Machine Translation Post-Editing Guide

AI is great.

It powers our social media, answers our questions, and even drives our cars.

But sometimes AI can translate “President Xi Jinping” to “President Shithole.”

Facebook learned that the hard way.

We’re here to make sure you don’t.

In this article, we'll guide you through the importance of Machine Translation Post-Editing (MTPE), its types, and best practices. What’s more, you’ll get a practical post-editing checklist to improve your machine translations.

Pro tip: Use a professional translation management system, Centus, to balance AI power and human precision for impeccable translations. Learn more.

What is machine translation post-editing (MTPE)?

Machine translation post-editing, or MTPE, is the improvement of machine-translated content by a human translator. MTPE involves checking and correcting errors that the machine translation (MT) system may have made, ensuring that the final output is not only accurate but also reads naturally in the target language.

The MTPE process usually involves two main steps:

• Machine translation: An initial automated translation is produced by a machine translation engine, such as a translation software or AI system.

• Post-editing: The raw machine translation is then reviewed and edited by a human translator. They correct mistakes, improve the quality of the language, and ensure contextually and culturally accurate translations.

MTPE vs. human translation vs. machine translation

There are three major approaches to translation: machine translation (MT), human translation (HT), and machine translation post-editing (MTPE).

Let’s look at their strengths and limitations below:

In summary, MTPE brings together the best of both worlds: the speed and scalability of machine translation with the accuracy and cultural sensitivity of human translation. This union is what makes MTPE a crucial component in any translation or localization project.

The two types of MTPE

Machine translation post-editing is not one-size-fits-all. Depending on the need for quality, context, and the purpose of your text, you might choose between two main types: Light Post-Editing and Full Post-Editing. Let's take a closer look at what each involves.

Type 1: Light post-editing

The purpose of light post-editing is to make the text understandable and accurate but without a deep focus on style or tone. As such, it produces a serviceable translation that falls short of the publication quality.

In this example, light post-editing includes two key changes: inserting the definite article "The" before "weather" and adding an indefinite article "a" before "walk". These adjustments ensure the sentence is grammatically correct and clear, without significantly altering the machine's output.

Type 2: Full post-editing

Full post-editing, in contrast, is a comprehensive process that ensures the translated text matches the quality of a professionally translated text. The aim is to eliminate all grammatical errors and improve the style, tone, and flow of the content to make it seem as though it was originally written in the target language.

In this full post-editing example, the changes are more comprehensive. "Weather is agreeable today" was substituted with the more natural phrase "It's a delightful day. The phrase "good day for walk" was also transformed into the less awkward expression "perfect for a stroll."

When to use post-editing of machine translation

Machine translation post-editing (MTPE) helps businesses by merging the speed of machine translation and the accuracy of human translation editing.

Let's look at five common scenarios where MTPE can be particularly useful.

Global product launch

A global product launch is a colossal task, with countless moving pieces to consider. One of them is to translate product details and marketing materials into several languages.

Traditional methods would involve waiting for human translators to craft the product's description in multiple languages, a process that could span weeks.

With MTPE, this process is expedited.

The machine produces swift translations of the product description into various languages, and a human editor adjusts these translations to their respective cultures. This method significantly reduces turnarounds and quickens your launch.

Bulk translation

Industries like healthcare and law often grapple with translating large quantities of documents. For instance, hospitals may need to translate patient records for an international team of experts. Similarly, legal cases with international implications might demand the translation of numerous documents.

Manual translation in such cases can be expensive and lengthy. MTPE, on the other hand, can handle these bulk translation tasks without compromising on quality.

Translating customer feedback

Businesses with international customers receive feedback in many languages. Accurate feedback translation is needed to understand the customers and address their suggestions.

MTPE couples the machine's speed with the human's precision, allowing businesses to translate large volumes of feedback efficiently.

Localizing websites and software

When venturing into new markets, businesses need to adapt their digital content, including website text, software interfaces, and help articles, to fit local culture, expressions, and norms. Using MTPE for this process guarantees that your translation is quick and appropriate.

The algorithms provide a broad strokes translation, while human editors add the fine details that truly resonate with the local audience, making their interaction with the software or website seamless and enjoyable.

Technical or Specialized Content

In industries like technology or pharmaceuticals, sharing research findings globally often involves translating complex technical or scientific documents. Machine translation can struggle with this specialized language, while human translation can be slow and tortuous.

With MTPE, human editors make sure all the complicated scientific terms are translated correctly, but they can also use the speed of machine translation.

Best practices for machine translation post-editing

Successfully implementing machine translation post-editing (MTPE) in your business requires adherence to the following best practices:

1. Define your project parameters

Every translation project is unique, with its own demands and challenges. Before starting, outline the specifics. Are you translating a technical manual or a novel? Is the target audience teenagers or professionals? What languages are involved? What level of translation quality do you need?

For instance, if you're translating an informal blog post intended for a younger audience, the language style may lean towards slang, colloquial expressions, and light editing. On the contrary, a technical manual would require full post-editing to ensure precision and clarity in every translated term.

Understanding the project's scale and requirements upfront will save you from unnecessary corrections and rework later.

2. Invest in editor training

Well-trained editors are critical for a successful MTPE. They should have excellent language skills and a deep understanding of the subject matter. But being fluent in a language or knowledgeable in a field isn't enough.

Editors also need to be adept at the tools they'll be using for post-editing. This includes understanding the software, being able to identify common machine translation errors, and knowing how to correct them efficiently.

For example, an editor proficient in the use of an MT tool would know that if it continually mistranslates a particular phrase, it's more efficient to add that phrase to a glossary or a translation memory system. This addition saves them from repeatedly correcting the same mistake and eventually boosts overall productivity.

3. Establish a glossary and style guide

Creating a glossary and a style guide provides a roadmap for your editors, ensuring that they maintain consistency and adhere to your brand's voice.

A glossary comprises common terms, phrases, and their approved translations. A style guide outlines the tone, voice, and grammatical rules to be followed.

For example, an e-commerce platform might have a glossary that includes product categories, common technical terms, and promotional phrases. The style guide might specify that product descriptions should be written in an engaging and persuasive tone.

Together, these documents help editors deliver translations that align with the company's brand voice and maintain consistency across multiple languages.

4. Review, provide feedback, and improve

Continuous improvement should be the cornerstone of your MTPE process. Since machine translation technology is always evolving, so should your process.

Establish a feedback loop with your editors to identify issues and areas for improvement. Regularly review the performance and adjust your approach accordingly. This might mean providing additional training for your editors or tweaking your post-editing guidelines to address common issues.

For instance, if an editor flags that the machine translation system struggles with certain idioms, it might be time to adjust the system's settings or provide more training data to improve its performance. Regular reviews and adjustments help keep your translation process optimized and efficient.

5. Maintain consistency

Consistency is crucial when translating large volumes of text or working on documents that are regularly updated. You want the same phrases or sentences to be translated in the same way each time.

Translation memory tools can be a big help here. Translation memories store translated segments and then suggest them when the same or similar text appears, ensuring your translations are consistent across all documents.

6. Choosing the right translation management system (TMS)

Selecting a suitable TMS is crucial, as it serves as the backbone of your translation process. It should not just facilitate machine translation and editing but also provide an organized workflow, efficient management features, and easy integration with your existing systems.

For a simple yet efficient MTPE process, consider Centus.

Centus's translation management system acts as both a repository for your previously translated content and a user-friendly platform to manage both translation and localization.

Key aspects of Centus that enable successful MTPE are:

• AI-accelerated workflow: Centus leverages AI technology to automate and speed up the translation and editing process. This results in faster project completion times without compromising accuracy or quality.
• Centralized platform: Track progress, delegate tasks to editors, and manage multiple translation projects with ease. All in one platform.
• Integrated QA checks: Centus has built-in quality assurance checks to eliminate common errors in machine-translated content.
• Customizable workflows: Personalize your localization process according to your business's unique requirements.
• Easy integration: Centus provides hassle-free integration with your technology stack.
• Reliable support: Enjoy peace of mind with responsive customer service always ready to assist you.

With Centus, you're not just getting a tool for machine translation and post-editing. You're adopting an all-encompassing solution for your localization needs, positioning your business to achieve a global reach effectively and efficiently.