Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 5

The Symbiosis between Translator & Computer

Introduction
For the last three decades, the world has been a witness of a colossal technological expansion.
Translation Studies, translators and their activities were also impacted by these advancements.
The accelerated technological explosion enabled an interaction between translators and
computers. This interaction was displayed in many forms, the more distinguished and intricate
nowadays being CAT tools and machine translation (MT).
Both softwares have a unique relation with the translator. When in use, they fundamentally
influence and change the way in which translators approach their translations, ‘forcing’ the
translator to adapt to the new situation.

 Say what the topic of your essay is


 Say how the topic will be approached
 Signpost how the essay is organised
The topic of my assignment essay focuses on the translator’s relation with CAT tools and
machine translation and how
The essay is divided into three sections: a discussion regarding the kind of relationship which
the translator has w
It is the latter in translation technology that begs the question of whether the translators still
hold any power over the machine, or they have converted into their fuel and are simply living in
their own ‘simulated reality’.
CAT Tools
As mentioned above, technology has changed the way translators’ work. On one hand,
technology has become the standard – although not impossible, it is rare the translator who does
not use a computer as their main translation device. (Läubli and Green, 2019). On the other,
technology only exists and continues to evolve due to humans’ intelligence and skills. In return,
these improvements then increase the usage of technology in various aspects/tasks. Although
some of the technologies were more easily accepted than others, in particular if the benefits
were immediate (online dictionaries and encyclopaedias, word processing – rapid access to
information all in one apparatus), a dependency was created, thus birthing the symbiotic
relationship between human and computer (O’Brien, 2012; Zetzsche, 2019).
A perfect example of this symbiosis is between translators and CAT (computer-aided-
translation) tools. Like the name indicates, CAT, a computer software, assists the translator with
the translation, without taking over the translator’s main responsibility – translating the text.
CAT storages previous translations/information which can then be of used for newer
translations. However, this assistance only occurs after the software’s depository is ‘fed’ with
human produced translations, otherwise the tool is simply idle.
Before CAT tools, translators would mark down repetitive content in word processing software.
Then in previous translated documents, they would select the pertinent part of the text and copy
and paste into the new target text. This action would consume large amounts of time and many
times would cause unchecked errors which were then perpetuated into other translations.
Despite their situation, translators were not pleased when CAT tools began to emerge. Before
being able to use the software, translators had to learn how to use its interface, which not only
took time but there was also a risk of low productivity until mastering the computer (Bowker &
Fisher, 2010). Additionally, it was believed that with the use of CAT tools, translators would
lose ownership of their knowledge, training, skills and produced translation as majority of text
was being provided by a computer (Zetzsche, 2019). These, plus the lack of accessibility to the
software due to high prices frustrated the translator, especially freelancers, hence the initial
refusal.
Nowadays, CAT tools are a staple in the translation process as in the words of O’Brien (2012,
p.9) the computer comes with:
Realistic benefits including faster throughput, increased consistency, lower costs for
clients, possibly leading to higher volumes being translated, as well as increased access
to information in language not normally seen as being commercially important.
This is particularly accurate in long repetitive texts as the tools – translation memory,
terminology, concordance, etc. – enable a minimization of errors, relieve the translator of
translating repetitive sentences (which can become a boring task), while increasing quality,
productivity and reducing the time spent on a translation (Bowker & Fisher, 2010; O’Brien,
2012). Furthermore, since these tools only aid the translation, the translator is not obliged to
copy and paste every match (full or fuzzy) that is presented. The translator is who translates and
ultimately, it’s the professional’s decision whether it accepts, modifies or rejects the suggestions
provided by the computer (Bowker & Fisher, 2010).
However, there are some disadvantages with CAT tools which can complicate the translator’s
job. The long repetitive texts usually equal restrictive text domains or genre. It does not
diminish the helpfulness of a CAT tool, but it does limit the type of texts which CAT tools can
be useful (ex: CAT tools would not be helpful in translating a literary text due to the genre’s
ambiguous and shifting language). Moreover, if translators decide to accept reusing/recycling
previous translations (segments/terms), then the translators run the risk of “propagating one
mode of expression over alternative ones ... and eventually to a lower level of variability in
translated language” (O’Brien, 2012, p. 14). Clients can also ‘give’ their own tools (translation
memory), which can restrict variability even more. All of these negative elements have serious
implications in the translator’s livelihood as clients believe translators should be paid less when
using a CAT tool because of the less variability and because they believe translators did not
exert the same effort and hard work as if translating from scratch, which is simply not true
(Bowker & Fisher, 2010; O’Brien, 2012).
Nevertheless, even with these drawbacks, CAT tools, whether translators choose to make use of
the software or not, are held in high standards as they help the translator guarantee consistency
during the entire translation process, further solidifying the translator/computer symbiotic
interaction.
Machine Translation
Nowadays, the most prominent symbiotic interaction is perhaps between translator and MT.
Except, while CAT and translator’s symbiosis mutually benefit both parties (mutualism), the
symbiosis between MT and translator could be said that it has become parasitic.
In O’Brien’s (2012, p.7) words, the concept of MT is defined as:
A human using a computer, sends a sentence to the MT system and, within
milliseconds, the sentence has been translated by a program into another language.
Where high-quality published material is required, translators are called upon to fix the
still existing errors in the often imperfect computer-generated translation ... (post-
editing). [The] corrections are channelled back to the MT system so it can ‘learn from
its mistakes and from the human’s corrections.
Essentially, while the machine produces extremely quick translations, it also strips away the
need for a translator. A translator’s main responsibility no longer is to translate the text. Instead,
the professional assumes an editor role – a secondary position in relation to the machine – and
their main function shifts into proofreading and correcting/editing any errors generated by the
machine – post-editing (PE).
The translator is still part of the translation process, but no longer take any part in the
translation.
Although MT is accepted by some, the general consensus is not positive. The same is applied to
post-editing (Vieira, 2019).
Even when using a CAT, the translator’s role in the translation is still very active. That is not
the case with MT. The translator, post-editing a raw MT, is put in a passive/submissive role. In
this role, the translator continues to improve the machine by furthering closing the gap between
a error generated MT to the MT’s ultimate goal: being able to produce a complete automated
MT without PE which equals to a Human Translation (HT) (Vieira, 2019). So, just like a
parasitism relation, whilst the machine continues to benefit from the translator – learning and
improving from human translations plus the readily available information on the Internet – the
professional ultimately comes out harmed as their skills, talent and overall work become even
more devalued. A translator trains for years to obtain all of their competences only for the
machine to perform the task much faster and then have the translator not make full use of their
capabilities.
This also frustrates the translator, diminishing the satisfaction of the job, because not only the
professional cannot make use of all of their training but is also correcting raw MT generated
linguistic errors – unintelligent mistakes that the translator was taught to avoid and that MT
constantly incurs in (O’Brien, 2012). In theory, MT + PE is a way to finish a translation faster,
and thus enable the translator to focus on editing to achieve the best possible result. This would
then bring benefits such as allowing the translator to accept more work than a regular basis.
However, when in practice, that is not the case. In fact, in some cases, these errors can actually
cause post-editing to consume more time than translating from scratch. Similarly, the raw MT
itself is also a reason as to why post-editing can also take longer than translating from scratch.
Arenas & Toral (2021) conducted a study which compares raw MT, post-edited MT and human
translation. When questioned about post-editing MT, the general consensus amongst the
translators’ responses were that they felt their creativity was constricted and limited – they
could not really think outside the box when they had already been provided with a translation –
it also made them uncomfortable having to act as a fixer, instead of writing their own
translations, diminishing their sense of ownership and feeling like the final product of post-
editing is not as satisfactory as translating from scratch (Arenas & Toral, 2021).
Albeit the study focused on translating fiction, these testimonies fit the consensus surrounding
MT and PE – stifles creativity and leads the translators to not feel good about their edited final
product – further adding more to their refusal of the machine (O’Brien, 2012; Vieira, 2019).
Another aspect is that machine translation and post-editing have a negative impact in the
translator’s income. Having a machine in the place of a translator will significantly lower their
fees, as clients will no longer be paying a translator, but an editor to correct the errors generated
by the machine (O’Brien, 2012). As already shown, post-editing and translating from scratch
take around the same time, so, post-editing, contrary to what some may indicate, may not be a
financially sustainable alternative (Vieira, 2019).
In theory, given that human brains are often compared to computers, it is easy to understand
why people believed that computers would also be capable of producing the same translations
as humans. However, when put into practice, this is not possible because computers lead with
logic, and so they can compute the literary part of the language – ambiguity, different words
sharing the same meaning and idiomatic expressions (Christensen, 2018, cited in Melby, 2019).
Computers cannot replicate the source content and form onto the target text because they do not
have the tools to recognise the type of text which is translating, leading to incur in many of the
mentioned errors. Also, language is extremely ambiguous and in order to produce a minimal
error free translation MT needs to be adequate to their purpose, which it clearly is not (Arenas
& Toral, 2021).
Zaros (2010) quoted in Zetzsche (2019, p.170), perfectly describes how translators succeed
where machines fail:
[Translators] ‘transformed’ a message in the source language into a message that could
be readily understood by target-language readers. ... To do this well requires knowledge,
skill, training and, yes, talent. ... A ‘real’ translator crafted each document with care.
Although initially used as an argument to ‘attack’ CAT tools, this passage perfectly sums up
why, despite the continuous evolution of MT, translators will continue to be needed. It does not
also mean that translators will stop acting as editors. MT is here to stay. What it means is that
the raw MT just does not compare to human translations, and despite its benefits, they do not
out weight the disadvantages. Although translators, regardless of the professionals’ level of
experience, can acknowledge MT’s positive points, the scale ultimately favours translators and
their art – even though most of the time, the professionals still lack the much-needed
recognition.
Translators also recognise that even though the quality of raw MT is not sufficiently good to be
used in more critical/professional scenarios, Google Translate processing 143 billion words a
day in 2018 – a number which in 2022 has most likely risen – proves that for quotidian
purposes, the machine perfectly suits its function (Wood, 2018, cited in Pym & Simón, 2021, p.
41). Despite the generated errors, the message of MT can still be understood by users who
understand the target language (Forcada, 2010).
Zaros (2010) observation plus Arenas & Toral (2021) study answer why when studies are made
comparing MT+PE and Human Translation (HT), although HT always surpasses MT+PE, the
gap between the results in terms of time, quality, productivity is very slim. And it is all because
of the translator. Regardless of how a translation is created – human or machine – the translator
is always involved in the process. In both frameworks, translator inputs their skills and
knowledge, hence, the similarity in results.
Conclusion
O’Brien states that to improve the relation between human and computer, “what is needed are
efforts to promote symbiosis, rather than friction” (2012, p.19).
The fact is that there is already a symbiotic relationship between human and computer.
However, these vary depending the computer – or machine. CAT and translators have a relation
which mutually benefits both parties, whereas MT and translator’s relation leaves the translator
in ‘harm’s way.’
Despite the less advantageous situation in the relation, translator continues in a position that is
helpful to MT while running the risk of losing their job and ultimately their livelihood.
Machine translation is here to stay. There are no doubts of that. Due to globalisation and the
increasing speeding which world wide companies have begun to use computer technologies, the
machine has stopped being just a translation software and has evolved into a communication
tool for the day-to-day, where its users can easily understand the conveyed content of MT,
despite its errors (O’Brien, 2012; Folaron, 2010; Yamada, 2019).
However, it is also clear that MT will never take the place of professional translators (Forcada,
2010). So, in order for both to co-exist, and subsequently improve their symbiosis, their overall
interaction has to change. Simply put, MT has to adapt to the same level of mutualism present
between CAT tools and translator. And for that to happen, the machine then must return to the
translator the active role in their relation, and therefore, the main responsibility it comes with it
– translating the text. On the other hand, MT adopts the function of a more advanced CAT tool
– MAT (machine aided translation). The machine suggests possible translations of sentence,
rather than whole bodies of text, the translator consults it, knowing that the final decision of
whether to use it falls upon the translator (Zetzsche, 2019; Vieira, 2019).

You might also like