I'm probably not the first to notice that a large number of features of reconstruct Proto-Indo-European are typological irregularities. The most famous of these probably being the voiceless/voiced/voiced aspirated distinction found among the stops, which is of course the stated motivation for the glottalic theory (though I must admit, I don't really see how it helps). Another thing which, to my eye at least, looks extremely odd is the vowel system: /i/ and /u/ basically pattern with the nasals/liquids, and apparent /a ā/ can be entirely explained as actually originating from /eh2/ or /h2e/, so the only vowels which are actually necessary to reconstruct are /e ē o ō/. To put it bluntly, that's just weird. Beyond that, many of the proposals for the phonetic value of the laryngeals seem strange as well. What kind of fricative inventory is /s ɣʷ χ h/ or /s ʕ ʕʷ/? The syllable structure seems very strange as well: why can /h2/ be syllabic but not /s/, for example? Apologies for the flippancy, but some of these claims have always felt a bit unbelievable to me.

I'm not actually trying to cast doubt on the validity of the reconstruction, I understand that many aspects of it are very well substantiated, I'm just trying to get a better grasp on how I'm meant to conceptualize the reconstruction of PIE within the framework of modern typological knowledge. Have these oddities been addressed by any Indo-Europeanists or language typologists? Is there a proposed explanation? Or am I just wrong about what is and isn't a "reasonable" feature for a language to have?

  • 3
    The glottalic theory helps because /t t' d/ (or /t t' tʰ/ etc) is more typologically reasonable than /t d dʰ/, mostly. We see that sort of three-way contrast in e.g. Georgian, while I can't think of any living language that contrasts voiceless, voiced, and voiced aspirated stops.
    – Draconis
    Commented Sep 15, 2020 at 5:26
  • 3
    You may find this article by Martin Kümmel interesting. Commented Sep 15, 2020 at 8:08
  • 2
    The scarcity of vowels is not very much typologically unusual. Kabardian can well be analysed as having just a single vowel phoneme despite that phonetically all the 5 cardinal vowels are present, most of them are reflexes of the single vowel influenced by the articulatory features of the nearby consonants. For example, [u] is found only near the labial or labialized consonants.
    – Yellow Sky
    Commented Sep 15, 2020 at 9:22
  • 1
    @YellowSky especially just North of the Caucasus, which happens to be pretty much exactly the area most commonly accepted as the PIE Urheimat
    – Tristan
    Commented Sep 15, 2020 at 10:43
  • 3
    Besides the perennial problem of how to apply the findings of typological statistics to historical linguistics, there’s another undesirable assumption, PIE was a stable language that never changed? Very unlikely of course.
    – Alex B.
    Commented Sep 15, 2020 at 13:00

1 Answer 1


There are three reasons for that unusualness.

The first one is artefacts from the reconstruction method. Our knowledge of PIE is based solely on what the child languages preserved, so there's a lot of missing info. We're likely losing whole phonemes, and conjecturing phonemes that never existed. The "laryngeals" are a good example of that - was *h₃ really a single consonant? Maybe two or ten? Maybe something else, like a sequence of phonemes?

The second reason is that no language is uniform; and yet we need to pretend otherwise if we want to describe it. PIE changed over time, and it had multiple dialects, spoken by different groups; it was not a single thing. What we're doing with it is roughly like trying to reconstruct English as spoken by Chaucer, Shakespeare, and that random guy in NZ" as a single thing.

A third reason is that we're always contrasting PIE with the modern Indo-European languages, and we call "unusual" the features that we don't find in those languages - even when they're attested elsewhere. A good example of that would be the way that PIE handles *i *u - sure, it might look weird if you speak Sanskrit or Latin, but it's remarkably similar to how Ubyx did it - you don't get /i/ or /u/, and all instances of [i u] are surface realisations of either /ə/+semivowel or a nearby palatalised/labialised consonant.

  • Regarding 「A good example of that would be the way that PIE handles *i *u - sure, it might look weird if you speak Sanskrit or Latin」: Alternations [i̯] ~ [i] and [u̯] ~ [u] would not look very weird for Latin speakers, since they partially survived in Latin. Examples: <i> between vowels is pronounced as [i̯ː] (e.g. illīus with [iː] versus cuius with [i̯ː], those are unrelated words with the same suffix (genitive of pronominal declension)), duo versus du̯is (→ classical bis), se- + luōsolu̯ō
    – Arfrever
    Commented Sep 2, 2023 at 20:29

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.