How to… |
Automatic translation
In most cases, translation involves the localization of numbers, because number formats (more specifically, the grouping of digits and the decimal separator) can differ between the source language and the target language. Using the auto-translatables function, with the correct setup, memoQ can provide localized suggestions for all the numbers in the segment being translated in the Translation results pane. This is made possible by some auto-translatable resource items that are installed with memoQ itself. To use these, go to Project home, click Settings, and then click the Auto-translation rules button. In the list, you will see three items, and by ticking one of them, numbers will be formatted to the following formats. The language references in the resource names are just indications: use the one that corresponds to the number format of your target language.
· | Numbers-EnglishGroup: comma as the digit grouping delimiter, period as decimal symbol |
· | Numbers-GermanGroup: period as digit grouping delimiter, comma as decimal symbol |
· | Numbers-FrenchGroup: space as digit grouping delimiter, comma as decimal symbol |
· | Numbers-SwissGroup: apostrophe as digit grouping delimiter, period as decimal symbol |
· | Numbers-HungarianGroup: space as digit grouping delimiter, comma as decimal symbol. No grouping of digits in numbers below 10,000. |
Note: auto-translation rules are completely independent from, and do not affect the functioning of the number checks in quality assurance.
Quality assurance
Number formats
For its checks, the quality assurance function needs information about the correct number formats of languages. For this, there are two solutions: on the one hand, there are built-in defaults for all the languages, and on the other hand, you can supply custom number formats for each language in the QA configuration resources you use. If no custom format is found in the current QA configuration for a language, memoQ uses the built-in default when performing QA checks.
Number format check
The number format check verifies that numbers in the translated text are correctly formatted according to the user settings or the defaults. This helps you spot errors in the use of digit grouping or the decimal symbol.
Mismatched numbers check
The mismatched numbers check verifies that the same numbers are present in the source segment and its translation. How the mismatched numbers check works is fundamentally influenced by the “Strict interpretation on source side” and “Strict interpretation on target side” settings. Consider the following example, in which a sentence is translated from English to German. English uses the period as the decimal separator, while German uses the comma.
The daily dosage of the drug is 5.000 mg. |
Die tägliche Dosis des Mittels ist 5,000 mg. |
If strict interpretation is turned off for both the source and the target, QA checks the segment by performing the following:
1. | It collects all the numbers from both the source segment (5.000) and the target segment (5,000). |
2. | It creates a list of all the possible interpretations of these numbers in a completely language independent way. This means that regardless of the source and target languages, both “5,000” and “5.000” will get two interpretations: “five thousand” and “five”. |
3. | What the checker “knows” about the segment at this point is that both the English sentence and the German sentence contains a number that should be interpreted as either “five thousand” or “five”. Therefore the checker does not issue a warning. |
The weakness of this approach is that it can miss severe mistakes
: for example, it will not issue a warning when the English “5.000” (five) is “translated” into German as “5.000” (five thousand). Strict number interpretation in QA takes care of this by always interpreting the numbers it finds according to the number format settings that are in use for the language. With strict interpretation turned on, such a mistake will be spotted by QA and result in a warning, as long as the format settings are correct.
See also: