Machine translation has been popularised by tools like Google Translate on the consumer market. While professional translators would never use such a tool for commercial projects, the same technology behind Google Translate powers the machine translation tools used by language experts, including our own AI plus solution.
In this article, we explain everything you need to know about professional-grade machine translation and how language experts use this technology.
What is machine translation?
Machine translation (MT) is a computational translation technology that uses programmes to translate text or speech from one language into another.
There are three main types of machine translation widely used today.
Rule-based machine translation
Uses extensive sets of linguistic rules developed by language experts to translate source content into the target language.
Statistical machine translation
Analyses existing translations performed by professional translators, to determine the most likely translation for the source content.
Neural machine translation
Uses neural networks to “learn” from existing translations and its own previous translations to constantly improve its output.
While rule-based translation operates on a word-based system, statistical translation can convert entire phrases and neural machine translation is capable of understanding the wider context of content.
What are the benefits of machine translation?
Machine translation software has progressed immensely over the past few decades and the modern breed of AI translation tools, using neural networks, are now capable of learning and improving by themselves.
As a result, the list of machine translation benefits is growing and their value increases as the technology continues to improve.
The key benefit of machine translation is its ability to translate content quickly and, in some cases, almost instantly.
Enterprise machine translation systems can handle huge volumes of content tirelessly.
Speed, automation and increased scale produce a more cost-effective translation process.
Modern machine translation systems can manage dozens or hundreds of major languages.
Machine translation integrates with other technology, such as translation memory and terminology management to enhance speed, quality and consistency.
While machine translation has its limitations, the latest technology is constantly improving and increasing the value of its benefits.
When implemented with a post-editing step (especially light or standard) or a translation memory, machine translation can leverage the quality of the machine-translated output, leading to reduced costs, faster turnaround times and better content quality. Machine translation with post-editing helps translation agencies like ourselves to complete a given project faster, manage larger workloads and content, whilst passing the savings on to our customers.
Machine translation certainly has its benefits, but there are also certain limitations that come with the use of this technology, especially when a post-editing stage is not incorporated.
It is important to understand these limitations if using MT tools correctly is considered by businesses, to avoid running into costly mistakes.
Even the best machine translation tools cannot match the quality of professional human translators – and, possibly, they never will.
While huge progress has been made in word-for-word and phrasal translation, contextual understanding remains the biggest challenge for machine translation.
Without any human input at all, machine translation produces inconsistent results.
Some language combinations have more lexical and grammatical similarities than others, but machine translation can struggle with the lexical distance between languages of greater difference – e.g. English and Chinese.
Machine translation can run into problems with certain input formats, particularly with audio visual formats.
While AI and machine learning constantly improve the capabilities of machine translation, the challenge of accurately translating content from one language to another is immense. The truth is, we don’t know whether algorithms will ever become powerful enough to accurately interpret context and nuance, such as sarcasm or the intended meaning behind metaphors and other figurative elements.
This means machine translation may never be a suitable one-stop solution for translating content without human intervention.
How do language experts use machine translation?
Language experts know how to use machine translation to maximise its benefits and overcome its limitations. Machine translation should not be used as a standalone solution for translating content that is intended for public viewing and if used, post-editing should always be part of the process (light, standard or full).
At translate plus, we recommend using machine translation as the first step in a complete translation process and combining it with post-editing in most scenarios. In this way, MT will provide an instant first draft for our professional translators to review and correct before it is passed on to our post-editors.
This approach helps us save time when handling larger scale projects and reduce costs for our customers, whilst achieving better output quality and content consistency.
If you need help with implementing machine translation into your translation workflow, our MT consultants can help you find the right solution and discuss all the pros and cons of the different MT tools we currently work with. To speak to one of our team members about machine translation, please fill out the form on our contact page.