This page describes how to transliterate bilinguals using Oracc (the same principles apply to trilinguals, etc.).


Multilingual texts provide special challenges: the text for each language needs to be processed both on its own and with some level of alignment (word, line or text), and the layout of the text on the original artefact is also of interest. This page provides some initial guidelines for preferred practices in processing multilinguals in Oracc.


Several layouts are attested in multilinguals:

  1. In some cases, words are given as glosses and the translation may be only partial.
  2. In perhaps the most common case, the translations are given as interlinear text--typically a tablet of this kind will be ruled in blocks, with each block containing a pair of lines, the first in, say, Sumerian, the second in Akkadian.
  3. Another format for multilinguals is to place the translations alongside each other in columns--this is commonly used in lexical texts.
  4. A variant on the preceding is to bracket one of the languages with line segments in the other language: so, an Akkadian line may have a Sumerian translation which is split partly on the left of the Akkadian and partly on the right. Words in the Sumerian may themselves be split by the Akkadian column. We call these "bracketing translations".
  5. Texts may be given in whole separate passages or on separate artefacts, as on the Bisitun inscription.


There are several fundamental principles which inform the recommendations on transliterating multilinguals:

  1. It must be possible to recreate the layout of words on the original artefact
  2. It must be possible to create a running text of each language-version regardless of the layout of the transliteration
  3. Words within a language version should never be reordered in the transliteration; it may be acceptable to separate interwoven language-versions


The examples in this how-to section are keyed to the numbers in the Layouts section.

  1. Use the double-curly bracket notation for translation glosses, giving the language inside the brackets: iri{{a-li}}.

  2. For interlinear translations, number each block of lines giving the various translations of the text as one line, with each translation given using a == line with the language switch immediately after the ==:

    2'.	piriŋ abzu-ta me huš šu# ti-a
    ==%akk	nam-ru ina ap-si-i par-ṣi ez-zu-ti le-qu-u
    $	single ruling

    Be aware that when you lemmatise the text, lemmatisation of the Sumerian comes immediately after the Sumerian line. Other types of lines ($, #, #tr:) cannot come between Sumerian and Akkadian. This means that interlinear translation into a modern language is not possible in this format.

  3. For bracketing translations, transliterate the text as it appears on the tablet, separating the different language sections with a double equals-sign, then the language shift:

    piriŋ abzu-ta ==%akk nam-ru ina ap-si-i par-ṣi ez-zu-ti le-qu-u
    ==%sux me huš šu# ti-a

    If a word is split between left and right sides, use the ATF conventions for words split across lines:

    piriŋ abzu-; ==%akk nam-ru ina ap-si-i par-ṣi ez-zu-ti le-qu-u
    ==%sux -ta me huš šu# ti-a
  4. For translations of entire texts, or text sections, there are several strategies. One is to edit each translated version as a division of a single composite text (regardless of whether the versions appear on the same object or not):

    &X000009 = DPi
    #atf: use unicode
    #atf: lang peo
    @div Persian 1
    1. m-y-u-x : k-a-s-k-i-n : d-a-r-y-v-h-u-š : XŠ-h-y-a 
    	: vi-i-θ-i-y-a : k-r-t
    @end Persian
    @div Elamite 1
    1. %elx {aš}li-ki₂ {aš}ik-nu-maš-na {m}da-[ri]-ia-ma-u-iš# [{m}EŠŠANA 
    	{aš}ul-hi{meš}-e-ma] hu-ut-tuk
    @end Elamite
    @div Akkadian 1
    1. %akk sik-kat kar-ri {na₄}ZA.GIN₃ ina E₂ {m}da-a-ri-ya-muš 
    	LUGAL e-pu-uš
    @end Akkadian
28 May 2015 osc at oracc dot org

Steve Tinney

Steve Tinney, 'Bilinguals', Oracc: The Open Richly Annotated Cuneiform Corpus, Oracc, 2015 []

Back to top ^^

Released under a Creative Commons Attribution Share-Alike license 3.0, 2014. [] [] []
Oracc uses cookies only to collect Google Analytics data. Read more here; see the stats here []; opt out here.