop (processing speed at
transliteration to Roman script. This is not anglicization.l Thus I 400MHz) indicated that the software runs excessively slow to the
typed the Pinyin romanization letters into my Chinese-capable point of being impractical. Once I have my desktop computer
Apple Macintosh computer and converted them to the appropriate configured to 750MHz, I plan to retest the software before deciding
characters in Chinese. I then conducted an Internet search using whether to purchase a license. An interesting point with this program
r/' the Chinese Yahoo! search engine, but alas, I could still not find an is that translation memories [see Chapter 4] can be set up for each
English equivalent. I ended up having to go to the Japanese client and shared with them. Memories can also be imported/exported
company that had issued the press release to ask. tolfrom Trados Translator's Workbench and other similar software.
One thing I am slightly concerned about is how applicable a
i
Conclusion
Internet began to be
U ntil the early 1 990s the time when the
given in Hutchins and Somers (1
worldwide, the translation types
decade later, the boundaries
were certainlY applicable More than a
these four translation types have become more blurred. Although
clear distinctions, these have
wrlters ln the field still make ;
increasinglY
harder to maintain AS technologY becomes fr
and more multitasking. The pace of change ln the develoPment
what rs current todaY
li,;
translation technologY IS extremelY rapid;