Module wconv

Conversions between scripts and dialects in Serbian.

Serbian standard literary language can be written in two dialects, Ekavian and Ijekavian, and two scripts, Cyrillic and Latin. Dialects and scripts can be freely combined, resulting in four official writing standards: Ekavian Cyrillic, Ekavian Latin, Ijekavian Cyrillic, and Ijekavian Latin. Some automatic and semi-automatic conversions between them are possible.

Script Transliteration

For plain text containing only Serbian words (including well adapted loans), it is trivial to transliterate from Cyrillic to Latin script. It is only necessary to take care when converting Cyrillic Љ, Њ, Џ into Latin digraphs Lj, Nj, Dž, because sometimes they should be full upper-case (e.g. Љубљана→Ljubljana, ЉУБЉАНА→LJUBLJANA). But this is easily algorithmically resolvable, by checking if the previous or the next letter are upper-case too.

To transliterate from Latin to Cyrillic is somewhat harder, because in rare cases digraphs nj, lj, dž may not be single, but standalone letters; i.e. they do not map Cyrillic to љ, њ, џ, but to лј, нј, дж (dablju→даблју, konjunkcija→конјункција, nadživeti→надживети). The only way to handle this is by having a dictionary of special cases.

Furthermore, in today's practice texts are rarely clean as assumed above. They are frequently riddled with foreign Latin phrases (such as proper names) quasiphrases (such as electronic addresses), and constructive elements (such as markup tags). On the other hand, foreign Cyrillic phrases are quite infrequent (may be found e.g. in texts on linguistic topics). This means that in practice transliteration from Cyrillic to Latin remains straightforward, but from Latin to Cyrillic decidedly not so.

Script Hybridization

Sometimes the result of direct transliteration from Cyrillic to Latin is against the established Latin practice in a certain field, even if valid according to official orthography. Then it becomes necessary to specially handle some parts of the text (e.g. transliterations or lack thereof of foreign proper names).

Alternatives directives are a way to compose "hybrid" Cyrillic-Latin text, out of which both ordinary Cyrillic and non-directly transliterated Latin texts can be automatically derived. For example, this hybrid text:

   Различите ~@/линукс/Linux/ дистрибуције...

can be automatically resolved into:

   Различите линукс дистрибуције...
   Različite Linux distribucije...

String ~@ is the head of alternatives directive. It is followed by a single character, which is then used to delimit Cyrillic and Latin parts, in that order, out of surrounding text. (For all details on format of alternatives directives, see resolve_alternatives()). Transliteration from Cyrillic to Latin is performed only on text outside of alternatives directives.

Dialect Hybridization

Both Ekavian and Ijekavian dialect may be represented within single text. Such hybrid text is basically Ijekavian, but jat-reflexes are marked by inserting one of the jat-reflex ticks ›, ‹, ▹, ◃:

   Д‹ио б‹иљежака о В›јештичјој р›ијеци.

Clean Ijekavian text is then obtained by just removing jat-reflex ticks preceding valid jat-reflexes, and Ekavian by applying the jat-reflex map:

   Дио биљежака о Вјештичјој ријеци.
   Део бележака о Вештичјој реци.

The jat-reflex mapping rules are as follows, grouped by tick:

• ›ије→е, ›је→е
• ‹иј→еј, ‹иљ→ел, ‹ио→ео, ‹ље→ле, ‹ње→не
• ▹ије→и, ▹је→и
• ◃ијел→ео, ◃ијен→ењ, ◃ит→ет, ◃ил→ел, ◃јел→ео, ◃тн→тњ, ◃шње→сне

For very rare special cases, it is possible to directly provide different forms for Ekavian and Ijekavian, in that order, by using alternatives directive:

   Гд›је с' ~#/то/ба/ пошо̑?

Compared to alternatives directives for scripts, the only difference is that here the directive head is ~#. Alternatives directives for script and dialect can thus be mixed without conflicts, in single text and even interwoven (when interweaving, different delimiters must be used).