I’ve been reading about PIE and i’m confused. As I understand, it is assumed to be the language spoken by Europeans 6,000 years ago. No written record of the language has ever been found so the language has been reconstructed through seemingly arbitrary means. So, In all likeliness actual PIE sounded very different. What makes this language (as it exists today) useful? This is essentially a conlang that is too complicated to learn. What am I missing? Sorry if I’m coming off as negative. I find PIE both confusing and fascinating.

  • Lvxferre [he/him]@mander.xyz
    2·
    5 hours ago

    Shamelessly plugging !linguistics@mander.xyz here. This topic fits well in that community.

    Also, I apologise for the wall of text, it’s just this topic interests me by a lot.

    Proto-Indo-European is a reconstruction of the common ancestor (actually two; more on that later) of multiple languages spoken in India and in Europe, that are clearly related to each other.

    We know there are a lot of gaps in our knowledge about it, that people have been trying to fill since the 19th century, but ideally this reconstruction should be as close as possible to the real deal that we can reasonably get. For example:

    • We don’t really know the true phonetic nature of the three series of occlusives of the language. But we know they were three; certainly not just two, and likely not as many as four.
    • We don’t know all suffixes of the language. But we do know the language had ~eight noun cases, those cases were marked by suffixes (unlike position, as in English… or particles, as in Japanese), and worked in a similar way as they do in Sanskrit, Latin, and Ancient Greek.
    • We might not have a fully solved family tree for the PIE descendants, but we have a pretty good guess on who’s a descendant (like Spanish, Italian, Hindi, German, Swedish) and who’s not (like Basque, Etruscan, Tamil, Hungarian, Finnish).

    About the method:

    The method used is called the comparative method. It boils down to looking for regular sound correspondences in the child languages and, based on sound changes we attested, coming up with hypotheses that would explain those sound correspondences.

    It is not just assumptions; an assumption pops up when you treat the uncertain as certain. When using the comparative method you’re expected to come up with multiple competing hypotheses, keeping in mind they might be wrong, and look for info that helps you to ditch one or another.

    And as messy it is, it has predictive power. A good example of that are the laryngeals, annotated *h₁ *h₂ *h₃. When Saussure (a linguist) proposed PIE had those sounds, nobody had direct evidence they existed; it was only indirect, based on sound correspondences found in the descendant languages (mostly the nearby vowels). Those laryngeals were shown to have actually existed once Hittite was discovered, because it actually preserved sounds in their place.

    Another example comes from Proto-Romance. Proto-Romance is what you get when you apply the comparative method to the modern Romance languages, as opposed to their real ancestor, Latin. So by comparing the theoretical Proto-Romance with the real deal Latin, we can gauge how well the method works. And well, Proto-Romance is surprisingly close to what’s attested for Late Imperial Latin, specially the one used by poor people. (I can go further on that if you want).

    About conlangs vs. reconstruction: PIE is not a constructed language. It’s a reconstructed one.

    The key difference is that, when you’re building a conlang, you’re free to decide everything about it. This freedom does not exist for Proto-Indo-European or other reconstructions, you need to do it based on the data at hand.

    You can of course create a conlang using that reconstruction as a basis; plenty people do*. And if you ever see “PIE being spoken” out there, it’s probably something like this. But note that you’ll need to add stuff not found in the reconstruction, if you want it to be actually usable, already crossing the line between science and art.

    *just in this case, please PLEASE make sure to fix that unholy notation. I have an alternative one if you want, it makes the language look a bit more reasonable.

    On being two languages:

    When people started noticing the obvious similarities between Sanskrit, Latin and Greek, in the 19th century, the idea they had a common ancestor came up naturally. And then, when Hittite was discovered, it was obvious that common ancestor of Sanskrit/Latin/Greek was also the ancestor of Hittite.

    But that doesn’t tell you the whole picture. What happened was something like this:

    • you have a language. Let’s call it “A”.
    • that language has two children. Let’s call them “Hittite” and “B”.
    • language “B” has a thousand children. And grandchildren, and grand-grandchildren. Among their descendants you see Latin, Sanskrit, Greek, Russian, English, and a lot more.

    Now. Which language is Proto-Indo-European: “A” or “B”? …tricky question! People use the term to refer to both languages. Even if there’s a gap of 1~3 millenniums between them. Yay, scientific precision! /s

    When they see the distinction, and want to specify which one, they call A “Early PIE” and B “Late PIE”. But the mess still affects reconstructions a fair bit; it’s like trying to reconstruct Ancient Greek and Modern Greek as if they were the same language, you know? (Or, dunno, English and Old English.)

    I believe some really weird shit we see in modern reconstructions is caused by this. Such as the weird *e *o *e: *o: vowel system; length is likely a late PIE feature, while this two-qualities (or three) system is from early PIE.