Loading...
 
(Cached)

OmegaT Translation Memory Tool

OmegaT

A Memory Translation Tool for your Document Workflow

I've noticed a lot of inefficiencies in my organization when translating documents into other languages. Here's a few:

  • The formating is lost in translation and each translated document must be formatted all over again.
  • When revisions are made to the master document, the translated documents must be retranslated or else there is a very painstaking document compare process to identify the parts that need to be revised in the translations.
  • The expense of professional translators, while part of the cost of doing business, is difficult to manage with our non-profit budget. So often documents are left untranslated.

OmegaT is a good solution to these problems.

  • It preserves the formatting in the original documents and lets the translator work on the embedded text. When the translation is finished, it reconstitutes the original document in the other language. Some formatting adjustments will need to be made due to the different phrase lengths in the target language.
  • Revisions can go very quickly because the memory translation database will pass over phrases that have not changed in the source, allowing the translator to focus only on the differences.
  • OmegaT is free and open source.
  • OmegaT can use free and commercial machine translation services in conjunction with its memory translation.

This tutorial will show you how to set up Microsoft Translation to work with OmegaT to provide automated translation when working with language pairs that Microsoft Translation supports.

How to Configure and Use OmegaT

Download OmegaT

The download page is found here: http://www.omegat.org/en/dl_overview.php

I find it confusing to select the correct download, so the link above is to the page with all of them listed. I recommend chosing the Latest version (Currently version 3.6.0). If you know you have the Java Runtime Environment installed you can download the smaller file "without JRE". If you are not sure, just download the larger file that has JRE included.

Install OmegaT

Double-click on the downloaded file and follow the instructions to install.

Creating a Project

Start OmegaT

From the File menu, select "Create a new project".

Right away you need to give the name of a folder and also decide on the placement of that folder. It is recommended to create the project folder inside your normal Documents folder.

  1. Open File Explorer
  2. Create a folder called OmegaT in Documents
  3. Create a sub-folder named according to the language pair you will be working with. The suggested location would look something like this:
Copy to clipboard
..\Documents\OmegaT\English-Français

The above folder is for translating documents from English to French. You would need a separate project folder for going in reverse. A different sub folder is required for each language pair and direction that you will be translating.

Project Settings

The project settings window opens, but there is not much that needs to be modified.

  • Select the language of the source text
  • Select the language of the target text.
  • That's all! Click OK.

Document Management

It is a good practice to keep all the translation documents in this folder to make the program effective. As you translate, phrases (not words) that you translate get stored in OmegaT’s memory and it grows with each document you translate.

  • They are inserted automatically if a 100% phrase match occurs.
  • They are suggested if an 80% match occurs.

This is especially important in the case of the revision of an existing document. At that point, if you use the same OmegaT translation folder, you can retranslated the revised document quickly, only needing to spend time on the phrases that have changed and quickly passing over the ones that have not.

Configuring the OmegaT Workspace

If you are beginning to work with OmegaT, the following setup is recommended. Later you can develop your own working preferences.

Menu: Options

Use TAB to Advance

The default setting in OmegaT allows you to go from one phrase to the next using the Enter key. That is OK, but then to go backwards you have to use Ctrl-Enter. We prefer to use the option of Tab to go forward and Shift-Tab to go backwards, the same way you might be used to working in a spreadsheet between cells.

Machine Translate

Select "Machine Translate". In the list of translation services that appear, select "Microsoft Translator" and we will show you how to configure that later.

TransTips

Select "Enable TransTips". This will cause words that are found in the glossary you create to be underlined in blue.  The glossary is a custom dictionary and is different from the full language dictionary you may also install. The glossary is a good place to keep track of key terms that you want to translate consistently.

Spell Checking

Spell checking will help find errors in the spelling of the target translation by underlining them in red if a dictionary exists for the target language. You can download dictionaries in many languages from here:

http://download.services.openoffice.org/files/contrib/dictionaries/

However, if you have LibreOffice installed, the dictionary files may already be on your hard drive. The dictionaries are in the same format for both LibreOffice and OmegaT. On Windows, you can find the LibreOffice dictionaries here:

Copy to clipboard
C:\Program Files (x86)\LibreOffice ?\share\extensions\dict-???

(The question marks above will be the version number of LibreOffice that you have, and the three letter code of the language dictionary)

You don't need to copy or move those files, you can just select them at the path above from the Spell Checking dialog box

Editor

By default, OmegaT copies the source text into the target window. I'm not sure why that is. One possible reason is that it will copy the source tags from the source text, allowing you to keep them in the right place when translating. I recommend starting with the text blank and inserting the source tags where they belong using keyboard shortcuts. In the Editor dialog box, select "Leave the segment empty".

Check the box next to “Insert the best fuzzy match”. When OmegaT finds a phrase in its translation memory that matches the current phrase by the specified percentage, it will insert its translation with the prefix fuzzy for you to review.

Machine Translation with Microsoft Translate

The Microsoft Translate service can generate a machine translation for you phrase by phrase as you work. Rather than send the entire document through a machine translation and being dissatisfied with the results, you can preview the translation in a separate pane before copying it to the target. If you are pleased with the translation you press Ctrl+M to insert it. Even if you are an excellent translator, this machine translation can save you a lot of time.

Cost

The Microsoft translation service is free for up to 2 million characters per month. This is the equivalent of 700 pages of text per month! If you go over this amount it is because you are a full time translator and are capable of paying for the service. For the rest of us, their limit of 700 pages is plenty.

Registration

You will need to register online with Microsoft and create an account in order to benefit from this service. Follow these steps:

  1. If you don’t already have a generic Microsoft account, you will first need to create one. You may already have a Microsoft account if you use Microsoft office and use any of their online storage. Go to https://signup.live.com(lien externe) The username must be an email address. It is important to pick a new password when creating the account and to not use the same password as the one for your email account.
  2. Once you have a working login for Microsoft, go to Microsoft Azure Marketplace and click on the button to Sign in with your Microsoft account.
    Image001
  3. You will be prompted to create a marketplace account. When you do, make sure to choose a country of residence that will give you an option to do an online banking transaction if the service ever requires payment and you wish to continue.
    Image002
  4. Accept the license terms.
    Image003
  5. Choose the plan with 2,000,000 free characters per month.
    Image004
  6. Final page. Agree to terms and click "Sign Up"
    Image005
  7. Now click on My Account at the top of the page. Your page will look like this:
    Image006

    Near the bottom you will see entries values for:Take note of these for step no. 10 below

    1. Primary Account Key (corresponds to the OmegaT parameter microsoft.api.client_secret)
    2. Customer ID (corresponds to the OmegaT parameter microsoft.api.client_id)

Edit the OmegaT Configuration File

  1. Close OmegaT if it is open.
  2. Find the notepad icon on the desktop or start menu.
  3. Right-click it and select “Run as Administrator”. 
  4. On the File menu of Notepad select Open.
  5. Navigate to the OmegaT installation folder, normally found here: C:\Program Files (x86)\OmegaT\   
  6. Select “All files (*.*)” in order to see the file OmegaT.l4J.ini and open it. 
  7. By default, notepad hides all the files that do not end in *.txt 
    Select “All files (*.*)” in order to see the file OmegaT.l4J.ini and open it.
    If you don't see OmegaT.l4J.ini but instead see OmegaT.l4J - it is because Windows is hiding the ini extension. It is the same file.
  8. Open the file and find these two lines:
    #-Dmicrosoft.api.client_id=xxxxx
    #-Dmicrosoft.api.client_secret=xxxxx
  9. Remove the hashtags # at the beginning of these two lines.
  10. Replace the xxxxxxx with the codes you noted earlier.
  11. Save the file and exit.

Open the Automatic Translation Window

Open OmegaT.

At the bottom of the screen there are tabs for windows that are not open. If you hover over them, the window appears for the time that your mouse is in the zone then closes again. We would like for the automatic translation window stay open while you translate. To do this:

Hover over the tab “Machine Translation”. While you hover, the window opens.

Machine Translation Tab

Slide the mouse to the top right of the machine translation window and click on the Restore button.
MT Restore Button

Now the automatic translation window will stay visible and you can arrange it into the position you would like it to stay in. As you translate, it will continually suggest translations that it automatically generates from the translation service. If you like the translation that it is proposing, just click Ctrl-M to copy the translation to the target text.

If the translation does not appear, it is possible that your Internet connection is not fast enough. It should display an error message after enough time has elapsed.

Ready to Translate!

  • Place source documents in the source subdirectory of your project folder.
    • You can use the command on the file menu: Copy Files to Source Folder.
      • Documents must be in OpenDocument or Microsoft XML format
  • In OmegaT:
    • Use Tab to advance and Shift-Tab to go backwards
    • Use keyboard shortcuts to
      • Insert format tags in the target
      • Insert source phrase in the target
      • Insert selected portion of fuzzy match in target
      • Insert machine translation in target
  • When complete, generate the translated document:
    • Project | Create Current Translated Document

Follow the Eyes

  • Look at the source text line
    • If a memory match was found, accept it and tab to next
    • If a fuzzy match was found, edit it and tab to next
    • If no match was found
  • Look at machine translation. Like it?
    • Press Ctrl-M to insert it
  • Look at partial fuzzy match. Like it?
    • Select phrase you want to copy, press Ctrl-I
  • Not happy with any of these?
    • Type in your translation
  • Insert missing tags one at a time with Ctrl-T
  • Tab to next

Contributors to this page: dhigby .
Page last modified on Monday April 11, 2016 05:45:29 GMT-0000 by dhigby.

License

Creative Commons License
All content on this LingTranSoft wiki are by SIL International are licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.