Translation Quality

Types of Quality Checks

Account Owner and Project Managers can manage Quality Checks individually within each Quality Check Profile to decide what level of translation consistency is appropriate for the  content.

There are three types of Quality Checks:

  1. Translation Consistency
  2. Spacing
  3. Other

To configure quality check severity:

  1. Do one of the following: 
    • In the New Experience, go to Account Settings > Linguistic Assets and then click Quality Checks.
    • In the Classic Experience, go to Assets > Quality Checks.
  2. Click the name of the Quality Check Profile to open it.
  3. (Optional) Use the drop-down filter to sort the list by Quality Check type.

Translation Consistency Quality Checks

Tag Consistency

(Default severity – High)

Check that source tags are correctly transferred to the translation.

Options include: 

  • Spaces around tags – Spacing around source tags is not the same as around target tags. Note, this check does not apply to Chinese, Japanese, or Thai languages.
  • Tag added – A tag that is not in the source has been added to the target.
  • Tag deleted – A tag that is in the source has been deleted from the target. 
  • Tag order changed – The tag order has changed in the target. For example, the source has a word in bold, followed by a word in italics, but the target has that order reversed.
  • Allow extra formatting tags – Extra formatting tags such as <b> or <i> can be added to the translation even if they were not in the source. 

Because of its unique markup language, all tags in Android XML files are ingested as placeholders in Smartling. This means that they will be checked under the Placeholder Consistency Quality Check option, and not Tag Consistency.

Number Consistency

(Default severity – Disabled)

Check that numbers in the source are correctly transferred to the translation.

Options include: 

  • Number added – A number that is not in the source has been added to the target.
  • Number deleted – A number that is in the source has been deleted from the target. 
  • Incorrect number format conversion – Considers how numbers are localized, for example, French uses a comma for a dollar value instead of a period.
  • Ignore positive / negative sign – If selected, ignores the Plus(+) or Minus(-) signs. 

Segment Completeness

(Default severity – Medium)

Check that a translation has a similar length to the source string. By default, if the target is 50% shorter or 250% longer than the source, a warning is generated. These percentages are configurable. 

Options include: 

  • Translation is too long compared to source
  • Translation is too short compared to source


(Default severity – Medium)

Check that placeholders in the source are correctly transferred to the translation.

Options include: 

  • Placeholder added – A placeholder in the target wasn't in the source.
  • Placeholder deleted – A placeholder from the source is missing from the target.
  • Incorrect placeholder format – The translator typed out the placeholder in plain text rather than copying it from the source. 
  • Incorrect spacing around placeholders – The spacing around placeholders in the source differs from what's in the target.

Emoji Consistency

(Default severity – Low)

Check that emojis in the source are correctly transferred to the translation.

Options include: 

  • Emoji added – An emoji that is not in the source has been added to the target.
  • Emoji deleted – An emoji that is in the source has been deleted from the target. 
  • Do not allow emoji in translation – Generates an error if there are emojis in the target. 

Insertable Consistency

(Default severity – Disabled)

Check that insertables in the source are correctly transferred to the translation. Insertables are unicode characters such as new line hyphens.

The insertable consistency check tracks the presence of the following characters:

  • Break tag (<br>).
  • Non-breaking space (a space character that prevents an automatic line break at its position).
  • New line character.
  • Word joiner (a code point in Unicode indicating that a word separation shouldn't occur at a position when using scripts doesn't explicitly specify spacing).
  • Zero width (invisible, "non-printing" characters that aren't displayed by most applications).

Options include: 

  • Insertable added – An insertable that is not in the source has been added to the target.
  • Insertable deleted – An insertable that is in the source has been deleted from the target. 

Target/Source Consistency

(Default severity – Low)

Check for inconsistent translations.

Options include: 

  • Same source string is translated differently – For example, "the red car" has two different translations in French.
  • Same translation is used for different sources – For example, "the blue car" is used as a Spanish translation for two different source strings. 
  • Ignore tags and placeholders – Tags and placeholders can be considered for this check, but are ignored by default. 
  • Ignore case – Case can be considered for this check, but is ignored by default. 

This check only appears in the CAT tool's quality check report and is not displayed on the sidebar.

Blocklisted terms

(Default severity – Disabled)

Check for blocklisted glossary terms and display an error if one appears in the translation. 

Glossary Compliance

(Default severity – Disabled)

Check whether glossary term translations are used. The check will only run if the term has a translation, or is a "do not translate" term. If it is a "do not translate" term, the check will see if the term was used in the target. 

Options include: 

  • Match by Percentage – This option allows for word variation. By default, the check is set to match the first 50% of the glossary translation and check if that series of characters appears in the target. For example, if a glossary term for "drive" has a French translation of "conduire," the first 50% of that word is "cond". A conjugation of the verb in the target, "conduit," would be a match.
    Match percentages can be set to 25%, 50%, or 75%. Even when match by percentage is configured words under 6 characters will use the Exact Match rule to prevent false positives.
  • Lexical Analysis - This option will use lexical analysis algorithms to determine if a word is a term in the glossary, even if it is not written (spelled) exactly the same way. Smartling will apply so-called word stemming and/or lemmatization strategies to determine that a glossary term is present. Lexical analysis is not available for all languages. If lexical analysis is not available for a language, 50% match will be used as described in the Match by Percentage setting.
  • Exact Match – Checks whether the exact translation for the glossary term was used. If it isn't used, an error is generated.
  • Term count match: choose to check that the number of terms used in the source is the same in the target by selecting the Require translation of term the same number of times as in source checkbox.

As a Translator or Editor, you will only see Glossary Compliance checks if you click the Quality Check tab to run quality checks on all strings in the Job, or when you're saving a string.

(Default severity – Low)

Check that legal symbols in the source like trademark and copyright are used in the translation.

The list includes:

  • Trademark: ™
  • Copyright: ©
  • Service mark: ℠ 
  • Registered: ®

Capitalization Consistency

(Default severity – Low)

Check for consistent capitalization between the source and target. For example, if the source begins with a lowercase letter, so should the target. If the source is all caps, the target should be all caps.

Punctuation Consistency

(Default severity – Low)

The punctuation check ensures that certain punctuation marks that are included in the source are also in the translation; and that similar characters are not introduced in the translation if they were not present in the source (with some specific exceptions).

The punctuation marks (characters) that are checked for include "brackets" and common "full stops".

The most common characters are:

. ! ? [ ] ( ) { }

A number of variations (like ideographic, AKA full-width versions) of these characters are included in the checks.  Additionally, the check supports "upside down" punctuation marks for Spanish as well as Right-to-Left versions of some characters for Arabic.

Users can add brackets if not present in the source without a warning, but not the other characters. If the source has brackets, then the translation must contain the same number of each character to pass the check. Similarly, the translation can have more full stops or other checked characters than the source.

All the characters in Group A (Sentence Terminal) of the Unicode Terminal punctuation characters, and other characters (some already previously noted above) are supported by this check.

Additional options include:

  • Closing punctuation is inconsistent: checks that the closing punctuation in the target is consistent with the closing punctuation in the source
  • Brackets / parentheses type or quantity is inconsistent: checks that the number of brackets / parentheses of the same type (e.g. {} [] () ) in the source is consistent in the target
  • Quote type or quantity is inconsistent: checks that the number of quotation marks of the same type in the source is consistent in the target

Spacing Quality Checks

Spaces include the tab advance function (tab key).

Leading or Trailing Spaces

(Default severity – Low)

Check that spacing at the beginning/end of the translation is consistent with the source. For example, there is an extra space at the beginning or end of a segment. 

Spacing Error

(Default severity – Low)

Check for extra spaces inside a segment.

Target Length Limit

(Default severity – High)

Check that the translation is within the character count limit.


Other Quality Checks

Original String Equals Target String

(Default severity – Low)

Check if the source string is exactly the same as the target string, e.g. the source is "hello world" and the translation is also "hello world."

Spell Check

(Default severity – Low)

A word is misspelled in the target language. Spell check is provided for these languages

Options include:

  • Skip if word starts with a capital letter
  • Skip if word appears in source string
  • Skip if word is valid in source locale
  • Run spell check for hyphenated words

Repeated Word

(Default severity – Low)

Check if words in a translation repeat in succession. For example, "A word is repeated repeated." 

Target in the Wrong Language

(Default severity – Disabled)

Check if the translated string is in the correct target language, e.g. the target should be in French, but a segment has been translated into Spanish.

Source in the Wrong Language

(Default severity – Disabled)

Words in the source are in a different language than what needs to be translated, e.g. the source should be in English, but a source string or segment is in Spanish.


(Default severity – Medium)

In a target language, plural translation forms are the same even though they are different in the source language. This check does not run if the target language requires only one form.

When this check is triggered, review the CAT tool for information about the plural source string and required translation forms. Check the Additional Details panel for Plural Forms in Source Language, as well as the language-specific plural instructions in the yellow bar above the translation entry field.

Custom Quality Checks

(Default severity – Medium)

Create customized quality checks using regex to target specific areas of translation that you need monitored and evaluated. Once created, you can configure language-specific settings.

Was this article helpful?