30 December 2021 · William Milsom
The robots are taking over.
Today artificial intelligence can do some pretty awesome things.
We all know AI is capable of telling us the weather, predicting what songs or movies you want to put on, and setting reminders.
However, AI technologies have some mind-blowing capabilities, such as, learning from experience, repairing themselves, driving us around, understanding our emotions, and even detecting illness via the smell of a human’s breath!
Nevertheless, just like the countless Alexa or Siri mistakes we all experience in our homes, linguistic artificial intelligence has often failed to achieve consistently accurate machine translations.
This article will explain machine translation, explore the latest developments in its process and finally evaluate its ability to replace human translators.
Machine Translation (MT) is a process where artificial intelligence is used to automatically translate content from one source language to another target language.
This process relies purely on computer software and does not require human involvement. However, whether their effectiveness matches human translation is still up for debate.
Machine Translation translates material by comparing the source language to the target language and then finding matching similar words.
This process is often conducted via 3 main types of Machine Translation:
1) Rule-Based Machine Translation – This is the earliest form of Machine Translation, and it uses grammatical and linguistic rules to determine meaning. It relies heavily upon Human coding to input the languages and edit the finished translation.
2) Statistical Machine Translation – This learns to translate material through analysing existing human translators and consequently learning the process human translators usually make when translating text.
3) Neural Machine Translation – This Machine Translation much like human translators learn via experience and will constantly improve their knowledge once appropriately trained. This neural network works like our brains and once trained delivers high-quality time efficient results in unlimited languages.
The technological developments in AI in the last 10 years have resulted in Machine Translation becoming a more efficient and accurate tool for language translation.
However, the robots may not have beaten us just yet. We humans, have a more comprehensive understanding of how language is reliant on context, tone, and register, factors that machine translation often misinterprets.
To settle this debate once and for all we will evaluate the process and results of machine translation vs human translation.
The main advantage of Machine Translation is its speed. Machine Translation can simultaneously translate millions of words. Therefore, for large projects with a high volume of text, it is more viable and cost-effective to use Machine Translation rather than human translation. It is also worth noting that some Machine Translation improves with experience so large projects are likely to train the Machine Translation to a high level of accuracy.
Extensive Language Selection
Another key advantage is that most major providers of Machine Translation can accommodate 50 – 100 languages or more and can instantly translate all of them simultaneously.
Automated Integration into translation services
Machine Translation’s cost-effective and time-efficient nature means it can be included as an additional extra alongside Human Translation.
Misinterpretation of contextual dependent language
The English language is confusing, ever-changing, and constantly dependent on the context of each sentence.
Take the following examples:
“I ran to the store because I ran out of milk “
“Can I run an idea passed you?”
“The house is looking run down”
The meaning of these sentences relies upon an understanding behind the context of the statement and can often lead to errors in machine translation.
The errors of google translate
Although Google translate is a saving grace when lost abroad and a useful tool in facilitating basic communication it often fails to accurately translate specialised terms, jargon, or creative sayings, such as idioms and metaphors within different languages.
Interestingly the errors of google translate have recently been highlighted within a legal trial where officers used google translate to ask for soliciting consent for a search, the non-native speaking civilian claimed to the court the google translation was inadequate and the court, after much debate, stated that it “does not believe it reasonable to rely on [Google Translate] to obtain consent to an otherwise illegal search”.
At PAB Languages we combine the best of Machine and Human Translation. We understand that although machine translation is time-efficient, somewhat accurate and constantly improving, it still requires development before it achieves parity with human translators.
The solution is to use ‘Hybrid Translation’ which uses the benefits of machine translation whilst maintaining the accuracy and thoroughness of our trained multilingual linguists. The use of our ‘Hybrid Translation’ technology can be applied to quickly translate technical texts, such as regulatory documents in the field of chemicals management, environmental control/waste/health, and safety, as well as, in the field of technical manuals and descriptions.