Sei sulla pagina 1di 7

沈瑞芝

201819600466
英语笔译

Post-Editing as an Emerging Practice and Surrounding Issues


Abstract: Our increasingly globalized world results in more demands for contents to be
translated than ever. MTPE emerges as a productive and efficient solution to satisfy the
translation needs in place of human translation. Post-editing is an improvement step by human to
ensure the quality of MT output. As MTPE model becomes more widely used, this essay will
explore and discuss some issues surrounding post-editing. Considered a fairly new practice, post-
editing appears to be recognized mostly by those closely engaged. Despite gradually wider use of
MTPE, many people are still unaware of or have misunderstandings towards post-editing. This
lack of recognition of post-editing as an established practice and profession not identical to
translation seems to be an underlying theme to the community‟s confusion or reluctance towards
itself, an unclear standard in quality control, and insufficiency in post-editor professional training.
This essay also discusses what kind of improvements can be implemented in such context.

Keywords: post-editing, machine translation, translation industry, MTPE

I. Introduction
To keep up with the paces and demands in cross-cultural communications in our globalizing era,
the translation process has become more and more industrialized, leading the practice of machine
translation (MT) to become a significant part of the translation and localization industry. Its
ability to complete a large amount of translation tasks in a short span of time allows it to become
more widely utilized by both professional translators and general community. While in a more
“traditional” area of translation such as literature or poetry, some translators may still rely on the
conventional way of human translation or computer-assisted translation; people are increasingly
turning to employ machine translation for tasks that demand a rather formulaic, standardized
style of language. Machine translation‟s advantages in creating an output lie in its speed and
quantity, in other words, productivity, but there is one more important aspect that should be
concerned: quality, which brings forth the concept of post-editing (PE).
According to ISO 18587:2017, “[…] there is no MT system with an output which can be
qualified as equal to the output of human translation and, therefore, the final quality of the
translation output still depends on human translators and, for this purpose, their competence in
post-editing.” As such, in order to obtain the desirable translation result, both in terms of
linguistic quality and faithfulness to the source text, post-editing is a practice indivisible to the
process of machine translation; one may say it even serves as the determining factor in producing
good quality output. At the same time, though post-editing as a service has gained more attention
along with the emergence of machine translation and its increasing quality, has it gained an
established status in the industry? With MTPE translation model becoming more and more
regularized, this essay would like to explore and discuss post-editing as a practice in today
translation and localization industry, issues surrounding it, as well as what improvements can be
made to benefit the efficiency of post-editing, the quality of translation output, including the

1
沈瑞芝
201819600466
英语笔译

translation industry as a whole.

II. Definition and Types of Post-Editing


In terms of definition, many views seem to agree upon what is post-editing. Though some slight
differences are prevalent, one common concept is the human participation in the process that
brings improvement to the machine output. We can take a look at some of the existing
definitions:

“Post-editing means checking, proof-reading and revising translations carried out by any
kind of translating automaton.” (Gouadec, 2007)

“the process of improving a machine-generated translation with a minimum of manual


labor” (TAUS, 2010, as cited in Zaretskaya, 2017)

“[post-edit] edit and correct machine translation output” (ISO 18587, 2017)

From the three definitions above, it can be seen that the one given by the Translation
Automation User Society (TAUS) focuses on the bare involvement of human translators/post-
editors, while the other two do not refer to any scale in particular. Such departing details bring us
to two narrowed-down concepts: light and full post-editing. ISO 18587:2017 states that light
post-editing aims “to obtain a merely comprehensible text without any attempt to produce a
product comparable to a product obtained by human translation”, while full post-editing is the
process “to obtain a product comparable to a product obtained by human translation”. Here, a
distinction line between the two is drawn according to the extent of how much human
involvement in the process. As for which kind of post-editing should be applied, it depends
largely on clients‟ intended purposes. Light post-editing aims for comprehensible results, and
thus mainly deals with the major grammatical errors and factual correctness, so long as the
translation can convey the overall meaning of source text. On the other hand, full post-editing is
definitely a more heavily modification process, since it aims for the same quality as human
translation. Thus, other than grammatical and factual precisions, texts undergone full post-editing
should be “faithfully” as well as stylistically close to source text, terminologies must be applied
accurately and consistently, and appropriately formatted. The implication here is that such
differentiation has an impact on the time spent on project, the quality of the translation, as well
as pricing, all of which are directly related to satisfying clients‟ demands. Light post-editing is
apparently a less taxing task, thus demands less time, labor, and cost (arguably, light post-editing
can be a more difficult job for very detailed-oriented translators, as it may be hard for them to
overlook some mistakes [RWS, 2014]); whereas full post-editing should be viewed in the same
light as human translation, and thus priced at the same rate.

In addition, another issue worth exploring is the type of texts that are appropriate for the
MTPE model, as it determines what kind of post-editing should be implemented. As post-editing

2
沈瑞芝
201819600466
英语笔译

is an integrated part to machine translation, texts that undergo post-editing are often limited to
certain types suitable for machine translation, for example, texts that display a lot of repetitive
information or terminologies; texts with standardized, formulaic pattern that can be readily
substituted by translation software with the help of translation memory; in some cases, texts that
clients only want a sufficiently “understandable” translation. The range may include weather
reports, organization‟s annual reports, formal government documents, instruction manuals, and
even information from online content, etc.

III. Issues Surrounding Machine Translation and Post-Editing


As the increasingly globalized world results in rising demands for machine translation, it in turn
gives rise to post-editing as a service. More people turn to favor the MTPE model over
“traditional” human translation for its time-saving and cost-cutting advantages. At the same time,
post-editing also becomes more widely used through wider uses of free, online machine
translation systems such as Google Translate, Microsoft Translator, etc. To those already in the
translation industry and willing to embrace new technology, it seems the era of post-editing has
arrived. With machine translation systems‟ relentless advancement, it is not difficult to imagine
MTPE‟s potentially increasing significance in the years to come. Yet for those outside the
industry, they may not be aware of the concept at all. This section will further explore some
issues surrounding the subject of post-editing; while the suggestions on possible improvements
will be discussed in the next section.

Firstly, the general community‟s recognition of post-editing, or lack thereof, may result
as a professional implication for those in the industry. In general, people‟s acknowledgements
and views on the concept of post-editing seem to vary extensively despite it becoming used in
wider community. In some cases, those completely unfamiliar with the translation industry may
be unaware of the term post-editing altogether. While some, should the need to hire a translator
arise, may want to have the impression that post-editing is somehow inferior to translating, since
it is integrated to machine translation, and so implies a lighter, more trivial load of work. While
certain light post-editing work may fit the description, such unawareness or misunderstanding
about post-editing matters because it has an influence on the pricing of a translation project. If
clients are not aware of differences between translating and post-editing and the range of tasks
each can entail, they may assume that the cost of post-editing should always be lower, which
would not benefit anyone involved.

On the other hand, though post-editing is starting to become more widely accepted and
practiced, some linguists that have been in the industry for a long time shows reluctance towards
doing post-editing work (Zaretskaya, 2017). In the context of this particular case, Zaretskaya
states that the linguists‟ skepticism is due to their lack of understanding about the company‟s
machine translation system, that it may cause a slower working pace; a misunderstanding that
post-editing work would pay lower; and some think that post-editing is similar to proof-reading,

3
沈瑞芝
201819600466
英语笔译

and therefore not their acquainted area. Although these causes may not necessarily be true
outside this case, it still demonstrates that even veterans in the industry may not fully grasp the
concept of post-editing. This goes back to the issue of awareness, though here it presents another
kind of consequence. Although there are practical differences between translating and post-
editing, certain aspects are still overlapped and those who were trained in translation should be
able to navigate post-editing to a certain extent. If these experienced linguists reject post-editing
work altogether only because they do not see it as a viable practice, their invaluable expertise
and experiences would become unutilized and potential projects would be lost.

The second issue to explore here is the quality of post-edited work and what determines
the desirable outcome. There are several factors to be considered in regards to this: the quality of
source text, the quality of “raw” machine translation output, the adjustment of machine
translation system and its relevant functions, whether light or full post-editing is applied, post-
editor‟ skills, as well as client‟s expectation. With so many involving factors, which are not only
limited to post-editing process but also a process before it, post-edited work is more vulnerable
and susceptible to errors, which would lower its desired quality. As such, in order to achieve a
desirable outcome, it is essential that these factors are controlled and optimized. Yet the aim to
quality control post-editing can become a challenge, as post-editing is an emerging practice
relying on a developing technology. To illustrate, ISO 18587:2017 interestingly states that “The
rate at which MT systems are changing renders it impractical to produce an overarching
International Standard on these systems, which could stifle innovation or be ignored by the
translation technology development industry.” Thus the on-going improvement of machine
translation technology is a reason not to have an “overarching” international standard for
machine translation systems. While its practicality is understandable, this may have an effect on
post-edited product quality, since such standard and the system itself have a direct impact on
machine output, which is one of the factors determining the quality of post-edited work.
Moreover, the fact that post-editing is very integrated to machine translation makes it rarely
recognized as something distinct to translation as a practice. Consequently, in cases where clients
expect to achieve quality close to human translation, the quality standard requirements for post-
edited work may share a lot of similarities to that of human translation. Although for the time
being, post-editing can be subjected to standards used in translation, but in the long run the lack
of its own formal quality standard would not benefit the quality of the MTPE output.

Thirdly, partly in relation to the lack of acknowledgement of post-editing work and what
tasks it entails, post-editing as a profession is still largely not recognized and regularized, or even
not systematically trained. From the aforementioned case, some linguists misunderstood post-
editing as similar to proof-reading, while many others would see it as a job any translator can do.
While such view is not completely incorrect, this certainly resembles the matter of quality
control, where post-editing is assumingly the same as translating. Although translation and post-
editing goals are similar in that they both aim to produce a product that satisfies target audiences‟

4
沈瑞芝
201819600466
英语笔译

needs, and certain aspects of the two tasks are overlapping, the fact is still that they require
different sets of skills (O‟Brien, 2019). For examples, while translators work with only one
source text, post-editors have to work with two: the actual source text and the machine output;
not to mention the fact that post-editors have to deal with machine translation output, which in
itself presents a different kind of linguistic style and errors, etc. At present, translators may be
able to take up post-editing work to a certain extent, but it would be more effective in the long
term if they are trained specifically for the task. Moreover, it is foreseeable that in the near future
the world will become more globalized and interconnected; the use of MTPE will potentially
increase and become widespread in order to satisfy the needs that human translation is unable to.
It will be beneficial to the industry to train sufficient workforces who have specialized post-
editing skill, as well as can readily satisfy and customize to the demands of clients.

IV. Possible Improvements on the Issues Surrounding Post-Editing


From these aforementioned issues, they seem to point to a common underlying theme: the lack of
awareness, acknowledgement, or recognition of post-editing as a practice not identical to
translation. While it is not uncommon for post-editing to remain as such since it is an emerging
professional practice, the trend in the translation industry seems to show post-editing is here to
stay. This section will focus on some improvements and what can be done that may help address
the issues surrounding post-editing.

First of all, it will be beneficial to all parties involved if awareness, acknowledgement,


and recognition of post-editing are promoted, not only among the general community, potential
clients, and practitioners, but also among those already in translation and localization professions.
In other words, if possible, post-editing as a profession should become more firmly established
and attain a more formal status than at present. It can help smooth out misunderstandings and
confusion towards post-editing and MTPE in general; give out a clearer scope of post-editing
process; allow post-editors, translation agencies, and clients to set up realistic expectations and
common, desirable goals, which will most importantly result in satisfaction with the quality of
final product and wage negotiation. In a professional context, an establishment of an association
of post-editors can help provide a form of community; set up standards, codes of conduct, or
work ethics; or provide them a professional platform to share knowledge and experiences. On the
other hand, it will also shed light on what kind of tasks do post-editors engage, encouraging
potential students, translators, or anyone interested in this line of work to enter the professional
pool with a clear idea, expectation, and preparation.

Next, there are certain improvements that can be done in regards to the quality of post-
edited work. As mentioned in the previous section, post-editing is often put in the same box as
translating, and that seems to include the matter of quality control as well. However, there are so
many involving factors that make post-editing work very susceptible to errors, and in turn
creating less than desired outcome. Here, we can make some efforts to control these determining

5
沈瑞芝
201819600466
英语笔译

factors to ensure the desired quality of the final product. For examples, the source text should
from the beginning be written in good quality and organized in a format appropriate to be
processed by MT system. At the same time, the MT system should be customized, e.g. equipped
with relevant functions or ensuring the application of terminology management through the use
of translation memory, in order to optimize the raw machine output. In the past few years, there
have been discussions about post-editing technology development, e.g. a post-editing function
incorporated to translation software (O‟Brien et al., 2014). This could mean that future MT
systems would allow post-editing work to become even easier, though also implies that human
post-editors may become obsolete at one point. But for the time being, post-editors can rely on
their skills and the available technology to complete tasks at hand, while making an attempt to
keep up with the ever evolving MT technological developments. In addition, the quality of the
finished product will be more likely to be satisfactory, if post-editors can discuss specific
requirements or acquire style guides from clients before starting the task.

Finally, the need to systematically train professional post-editors and cultivate post-
editing skill should be addressed and responded, since it appears to be a more sustainable way of
ensuring quality standard, as well as creating wider recognition of post-editing. Moreover, the
industry will also have available workforce that can readily satisfy the ever growing demands for
content to be translated. As of present, more institutions of higher level education, translation
associations, as well as MT system firms are starting to offer courses and programs in MT post-
editing, although the availability is undoubtedly not as expansive as translation courses. In terms
of what should be taught in post-editing training programs, O‟Brien (2019) has suggested a
number of skills that post-editors should cultivate: knowledge of MT, terminology management
skills, pre-editing/controlled language skilled, programming skills, and text linguistic skills. The
need of these skills only serves to highlight differences between translating and post-editing. As
post-editing comes as a package with machine translation, one of the necessary qualifications of
post-editors then is the willingness to embrace MT and relevant technology. In other words,
while translators can choose for themselves to what extent they want to employ MT systems in
their work process, post-editors must work closely with it and become familiarized with its
output. Again, if those interested in post-editing job are made clear of such differences from an
early stage, it will better prepare them before signing up for the profession. Here, O‟Brien also
suggests that post-editing skill is developed gradually. This indicates that the training will take
some time before post-editors become accustomed to their work process. If the translation
industry is gearing up for even more demands for MTPE and post-editors of the years to come,
systematic professional training of post-editors needs to be implemented as soon as possible.

V. Conclusion
Our increasingly globalized world means there are more cross-cultural communications and
demands for contents to be translated than ever before. Machine translation emerges as a
productive and efficient solution to satisfy the translation needs that traditional human translation

6
沈瑞芝
201819600466
英语笔译

cannot keep up with. Its emergence in turn gives rise to post-editing: a human touch of
improvement to ensure the quality of MT output. As MTPE model becomes more widely used,
both in the general community and the translation and localization industry, this essay has
explored and discussed a few issues surrounding post-editing and what seems to be their
implication. Considered a fairly new practice, post-editing appears to be recognized mostly by
those closely engaged with it. Despite the gradually wider use of MTPE model, many people are
still unaware of or even have misunderstandings towards post-editing. This lack of recognition of
post-editing as an established practice and profession not identical to translation seems to be an
underlying theme to the community‟s confusion or reluctance towards itself, an unclear standard
in quality control, as well as the insufficiency in post-editor professional training. This essay also
discusses what kind of improvements can be implemented in such context, in hope that post-
editing as well as MTPE model will become as a sustainable solution to the future demands of
translation.

V. References
Gouadec, Daniel. Translation as a Profession. John Benjamins, 2007.

ISO 18587:2017, Translation services — Post-editing of machine translation output —


Requirements

“Light and Full MT Post-Editing Explained.” RWS. 19 AUG 2014, www.rws.com/insights/rws-


moravia-blog/bid-353532-Light-and-Full-MT-Post-Editing-Explained. Accessed 16 Jan.
2019.

“MT Post-editing Guidelines.” TAUS (Translation Automation User Society), November 2010,
www.taus.net/academy/best-practices/postedit-best-practices/machine-translation-post-
editing-guidelines. Accessed 16 Jan. 2019.

O‟Brien, Sharon. “Teaching Post-Editing: A Proposal for Course Content'”. ResearchGate. 2019,
www.researchgate.net/publication/228787033_Teaching_Post-
Editing_A_Proposal_for_Course_Content'. Accessed 17 Jan. 2019.

O‟Brien, Sharon, et al. Post-editing of Machine Translation: Processes and Applications.


Cambridge Scholars Publishing. 2014.

Zaretskaya, Ana. “Machine Translation PostEditing at TransPerfect – the„Human‟ Side of the


Process.” Revista Tradumàtica. Tecnologies de la Traducció, vol. 15, pp. 1-11. December
2017, ddd.uab.cat/pub/tradumatica/tradumatica_a2017n15/tradumatica_a2017n15p116.
pdf. Accessed 16 Jan. 2019.

Potrebbero piacerti anche