I've been pondering a conlang with a rather unusual orthography. I'm only stating this because no natural language has this sort of writing system. Essentially, its a system that only writes consonants like abjads do, but vowels are far easier to predict. Most of the time, the vowels can be guessed by things such as part of speech and accompanying particles and affixes (such as articles and tense/aspect markers). Syllable structure too can easily be determined too, mostly thanks to the language's phonotactics (a word like pst could only be pVstV, there is literally no other valid vowel pattern for this set of consonants, even in the worst case scenarios there's only at most 2 possibilities which are conveniently distinguished by number of vowels).

When I first mentioned this over on the conlang board, I was told it wasn't really a true abjad since its possible to deduce the vowels even when you've never encountered the word before. So I made another question. However, now I'm thinking this system can't be 100% reliable. You may not need to know the vowels for verbs and nouns and whatnot, but what about the affixes and particles that hint at what they are? They would have to be arbitrary. Of course, the consonants alone would give the vowels away, but it wouldn't be possible to deduce this from the consonants themselves; you just have to be remember which vowels go with each particles and affixes.

This makes me question whether or not this is truly an abjad. Are the particles 100% unpredictable? If you need to just remember that t comes with a and b with u and whatnot, would that classify as predictable or not? Keep in mind, the number of morphemes where this is the case would be quite minimal, so for most words it would still be obvious. You'd have to just learn each of the particles and affixes one by one, but for everything else this would be unnecessary.

Of course, in the real world abjads are mainly defined by ancestry. If we went by how they worked, they would be one and the same with abugidas, minus the vowel markers being optional. Also, I don't know any Semitic language, but I do see Hebrew speakers insisting a lot that predicting the vowels in their language isn't actually that hard, with the exception of recent loanwords and rarely used technical terms of course.

So, would a language like this classify as a true abjad or is it just an analogue form of a compression algorithm? I am unsure which it would be, and its giving me problems discussing it online.

  • 11
    I don’t think the conlangers were right: there is absolutely no requirement that vowels must be unpredictable for it to qualify as an abjad. An abjad is simply a system that only represents consonants as primary graphemes, relegating vowels to (optional) diacritics or not representing them at all. Abjads and abugidas are indeed, at an abstract level, defined by whether or not the vowels are optional: in an abugida, consonant graphemes have an inherent vowel, which they don’t in abjads. Ancestry plays no part (or the Latin alphabet and most abugidas would all be abjads). Commented Jan 6 at 9:15
  • You can't infer anything from dead languages, see Phoenician.
    – Lambie
    Commented Jan 6 at 22:38

1 Answer 1


"Abjad", like other labels for writing systems, seems easy to intuitively define but is really difficult to be precise about. Look at examples like Ugaritic, which is pretty clearly an abjad…but also has glyphs for three consonant-vowel syllables. The glyph for /ʔ/ changes depending on what vowel comes after it. If that's an abjad, then what about Old Persian, which has a handful more consonants that change like that (sort of like a syllabary), and also has letters for some vowels, but not for some others (there's no sign for ī for instance)?

So I would call it an abjad if it's useful, in one way or another, to analyze it that way. If you have one glyph for each consonant and no glyphs for vowels, you'll want to teach people the writing system just like you would an abjad, so I'd call it an abjad regardless of predictability.

  • 1
    That would make sense, regardless of predictability. I mean, if you have to just remember which vowels go with each of the dozens of particles/affixes, or what declension/conjugation each word belongs to, why not just say its an abjad for simplicity?
    – user43283
    Commented Jan 6 at 6:13

Your Answer

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