Client Review - Polyglot features

This article discusses Polyglot features more in depth. Some of these may come in handy when conducting the review but should also be used with caution as they may introduce changes in the translation at large scale. Because these features are shared with editors, they are linked to articles which can provide more context.

 

Polyglot is editor interface made by Unbabel. It’s fairly simple and intuitive. This page is designed to address questions you might have about particular functions. Alternatively, you can watch a video overview of Polyglot below.

 

 

Polyglot anatomy

There are 3 main components in Polyglot:

1. Source text.
2. Target (translation). You can edit a translation in any way you want.
3. Translation instructions. They are made by customers to help translators.

 

 

If you want to finish a translation later, click on the Save and close button in the upper right corner.

If you want to submit the translation, click Submit in the bottom right corner instead. After you submit the translation, it will get delivered right away.

 

Segment types

In each task you will see segments of grey and white background:

 

 

A segment will have a grey background if:

• It is 100% matching with translation memory. In this case they were blocked by translators who was working on this file, but you can edit them anyway. Your edits will override existing translation memories.
• It is considered to be a high quality translation by Unbabel's automatic QE model. Some of these segments were blocked for translators, some not – depending on the level of confidence of the model. You also can edit any of those segments.

Segments that have white background are not matching translation memories and their AI confidence level is not high.

 

Segment groups

If two or more sentences from the source text are grouped together, such as when they're part of the same paragraph, they'll be marked with a blue line on the left side of this segment group.

 

 

Character limits

Some segments have a limit on number of characters they can contain. These limits are imposed by  the customers when they submit the project. They are usually needed when text is displayed in a limited real estate, like a table or a button.

When the editor has less than 10 characters left before character limit is exceeded, the amount of remaining characters is shown in the right bottom corner of the active segment.

 

 

If you exceed the character limit, the segment or segment group will get highlighted in red, and you will see the number of excessive characters. Note that if dealing with a group limit, the total refers to the sum of characters for all segments in a group.

 

 

A task can't be submitted if any segment or group is over the limit.

 

Annotations

Annotations are words that, before translation, were identified as being particular for different reasons. They come in different types:

 

Glossaries: Glossaries are client approved annotations. In the text they are underlined in green in source and target. If you click on the segment, you’ll see more detailed information on the side panel.

 

Context dependent glossaries: In case certain term has multiple possible translations, this term will be highlighted with a green dashed underline in the source. If you click on the segment, you’ll see all possible translations of the term and instruction in which case which translation should be applied.

 

Named entities: When, during machine translation step the algorithm recognises a city name, telephone number, measure, etc, it highlights it as a named entity annotation, to maintain the standard of data displayed for the target language. It will be highlighted with a grey underline in both source and target.

 

Non-deletable and non-editable annotations

• No-translates
• Anonymisations: Names, credit card numbers etc

Non-deletable annotation will be restored at the end of segment if deleted. Polyglot shows them as greyed boxes.

 

Markup tags

Markup tags are a set of symbols that are inserted into a rich-text document to control how it looks. They can stand for text formatting (bold/italic/underline), hyperlinks, images, etc.

Markup tags appear as purple boxes with numbers inside:

 

We have three types of markup rendered:

• Open/close pairs: [5] and [/5]
• Ranges of tags: [2-4] and [/2-4]
• Self-closing tags: [1]

 

How to work with markup tags:

• Tags in target have to be placed in the same place as in the source.
• Closing tag always follows corresponding opening tag. Wrongly positioned pair of tags will be highlighted with red. It’s impossible to submit the file until this is fixed.

 

 

Copy source to target

Sometimes it’s convenient to copy a source segment content to target (for example to handle complex markup) or multiple other cases.

It can be done by clicking the button on the middle of the segment or using shortcut Command+I. The button will only appear if the segment is focused.

 

 

Restore segment

By restoring a segment you can restore the translation of a particular segment before any edits you did. It is a way to restore deleted tags or annotations.

 

 

Propagation of repetitions

If there are repeated source segments in the task, translations for such segments can be propagated.

When you click on the first segment of a repetition group, you’ll see a suggestion to update other repetitions of this segment:

 

 

If you clicks the button “Update”, the translation of all identical segments will get updated across the file.

 

 

Find and replace

Provides editors with a quick interface for searching and replacing across the whole task, even if split across multiple pages.

Find and replace is located on right side panel, and can also be opened with the cmd+F shortcut.

 

 

• Finding is done by typing into the “Find” text box, no need to press any button. Polyglot will immediately highlight any matching results on current page and focus on the first one.
• Highlight can be done on any type of text except markup.
• Focus highlights a matching term. When a match is focused, the segment itself is also focused and auto scrolled into view.
• If there are no results on current page, but are on other pages, it will move to the first result.
• Editor can navigate between matched results using the arrows and can click “Replace” to do a single replacement, or “Replace All” to replace all matches in the task.
• If a match is in blocked segment, in a locked page, or overlaps with a non-editable annotation, it will not be replaced, but still matched and highlighted.

 

Pagination

When the translation is very long, it is split on multiple pages, and you will see it paginated at the top of the screen:

 

In order to submit the translation, you have to mark each page as done. To mark a page as done, you have to open that page and click on its number. Normally the page will become uneditable, and the page number will change to a checkmark:

 

 

In order to edit this page again, click Undo, or click the page number again (which now looks like a checkmark).

 

If the page you’re trying to mark as done has any critical error that prevents task from submission, you will have to fix them first:

 

 

Critical errors that prevent task from submission:

• Segment left empty
• Error in order of markup tags (like on the screen above).

 

If you’ll try to submit review with some pages left undone, you’ll see a corresponding message. Pages numbers are clickable here.

 

 

Shortcuts

 

Task level shortcuts

 

Mark page as done	cmd + D
Previous Page / Next Page	ctrl + option + arrow left/right
Open Find&Replace (Auto fill with focused text)	cmd + F

 

Segment level shortcuts

 

Report blocked segment	cmd + R
Copy source to target	cmd + I
Restore segment	cmd + U
Jump to segment by number	cmd + J