You are currently viewing 5 Ways to Determine Machine Translation Post Editing Rates 

5 Ways to Determine Machine Translation Post Editing Rates 

Knowing how to determine your machine translation post editing rates is a necessity as post editing of machine translation is increasingly becoming popular. 

Machine translation has always been around in the translation industry, however, it’s had its drawbacks. Despite it being very affordable and fast, it isn’t always efficient. 

However, MT efficiency has increased over the years and has made post editing of machine translation easier and faster. 

To accurately determine your machine translation post editing rates, it is necessary to understand how machine translation works. 

Machine Translation Explained

Machine Translation, when used correctly, can help you save time and money. But what if the MT’s final output is of poor quality and does not meet your minimum requirements? That is where post editing machine translation comes into play.

Implementing MT has become a reliable and quick method of producing rough drafts for translations over the last decade. 

Now, thanks to CAT tools that rely on translation memory (TM) databases and fuzzy match, it is more common for businesses and translators of all types to perform Machine Translation on their own, making their jobs easier, increasing productivity, and speeding up the translation process.

Nonetheless, despite all of the necessary upgrades and learning that Machine Translation has undergone, it remains flawed. 

But Machine Translation Post Editing creates a solution when the MT output is raw and contains linguistic errors

Machine Translation (MT) is a branch of computational linguistics that uses software to translate texts from one language to another.

In its prime, MT simply substituted every word of the passage with its most commonly used equivalent in the other language – a system known as rules-based machine translation. 

This, predictably, resulted in numerous inaccuracies in the translated document. They were frequently complete gibberish that made no sense at all.

While Machine Translation has improved exponentially since the days of rules-based machine translation, it is still no match for a human-generated translation.

How Does Machine Translation Post Editing Work?

It is important to note that MT is typically used to create an initial, quick, and raw version of a translation. Hiring human translators usually depends on the accuracy and precision of the final machine-translated project.

Soon after, a human translator will attempt to compensate for translation flaws by filling in the cultural and linguistic gaps left by MT.

MT Post Editing Process

In the legal world, it is frequently necessary to translate a large number of documents for a large case. When considering these business requirements, it is not surprising that machine translation (MT) is frequently mentioned. Sure, an MT engine can translate faster than humans — think tens of thousands of words in an hour. 

There is no doubt that MT can translate more inexpensively – with no human intervention, the only cost is that of the engine. However, without a human touch, important legal implications or key pieces of information may be overlooked.

It is critical to discuss how to achieve the desired level of output quality when using MT.

To bring raw MT up to the required level of quality, we generally discuss and implement one of two levels of post editing: light or full.

Light Post Editing

This entails making as few changes to the raw MT output as possible in order to make the translation understandable, factually accurate, and grammatically correct.

Light post-production duties include:

  • Correcting grammar and spelling errors only when they affect meaning;
  • Rewriting confusing sentences in part or entirely
  • Repairing machine-caused errors and
  • Getting rid of extra or unnecessary translation alternatives generated by the engine.

The translated text must accurately convey the meaning of the source concepts. Only major errors (incomprehensible content) and critical errors (that may result in legal ramifications or the use of incorrect terminology) are covered. The resulting content may sound robotic or a little ‘off,’ but it is fluid enough for the reader to grasp the meaning.

Full Post Editing 

A light post editing, on the other hand, would be useless for a truly poor MT output. If the content appears to be riddled with incomprehensible errors, the editor must improve fluency, style, format, consistency, and contextual relevance.

A thorough process and numerous linguistic changes are required for an initially poor MTPE project to come close to a high-quality human translation. Sometimes the MTPE editor will have to completely re-translate the original text.

Full post editing is common in these situations:

  • When the raw output is the result of the use of outdated or low-cost MT engines.
  • When the MT is not given enough specific context, the engine is forced to perform an automated translation, blind to context and with little reference material.

How to Determine Machine Translation Post Editing Rates?

Post editing machine translation rates are primarily determined by the client’s objectives, the quality of the source text, and the quality standards in place.

Depending on the quality of the source material, the translator will need to employ a variety of methods and approaches in order to deliver a precise and accurate PEMT translation.

But here is a guide to help you determine your MT post editing rates:

Factors That Affect Machine Translation Post Editing Rates

Factors to consider when post editing machine translation rates: 

  • Language pairs
  • Workmanship
  • Machine Translation Software 
  • Industry
  • Word Count

Language Pairs

Some languages managed by MT have higher quality than others, though this may vary depending on the engine used. If the original text necessitates a Machine Translation from a language pair derived from a major language, such as English or Spanish, the post editing rates will be lower.

However, because some languages, such as African or Japanese, have yet to be polished in MT, it would necessitate even more effort from the translator or editor, raising the MTPE rate.

Workmanship

When machine language software is unable to translate documents with 100% accuracy, you will be required to do more work as a post editor. Consider your work hours as well as the effort put in to improve the quality of the machine translation when determining your machine translation post editing rates.

Reduce your rates if the software provides translations with an accuracy of 80 – 90%, leaving you with little to no work as a post editor.

Machine Translation Software

The price of post editing may vary depending on the machine translation software you choose. Some MT software is more effective than others, so the documents it produces don’t need as much post editing from a person.

The post-editor can complete their work much more quickly and for less money, if you/ your client use dependable MT software to create high-quality translations.

Industry

Depending on the type of content, machine translation post editing effort may also change. Different contents require different types of post editing levels of expertise because highly specialized translations with specific terminology are still a challenge for MT technologies.

For example, MTPE for the legal and medical fields requires familiarity with particular terms that are foreign to regular translators outside of these fields. As a result, since lives could be at risk, it would be necessary to use medical or legal translators.

Word count

Any effective post editing procedure for machine translations should incorporate:

  • Examining and reading the source material.
  • A comparison of the output and the source text side by side.
  • Verifying the translation, revising the text, and occasionally retranslating sections of text are all important, particularly for contextual and cultural factors.
  • Strict quality assurance (QA) inspections.

Costs can be influenced by a variety of factors, most notably how long it takes to complete the translation project. And much of what occurs with the per-word editing rates for translations also applies in this situation. The specific content, language input, and translator experience all affect hourly rates.

Choose iVoiceAfrica for Affordable Machine Translation Post Editing Rates 

iVoiceAfrica has provided hundreds of accurate MTPE for several clients across the world. Our Machine Translations are post edited to utmost accuracy, ensuring that information and context aren’t lost in translation.

We have translated over 80,000 words in the legal, medical and media industries to name a few. iVoiceAfrica also offers affordable post editing machine translation rates. Contact us. 

Leave a Reply