Post-edit points: Notes when registering machine translation in TM (Style Guide Edition)

Neural Machine Translation (NMT) was released in 2016. It has now become an essential tool in the translation industry. Many companies are using a combination of automatic translation and translation memory (TM) to streamline their translation process.

Translations from NMT are very easy to read, so it may seem like there are no issues at first glance. However, if you register the translation in the TM with errors that are difficult to confirm visually, it may lead to major problems later on. This is the issue with notation.

Some companies have defined writing rules in their style guide. Let's take the colon symbol as an example. There are two types of colons: full-width and half-width. ":" and "：". Can you tell which one is full-width and which one is half-width at a glance?

Table of Contents:
1. Pitfalls of NMT: Unable to Specify Notation
2. Points to note when registering MT+PE translations in TM
3. Even if you use a QA check tool...
4. Greatly reduce spelling errors with MTrans Post-Edit Booster!
5. Free Brochures/Trials Available

1. Pitfalls of NMT: Unable to Specify Notation

Currently, it is not possible to set the notation in NMT. Therefore, it is common to see situations where the colon on the first line is full-width, but the colon on the second line is half-width.

However, on the screen for editing translated text, it is often difficult to understand these differences, and there are many cases where they are overlooked.

This is the cause of mixed notation.

2. Points to note when registering MT+PE translations in TM

For example, what happens if full-width and half-width colons are mixed in TM?

First, the translator/post-editor referring to that TM may be confused about which colon to use. If there is a style guide and it includes rules about colons, the correct one can be applied. However, if there is no mention,

・Translator/Post-editor decides whether to use full-width or half-width characters, resulting in a situation where full-width and half-width characters are mixed

・Translator/Post-editor contacts the client for confirmation and cannot apply the correct writing method until a response is received, resulting in unnecessary time being taken until the work is completed.

It can be considered that such things are possible.

In other words, it can cause unnecessary work (such as checking styles and making final adjustments).

We have taken the colon as an example, but in Japanese there are various symbols such as slashes (/) and question marks (?) that exist in both full-width and half-width. If we leave these notation rules ambiguous, many inconsistent notations will be registered in the TM, resulting in a decrease in the accuracy of the TM.

3. Even if you use a QA check tool...

A QA check tool is able to detect such typographical errors. It is commonly used by translators in their daily work. There are various QA check tools available, from paid ones to open source ones.

For example, if there is a rule that states "always use half-width colons, full-width colons are not allowed", this will be set in the QA check tool. If there are full-width colons in the file being translated, their locations will be displayed.

However, most QA check tools do not automatically fix errors even if they are displayed.

Therefore, if there were 50 colon errors in the translated text, you would have to perform "find and replace" 50 times. Some of you may have had such an experience.

Not only colons and question marks, but also parentheses have full-width and half-width versions. For example, in the JTF Japanese Style Guide published by the Japan Translation Federation, rules for 9 types of symbols and 8 types of parentheses are listed. Translators/post-editors must be aware of these rules (and potentially more) and make necessary revisions and improvements to the translated text.

However, with this many symbols and brackets, it becomes difficult to focus on revising and polishing the translated text. Ideally, any errors that can be fixed mechanically should be automatically corrected.

4. Greatly reduce spelling errors with MTrans Post-Edit Booster!

This is where Human Science's MTrans Post-Edit Booster comes in, which automatically applies such formatting rules.

This MTrans Post-Edit Booster is an efficiency tool (SDL Trados Studio plugin) that supports post-editing work by automatically correcting errors specific to machine translation.

Usage is very simple.

1. Install MTrans Post-Edit Booster.
2. Set the notation rules (for example, define a rule to search for full-width colons and replace them with half-width).
3. When creating a project in Trados Studio, specify MTrans Post-Edit Booster as the MT engine and execute batch translation.
4. Automatic search and replace will be performed on machine translation based on the settings of MTrans Post-Edit Booster.
5. When you open the file, all the settings configured in MTrans Post-Edit Booster will be corrected ☺.

With this, translators/post-editors will be able to significantly reduce the number of corrections for symbols and parentheses.

Search and replace can be individually configured by users, so depending on updates to the style guide and changes in the output results of MT, the settings for search and replace can be flexibly changed. Additionally, regular expressions can be used, allowing for advanced automated editing.

In addition, with MTrans Post-Edit Booster, exporting and importing search and replace settings is possible. This allows for standardization of post-editing results among workers. For translation departments and companies that perform post-editing as a team, it becomes an essential collaboration tool.

MTrans Post-Edit Booster is a recommended tool for anyone working on post-editing machine translations in Trados.