Translation is expensive; why don’t we use Google Translate instead?

by Tasos Tzounis, Project Manager at Commit

We all had texts that needed translation at some point in our lives.

In those cases, certain questions have arisen:

  • How much will it cost?
  • Will it be good but also affordable?
  • Will I have it on time, at a low price and in excellent quality?

While searching for the best solution, there are various alternatives to choose from, in an effort to settle on either the most affordable one or the one meeting our needs. But do we have all the necessary information to end up with an informed decision?

Those who are familiar with the Internet and its capabilities know Google Translate. Google Translate is a Google service that provides a translation of words or sentences from and to almost all languages. You just type or paste your text in the appropriate field and then choose the source and target languages. It has become such a large part of our lives that we have all heard the following phrases in some wording or another: “I’ll look it up on Google Translate”; “why don’t you use Google Translate?”; “translating a simple text is very expensive, so I’ll do it myself, and with Google Translate I will pull it off”; “why do translators ask for so much money since there is Google Translate?”. If we explore the subject more closely, there is a large percentage of buyers believing that translators either use Google Translate or mistake Google Translate for translation memories. And the question remains: why pay for translators when there is Google Translate? Can it take the place of a professional translator?

In recent years, due to the reduction in cost and delivery time, considerable progress has been made in the training of translation engines, growing the demand for automatic translation. But can this become a reality? In fact, the translation quality of Google Translate has improved quite a lot, particularly in language combinations that are widely-spoken, such as French or English, and remarkably when the target language is English. But what happens with not so widely spoken languages or languages ​​with complex grammar and syntax? Greek, for example, uses cases, specific rules and demonstrates peculiarities that at this moment a computer cannot work out on its own. Also, in many languages one word has more than one meaning or changes its meaning depending on the syntax; and this is where the famous Google Translate falls short compared to a professional translator.

Many now realize that Google Translate is not the solution and that the automatic translation it provides cannot replace the human factor. Nevertheless, the issue of cost and time remains, and many claim that translation should be performed with Google Translate and then get edited by a translator. However, this solution also seems ineffective. Most of the times, for the reasons mentioned above, the translator ends up translating from scratch and, of course, being remunerated for translation and not editing services. The cost then is the same for the client and significant time has been needlessly spent with pointless experiments.

But what happens when the text to be translated is technical and contains legal, economic, or medical terminology? Can Google Translate detect the corresponding terms and render them properly in the target language in order to create a meaningful text that has cohesion and coherence? Can it inspire the same trust as a translator? In these texts, the terminology is specific and often provided by the client. In other cases, the translator has compiled a terminology library from previous projects. Google Translate doesn’t have the ability to integrate this terminology. Besides, most of the times it fails to render these terms correctly or understand if a word refers to a Product Name or a Trademark that doesn’t need to be translated. Therefore, with texts that require particular attention and baffle even an experienced translator, the use of Google Translate is lurking dangers. Medical reports or case studies do not leave room for mistakes. The use of Google Translate for cost reduction might not be prudent as the consequences of an error exceed the cost of a translation by a professional. In every transaction, there is trust that is built over time. So, when we have a technical text, we have to do research, assess translators and choose the right person for the job. Especially in cases where more than one text needs to be translated, and we have to reach out to a translator many times, we need to choose the most suitable one that will meet our needs; something that is impossible with Google Translate as we cannot trust it blindly.

Another drawback of this “all-in-one translation engine” is that it cannot follow any instructions provided. Technical texts are usually accompanied by several directives, such as the translation or not of measurement units, chemical compounds etc. In these cases, a specialized translator outweighs Google Translate for the following reason: the translator can also perform research while Google Translate memorizes terms and places them in the text without understanding their meaning or the outcome created by this “mishmash”.

However, the main issue of using Google Translate is confidentiality. Working with a translator, the customer ensures the privacy of their personal data through contracts. This is not the case with Google Translate since Google keeps the data collected in the event you choose to download, send or store the content of your file and has the right to use and reproduce your text. This is also clear in Google’s terms of service:

“When you upload, submit, store, send or receive content to or through our Services, you give Google (and those we work with) a worldwide licence to use, host, store, reproduce, modify, create derivative works (such as those resulting from translations, adaptations or other changes that we make so that your content works better with our Services), communicate, publish, publicly perform, publicly display and distribute such content.” Lastly, the use of Google Translate does not only affect the translation outcome in terms of quality, but it also raises copyright issues as it can be modified and republished.

Having explored the negative points of Google Translate, in my opinion it has one very positive aspect. It can be used as a dictionary to search for individual words as it provides a variety of interpretations. When searching for the translation of a term, it offers more than one rendering. Also, the translations of individual terms are correct, and surprisingly it seems to be more comprehensive than other online dictionaries. However, it cannot be used as a CAT TOOL or a translation memory, but it works perfectly as a multilingual online dictionary.

In conclusion, automatic translation is indeed free, but it has not yet succeeded in replacing the value and quality of a human translation. We will just have to wait and see what the future holds!

In-country Review: a must, a pain or both?

by Clio Schils, Chief Development Officer at Commit

When we look at the history of the process of translation and localization, primarily at the quality assurance step, we have come a very long way in the past 3 decades. I vividly remember a story from a good friend and now retired localization manager from a medical company. He mentioned to me that when he started running the translations for that company, the team literally had to “cut – with scissors – and paste” pieces of text for reuse into new updated versions of manuals. The content was then finalized with new to be translated text. The new text was translated and “more or less” reviewed back then as well, but one can imagine the challenges and risks of such a process.

The concept of Quality Assurance since then has been further developed, refined and optimized by industry stakeholders on client and vendor side, and the process of refinement is still ongoing: translation software programs have emerged and are still emerging, QA standards are being implemented, numerous commercial QA tools are being marketed and sold to those who understand that high quality is key. Still, in addition to all the tools and standards, there is one historical component in the process that is still there and offers the essential added value to any QA process, the human reviewer.

We all know that the essence of good translated output is a well-written source, the known “garbage in, garbage out” theory. However, for the sake of argument, let’s assume we have a translation that was based on a perfect source. We now move to the next step in the process, the review. Leaving aside the question “why we need a review in the first place, when we have a “perfect” translation, since it was based on a perfect source?” we go straight to the review step itself.

There are many criteria that co-define the type or depth of a review. As a rule of thumb, one could say that the higher the risk impact of a wrong translation, the more in-depth review is required. A mal-functioning vacuum-cleaner will not have the same impact as a wrong interpretation due to a bad translation of a patient’s medical-technical manual. In the latter case, a mistake in instructions could potentially have fatal consequences. Therefore, the in-country review is a must.

As per the example above, the in-country subject matter expert review is mandatory for highly regulated content “to the extent possible”. This step is conducted after the linguistic review by a subject matter expert. The emphasis lies on the technical aspects, functioning, use and terminology of the product rather than the linguistic elements.

Unfortunately, the in-country review step is not without challenges:

  1. The ideal subject matter reviewer is the in-country expert on client side. In most cases, these experts have other responsibilities and reviewing product content comes on top of their core responsibilities. It is a challenging act to balance.
  2. More and more “exotic” languages are required. Clients and buyers of translation services do not always have experts readily available in these countries.
  3. The limited availability of expert reviewers poses challenges on the overall TAT of a translation project and could endanger market release date of a client’s product.
  4. High turn-over among in-country reviewers of some companies, lead to longer lead times and potentially less reuse efficiencies due to differences of opinion regarding translations.

There are ways to ease the pain to some extent, some of which are:

  1. facilitate the process by providing specific proofreading guidelines and by providing validated “do-not-touch” technical glossaries. This will also be useful in cases of an instable reviewer pool.
  2. come up with other ways to execute this important step, i.e. use of specialized third party in-country review companies, use of the best specialized linguists who are being offered product training to master the features and function of the product.
  3. allow for reasonable time to execute a specific subject matter review task and document these pre-agreed lead times in a binding SLA, for example “up to 10k words, review time 3 working days”. When the generous deadline is not met, the project manager has the go-ahead to continue the process without any repercussions.

Finally, to summarize the answer to the question “In-country Review: a must, a pain or both?”. My answer is “both”, but the job needs to be done. Even today, and despite the challenges, an in-country review by a highly qualified subject matter expert offers a substantial contribution to the process. It will not only reflect on the overall quality of content but also on the company’s branding and reputation. Translated product documentation remains a very powerful marketing tool. It allows for deeper local market penetration thus bringing the product within reach of local end-users.