I shared this Oracc RINAP JSON example in my last post, but now I'm focusing on how to automatically generate an IPA version of each Akkadian "word", from some sort of input word/text.

It appears there are two types of texts, two syntaxes:

  • The "Cuneiform"-transformable syntax (glyph-syntax let's call it). This is used to be able to cuneify texts. So from the forms.n we have like a-bu-ka, and that would be like this: 𒀀 𒁍 𒅗.
  • The pronunciation-syntax. This has letters like â and š and apparently there is a 1<>1 mapping between those letters and IPA sounds, as far as my understanding goes.

But do I have all this correct so far? I'm not sure. Can you not use the "glyph-syntax" for pronunciation?

  • Am I to use the glyph-syntax only for converting to Cuneiform (are there any docs on this too)?
  • Am I to assume the cf field form like abāku or abūbāniš or baʾāšu are for pronunciation? Or is the glyph-syntax somehow also for pronunciation, or is it somewhere else I haven't found?

1 Answer 1


The cuneiform writing system doesn't represent the sounds of Akkadian especially well. For example, Akkadian had three vowel lengths, but there were only two ways of writing these in cuneiform: plene vs non-plene spellings. Short vowels are rarely written plene, long vowels are sometimes written plene, and overlong vowels are usually written plene. Similarly, the language had separate /i/ and /e/ vowels, but the same cuneiform sign was used for both zi and ze (and also for ṣí and ṣé and…).

So what you call the "glyph-syntax" is a representation of the cuneiform spelling, and what you call the "pronunciation-syntax" is properly called a normalized form—a representation of how we think the word actually existed in the mind of an Akkadian-speaker, unbound by the limitations of the writing system. The cuneiform transcription is what you should look at for the writing; the normalized form is what you should look at for the pronunciation.

  • Perfect, thanks, that clears things up. Can you elaborate more on what would go wrong when trying to use the normalized form for pronunciation?
    – Lance
    Commented Sep 2, 2023 at 23:45
  • @Lance The normalized form is what you should use for the pronunciation.
    – Draconis
    Commented Sep 2, 2023 at 23:49
  • It must not be the cf then (still need to understand what that is exactly), so I must need to look somewhere else in that JSON there? Maybe that's what the norms section is, hmm, I'm not sure. AH, you called it the normalized from, so it must be that!?
    – Lance
    Commented Sep 3, 2023 at 0:00
  • @Lance Yes, norms is normalized forms.
    – Draconis
    Commented Sep 3, 2023 at 0:34
  • 1
    @Lance Sumerian phonology is much less understood than Akkadian.
    – Draconis
    Commented Sep 3, 2023 at 1:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.