The Google Translation Engine is a key component to the Google Empire and an integral part of the way we use the internet. However, the way that this tech has been implemented through their apps has raised questions by users and developers alike. The Google translation engine is an automated translation system that has the capacity to deal with dozens of languages in an effective and mostly, unproblematic manner. This appealed to web developers as it allowed their pages to be translated quickly and web sites to be promoted globally.
You may even remember the times when it was commonplace for a site to have the Google Translate drop-down box before it became an integral part of the Google Toolbar;
The Google translation engine works by generating translations through pattern recognition, creating a database that is used to create the most accurate translations possible. This method entails processing documents which have been manually translated by professional human translators, using this as a benchmark to their learning. The system seeks specific patterns in large chunks of text in what is known as “statistical machine translation”, making it able to ascertain more and more of the language as time, and the number of documents processed, increases.
As the tool had become so popular Google launched the Translate API which allowed web developers to process translations without having to use the online translation interface. This opened up the technology for large scale translations for educational text, online libraries and website content. This integration went further as developers used the translation API in many apps and online programs. With the widespread use of the Translation engine Google soon realised that any monetary value gained by the service was quickly nullified through the costs required to host and maintain the engine. They also questioned the benefit of maintaining the API as it was being used heavily by spammers translating copyrighted content.
Consequently, on May 26, 2011 Google, seeing this as a potential monetary problem, announced that the Google translate API would cease to function, due to cost and wide spread abuse, on December 2011. The reaction to this announcement showed that the Google translate engine was integral to the way many used the internet, as the whole of the internet seemed to rebel, causing Google to announce in June of the same year that they would keep the API in circulation, but as a tiered paid service.
Some people reacted well to this announcement but it did cause issues with many developers who instead started taking advantage of the free, non-API version of the engine. The paid API access was proven to be fiscally viable because, even with the somewhat exorbitant price tag attached to it; people still valued its speed and accuracy.
The public option was much more than accessible in financial terms, but it could be time consuming to use, and was not as linguistically robust as the paid option.
Many spammers also turned away from the Google Translation engine altogether as alternatives were available including a Microsoft version amongst others (later renamed to Bing Translator).
Part of the reason that Google began to infiltrate the automated translation market with a paid service stemmed from the presence of spammers, leading to the introduction of the Google translate apps for the iPhone and Android, and even the Google Chrome translate options.
This essentially introduced a split in the Translate engine; a premium version existed for those who paid for the privilege, and a basic version available as a freebie for websites, apps and other services.
At this point the differences between the versions really started to take shape, with the official licensed version standing out with incredible translation accuracy, the public version being fairly limited and still containing minor inaccuracies, especially in some of the more obscure languages.
The official version could be integrated into a website easily but it did pose a challenge in terms of cost. With the paid version, one had to consider the importance of the investment. If you weren’t targeting an international audience who spoke a language covered by the engine, it might not have been worthwhile. With the public option, cost was a non-issue which freed you from financial considerations. However, in terms of accuracy and reliability users found the public option wanting.
Some of the fundamental issues when it comes to the use of the free Google Translation engine include accuracy and reliability, both of which manifest in particular ways. First, there is name identification issue. The Google translation engine has a history of mistakenly translating names into phrases that may be equivalent in other languages. For example, Apple (the multi-national tech company) will be translated into the Spanish word “Manzana” (meaning the edible apple), which can be potentially problematic for those worried about the integrity of their brand, or for those concerned about a specific title of a product or trademark.
Then there is the issue of feminine and masculine identification. This is another example of a machine not being able to differentiate cultural nuances, and an example of a problematic feature in the public version of the program. Still, it’s not the fault of the tool, but an issue with comparative language and yet another deficiency of the software’s limitation and its not the only one. For example, there is the particular mistake of Hebrew being translated into “She is a great game,” when the more accurate translation would be “it is a great game”. However, again issues like these are more actively addressed in the paid version of the API, brand-names and trademarks will be maintained and masculinity/femininity is also considered when translating between languages.
These limitations and shortfalls have led the Google API to be held in high regards with automated translation as it can produce incredibly accurate text in a variety of languages. However, the evident inferiority of the spammer versions and public versions of the software is most apparent when they are put to the test when interfacing with other software. This is where the more obvious mechanical issues that have plagued the Google translate engine, such as spelling, have become such a problem for the public interface. When the internal Google translation tool’s output is put through the Microsoft Office spell check, it shows fewer mistakes than the public version. Other issues, between the internal (paid) and public tools have to do with fluency in less frequently used languages. In some instances when comparing languages like Spanish and Hebrew, Spanish was well translated even in the public version while Hebrew was only useful in the internal version. Even so, in both cases we can see obvious difference in the translations being served:
An example can be seen in the Google Chrome web store (which uses the paid version of the Translation engine) and the online Translator (which is based on the free Translation engine).
Here we have used a snippet from the Chrome app description for ‘Google Keep’ –
The snippet from the English version of the Chrome store –
Quickly capture what’s on your mind and be reminded at the right place or time.
Capture what’s on your mind
Add notes, lists and photos to Google Keep no matter where you are.
Find what you need, fast
Color code notes to quickly organize and get on with your life. If you need to find something you saved, a simple search will turn it up.
Always within reach
Keep works on your phone, tablet and computer. Everything you add syncs across all of your devices so your thoughts are always with you.
The right note at the right time
Need to remember to pick up some groceries? Set a location-based reminder to pull up your grocery list on your phone right when you get to the store.
And that same description for the Spanish version –
Apunta rápidamente todo lo que te pase por la cabeza y consúltalo donde y cuando quieras.
Captura tus ideas
Añade notas, listas y fotos en Google Keep estés donde estés.
Encuentra rápido lo que necesites
Las notas identificadas por colores te permiten organizar rápidamente tu día a día. Si necesitas encontrar algo que has guardado, solo tienes que buscarlo.
No vuelvas a perder ninguna idea
Accede a tu información en tu teléfono, tablet y ordenador. Todo lo que añades se sincroniza en todos tus dispositivos para que puedas acceder a tu información estés donde estés.
Using the Free version of the Translate engine we then went to translate the Spanish version to English – if they were using the same engine to create the translation in the first instance then we would have similar results the second time. However the results were surprising –
You can easily compare the two and see the glaring mistakes made.
These types of issues have dictated the structure of the Google API versions as we see them to today. If you are a developer and have previously used the public Google Translation tool and found it to be inaccurate, then its best not to dismiss the paid API with the presumption it too would suffer from the same downfalls. With two versions of the engine, one should consider the benefits and drawbacks of each before deciding which one to use for a project or on an on-going basis.
Always remember though that automated translation, no matter how accurate it is, is still automated. Now the Google Translation engine has been running for many years it still shows issues with specific language translations, really it should only be used as a guide or for non-important text.
If you are translating a legal document or technical guide we would definitely not recommend any automated route as certain phrases and terminology could be lost in translation – thus rendering them useless or even dangerous in certain circumstances. For this we would always recommend a professional translation company such as Kwintessential – with certification and accreditation from some of the top translation bodies you can be confident in the work any legitimate agency will provide.
So while automated translation is still improving there is always room for a qualified translator!