MTPE vs Machine Translation vs Human Translation
All businesses entering the international expansion have to take language differences and cultural gaps into consideration. They have to translate and localize a wide range of content so that their targeted customers and business clients can understand their messages and their value proposition. The translation must be flawless, precise, and culturally relevant to achieve the highest communication effectiveness.
Human translators have been used for so many years to adapt a business’s writings to other languages, but in recent years, a new trend is increasingly growing in popularity. MTPE – Machine Translation Post-Editing is now being adopted by so many translation agencies. It is slowly changing the translation industry and creating a more connected world.
In order to understand the concept of Machine Translation Post-Editing (MTPE), we first need to understand the concept of Machine Translation and Human Translation.
1. What is Machine Translation?
Machine Translation is the automated translation of the text from the source language to the target language. There are many ways to “teach” machines to translate. We have to prepare massive databases of linguistic materials, sometimes amounting to billions of words, for the machine to learn translation through Machine Learning technology. The larger the database, the faster the machine can learn, and the more accurate the translation will be.
There are many approaches to Machine Learning in translation. For example, we can build statistical models to select the most common meaning for any specific word. However, this approach proves to be ineffective and inaccurate. A more advanced and recent approach that the industry is taking is building Neural Machine Translation (NMT) systems, in which the algorithm is built to recognize the way words are associated with each other, then select the most suitable meaning for that association. This approach is a lot better than building statistical models because it takes the context into account.
A good example of Machine Translation is the all-too-familiar Google Translate that we all have come to love and hate.
Google Translate is one of the most well-known Machine Translation technology that we know. It offers lightning-speed translation for nearly every single language on the planet. But it is far from a fully developed technology. Google has the largest database of all languages in the world, yet Google Translate still can’t accurately translate cultural and linguistic nuances in the text you put in.
2. Pros and Cons of Machine Translation
Pros of Machine Translation:
- Extremely fast translation
- Extremely convenient
- There are many Machine Translation platforms available on the Internet such as Google Translate and DeepL for FREE. If you want a quick translation of an unimportant, simple piece of content, Machine Translation is the way to go.
Cons of Machine Translation:
- Poor quality of translation. Sometimes the translation is meaningless in the target language because it was too literal.
- Unable to handle complex content
- Unable to detect human aspects of the text, such as humor, sarcasm, tone of voice, personality, etc.
- Unable to recognize multiple layers of meaning
- Unable to recognize idiomatic expressions, slang, and contextual expressions
- Unable to retain brand voice
- Robotic, unnatural choice of word, leading to ineffectiveness in message conveyance
- Requires huge database of linguistic resources to develop
- Can’t achieve consistency in translation: Machine Translation tends to alternate from one meaning to the other for the same word, making it considerably harder to follow the content.
- Limiting pool of languages to choose from. Linguistic resources in English and other European languages are abundant, so it is easy to train machines to translate these languages. However, resources in Asian languages tend to be rarer, so machines can’t “learn” how to translate these languages properly.
- Unable to translate Japanese (Japanese is a unique language that Machine Translation fails to comprehend and translate)
3. What is Human Translation?
Human Translation is translation done by a human translator. Humans know humans. They know the context behind the words. Human translators can select from a wide range of meanings of the words they are translating. They can also comprehend words with multiple layers of meaning, allowing them to effectively convey all of the nuances hidden behind the literal, superficial meaning in the dictionary. Human Translation recognizes humor, sarcasm, tone of voice, personality, and a wide range of human emotions in the text, which Machine Translation fails to notice.
Of course, the quality of Human Translation can’t be achieved within a blink of an eye. Human Translation is slower than Machine Translation. Translation is an art in itself, and you can’t rush art.
4. Pros and Cons of Human Translation
Pros of Human Translation:
- Able to recognize and take into account human aspects of the translation
- Able to read into the deeper layers of meaning so that they can select the most suitable to convey the nuances
- Overall better quality of translation
- More consistency in translation
- Take cultural aspects into consideration
- Only human translators can deliver high-quality translations for complex, high subject knowledge documents
- Can translate Japanese language pairs
Cons of Human Translation:
- More costly
- Can make typos and other dictation errors that Machine Translation can’t make
5. What is MTPE - Machine Translation Post-Editing?
Linguists and developers have found a way to combine the good sides of 2 translation methods mentioned above. They combine the lightning-speed of Machine Translation with the human quality of Human Translation and create Machine Translation Post-Editing.
Machine Translation Post-Editing is when Machine Translation and Human Translation meets. It enables faster translation while making no compromise to the overall quality by adding an extra touch of human intelligence and emotion.
Machine Translation Post-Editing is when Machine Translation and Human Translation meets. It enables faster translation while making no compromise to the overall quality by adding an extra touch of human intelligence and emotion. Essentially, MTPE is the process of putting the document through a Machine Translation platform, then have a translator edit it to improve quality to an acceptable level.
6. The MTPE Process
A Machine Translation Post-Editing happens in the following process:
- The Language Service Provider (LSP) receives the document in the source language.
- The LSP runs the document through a Machine Translation system. Many large translation agencies have their own technology that they developed over the years. The more translation projects an agency successfully delivers, the more linguistic resource they have to train their Machine Translation technology.
- The LSP now has a poorly translated file filled with errors and word misuses.
- The LSP assigns that machine-translated file to a human translator with relevant expertise
- The human translator will scan through that machine translated to edit it. They will keep the parts that are good enough and make changes and edits where necessary.
- The human translator submits the final file for delivery. The client receives a translation completed in a shorter period of time for the same quality.
7. Pros and Cons of MTPE
Pros of Machine Translation Post-Editing
- Faster than full Human Translation
- Lower cost than full Human Translation
- Higher quality than full Machine Translation
- Affordable price for many clients
- Convenient for both the agency and the client
- Higher clarity
- Improved flow
- Greater cultural sensitivity
Cons of Machine Translation Post-Editing:
- Slower than full Machine Translation
- Potentially lower quality than full Human Translation
- Sometimes the machine-translated text is too bad and meaningless that the translator can not understand what the text even means, at all. This leads to potential misunderstanding on the translator’s side, yet it was totally not their fault.
- The word order and expressions of the translated document can be a little bit weird and unnatural because, after all, it is still Machine Translation.
- The translation quality is acceptable at first glance, but a lot of linguistic nuances can be lost in the process. If you aim to retain these nuances, MTPE might not help you.
- Inefficient for too complex documents. If the amount of editing the translator has to perform for these documents is too much, they are better off translating everything from scratch. The quality the client receives in these cases is only acceptable, not excellent.
- Inefficient for literary writings. Only human translators can translate literary writings, and it is going to be a very long time before machines can start translating literary writings.
- Inefficient for rare languages. There are currently not enough linguistic resources to train the machines to translate these rare languages. The translator is better off translating everything from scratch than editing a too-poorly-translate document.
- Many translators don’t want to do MTPE because, from the translator’s side, it is much harder to re-translate a poorly translated text than translating everything from scratch. Professional, trained, experienced translators usually prefer full human translation requests.
Considering all of the pros and cons listed above, you can choose the suitable approach to your translation order based on your needs.
8. When should you use Machine Translation?
You should use Machine Translation when:
- You need lightning-speed translation
- You don’t need exact translations. You only need to know the gist of the content
- Your content doesn’t have high visibility i.e. not many people will see your translated content
- Your content is not of high importance and does not affect the well-being of many people
- You have a limited budget
Recommended documents to use Machine Translation: highly personal documents, simple documents, or phrases and words that you encounter on the go. If you want to use Machine Translation, you can just go to Google Translate and input your text to receive the translation. It is not perfect, but it is good enough.
9. When should you use Human Translation?
You should use Human Translation when:
- Your document is of high importance and affects the well-being of many people (for example: legal papers, medicine documents, safety documentation, financial papers)
- Your document has high visibility (for example: marketing content, brand value proposition, IPO whitepapers, etc.)
- You need high-quality, professional translation
- You need to understand accurately the content
- You need precision
- You need a lot of human intelligence involved in the translation
- Your documents have a lot of “human” linguistic aspects, such as slang, sarcasm, or have deep layers of meaning. These usually appear in literary writings
- You need a lot of cultural understanding and sensitivity in your translated documents. These documents can have a multilingual, multicultural audience.
- You have a decent budget and ready to invest in your translation
10. When should you use MTPE - Machine Translaton Post-Editing?
You should use MTPE (Machine Translation Post-Editing) when:
- You need fast translation and decent quality while wanting to be cost-effective
- You expect lower translation quality. MTPE quality is not as low as full Machine Translation, but it is hardly going to be as good as from-scratch translation
- You need to translate large volumes of simple content in a short period of time
- Your content doesn’t have too high visibility i.e. it is personal and internal
- Your content doesn’t affect the well-being of many people (i.e. legal, medical, engineering documents). If you want to translate those critical documents, we recommend that you choose Human Translation instead
- You only want to understand the gist of the content. You only want it to be comprehensible with no typos and grammatical mistakes
- Machine Translation is too low-quality for you, while Human Translation is too high-quality for your needs
- You have a medium budget to spend on translation
Considering all of the factors mentioned above, we recommend that you only choose Machine Translation Post-Editing when you expect an acceptable, not excellent, quality of translation. In short, MTPE is in the middle on the quality spectrum, with Machine Translation on the lower end and Human Translation on the higher end. If you want to translate high volumes of documents in a short period of time while sacrificing a little bit to the overall quality, MTPE is the way to go.
We want to educate our clients on the risks and benefits involved when choosing each of these options. They have their pros and cons. You need to evaluate carefully your needs and wants before picking the types of services you want.
Want to have your documents professionally translated?
pTranslate helps you achieve global success with our world-class translation services
11. Does pTranslate offer MTPE services?
pTranslate offers MTPE, but we don’t include it in our official packages.
pTranslate offers 100% human translation for all of our projects, but we are ready to deliver MTPE service when our clients request it. We usually deliver MTPE services to clients who want to urgently translate up to 50,000 or 100,000 words, or clients who only want to translate internal communications documents with low importance. In these cases, MTPE is the perfect answer. In other cases, we advise our clients to take the slower but better route of Human Translation. The result is worth the investment.
If you have any questions, feel free to contact us.
Become a True Global Leader
Subscribe to our Newsletter to receive the latest insights on Global Business.
pTranslate is a translation and localization agency for businesses and individuals. We offer world-class language services at a competitive price and exceptional quality. Our goal is to eliminate the language barriers and connect the world.
Connect The World
Translate your documents, writings, books, and more!