Five Things You Need to Know About Machine Translation and Post-Editing

When interest in machine translation (MT) was first generated in the 1950s, no one could have imagined how far it would come. Between the introduction of neural machine translation and its integration with artificial intelligence (AI), MT has come a long way.

As MT has evolved in the translation industry, so requires post-editing. To help you navigate through the information about MT and post-editing, here are five vital points to know about the services.

1. Machine Translation Still Needs a Human Component

The greatest aspect of machine translation is its speed; it can translate millions of words per hour, making the translation process considerably more efficient. However, machines do make their share of errors. For example, they do not recognize unique expressions, such as idioms and colloquial phrases.

Other mistakes that machine translators tend to make include:

• Using the wrong punctuation.
• Putting words in the incorrect order.
• Mistranslation of gender or number.
• Leaving words in the target language instead of translating them.
• Omitting articles or using the wrong ones.

To correct the errors that machines make, it is clear that there needs to be a human aspect to translation. That is where machine translation post-editing (MTPE) comes in. MTPE combines the high speed of machines with the precision of humans to create an effective translation process.

During MTPE, a human translator takes over after the machine translation service has completed its work. The linguist focuses on editing the content for accuracy, clarity, flow, and cultural appropriateness.

2. There are Two MTPE Options to Suit Your Budget and Needs

If you partner with an AI translation service, make sure the company offers both light and heavy post-editing. This allows you to choose the service that meets your requirements while remaining within your budget.

To decide which MTPE option is right for you, discuss your needs with your translation service. Take into consideration your turnaround time and quality requirements, along with the cost.

Light post-editing is the ideal choice to verify the overall machine translation accuracy, as well as to find and correct general mistakes. Light post-editing is ideal for:

• Situations in which you are certain that only a minimal number of edits will be required.
• Translations that you need to be completed quickly at a lower cost.
• Communications that need to be accurate and legible but are not complex.

Conversely, select heavy post-editing when you require a thorough review of the machine translation’s output, such as for documents that are complicated and will reach more people. This includes situations in which:

• The machine translation includes a large number of grammatical mistakes.
• The content has issues with cultural appropriateness and sensitivity.
• Improvements need to be made to the consistency and tone used in the document.
• Stylistic changes have to be addressed.

When you select heavy post-editing, you can expect your output to sound as though it was originally written in the audience’s native language. Although this is a more expensive option, heavy post-editing will ensure top-notch results.

For most translations, post-editing is always recommended. However, if you decide to skip this step, only do so when:

• You are confident that the AI automatic translation will be acceptable.
• The documentation is for internal use only and just requires a basic translation.
• The chance of a bad translation is low because the document only uses simple sentences and general terminology.

3. Machine Translation Does Not Work Well for Some Languages

One of MT’s limitations is that it does not work as effectively with some language combinations. Overall, the technology is better suited for translating from English to European languages because they are more closely related than other language pairs.

Conversely, machine translators can have more challenges when translating dissimilar languages, such as English and Japanese. This is also the case when different dialects are involved. In addition, languages that have more complex syntax and grammar, including Russian and Turkish, can pose difficulties for MT technology.

However, post-machine translation editing will ensure that the final product is accurate, regardless of the language combinations.

4. MTPE is Effective for Several Content Types

Without post-editing, machine translation works best with content that is concise and uses simple language, instead of having complex, lengthy sentences. In addition, machines can produce better outputs when working with straightforward text, rather than content that includes nuances, such as humor or sarcasm.

Although certain types of content should not be translated solely by machines, most types of documentation can be successfully translated when post-editing is added into the process. MTPE is especially useful for content that:

• Is culturally sensitive: No matter how well the actual words have been translated, it will not matter if cultural insensitivity ends up in the final product. Cultural appropriateness is one of the most essential requirements for a translation to be successful, and MTPE can ensure that this is not overlooked.
• Will be viewed by the public: Any machine-translated material that will be seen by clients should be posted-edited, including packaging inserts, product descriptions, various marketing materials, and website content.
• Includes specialized terminology: Whether you are translating content for medical, financial, or other products and services, MTPE can ensure that all specialized and technical information is translated properly.

5. Machine Translation Saves Money 

Because MT produces fast results, it saves both time and money on the entire project. Initial machine translation reduces the time that human translators spend on projects by at least a few hours, depending on the size and scope of the work.

Even when post-editing is included, clients will pay less than if human translators conduct the entire translation on their own. With MTPE, customers can expect potential savings of up to 30% on their translation and localization costs.

MT and MTPE save money in the following ways:

• Using a translation management system (TMS), which organizes the translation workflow, assigns tasks, and automates processes.
• Implementing a translation memory system, which keeps track of phrases that were already translated. In addition, it generates rules for capitalization, spacing, and terminology, which streamlines the translation process.
• Managing large-scale and multinational projects efficiently. Without the use of machines, translations that are large in scope would be expensive and take a considerable amount of time to complete.

When considering MT, remember that, while some machine translators are free to use, they have limitations. For example, if you have a small project, you might be thinking about using an online tool. If so, you will have a limited number of characters you can translate at no cost.

Alternatively, if you use a machine translation site, you will pay a low rate to translate up to one million characters, then an additional price for post-editing.

As technology continues to progress, machine translation will evolve to become even more advanced than it is now. However, that will not replace the human aspects that linguists bring to translations. When you embark on your translation project, partner with an automatic translation service that will pair you with a native speaker of your target language to ensure the best results.

Summary                                      

If you are considering using machine translation and post-editing, it would be helpful for you to know some essential information about the services. Here are five vital points to consider as you decide on the best ways to utilize machine translation and post-editing for your project.