Translation technology

Translation: is an act of writing that conveys the meaning of a text

written in one language in another language.
Evidence exists of translations of a Sumerian epic dating from the
second millenium BC.
A journey through the centuries reveals significant translation
activities in Ancient China, Mesopotamia, Rome, Toledo and the court
of King Alfred the Great.
Translation brought knowledge from the Orient to Europe, and the
translations of the Bible into German and English were instrumental
in changing the course of history.
The second half of the twentieth century saw the professionalization
of translation. Institutes and organizations were founded, standards
and qualifications were developed, and universities introduced
courses designed to prepare their graduates for admission to this new
profession. Newly qualified translators will start their careers as
freelancers working for Language Service Providers (LSP's) or as staff
translators for a government or commercial organization. Some will
eventually set up their own businesses, while others will remain in
employment throughout their working lives.
The growing demand for translation world-wide has focused
commercial and government organizations on the cost of translation
and the need to maximize translator productivity. Translation
technology has been identified as a means of boosting productivity.
Technology for translators
Over the past thirty years, translators have benefited from the
development of a range of computer-assisted translation tools and
other technology.
• These tools include:
• Translation memory software
• Machine translation software
• Electronic dictionaries and online glossaries
• Voice recognition software
• OCR and PDF conversion software
• Spelling and grammar checkers
Translation memory software (commonly known as TM software)
Translation memory software is likely to be the tool translators will
use most in their careers.
The basic function of the translation memory program
A translator should never have to translate the same sentence twice.
a translation memory application stores each translated sentence in a
database which it then consults to translate subsequent sentences. If
the source sentence has a translation in the database, there is a 100%
match, the reliability of which is dependent on the underlying
reliability of the translation memory. The translator may also be
presented with partial matches ("fuzzy matches") which can be
accepted, edited, or rejected.
In theory
translators can start with an empty database, or translation memory,
and build up their own memory sentence by sentence. Over several
years, the professional translator will build up a substantial database
of translated sentences, which will save a significant amount of time
and effort, particularly if he or she only undertakes work in a
specialised field. Translators will also receive existing translation
memories from clients or make use of freely available public
translation memories.
The European institutions and organizations such as TAUS provide
opportunities for sharing data in the form of translation memories
and their databases can be consulted free of charge.
The principal commercial translation memory programs used by
translators today are “SDL Trados Studio 2014”, “memoQ 2014”,
“Wordfast Pro” and “Déjà Vu X3”. there are at least 25 translation
memory tools available, including a number of open-source programs
such as “Omega-T” and “OpenTM2”. Several of these programs
started life as macros or templates to be run within Microsoft Word.
Nowadays, all the translation memory packages run as self-contained
programs within some kind of graphical user interface (GUI) or within
a browser.
Modern commercial CAT tools like “SDL Trados Studio 2014” can
handle over 50 different file formats.
The user typically works within a graphical interface with the source
and target languages displayed alongside each other. For each new
segment (usually a sentence) in the source text, proposed translations
are displayed in a separate pane, allowing the translator to select the
most appropriate translation which is then inserted in the editor
window alongside the source sentence. The translator may also
decide to use one or more machine translation plug-ins which will
come up with translation proposals if there is no corresponding
sentence in the selected translation memory.
Translation memory software is designed for interactive use.
once the translator has built up or obtained a large enough
translation memory he or she will achieve significant time savings by
automating part of the translation process with the software's "pre-
translate" feature. This feature will attempt to translate the
document by first retrieving 100% matches or fuzzy matches from the
selected translation memories and then by using a machine
translation plug-in if this option has been chosen by the user.
Freelance translators usually work with a translation memory
installed on their own PCs, even if it is provided by a client.
Staff translators or freelance translators working on large projects
involving several translators may make use of a server-based
translation memory. In such circumstances, translators are able to
share the benefits of each other's translation work and are more
likely to make consistent use of terminology.
The concept of the server-based translation environment has over the
last couple of years undergone a progression into what is known as
the cloud-based translation environment.
"Memsource", "XTM", "SmartCAT", "MateCAT", "Wordfast
Anywhere", "Smartling", "Easyling", "Lingotek", and "Wordbee" are
completely web-based solutions. While some of these solutions are
fee or subscription-based, "MateCAT" is free open-source software
released under the LGPL license.
MateCat (Machine Translation Enhanced Computer Assisted
Translation) has been developed with public funds by some of
Europe's leading researchers in the field of language technology and
has been designed with ease of use in mind. The translator can quite
literally enter "https://www.matecat.com/" in his or her browser and
start using the software, which supports 56 file formats and languages
extending from Afrikaans to Welsh, including languages with non-
Latin alphabets and right-to-left languages. User support is provided
free of charge via a Facebook group and via email. As its name
suggests MateCAT combines Machine Translation and Translation
Memory technologies. It automatically links to the freely available TM
Server My Memory and to Google Translate and will therefore come
up with translation suggestions for the Welsh-English/English-Welsh
language pair.
Machine translation, also known as "automatic translation or
computer-generated translation", is the branch of language
technology in which a computer program translates without any
human intervention. The usefulness of these translations is
dependent on a variety of factors including language pair, the
complexity ambiguity of the source text and the familiarity of the
system with the terminology and subject matter of the original. Some
languages are more amenable than others to machine translation.
Welsh, with its mutations and sentence structure, poses a significant
though not insurmountable challenge to developers.
Research and development in machine translation had its first big
boost in the Cold War period with much attention being paid to
Russian and Chinese.
The expansion of the European institutions brought about investment
in the development of the "commercial Systran software" to cover
the languages used at the European Commission.
International conflicts and the need for better intelligence have given
an impetus to the development of machine translation programs for
languages such as Arabic, Pashto and Farsi, particularly in the USA.
The need to translate many of the 684,478 pieces of content
generated every minute on the Internet is perhaps the most powerful
driving force behind progress in MT R&D.
In Wales, the introduction of machine translation by the National
Assembly for Wales has to be seen in the context of the Official
Languages Scheme. It is a tool to help people living and working in
Wales to communicate in either of the official languages.
Much of the earlier work in MT produced Rule-Based and Example-
Based systems.
Since the mid-1990's Statistical Machine Translation, where
translations are generated on the basis of statistical models derived
from the analysis of bilingual text corpora has become the
predominantly used approach, although researchers in this field are
now working on the introduction of syntactic rules into the
translation process.
Both publicly available machine translation systems offering Welsh-
English/English- Welsh - Google Translate and Microsoft Translator -
are fundamentally statistical systems. The successful use of these
systems is broadly dependent on the quantity and quality of the
bilingual data used to build the translation model. The partnership
between the National Assembly for Wales and Microsoft which
provides a mechanism for submitting a variety of bilingual texts to the
system is the best guarantee for achieving a steady improvement in
the quality of the computer-generated translations.
The general public can access Microsoft Translator directly via a
browser under any operating system (including Linux and Android) or
from within the Microsoft Office suite where there is greater
functionality than when simply using the browser.
An employee can receive an e-mail in Welsh and read an English
translation at the click of a mouse.
Professional translators may prefer to use the Welsh-English/English
Welsh MT service as a plug-in within their chosen CAT environment.
Most translation memory applications offer machine translation plug-
ins and allow the translator to choose when an MT-based translation
proposal will be made. The translator can also "pre-translate" an
entire document so that Microsoft Translator will populate all the
segments for which there is not a translation in the translation
memory. Whichever approach is used, the translator is able to review
or post-edit the machine translation proposal so that only an
approved translation is saved in the translation memory.
Studies have shown that significant productivity gains can be made by
combining machine translation with post-editing in a translation
environment tool.
Spelling and grammar checkers:
Proofing tools such as the Cysill Welsh online spelling and grammar
checker (http://www.cysill.com/arlein), the spellchecker in Microsoft
Word, E-gyhoeddwr (Epublisher) are quality assurance tools that
enable a translator to deliver professional work without embarrassing
spelling or grammatical mistakes. These tools are useful quality
safeguards for those translation jobs where translation automation is
impossible, inappropriate or offers no productivity gain.
Voice recognition software:
Voice recognition is not yet available for translators working into
Welsh, but there are several voice recognition packages on the
market which will enable those working into English to achieve
significant productivity gains. Nuance's Dragon Naturally Speaking
range is the market leader in speech recognition. The Home version is
an entry-level product that is adequate for occasional use.
OCR & PDF conversion software:
Optical Character Recognition (OCR) software is an essential tool in
the translation kit for those translators who want to use CAT tools but
receive their jobs in the form of hard copy. This software allows the
translator to scan the text to be translated and save it in an editable
format (e.g. as a Word or Excel file). Two decades ago such software
cost thousands of pounds, though today there are many free
packages available on the Internet. However, it is important to use
OCR software with a proven level of accuracy on medium-quality
documents, otherwise, the productivity gain achieved by not having
to type the text will be lost through the need to correct pages of "OCR
garbage". "Nuance's entry-level Omnipage 18" and "the Abbyy"
FineReader are considered to be a reliable applications for turning
printed documents into electronic files that can be processed by
machine translation and/or translation memory software. Dedicated
PDF conversion programs such as Nuance's PDF Converter or the
Abbyy PDF Transformer will generally provide results that need little
or no manual correction. Many "free" OCR and PDF conversion
packages are available on the Internet.
Translators and security:
Simple precautions such as locking up confidential documents and not
leaving unprotected computers switched on unattended need to be
complemented by effective cybersecurity measures. Judicious use of
public WiFi, encryption of sensitive documents sent via e-mail,
precautions against third-party use of USB sticks and smart phones in
the vicinity of computers holding confidential or sensitive
information, effective removal of files from hard drives are common-
sense measures which are often ignored.

Translators working throughout the United Kingdom have to comply

with the Data Protection Act 1998 if their work involves documents
containing personal information on third parties. They are not
permitted to hold this data longer than is strictly necessary for their
work and must keep it safe and secure.

Similar considerations apply to the use of cloud-based machine

translation and translation memory systems. Google clearly states
that it stores all the information submitted to its translation servers.
SDL states that content submitted to the SDL Language Cloud API is
not stored on the server after a translation is completed. The
Microsoft Translator service is governed by the Microsoft Service
Agreement and the Microsoft Privacy Online.
Translator training:
A good step for those who find translation tools a challenge would be
to take some basic IT courses.

“Computer Courses Near You” is a Welsh Government/European-

funded online resource that is used by staff in libraries and post
offices to signpost enquiries to suitable IT courses.

Cymdeithas Cyfieithwyr Cymru (The Association of Welsh Translators

and Interpreters) and the Institute of Translation and Interpreting (ITI)
organize reasonably-priced workshops and webinars on the use of
translation tools as part of their CPD programmes.

Major vendors of translation tools such as SDL (SDL Trados Studio

2014) and Kilgray Translation Technologies (memoQ) provide online
training in their products and face-to-face training through their
accredited trainers.

Most vendors provide some form of support through forums and chat
rooms, and details of many of the available training courses and
events are given under the Education tab of the website of proz.com -
an online marketplace for translators.

Users of translation tools will testify that taking some basic training is
the most effective way of getting up and running with a new CAT
package.
A typical TMS consists of two parts: a Translation Memory (TM) and a
search engine. The TMS is a type of CAT tool that is able to store the
previously translated text and the corresponding source text in a
database, and allows the translators to reuse them or part of them in
a new translation task. The TM is in its nature a database that store
translated texts together with the corresponding original texts. The
user can make use of the previously translated text and the source
text to set up one or more TM database(s). Then, in a new translation
task, the system can use the search engine to automatically search in
the existing TM database for the same or similar translation material
(such as a sentence, a paragraph, etc.) and provide a translation
suggestion to the translator. As a result, the translator does not have
to do any repetitive work and can focus his/her energy and time on
the translation of the new materials. At the same time, the TM
database is going on storing newly translated materials and
expanding the volume of the database. As for the translation
suggestion, the translator can choose to follow it, edit it or discard it.
Some basic concepts in the technology of the TMS are briefly
explained as follows.
1. Segmentation
When texts are added into a TM database, they have to be first
divided into some small chunks that are called segments. A segment is
the smallest unit that can be processed by the TMS, and is also the
base of the alignment. Generally speaking, a segment is a sentence or
a sentence-like structure and is usually ended with a punctuation
mark (such as a period or a question mark). However, according to
Bowker (2002a), it is also possible for the user to specify what should
or should not represent the end of a segment.
Alignment
The alignment is the second step in storing and organizing the texts.
Aligning texts means matching the source text segment with the
corresponding target text segment, and storing them in the database
as an integrated unit for future processing. The correct alignment is
the key to the successful reuse of the stored information.
Retrieval
Once the aligned texts have been stored in the TM database, they can
be reused by the translator in a new translation task via the automatic
search and retrieval of the system. When a new text is opened in the
translation editor of a TMS, the system first divides it into segments
as was introduced before. Then the system takes the first segment
and compares it with those segments that have been stored in the TM
database. The comparison is carried out on the basis of some preset
rules. The establishment of these rules is also an important topic of
researches and studies on the TMS. Once a match is found out in the
comparison, the system retrieves the relevant information (a unit of
the aligned texts) and presents it as a translation suggestion to the
translator. There are a number of types of matches that a TMS can
retrieve, so as to find not only identical translation units but also
similar ones.
Perfect or Exact Match
A perfect or exact match means that the new segment in the source
text is exactly the same as a segment that is already stored in the TM
database (100 percent match). According to Austermuhl (2001), there
is another type of match that is similar to the exact match called a full
match, which means the new segment is identical to a stored
segment in all the other aspects except for some variable elements
such as dates, numbers, time or measurements. In most TM tools
such as the Trados, these elements can be automatically treated by
the system by replacing the old dates, numbers, time or
measurements with new ones.
Fuzzy Match
Fuzzy match is an important and powerful function of the TMS,
meaning that the new segment is not identical to but similar to a
stored segment. It is more often encountered by translators in the
real practice than the exact match. Translators can edit the fuzzy
match so as to produce a valid translation for the new source
segment. In most TM tools, different parts between the new segment
and the stored segment are highlighted with a different color to
indicate which elements in current translation need to be changed or
adjusted to meet the source text. Another important thing in using
fuzzy match is the degree of fuzziness, or on the other hand, the
degree of similarity, which respectively means to what degree a
stored segment found by the searching engine is different from or
similar to the new source segment that needs to be translated. In
most TM tools, there is a user- defined minimum match value which
can be adjusted by the translator before translation according to the
desired degree of fuzziness or similarity. Generally speaking, the
higher the match value is, the stricter the retrieval will be, and vice
versa. However, according to Bowker (2002b), the translator must be
very cautious in selecting the degree of fuzziness so as to avoid the
problem of silence and noise. That is to say, if the match value is set
too high (e.g., 95%), the system may not retrieve matches that could
be potentially helpful to a translator. In information retrieval, this
type of problem is known as "silence." In contrast, if the match value
is set too low (e.g., 10%), then too many segments, or segments that
are not useful, could appear leading to a problem known as "noise,"
which may actually hinder the efficiency of the translation process.
There has been a significant increase in the volume of text that needs
to be translated into a wide variety of languages largely as a result of
globalization. In addition, new types of texts, such as web pages, have
appeared and require translation. These demands of our fast-paced,
globalized knowledge society have left translators struggling to keep
pace with the increasing number of requests for high-quality
translation into many languages on short deadlines. In today's
translation market therefore, the use of technology by translators is
no longer a luxury but a necessity if they are to meet rising market
demands for the quick delivery of high-quality texts in many
languages.
Translation Technology
The invention of the computer led very quickly to attempts to use it
for translation of natural languages. In the period immediately
following World War II.

initial attempts were made to develop fully automatic, high-quality

machine translation systems intended to replace translators.
However, researchers soon came to appreciate that translation is a
highly complex task that consists of more than mere word-for-word
substitution. It proved very challenging to programme computers to
take into account contextual, pragmatic and real-world information.

Major important advances occurred during the 1980s. The

administrative and commercial needs of multilingual communities
stimulated the demand for translation, leading to the development in
countries such as France, Germany, Canada and Japan of new
translation systems such as Logos (from German to French and vice
versa) and the international system created by the Pan-American
Health Organisation (from Spanish to English and vice versa), as well
as a number of systems produced by Japanese computer companies.

Again, the beginning of the 1990s saw vital developments in machine

translation with a radical change in strategy from translation based on
grammatical rules to that based on bodies of texts and examples (for
example, the Reverso Program). Language was no longer perceived as
a static entity governed by fixed rules, but as a dynamic corpus those
changes. according to use and users, evolving through time and
adapting to social and cultural realities.
To this day machine translation continues to progress. Large
companies are now using it more, which also increases software sales
to the general public.

Again, with the advent of new age software tools such as Google
Translate and Translation Memory tools like Trados, translation is
becoming much easier to manage. This situation has led to the
creation of on-line machine translation services such as AltaVista,
which offer rapid e-mail services, web pages, etc in the desired
language, as well as to the availability of multilingual dictionaries,
encyclopedias and free, direct- access terminology databases. These
tools do not pose a threat to the translators and interpreters to do
their job better and aren not in any way replacement for them but are
to help them to function better. We believe in the power of
technology for progress and acknowledge that when misused it can
bring about evil, but then, that is not the problem of technology but
that of users and abusers of it.
machine translation (MT) and computer- assisted translation (CAT).

These two technologies are the consequence of different approaches.

They do not produce the same results and are used in distinct
contexts.
Machine Translation (MT)
MT which relies solely on software aims at assembling all the
information necessary for translation in one program so that a text
can be translated without human intervention. It is that sector of
automated activity that consists of translating electronic data
automatically (the machine does all the translating with no user
input) from one language to another. It exploits the computer's
capacity to calculate in order to analyze the structure of a statement
or sentence in the source language, break it into easily translatable
elements and create a statement with the same structure in the
target language. It uses huge plurilingual dictionaries, as well as
corpora of texts that have already been translated.

MT which in the 1980s held great promises has been steadily losing
ground to computer-assisted translation because CAT responds more
realistically to actual needs since it has a human touch.
Computer-Aided Translation
CAT systems are systems which actually perform the task of
translation but rely on the intervention of the human translator at
various stages in the translation process. CAT uses a number of tools
to help the translator work accurately and quickly, the most
important of which are terminology databases and translation
memories. recent decades have witnessed considerable changes in
our society in general and in the translation market in particular.
Most texts are now produced in a digital format, which means they
can be processed by computer tools.
Types of Modern Translation Technologies:
Blatt et al, (1985), distinguish three types of computerized
approaches to the translation process: “machine aids for translators”,
“machine-aided translation” and “machine translation”.

“machine aids cover systems” such as word processors, dictionary

management tools, TERM BANKS and various look-up facilities which
support the translator but do not actually perform the translation
task.

Machine -aided translation systems, on the other hand, are systems

which actually perform the task of translation but rely on the
intervention of the human translator at various stages in the
translation process.

The difference between machine-aided translation systems and

machine translation systems, in Blatt et al's view, is that the latter are
intended as fully automatic translation systems, though their output
can of course be passed on to a translator for post-editing.

One can also categorize the types of computerized translation using

the more recent approaches. These more recent ways of categorizing
the types of computerized translation take as their principal criterion
the degree of automation, that is relative contribution of the machine
and the human translator to the translation process.

Resulting in a classification which distinguishes among machine-aided

human translation (MAHT), human-aided machine translation (HAMT)
and fully automatic machine translation (FAMT)
Balkan (1992), [3], makes a binary distinction between machine
translation (MT) and machine-assisted/ computer-assisted translation
(MAT or CAT), using machine translation to refer to „any system that
actually performs a translation‟ and classifying „any other
computerized translator tool which falls short of translating as a CAT
device‟.

Again, the term 'machine-aided translation' is used in a broad sense

here to cover all kinds of software systems especially designed and
developed for use as part of a translator's work-station, but not
themselves performing the task of translation as such.

In other words, the systems discussed here are not designed to

undertake any syntactic or semantic analysis of a source text nor to
generate a target language equivalent of the source text or any part
of it not included

also in the definition of machine-aided translation here are standard

software systems used in a modern office environment in general
rather than specifically by translators, examples: standard word-
processing software, universal database systems and other tools used
in performing administrative tasks.

Computer-aided translation in our definition further assumes that the

source-language is in machine-readable form [3]. Thus, machine-
aided translation as presented here occurs in any situation where a
machine-readable source text is processed by computerized tools in
order to produce a target-language translation, with the translator
being in control of all stages of this process and performing the
intellectual process of translation.
The Impact of Modern Translation Technologies in Today's World
and on Translators
The continued expansion of the global market and the realization that
language and the ability to communicate in multiple languages is a
primary driver as business compete in the global marketplace has
placed translators and interpreters in higher demand than they have
ever been in history and their services are being recognized by
mainstream as "vital" to the functioning of this new multilingual and
multicultural world. An indication of the increasing globalization is the
growth of language translation and the localization market in 2006
from 8.8billion to a projected increase of 9.6billion by 2012

"Common Sense Advisory", an independent analyst firm that focuses

on this area estimates that the demand for translation services will
grow 12% annually. but human translators do not have the capacity
to meet this demand and in such instances where the translation is
highly required, translation technology makes the process easier.
Translation technology solves this problem in two ways. First, it
enables translation at a level where it does not need to involve a
human. Secondly, we have seen translation technology increase
human translator productivity by up to 400%.
Globalization has made it that for some organizations, there is no
other option other than to use new technology. A very good example
is looking at what “Angelique Petrits” says.

"Angelique Petrits", a language officer at the European Commission

says "the organization is responsible for translating 2million pages
into 24 different languages every year and it wouldn't be able to fulfill
its mission without up-to-date translation technology". She continues
by saying: "Technology is a tool that helps dealing with scarce
resources of translators, by speeding up their work and allowing them
to concentrate on the essential". It also contributes to the consistency
of terminology, crucial in EU texts".
"Andy Way" says "There is just so much translation to be done-
people have estimated that only around 5% of what needs to be
translated actually is-that good translators will never be out of a job".

" Professor Philipp Kochn" chair of the machine translation school of

informatics at the University of Edinburgh adds his part by confirming
thus: "What's happening is their jobs are changing". He continues by
saying: "They are increasingly more about being a content editor than
a translator..."

This assertion by "Koehn" has confirmed a subject that has been a

long discussion, which is whether machine translation and computer-
assisted translation could convert translators into mere editors,
making them less important than the computer programs.The fear of
this happening has led to a certain rejection of the new technologies
on the part of translators, not only because of a possible loss of work
and professional prestige, but also because of concern about a
decline in the quality of production.
However, there is clearly the development of new capabilities, which
leads us to point out a number of essential aspects of the current
working situation.

Translating with the help of the computer is definitely not the same
as working exclusively on paper and with paper products such as
conventional dictionaries, because computer tools provide us with a
relationship to the text which is more flexible than a purely lineal
reading.

We also share the conviction that translation has not become a new
profession with the new technologies, but the changes have come to
stay and will continue to evolve. Translators need to acquire totally
new skills, to join the digital revolution (not fight it)...to understand
mobile technologies (not discard them)...to use the tools of the trade
in the 21. century (and learn what they are and what they can do for
them). There is no stopping technology.
How Translation Technologies Work:
In reality, the translation process comprises two stages:
First, the production of a rough text or preliminary version in the
target language, in which most of the translation problems are solved
but which is far from being perfect; and second, the revision state,
varying from merely re-reading the text while making minor
adjustments to the implementation of radical changes. It could be
said that MT aims at performing the first stage of this process in an
automatic way so that the human translator can then proceed directly
to the second, carrying out the meticulous and outstanding task of
revision.
Machine Translation Strategies
Machine translation is an autonomous operating system with
strategies and approaches that can be classified as follows:
the direct strategy
the transfer strategy
the pivot strategy
Computer-assisted Translation
Computer-assisted translation is in practice a complex process
involving specific tools and technology adaptable to the needs of the
translator, who is involved in the whole process and not just in the
editing stage.

The computer becomes a work station where the translator has

access to a variety of texts, tools and programs: for example,
monolingual and bilingual dictionaries, parallel texts, translated texts
in a variety of source and target languages and terminology
databases. Each translator can create a personal work environment
and transform it according to the needs of the specific task.

A wide range of electronic tools and resources are of interest to

translators to help them carry out various translation-related tasks;
however, CAT tools are typically considered to be those designed
specifically with the translation task proper in mind, rather than tools
intended for general applications (e.g. word processors, spelling
checkers, e-mail, work flow and project management). Tools
commonly considered to fall under the CAT umbrella include
translation memory systems, terminology management systems, term
extractors, concordances, localization tools and even machine.

Thus, computer-assisted translation gives the translator on-the spot

flexibility and freedom of movement, together with immediate access
to an astonishing range of up-to-date information. The result is an
enormous saving of time.
The following are the most important computer tools in the
translator‟s workplace, from the most elementary to the most
complex:

Electronic Dictionaries, Glossaries and Terminology Databases:

Electronic dictionaries are available in several forms: as software that
can be installed in the computer, as CD-ROMS and most importantly
through the internet.

Concordances: They are word-processing programs that produce a list

of all the occurrences of a string of letters within a defined corpus with
the objective of establishing patterns that are otherwise not clear.

On-line Bilingual Texts: A bilingual corpus normally consists of a source

text plus its translation, previously carried out by human translator.
This type of document, which is stored electronically, is called a bi-text.
It facilitates later translations by supporting ready solutions to fixed
expressions, thus automating part of the process. The growth of the
translation market has led to increased interest on the part of
companies and international organizations in collections of texts or
corpora in different languages stored systematically online and
available for immediate consultation.

Translation Memories: Translation Memories represent one of the most

important applications of on-line bilingual texts, going back to the
beginning of the 1980s with the pioneering TSS system of ALPS, later
Alpnet. Translation memory programs are based on accumulation and
storing of knowledge that is recycled according to need, automating the
use of terminology and access to dictionaries. A translation memory
can be used in two ways: in interactive mode and in automatic mode.
Limits of Modern Translation Technology: Lack of sufficient linguistic
resources for many languages and domains has been identified as
currently being one of the major obstacles in further advancement of
automated translation. Language reflects society's constant evolution,
which is why it's still easier for humans to keep up than it is for
machines. Natally Kelly, author of Found in Translation explains:
"Professional translators take care to ensure that the message
resonates with a foreign audience as the original author intended it to.
Machines still lack the ability to do this. A machine doesn't have a sense
of humour or the ability to choose the perfect words for a target
audience".

It is important to stress that automatic translation systems are not yet

capable of producing an immediately useable text, as languages are
highly dependent on context and on the different denotations and
connotations of words and word combinations. It is not always possible
to provide full context within the text itself, so that machine translation
is limited to concrete situations and is considered to be primarily a
means of saving time, rather than a replacement for human activity.
Some people are also confirming the fact that Google translate
produces nonsense most times, this is actually fueling a demand for
professional human translation.

Again, machine translation requires post-editing in order to yield

quality target text. So, despite the effectiveness of machine
translations, professional translators must still review the final
translation in order to ensure accuracy and cultural correctness. Also,
the expense associated with translators is excessive. Take for instance,
the European Union must translate over one million pages per year and
as new members are added, more languages require translation [4-6],
also not the expense associated with the translation due to the
concentrated use of human resources. This is as a result of the fact that
machine translation alone does not provide the type of accuracy that is
needed due to its failure to recognize the subtleties within languages
and various companies are attempting to address the deficiency.
According to Belluomini (2006), [7], the ultimate intent of multilingual
communications is being able to not only translate, but provide cultural
interpretation as well.

Contributions of Modern Translation Technologies: Technology has

broken down the barriers of time and space, enabling global business
via constant connectivity. Over the two decades we have witnessed the
enormous growth of information technology with the accompanying
advantages of speed, visual impact, ease of use, convenience and cost-
effectiveness. At the same time, with the development of the global
market, industry and commerce function more than ever on an
international scale, with increasing freedom and flexibility in terms of
exchange of products and services. As more nations open their doors
for noncitizens, translation has become an important tool to bridge the
gap between global and local communication. The nature and function
of translation is inevitably affected by these changes.

Furthermore, the internet with its universal access to information and

instant communication between users has created a physical and
geographical freedom for translators that were inconceivable in the
past. IT has produced a screen culture that tends to replace the print
culture, with printed documents being dispensed with and information
being accessed and relayed directly through computers (e- mail,
databases and other stored information). These computer documents
are instantly available and can be opened and processed with far
greater flexibility than printed matter, with the result that the status of
information itself has changed, becoming either temporary or
permanent according to need. In effect the language barriers to further
globalization are being surmounted by translation technology.
"Automated translation systems may be either a centrifugal factor that
fragments the world, or a centripetal force that binds cultures closer
together. Both trends could well occur simultaneously" [7].

CONCLUSION
It did not take long following the development of the first computers
for researchers to turn their attention to applying these computers to
natural language processing tasks. The translation market has changed
dramatically in the last decade because there has been an increase in
demand of translation services. The sharing of information is at every
level of the human experience today, Skype, face book, create video,
etc. "According to UNICEF more people die from lack of knowledge
than from diseases. People in poor countries are simply unable to
access global knowledge in a language they understand. modern
translation technology may be bringing this knowledge closer to the
bottom three billion....."

Ultimately, translation is important because it facilitates multilingual

communication and allows people from around the world to better
understand one another culturally, economically and socially. However,
different aspects of modern life have led to the need for more efficient
methods of translation. Translators therefore envision a world where
knowledge knows no language barriers. At the present time the
demand for translations is not satisfied because there are not enough
human translators. Furthermore, because companies want to get their
products onto the shelves in all corners of the world as quickly as
possible and because electronic documents such as web pages often
contain content that needs to be updated frequently, deadlines for
completing translation jobs seem to be growing ever shorter.
Translation technologies have provided the answers.
In this paper, we have been able to establish that the discussion should
not continue to be "if" but must shift to "how" translators will
participate (that is to say, not if they should use these technologies but
how can they use them and what do they need as an industry for
consistency). This calls for a positive career growth for the people in the
industry. Translators have to become proficient using the tools of their
trade today (which include devices everyone uses!), for example, smart
phones are now working devices, not just "common communication"
tools. To this, we contribute that translators should accept the new
technologies and learn how to use them to their maximum potential as
a means to increased productivity and quality improvement.

Localization is an umbrella term that refers to the processes whereby

digital content and products developed in one locale are adapted for
sale and use in one or more other locales. Although the term
'localization' has been in use since the early 1980s, confusion persists as
to what exactly it means. To understand localization, it is necessary to
consider when, why and how it arose, the ways it has changed over
time, and its relationship to translation and internationalization. Thus,
this chapter will examine localization and its evolution from the 1980s
to present.

The practice of translation remained relatively unchanged from the

dawn of writing until the commoditization of the PC and the advent of
mass market software ushered in the digital revolution in the 1980s. As
increasing numbers of computers appeared in homes and business
offices, 'typical users were no longer professional computer
programmers, software engineers or hardware engineers' (Uren,
Howard and Perinotti 1993: ix). U.S.-based software companies quickly
realized that by developing products such as spreadsheet programs and
word processors that average people could use for work or leisure, they
could sell to a vastly larger potential market. Targeting average users
instead of computer professionals was not without challenges,
however.

While experienced professionals had become adept in detecting bugs

and working around them, the new users expected, indeed demanded,
that the software they bought operate exactly as described in the
manuals. Benign acceptance of anomalies in the operation of software
could no longer be tolerated.
(Uren, Howard and Perinotti 1993: x)

Initial efforts by software publishers to develop this embryonic mass

market thus focused on improving software reliability and user-
friendliness.

U.S.-based software companies soon broadened the scope of their

marketing efforts beyond the domestic market to target international
users. Expansion into international markets required that software
publishers offer products in languages other than English. 'For a
software product to have wide market acceptance in a non-English-
speaking environment, it was essential to convert the software so that
users saw a product in their own language and firmly based in their
own culture'
(Uren, Howard and Perinotti 1993: x). Software publishers thought that
adapting their products for international markets was merely a matter
of 'translating' software. As a result, initial attempts to adapt software
for international users were characterized as 'translation on the
computer for the computer' (van der Meer 1995). However, it soon
became clear to practitioners that this work was 'related to, but
different from and more involved than, translation' (Lieu 1997). Indeed,
the scope of the undertaking was not confined to translation of text in
the user interface but rather encompassed all target market
requirements for culturally dependent representation of data, including
but not limited to the following:1

character sets, scripts and glyphs for the representation of various

writing systems

encodings to enable the storage, retrieval and manipulation of

multilingual data

text comparison, searching and sorting (collation)

.. line and word breaking

calendars (e.g., Buddhist, Coptic, Gregorian, Hebrew lunar, Hijri,

Japanese Emperor Year, Julian, Year of the Republic of China and
Tangun Era calendars)

date formats (MM/DD/YYYY, DD/MM/YYYY, YYYY-MM-DD, etc.; for

example, 5/11/2014 would be read as May 11, 2014 in the United
States but as November 5, 2014 in Italy)

..time formats (12-hour vs. 24-hour clock; use of AM and PM)

number formats, digit groupings and decimal separators (period vs.
comma)

paper sizes (A3, A4, legal, letter)

.. units of measurement (metric vs. imperial)

In software engineering, these local market requirements are referred

to using the hypernym 'locale'. Locales are expressed as language-
country pairs. Thus, French-Canada is one locale, while French-France is
another' (Cadieux and Esselink 2002). The need to account not only for
translation but also for 'locale' explains why and how the process of
adapting software for international markets came to be known as
'localization' in the early 1980s. The scope and scale of this new activity
expanded so rapidly that within less than a decade localization was

perceived as an industry unto itself, as reflected by the creation of the

Localization Industry Standards Association in 1990 (Lommel 2007: 7).

The costs of adapting software products for other locales seemed like a
small price to pay given the sizable international markets and potential
revenues to which localized products could enable access. Software
publishers approached localization in different ways: some performed
the work using in-house teams and some outsourced the work to
specialized service providers, whereas others assigned responsibility for
localization to in-country subsidiaries or distributors (Esselink 2003b: 4).
Despite the ostensible differences between these approaches, they all
shared one fundamental characteristic: in each case localization was
performed apart from, and subsequent to, the development of the
original, domestic- market products. 'This separation of development
and localization proved troublesome in many respects,' observes
Esselink (2003b: 4).

First, the software provided to localization teams often could not be

localized because it lacked certain fundamental capabilities, such as the
ability to display target-language scripts and writing systems. In such
cases, the localization teams had to send the software back to the
development teams for implementation of the necessary capabilities,
such as support for the display of Asian languages or of right-to-left
scripts for languages such as Arabic and Hebrew. Second, translatable
text was typically embedded in the software source code. Identifying
and locating translatable text was very difficult for localization teams
that had not participated in the development of the software (see
Figure 33.1). Finally, and perhaps most critically, localization required
that changes be made directly to the source code of the software. To
understand why and how working directly in source code caused
problems, it is important to note that software source code is the raw
material from which the running copy of a program is created. In other
words, source code must be compiled, or built, into a machine-readable
(binary) executable file, which in turn must be tested (and debugged, if
any bugs are found) before the software can be released for use (see
Figure 33.2).

Identifying translatable text can be difficult for non-programmers. In

the example shown in Figure 33.1, items in quotation marks are
translatable except for the name of the default font (MS Sans Serif).
Each group of four digits separated by commas represents layout
coordinates.
Working directly in source code had profound ramifications for
localization. Indeed, the adaptation of software products for other
locales did not merely entail a few changes to compiled, tested, and
debugged versions of programs that had already been released to the
domestic market. Instead, localization of a given program required that
a separate set of source code be maintained and that a different
executable be compiled, tested and debugged for each target locale.
Consequently, creating N localized versions of a program required that
the publisher maintain N + 1 sets of source code: one for each target
locale plus one for the domestic market. In addition, each set of source
code had to be localized, compiled, tested, debugged, updated and
managed separately (Luong, Lok, Taylor and Driscoll 1995: 3). For
instance, a U.S.-based publisher that wanted to market a product in
three international locales, such as German-Germany, French-France,
and Japanese-Japan, was required to manage four different versions of
source code in parallel, one set for the domestic English-United States
locale plus one set for each of the three international locales (see
Figure 33.3). The process of compiling and testing localized software
products as distinct from source-locale versions is called localization
engineering (Esselink 2002).

Creating, maintaining and supporting multiple localized versions in

parallel proved to be time-consuming and expensive. Testing and
debugging software was inherently labor-intensive and costly even
without adding localization as a variable. Seminal work on software
engineering economics by Boehm had demonstrated that 'uncorrected
errors become exponentially more costly with each phase in which they
are unresolved (1981: 8). The exponential cost increase of error
correction was exacerbated by the post hoc approach to localization in
which the adaptation of software for other locales and thus the
multiplier problem was compounded by the management of a distinct
set of source codes for each target locale, since a bug discovered in one
set of code might need to be fixed in all other sets. Indeed, localization
engineering has traditionally involved quite a bit of bug fixing,' as
Esselink observes (2002: 4). Not surprisingly, complexity soon
established itself as a hallmark of localization (Esselink 2000b).
Ultimately, most of the problems posed by early localization efforts
stemmed from a single root cause: the failure to effectively plan for the
reuse of software source code across multiple locales.2

Most software and hardware companies that made forays into

international markets quickly concluded that localization and
translation were not an integral part of their business. As Esselink
(2000a: 5) observes, '[t]he increasing size and complexity of localization
projects soon forced companies to an outsourcing model. Most
software publishers simply did not have the time, knowledge or
resources to manage multilingual translation or localization projects. As
a result, most companies decided that it would be more efficient to
outsource the adaptation of software products for international
markets to external language service providers as project work. In
addition to the adaptation of the software application, a typical
software localization project might also involve the translation and/or
adaptation of various other components such as sample files, demos
and tutorials, Help systems, printed and online user documentation, as
well as marketing collateral (see Figure 33.4). The fact that these
components were authored in a variety of digital formats, some of
which needed to be compiled and tested prior to release, meant that
localization involved a number of new forms of work in addition to
traditional translation, including software and online Help engineering
and testing, conversion of documentation to
different formats, translation memory creation and management, as
well as project management (Esselink 2000b; Esselink 2003a: 69; Dunne
and Dunne 2011). Localization thus required that translators possess
strong instrumental and technical skills in addition to traditional
translation and domain expertise. Throughout the 1990s, the
localization industry tried to turn translators into semi-engineers',
recalls Esselink (2003b: 7).

Outsourcing shifted the challenges of managing complex multilingual

localization projects to external service providers but did not address
the fundamental problem of the duplication of effort required to
manage multiple sets of source code in parallel. Faced with the
challenge of controlling the complexity and cost of the localization
process, software publishers in the late 1980s and early 1990s began to
realize that 'certain steps could be performed in advance to make
localization easier: separating translatable text strings from the
executable code, for

example. This was referred to as internationalization or localization-

enablement (Cadieux and Esselink 2002). Internationalization is an
engineering process that precedes localization and entails the
separation of '[a]ll the culturally and linguistically sensitive software
components... from the core of the application' (Hall 1999: 298). In
practice, the scope of internationalization is typically confined to the
linguistic and culturally dependent contents of the user interface that
may require adaptation, which are collectively designated using the
hypernym 'resources. When a piece of software is properly
internationalized, '[t]here is no programming code in the [resources]
nor is there any [translatable] text in the program code' (Uren, Howard
and Perinotti 1993: 60). Resources in a typical desktop software
application may include the following:
What is translation? And when did it started?
Translation is an act of writing that conveys the meaning of a text
written in one language in another language. Evidence exists of
translations of a Sumerian epic dating from the second millenium BC.
When did the professionalization of translation started? And why?
The second half of the twentieth century saw the professionalisation of
translation. because of the need to maximize translator productivity
and boost productivity.
What is translation memory software? = what is the basic function of
the translation memory program? = explain theoretically how does
TM work?
this software commonly known as TM software is likely to be the tool
that all translators will use most in their careers.

In theory, translators start with an empty database, or translation

memory, and build up their own memory sentence by sentence. The
software stores each translated sentence in a database which it then
consults to translate subsequent sentences. which means that a
translator will never have to translate the same sentence twice. Over
several years, the professional translator will build up a substantial
database of translated sentences, which will save a significant amount
of time and effort.
Give examples of commercial translation memory programs used by
translators today.
The principal commercial translation memory programs used by
translators today are "SDL Trados Studio 2014", "memoQ 2014",
"Wordfast Pro" and "Déjà Vu X3".
Explain how can modern commercial CAT tools like “SDL Trados
Studio 2014” help translators.
Modern commercial CAT tools like "SDL Trados Studio 2014" can handle
over 50 different file formats and this enables translators to translate
text in applications without needing to have those programs installed
on their PC. Also the software will support any language combination,
the translator will be able to translate between languages with
completely different scripts.

"SDL Trados Studio 2014" works with the source and target languages
displayed alongside each other. For each new segment (usually a
sentence) in the source text, proposed translations are displayed in a
separate pane, allowing the translator to select the most appropriate
translation which is then inserted in the editor window alongside the
source sentence.
What is the “pre-translate” feature?
The "pre-translate" feature can help the translator by translating the
document by first retrieving 100% matches or fuzzy matches from the
selected translation memories. but it works once the translator has
built up or obtained a large enough translation memory, thus he or she
will achieve significant time savings by automating part of the
translation process with the software's "pre-translate" feature.
Define machine translation (=automatic translation or computer-
generated translation)
Machine translation is the branch of language technology in which a
computer program translates without any human intervention.
What are the factors that the usefulness of machine translation
depends on?
The usefulness of machine translation depends on various factors
including language pair (= the equivalence), the complex ambiguity of
the source text, and the system's familiarity with the terminology and
subject matter of the original.
What is the first big boost that led to the development of machine
translation?
Research and development in machine translation had its first big boost
in the Cold War period.
What are the reasons behind research and development in machine
translation?
1. The expansion of the European institutions
2. International conflicts and the need for better intelligence
3. The need to translate many pieces of content generated every
minute on the Internet
What is the statistical model and what is syntactic?
statistical models are what the translations of the Statistical Machine
are generated on.
when has Statistical Machine Translation become the predominantly
used approach?
Since the mid-1990's Statistical Machine Translation has become the
predominantly used approach.
Are Google Translate and Microsoft Translator statistical or syntactic
systems?
Both Google Translate and Microsoft Translator are fundamentally
statistical systems.
what is the successful use of statistical systems depending on?
The successful use of these systems is broadly dependent on the
quantity and quality of the bilingual data used to build the translation
model.
What is GUI stands for?
What is CAT Tools stands for?
How can translators working on the same project make a consistent
use of terminology?
What is TAUS? What does it do?