Wednesday, 1 June 2011

Greek localization and inflection: the case of short strings ambiguity

First published as: Why contextual information is necessary for the localization into highly inflectional languages like Greek

There are many issues that affect localization which have been mentioned in some detail in a previous post.

Quite often, major companies or major localization service providers, deliver lists of English strings that need to be localized (or reviewed, edited, QAed) in highly inflectional languages, the case with Greek.

What do we mean by inflectional language or inflection?

In grammar, inflection or inflexion is the modification of a word to express different grammatical categories such as tense, grammatical mood, grammatical voice, aspect, person, number, gender and case. Conjugation is the inflection of verbs; declension is the inflection of nouns, adjectives and pronouns.

Back to our case with Greek. The suffixes (and often the prefixes) of Greek nounsGreek adjectives and Greek verbs (to mention but a few) change drastically depending on the above. Since Greek has three genders and English, well, none, there is no problem say, when we have a string like Custom—for the English language that is. In English it could refer to any setting (or settings) for any component (or components). In Greek however, the case is completely different. It could be any of the below, depending on the context:

Προσαρμοσμένος
Προσαρμοσμένη
Προσαρμοσμένο
Προσαρμοσμένοι
Προσαρμοσμένα
Προσαρμοσμένες
Προσαρμογή


For example, if we are talking about Settings (Ρυθμίσεις, in Greek, plural and feminine) it would be Προσαρμοσμένες. If we are talking about Setting (Ρύθμιση, in Greek, singular and feminine) it would be Προσαρμοσμένη.

To get an idea of the variety, refer to Microsoft translations:

custom show -> προσαρμοσμένη προβολή
A presentation within a presentation in which you group slides in an existing presentation so that you can show that section of the presentation to a particular audience.

custom form -> προσαρμοσμένη φόρμα
A form that you customize to better fit your needs.

custom color -> προσαρμοσμένο χρώμα    
A color in the Visio application that is stored with a shape as an RGB or HSL value rather than as an index to the document's color palette. A custom color is saved only with the shape to which it is applied.

custom table -> προσαρμοσμένος πίνακας
A type of filter that enables users to choose from a list and then drive dashboard content from multiple data sources.

custom order -> προσαρμοσμένη διάταξη
User-defined sort order. For example, you could define a custom sort order to display values in the EmployeeTitle column on the basis of the title's seniority.

custom price -> προσαρμοσμένη τιμή
The price calculated according to specific pricing rules that apply to products in a virtual catalog. There are three types of custom prices: percentage off, fixed amount off, and explicit price.

custom field -> προσαρμοσμένο πεδίο
A field you can create. A custom field can be a blank, combination, or formula field.

custom group -> προσαρμοσμένη ομάδα
An item of a custom group field. A custom group contains two or more items from a row or column field.

custom editor -> προσαρμοσμένο πρόγραμμα επεξεργασίας
An editor that allows you to view, edit, compare, or merge content inside a SourceSafe database.

custom device -> προσαρμοσμένη συσκευή
A microphone and speaker combination configured by the user in Microsoft Office Communicator and Microsoft Lync.

custom phrase -> προσαρμοσμένη φράση
A word or phrase that the administrator sets for the Content Filter agent to evaluate the content of an e-mail message and apply appropriate filter processing.

Or even the single word hits:

Custom    προσαρμoσμένο    .NET Framework 3.0
Custom    Προσαρμoγή    Access 2007
Custom    Προσαρμοσμένη    Access 2007
Custom    Προσαρμοσμένο    Access 2007


Note how they change. And they change according to context.

Hence, there is no way for a localizer to localize correctly, or an editor to edit correctly, when presented with out-of-context strings like:

Custom
None
Time


The question is: "Custom" what? "None" of what? "Time" as "12.34am" or as "10 minutes"?
You catch the drift.

QA of terminology

A few more examples can be derived from automated QA systems which check the text against a list of terms. Sometimes a term can be a very general word, like "of" for example. See this QA report:

Photo of the Day   Φωτογραφία της ημέρας   Against Term|EN-US: of -> EL: από
Next photo   Επόμενη φωτογραφία   Against Term|EN-US: Next -> EL: Επόμενο
Most Popular Photos   Πιο δημοφιλείς φωτογραφίες   Against Term|EN-US: Most POPULAR -> EL: Πιο ΔΗΜΟΦΙΛΗ
Enter Symbol   Εισαγωγή συμβόλου   Against Term|EN-US: Symbol -> EL: Σύμβολο
Enter search term   Εισαγωγή όρου αναζήτησης   Against Term|EN-US: search term -> EL: όρος αναζήτησης


What happens here is that it is taken for granted that the target language has no inflections whatsoever, which of course, is hardly the case.

100% matches and CAT pricing policies

Moreover, it is not fair to expect a translator not to be paid for 100% matches, or to be paid a lower rate, when these apparent "100% matches" need to be thorougly reviewed. Sometimes, the research that one has to do for single-word "100% matches" (i.e. searching in Help Files or running the non-localized version of the application) takes much more time than translating a full paragraph.

Kindly refer your clients to this post in order for them to gain an understanding of Greek localization and how they can help to improve quality.


Declension made easy—in English cat or cats. In Spanish:



Inflection of the Portuguese or Spanish lexeme for "cat".
Blue represents the masculine gender, pink represents the feminine
gender, grey represents the form used for mixed-gender, and
green represents the plural number; the singular is unmarked.

No comments: