Human translation versus machine translation tools: Are human translators using them? A translator survey
With technology advancing at light speed these days, we see more and more that, in various industries, humans have been replaced by machines that are much faster and more precise at repetitive tasks. In some sectors, devices, including the translation industry, aren’t ready for human work. There have been numerous articles on this, including our own Is Google Translate helpful in translations, where we have asked our translators if they use this tool and if it will replace human translators. Yet, we’ve decided to create another survey for our translators and asked them if they are using machine translation tools and how human translation is affected by the introduction of such devices.
Human translation vs machine translation tools
So, here it goes human translation versus machine translation tools. Are human translators using such devices in their daily work? Over 200 translators gracefully responded to our survey, and the results are pretty interesting.
The survey
We’ve decided to consider three types of machine translation tools:
- Online tools: Google Translate, which is probably one of the best stand-alone tools you can use online or as an app on all smartphones;
- A combination of CAT tools and machine translation: SDL Free Translation
- Built-in translation tool: Microsoft Word Translator
Although numerous other machine translation tools exist, the above have been chosen from different categories as the most representative of each type. Since human translators usually work with documents, they may find a built-in tool more valuable as the translator doesn’t have to copy/paste bits of text into an online tool. SDL Free Translation has been chosen to represent its category, combining machine translation with the advantages of computer-assisted translation tools.
Following are the questions we have sent to our translators along with their answers. Click any of the images below for their larger versions.
Are human translators familiar with such tools?
Some say that human translators hate machine translation tools, sometimes for good reasons. Anticipating a bit, some of the replies we’ve got from our translators indicated that the linguists would instead translate from scratch instead of editing a pre-translation or a translation done by a machine.
While we believe linguists would be the most familiar with translation tools, most of our team decided to pass on the survey because they don’t use such tools. However, from the ones who took the survey, it looks like they are fairly familiar with all three tools, with the most-known one being Google Translate.
Which tools are primarily used?
To understand this question, an explanation is required. There are two types of human translation:
- Normal translation, where the linguists are using text editors and dictionaries to translate;
- Computer assisted translation, where the translator uses specialized tools which (despite public belief) help the translator in his work by indicating repetitive text and using past human translation in current projects. This does not mean that the machine does the translation as a pre-translation; it is just a way to help translators in their tasks.
By the looks of it, although the previous question indicated that Google Translate is the most-known tool, it is also the least-used one. Translators use Microsoft Word Translator (probably as a dictionary) or SDL Free Translation to speed up the process and deal effectively with repetitions.
Are machine translation tools used to speed up the work?
Now, here’s the first contradiction we see in the results. While half the translators sometimes use machines to speed up their work, the other half don’t use technology. At first glance, you may find this a bit weird when it isn’t. Some linguists don’t want to use such tools, others may not know how to use them to their advantage, and some projects cannot be used with such devices. Let’s take small documents, like personal documents. When certified translation for small documents is needed, there is no point in using technology assistance as the papers are up to 200 words, and there are no real benefits in using memories with this type of translation.
Some of the notes we considered worth quoting:
- As a dictionary, to find alternative translations
- Terminology confirmation
- As a thesaurus
- At times, I use them only for some technical terminology
- Because many translation agencies require SDL Trados and help us correct mistakes
- Occasional quick translation of a word or phrase, but they are often very wrong – translation makes no sense
- I use Google Translate as a dictionary to know the meaning of some words in a particular context
- I only use Termium Plus, Linguee, dictionary, my skills, the context, my judgment
- I don’t use them. They are a waste of time.
- I don’t think machine translators are reliable at all. They do not consider the context. They perform a translation that can be downright misleading.
- To find other possible translations/wordings when I get stuck on something
- Only for private use since Google Translate is not reliable and makes lots of mistakes
Are translation tools a threat to human translators?
Human translation is considered the best as machines usually don’t believe the context or meanings of words, resulting in more or less of a mess than a translation. Indeed, humans are prone to mistakes, too, which is why we use an extra linguist to proofread the translation.
But, without jumping to conclusions, it looks like our linguists are not considering automated tools a threat. Are we done yet? No, there are quite a few more questions we’re asked our professional translators.
Did you already receive machine translations from translation agencies that you were requested to edit? Did you accept them?
Some translation agencies are trying to speed up the work and use cheap or free machine translation tools to produce a pre-translation. Human translators are then requested to edit such translations. We were wondering if our translators have accepted such projects in the past, and if not, why not.
Here are some comments from our translators you may find interesting:
- I would not accept machine translation if the work were translated from Japanese or Chinese because MT is not yet sophisticated enough to make the job quicker.
- There is usually too much to correct or re-translate for less money than for translating.
- The text is always garbage, and the client is always cheap on these jobs.
- I do not accept machine translations for editing, but I do for translating. Machine translations are horrible translations.
- MT’s quality was terrible. It was faster to translate it by myself from scratch.
- I used to, but now I always decline them because it is always more work to edit a machine translation than to translate from scratch and for much less pay.
- There is no grammar or real meaning/sense of the source in the target, especially in fields such as technical, CNC, medical, etc.
- They take more time than translating from a written source and thus pay a LOT less than translations. I also never buy a product when ads in newspapers or online are visibly machine-translated – they insult me as a consumer.
- They take so much time to correct that it would often take less effort to translate it from scratch than have some machine develop a bizarre translation.
- I have no interest in correcting the mistakes of a notoriously unreliable algorithm, especially if the client doesn’t include the source text for comparison. The client should have hired a human to do the translation in the first place.
- Machine translations are so poor that correcting them is more job than translating from scratch.
Are there any differences between the three tools? Are there any other instruments being used?
There are differences between the three machine translation tools we’ve selected. One is a machine translator, another is a CAT tool combined with MT, and the third is a thesaurus option in Word documents. Surprisingly, 54.77% of the translators didn’t differentiate between them, and that is because they mostly do human translation and only use such tools when they need to cross-check words or look for synonyms.
Human translators and their thoughts on machine translation tools
Is human translation going to disappear in the future with so many machine translation tools popping up? Will human translators remain with no work? Let’s see what our linguists have to say about this.
- It is not a threat necessarily, but the role of the translators will probably change in the future when there will be more editors/reviewers. Hence, they will have to adapt to their new roles and positions, the idea of what it means to be a translator, and those who cannot do so might see this as a threat. But, I also believe that some genres, such as literary, will take a long time before a machine can translate them. Thus, the traditional role of a translator will still be present for quite some time.
- At the moment, it makes things a lot easier and faster. Machine translation has improved quite a bit over the years, but if it can replace a human will have to be seen – especially in literary translation.
- Any machine-translated text I’ve seen was pure garbage, and I’ve seen quite a few. MT is far from reliable.
- I would rather translate the text for free than edit something translated by Google Translate, especially in my language combination.
While we won’t draw any conclusion from the human translation versus machine translation tools battle, get in touch if you need highly accurate, human translation services. We’ll give you just that at a price you will love!
Transforming Translation Role of AI and NMT
The evolution of the translation process has been significantly accelerated by advancements in Neural Machine Translation (NMT) and artificial intelligence, breaking down language barriers more efficiently than ever before. NMT leverages deep learning algorithms to provide automatic translation, enhancing the accuracy and contextuality of translated content across various language pairs.
A pivotal component of this technological advancement is the use of translation memory, a database that stores previously translated sentences or phrases. This not only streamlines the translation process by reusing relevant past translations but also ensures consistency and speed, particularly when dealing with a specific target language.
The integration of artificial intelligence into translation tools continuously refines the accuracy of translations, adapting to nuances and idioms of the target language, thus offering a more natural and reliable translation. This synergy between human expertise and machine precision is transforming the landscape of translation, making cross-lingual communication more accessible and effective.
AI’s Human Partnership Advancing Translation
Ensuring accurate translation is paramount in bridging communication across languages, and this necessitates a focus on translation quality. Quality translation involves a meticulous process of rendering text from a source language into a foreign language while preserving the original’s meaning, tone, and context. Although recent advancements in statistical machine translation have significantly improved the quality of machine translation, human intervention remains crucial.
The subtleties and nuances of language, often deeply rooted in cultural context, require a human touch to achieve truly accurate and high-quality translations. While machines can efficiently process vast amounts of data and provide quick translations, they may not always capture the essence of idiomatic expressions or cultural references. Therefore, the integration of human expertise with machine efficiency is essential for overcoming the limitations of purely automated systems and ensuring that translations not only convey the intended information but also resonate with the target audience on a cultural and emotional level.
Instant translation, powered by advanced machine translation services, has revolutionized the way we communicate across language barriers. Machine translation technology has seen significant improvements, enabling the production of high-quality translations across a diverse range of content types. The quality of translation offered by these services has improved remarkably, thanks to developments in neural networks and deep learning algorithms.
While the benefits of machine translation are manifold, including speed, cost-effectiveness, and the ability to handle large volumes of text, it’s the advent of high-quality translations that truly marks its advancement. This technology makes it possible to rapidly disseminate information and ideas across the globe, facilitating international collaboration and understanding. However, for nuanced or highly specialized content, human oversight remains essential to ensure accuracy and contextuality. Nonetheless, the strides made in machine translation technology continue to expand its applicability, making instant and reliable translation more accessible to everyone.
The Balance of AI and Human in Language
Language models, powered by advanced computational linguistics and artificial intelligence, have become increasingly sophisticated, enabling a broader understanding and processing of various languages. Despite this technological prowess, language experts emphasize the importance of maintaining the nuances and subtleties inherent in the original language across different types of content and file types. The type of content, whether it’s legal documents, literary works, or technical manuals, often requires specific linguistic and cultural knowledge that goes beyond the capabilities of even the most advanced language models.
This is where human input and involvement become indispensable. By integrating human expertise with the computational efficiency of language models, it’s possible to achieve a balance that respects the complexity of language and ensures accuracy and fidelity to the source material. Human involvement not only enriches the translation process with cultural and contextual understanding but also refines and guides the development of language models, making them more attuned to the intricacies of human language.
This synergy between human expertise and technological innovation is crucial for advancing language processing and translation practices, ensuring that they remain sensitive to the diverse expressions of human thought and communication.
Frequently Asked Questions
Yes, over 200 translators responded to a survey, indicating a varying degree of use of machine translation tools such as Google Translate, SDL Free Translation, and Microsoft Word Translator in their daily work, primarily for tasks like speeding up the process and dealing with repetitions.
The survey considered three types of machine translation tools: online tools like Google Translate, a combination of CAT tools and machine translation like SDL Free Translation, and built-in translation tools like Microsoft Word Translator.
The survey responses suggested that linguists do not consider automated tools a threat to their profession. Instead, it’s believed that the role of translators might evolve, potentially requiring more editing/reviewing in the future.
Yes, some translators have received requests to edit machine translations produced by cheap or free tools. However, many choose not to accept such projects due to the poor quality of machine translations, which often require more effort to correct than translating from scratch.
The main differences lie in their applications: Google Translate is a standalone online tool or app, SDL Free Translation combines machine translation with CAT tool benefits, and Microsoft Word Translator is built into the word processor, potentially more convenient for document work. Despite these tools, many translators prefer human translation, using machine tools mainly for cross-checking words or finding synonyms.
