ICT/EPC May 2016.indd - Corporate Translations

Image: © everything possible – shutterstock.com
| Translation Tools
Lingua Franca
Trevor Korb
at Corporate
Translations
As the number of multinational clinical trials has grown over the past two decades, another
type of revolution has been taking place – namely the rapid development and adoption of
translation technology, such as computer-assisted and machine translation. What do these
new tools mean for pharma?
The translation industry has come a long
way, thanks to innovative technologies
like computer-assisted translation (CAT)
and machine translation (MT) that give
companies more options than ever before. This
tectonic shift is not simply a consequence of
globalisation, but instead has helped drive and
facilitate the increasing interconnectedness
of governments and large corporations. For
pharmaceutical companies and CROs, these
tools have their greatest potential in the
administration of clinical trials, as they allow
language service providers (LSPs) to keep
pace with the rapid increase in demand for
translation and localisation services – and
enhance the quality and comprehension of
translations for both patients and investigators.
How has translation technology developed
into a must-have for individual translators and
LSPs alike?
Global Expansion
There is no question that clinical trials
are now a global enterprise. As of 2005,
8
l
February 2016
nearly 40% of all trials were conducted in
emerging markets, such as Russia and India,
compared to just 10% in 1999 (1). Likewise,
a cursory glance of registered trials on
clinicaltrials.gov shows that approximately
one third are taking place outside Europe
and North America, and more than half
outside the US (2).
sponsors, however, the question ultimately
boils down to how LSPs can maintain
quality – while also keeping up with the
ever-increasing number of trials – as well as
find more efficient ways to save time and
money during the translation process. This
is where technology, as it has done so many
times before, finds its niche.
This expansion of clinical trials across the
globe has added much more complexity
to their implementation, presenting
sponsors and CROs with a host of logistical,
administrative, regulatory and – above
all – linguistic challenges. Language in
particular cuts through almost every phase
of administering clinical trials, from patient
recruitment and enrolment to reviews
by ethics committees, communications
between investigators and the quality
of data. Accurate and culturally sensitive
translations not only affect the rights
of patients, but also how investigators
understand the treatments themselves
and the results garnered therefrom. For
Past, Present, and Future
Not so long ago, a translator’s primary
toolbox consisted of dictionaries and
typewriters; however, today, the question
for translators is not whether to use
translation technology, but instead which
tools to buy, learn and use (3). Indeed,
“The advent of the information age has
sustainably altered the profession of
translation, especially with regard to the
type of electronic tools used, the type of
texts translated, and the types of skills
needed by today’s translators” (4). While
the information age has similarly impacted
many other traditional industries, the
| M&A Activity Report
as translators continue to become more comfortable using
technology, the future looks bright for the industry overall and CAT,
as well as MT tools, will continue to improve
Translation Memory
Before taking a closer look at CAT tools,
and MT specifically, it is worth mentioning
the backbone of translation technology,
which is translation memory (TM).
Essentially, TM is a database consisting
of previously translated sentences and
phrases for any given language pair.
Translators using TM within software
applications are provided with possible
matches (for example, a 100% match
means an entire phrase has been located
verbatim) based on how much of the
sentence or phrase has been located
in the database. For less than 100%
matches, translators can then choose to
either accept or reject what the memory
is providing, based on context or other
document-specific factors. TM systems
are inherently embedded in the more
sophisticated CAT and MT software tools,
and also allow translators to continually
update and collaborate on large TM
databases through what is called TM
exchange – an XML-based standard
exchange maintained by the Localisation
Industry Standards Association.
SDL Trados
For its part, computer-assisted translation
software first appeared in the 1980s, as
several companies began developing their
own prototypes. At first, these software
applications were too expensive for
individual translators to use, and were only
employed by large companies carrying
out extensive amounts of translation such
as IBM. One of these early programmes,
SDL Trados – which was first created in
Stuttgart in 1984 – has become the
most widely used software among
translators today.
10
l
www.samedanltd.com
CAT tools such as SDL Trados create
an interactive environment between
translators and computers. The software
will break down a document into
particular segments and will make
suggestions for possible translations based
on TM databases and other electronic
dictionaries. Translators can then make
edits within the software itself, and use the
CAT tool as a workbench instead of a fullyfledged translation substitute.
Google Translate
Machine translation, on the other hand, is
essentially computer-assisted translation
without the ‘assisted’ attached, in which
the computer will translate documents in
full – only to be ‘post-edited’ by translators
afterward. The most common example
of MT is Google Translate, which applies
a method known as statistical-based
translation, in which highly sophisticated
algorithms continually improve and learn
based on feedback and input from users.
Long before applications like Google
Translate arrived on the scene, however,
the discussion surrounding MT began with
Martin Kay’s paper on ‘The proper place of
men and machines in language translation’,
which foresaw the modern development
of machine translation, and cautioned
against using it as a catch-all solution. While
machine translation has slowly gained
respectability among translators, it is still
the subject of much research, debate and
skepticism among translators who simply
do not trust the methods involved, or find
it counterproductive as it entails a large
amount of time-consuming post-editing for
low-quality translations.
Image: © Stepan Kapl – shutterstock.com
changes underway in translation have the
potential to ripple through almost every
other sector. As the world becomes more
connected, the need and demand for
communicating across languages becomes
greater, placing the onus on translators
to keep pace and adopt these new
technologies.
| M&A Activity Report
CAT
Taking a closer look at CAT tools specifically,
the immediate advantages are obvious, and
have cemented their necessity in today’s
globalised world in which translation
is an integral part of the information
cycle. Certainly, if used properly, CAT
software greatly enhances productivity
in terms of speed and volume, and can
increase the quality of translation by
ensuring consistency of language within
one document or across multiple similar
documents. The consistency aspect is
especially important when it comes to
medical translation in general, as any
inconsistencies with highly standardised
translations – such as study titles for clinical
trials – can cause misunderstandings
between investigators and regulators. By
breaking up documents into segments, CAT
tools can also identify which text needs to
be translated and which does not, saving
translators and LSPs considerable amount
of time and money with repetitive texts.
Image: © lassedesignen - Fotolia.com
European Commission
Translation Service
A good example of how both CAT and MT
software can be properly implemented
in the production of documents and
the translation process is the European
Commission Translation Service (SdT)
(5). As Pérez explains: “The Commission
produces over one million pages a year in
the form of legislation, calls for proposals,
internal reports and intermediary drafts.
All this documentation is then handed
out to the SdT, which uses EC Systran
(as machine translation), translation
12
l
www.samedanltd.com
memories and other proprietary tools to
meet its clients’ needs” (5). The extent of
automation to apply to a specific translation
is, ultimately, determined by the purpose of
the document, as internal documentation
is typically translated entirely by MT
– whereas legislation is left to human
translation, given those documents’
degree of complexity and liability. A similar
approach could be applied to medical and
pharmaceutical translations, as the ultimate
use and purpose of documents varies
greatly, and sensitive patient materials –
much like legislation – still require human
expertise and judgement.
TMs and terminology databases are not
adequately maintained, it can essentially
create a snowball effect where an inaccurate
translation is unwittingly repeated
throughout a document.
User-Dependent
However, as translators continue to
become more comfortable using
technology, the future looks bright for
the industry overall and CAT, as well as
MT tools, will continue to improve. Currently
being discussed as the next step from
MT and CAT is controlled translation, or
controlled language. Controlled language
– although on the surface a contradiction
to the concept of language as fluid and
dynamic – essentially streamlines the work
of a computer. By “restricting vocabulary size,
applying certain grammatical constraints,
and encouraging clear and direct writing,
texts are more easily processed by MT. In
this sense it is expected that controlled
languages will expand to other areas
such as software localisation, information
technology, etc” (5).
Despite the obvious benefits to CAT tools,
it is important to remember that all tools
are still largely dependent on how we use
them. Any CAT or MT software can still fail,
causing translators to spend more time
with IT issues than attending to the actual
business. On top of this, incompatibility
problems can also arise between freelance
translators and LSPs, spurring more
research into creating open standards
between different software packages and
tools (6).
While consistency regarding terminology
and phrasing can be a strength of translation
technology, it can also be a weakness. As
much as 70% of translation errors found
by users of SDL Trados are, in fact, due to
incorrect terminology, placing an even
greater impetus on translators themselves
to maintain terminology databases (7). If
Thus, while CAT tools have made a
tremendous positive impact on the overall
translation production process, it is still
largely dependent on the correct usage
by translators and editors, making it
imperative that all individuals involved
with the translation process understand
and familiarise themselves with using the
tools correctly.
For the most part, however, machine
translation still requires extensive postediting and, while certainly there are
translation errors committed by humans
every day, these still pale in comparison
with errors made by machine translation
(8). Indeed, as Hutchins states, the “aim of
using computers for translation is not to
emulate or rival human translation”, but
instead to support it (9). As both CAT tools
and MT continue to see further adoption by
LSPs and other major companies, translators
have been forced to expand their skillsets
and have even begun to double as technical
writers, blending their roles to accommodate
the technological shifts and using their
cultural and grammatical expertise to fill in
the gaps where computers simply cannot.
What Does This Mean for Pharma?
Both the globalisation of trials and the rise
of translation technology overlap in a few
key ways: the sheer volume of translation
now required, and the demand for speed,
consistency of terminology, and – perhaps
most importantly for clients – reduction in
costs. This last aspect, of course, cannot be
overstated – as, despite the attractiveness
of offshoring clinical trials due to lower
costs, the extra regulatory and linguistic
hurdles are instead proving to be bigger
headaches than many companies originally
anticipated. However, recent trends
in translation can at least somewhat
mitigate these complexities. With better
technology and their responsible utilisation
and management, large and repetitive
documents (such as batch records) can
be easily churned out in high quality. As a
result, translation costs have been trending
downward (10).
Translation technology also has tremendous
potential – not only in facilitating the sheer
volume of documentation created over
the course of a clinical trial, but in ensuring
consistency and quality among translations,
and can also help overcome language
and cultural barriers. Still, not every
document should be translated entirely
by a computer, and in most scenarios for
medical translations, the best solution is a
blend between man and machine provided
by a CAT tool. For example, patient material
such as informed consents or recruiting
materials require a higher degree of cultural
understanding and are more nuanced than
other documents.
Technology in these cases should be used
more cautiously, as the language in such
documents undergoes more regulatory
and ethical scrutiny, and can be much more
fluid than the static terminology found
in protocols or investigator brochures. In
addition, regarding translation production
itself, it may be cheaper and more beneficial
to use an alternative method of document
comparisons (translating one document
in full and then using that translation to
update near-identical documents) rather
than applying TM.
In an ideal world, as Martin Kay points out:
“A computer is a device that can be used
to magnify human productivity. Properly
used, it does not dehumanise by imposing
its own Orwellian stamp on the products of
the human spirit and the dignity of human
labor but, by taking over what is mechanical
and routine, it frees human beings for what
is essentially human” (11). This could not be
more important than in the field of medical
translations, as the complex cognitive
activities of translators trickle down to the
lives of patients themselves, and can greatly
affect the outcome and results of clinical
trials as a whole.
Part of the Equation
While machine translation and CAT tools
have improved dramatically since their
introduction, claims that translators will
soon be replaced by technology are
premature. Machine translation may
provide the general gist of a document,
but in the business world, a rough
translation is not enough and high-quality
translations still rely on human expertise
(12). For pharma companies and CROs
especially, translation technology should
be used wisely. Although applications can
certainly help by translating large amounts
of documentation at a fast pace – and if
used appropriately can ensure linguistic
and cultural consistency – not every type
of medical document lends itself to such a
process. In the end, whether a document
is translated in full by a machine and later
post-edited, or is translated through an
interactive man-machine relationship
provided by a CAT tool, humans are still an
essential part of the equation.
References
1. Lustgarten A, Drug testing goes offshore,
Fortune, August 2005. Visit: http://archive.
fortune.com/magazines/fortune/fortune_
archive/2005/08/08/8267653/index.htm
2.Visit: www.clinicaltrials.gov/ct2/
search/map
3.Elimam AS, The impact of translation
memory tools on the translation
profession, Translation Journal
11(1), January 2007. Visit: www.
translationjournal.net/journal/
39TM.htm
4.Maylath B, Current trends in translation,
Communication & Language at Work 2:
pp41-50, January 2013
5.Pérez CR, Overcoming the language
barrier: Paths that converge in technology
and translation, Speaking in Tongues:
Language across Contexts and Users, 2003
6.G ough J, A troubled relationship: The
compatibility of CAT tools, Translation
Automation User Society, December
2010. Visit: www.taus.net/think-tank/
articles/translate-articles/a-troubledrelationship-the-compatibility-ofcat-tools
7.Morgan D, SDL Webinar: An introduction to
terminology management. Visit: https://sdl.
webex.com/sdl/onstage/g.php?MTID=e7b4
c5e7cc8258045bf016be091b3a942
8.Stejskal J, From the President: Translate
server error, The ATA Chronicle 38(1): p7,
2009
9.Piekkari R et al, Language in international
business: The multilingual reality of global
business expansion, 2014
10. DePalma DA and Stewart RG, Trends
in translation pricing: Falling rates
accompany changes in economy and
buying behaviors, Common Sense
Advisory, September 2012
11. Kay M, The proper place of men and
machines in language translation,
Machine Translation 12: pp3-23, 1997
12. Visit: www.economist.com/news/
business/21642187-technology-may-notreplace-human-translators-it-will-helpthem-work-better-say-what
Before working at Corporate
Translations as a Quality
Assurance Editor, Trevor Korb
received a BA in German
Studies and Philosophy, and
an MA in German Studies
from the University of Connecticut, US, as
well as a Graduate Certificate in Human
Rights. His interests are varied, ranging
from political philosophy, international
law and foreign policy to the philosophy of
language and linguistics.
Email: [email protected]
www.samedanltd.com
l
13