Improve Productivity with Machine Translation and Post-editing

Table of Contents

1. Did you know that we are currently in the midst of the Fourth Industrial Revolution?

2. AI is also being introduced in the field of translation.

3. However, how much productivity can be increased with machine translation?

4. Benefits of Increased Translation Productivity

5. How to Effectively Utilize Machine Translation

1. Did you know that we are currently in the midst of the Fourth Industrial Revolution?

The First Industrial Revolution occurred in the late 1700s to early 1800s in England. With the invention of steam engines and spinning machines, technology related to transportation and manufacturing rapidly advanced. The birth of spinning machines was driven by the need for solutions to meet the rapidly increasing demand for cotton textiles.

In addition, there was the Second Industrial Revolution centered around the United States and Germany from the late 1800s to the early 1900s (gasoline engines, electricity, etc.), followed by the Third Industrial Revolution (computers) in the late 1900s.

And in the 2000s, the Fourth Industrial Revolution centered around artificial intelligence (AI) arrived and is still ongoing.

The things and backgrounds developed by each industrial revolution may differ, but what is common in any era is the relentless pursuit of improving productivity.

2. AI is also being introduced in the field of translation.

This is machine translation.

While machine translation technology has existed for some time, it has not been very effective, at least in the context of translating from Japanese and to Japanese. This is where neural machine translation (NMT), which utilizes neural networks, comes into play. The accuracy has significantly improved compared to previous machine translation technologies. Starting with Google Translate, which was released by Google in 2016, multiple companies have announced NMT and begun offering services.

NMT is also highly accurate in Japanese translation, and can be expected to improve productivity in translation tasks.

3. However, how much productivity can be increased with machine translation?

We will give an example of professional translators who perform translations on a daily basis.

Generally, it is said that a professional translator can translate 2,000 words (250 words/1 hour) per day (8 hours) from English to other languages (*). For example, the English page of Human Science's machine translation is about 1,000 words, so a professional translator can translate this page twice in one day.

In some cases, in order to ensure the accuracy of the translation, a separate reviewer may conduct a review (translation review + correction) in addition to the translator. It is said that it takes about half a day to review 2,000 words (*), so if only humans translate a 2,000 word file, it will take a total of 1.5 days.

So, how much time can be shortened by utilizing machine translation?

When machine translation is integrated into the workflow, the translation process is done by the machine, so a translation that would have taken a full day for 2,000 words can now be completed in just 5 minutes.

However, there may be concerns about using or publishing machine-translated content as is. Therefore, we will add a post-edit process to the workflow.

Post-editing is the process of reviewing machine-translated content and making necessary corrections to achieve results similar to those of human translation.

The average processing amount for post-editing from English to Japanese is said to be between 3,000 and 4,000 words (*), but let's assume 3,000 words per day (375 words/hour) here. In that case, even with machine translation, a 2,000-word file would be completed in about 5 and a half hours.

*: The processing volume for translation, review, and post-edit per day may vary depending on the type of materials and content to be translated, the required quality level, and other conditions.

4. Benefits of Increased Translation Productivity

Many companies may have a lot of materials and content that have not been translated due to reasons such as lack of manpower or time. However, if they do not translate them, they may miss out on opportunities in the global business world.

A survey conducted in English found that 72.4% of users who shop online answered that "they are more likely to purchase when the product introduction is written in their own words," and 56.2% of users answered that "it is more important for the product to be written in their own words than the purchase price."

In other words, translating product introduction pages and articles can potentially lead to expanding business opportunities.

With the workflow of machine translation + post-edit, you can process about twice as much translation per month (20 business days) compared to when translating with humans alone.

In addition to the amount of translation, there is also the benefit of being able to shorten the time until publication. By being able to publish content and materials in half the time as before, it becomes possible to reach users quickly.

5. How to Effectively Utilize Machine Translation

In the traditional workflow where both translation and review are done by humans, in the workflow using machine translation, the translation is done by machines and the post-edit is done by humans.

Therefore, it is necessary to review the previous work process.

In addition, spending too much time on post-editing may result in overlapping corrections and reduce the benefits of efficiency through machine translation.

Therefore, it is necessary to define the level of post-editing and ensure that the workers have a common understanding of the scope of corrections in order to minimize unnecessary work and maintain a consistent level of translation quality.

Human Science Co., Ltd. supports the introduction of machine translation for our customers.

The HS consulting team will ask about the purpose and requirements for the introduction of machine translation, and we will propose the optimal workflow.

In addition, for those who have no experience with post-editing, we offer post-editing operation support and task outsourcing services.

Please feel free to contact us!

Translation Inquiry