They say nobody exactly knows how Indo-European words were actually pronounced since obviously there was no Sony sound recorder back then.

So, what are these phonetic symbols that they use to represent Proto-Indo-European words? If they don't know how they are pronounced, what are they writing? We write something we can "read" later.


It should also be noted that the point of those reconstructed phonemes is that they "oppose" each other. I.e. a string of reconstructed phonemes uniquely (more-less) identifies a word, say, domos in this exact configuration means "building (something which was built)", as opposed, to, say, tomos which would mean "something which was cut in pieces". We know for sure that tomos is different from domos because their sequences of phonemes differ. We don't know how it was pronounced exactly, the real pronunciation of domos could as well be something like "tamash", we'll probably never know. But we know for sure that those were different words with different sets of sounds, and they had different reflexes (continuations) in daughter languages. We don't know the exact pronunciation but we still have to settle on some way to represent them symbolically (how do we tell them one from another otherwise?), so we settle on the most logically likely variants.

  • 1
    OK, let's talk about 'tomos' you mentioned. You said that we have had to devise a way to differentiate between different words; and that we didn't mean to represent the exact pronunciation. So why, for example, didn't linguists demonstrate it as 'gomos' (for instance) or anything else like 'cumes', 'rwsmuis' etc. Why 'tomos' when we don't care nor know about the pronunciation? – Joseph_Marzbani Apr 10 '17 at 14:52
  • 1
    Because the majority of daughter languages have reflexes expectable from such sounds. domos gave dom in Russian, domos in Greek, domus in Latin, damah in Sanskrit. It would be strange to propose something like gomos here instead of domos. First, it's confusing; and, second, thanks to typology and observable historical changes in more modern languages (such as Latin=>Romance languages), we know that some phonetical changes are more likely than others (it's pure anatomy), for example, debuccalization S=>H is much more likely than H=>S; and D=>D is as straightfoward as you can get. – Constantine Geist Apr 10 '17 at 15:29
  • Thank you. To me, your answer was closer to acceptable... – Joseph_Marzbani Apr 10 '17 at 15:31
  • Thank you for making a point of the problems of incomplete information in regard to this exciting and useful endeavor of reconstructing a proto-language. – DukeZhou Apr 11 '17 at 20:31

There are very many opinions of the "meaning" of the letters used to represent PIE reconstructions. One approach treats them as algebraic abstractions, where e.g. bh represents some sound that corresponds to φ in Greek and f in some positions in Latin, and b in English: there is no claim as to phonetic value. The other approach is that these are approximate hypotheses regarding the most likely pronunciation, thus reconstructed m is probably pronounced [m], and so on. Within the latter approach, there can be competing theories, for example some people reconstruct [d] where other people reconstruct [t'], that is there are competing ideas of what the substance of the proto-phonemes are. The strength of the argument for particular phonetic reconstructions varies, so while n is fairly uncontroversial (the reflexes are [n]), gʷʰ depends on sophisticated argumentation, since no language actually directly attests [gʷʰ]. The so-called laryngeals are a particular mystery, since they are preserved in no living though do show up in Hittite (where we don't really know for sure how things were pronounced).

The reconstruction of phonetic values is based on overlapping analogies and estimations of phonetic likelihood. For example, gʷʰ (a voiced aspirate labialized velar) is justified by the fact that that sound behaves like other voiced aspirates (is voiced in Germanic and Armenian, not devoiced), and acts like the other labialized velars (kʷ, gʷ) which are better preserved.

In other words, the letters represent hypotheses as to pronunciations, though with a certain limit on ability to reconstruct exact phonetics. I don't think there is much hope for reconstructing a very exact phonetic value of r, in light of the very many phonetic kinds of rhotics that exist, and even is one of the laryngeals is "ħ", there are quite a number of phonetic types of ħ in the world's languages, so some questions are probably unresolvable.

  • Do you mean these symbols are, sort of, a "description" of how "possibly" those sounds were produced (i.e. by tongue, larynx, nose etc)? – Joseph_Marzbani Apr 10 '17 at 4:30
  • In the same way that the IPA symbol [ʃ] is a description of how a certain fricative is produced: the letters imply certain articulatory facts, and for historical reconstruction, since we haven't heard the sounds we can't claim direct evidence for how the sounds were produced, this is a claim as to how they were probably produced, within the limits of precision of any phonemic transcription. – user6726 Apr 10 '17 at 4:43
  • Thank you very much. So, the short one-word answer to the question I asked in the last comment is a YES, right? – Joseph_Marzbani Apr 10 '17 at 4:47
  • Yes, it is yes. – user6726 Apr 10 '17 at 5:06
  • 😀👍 I love StackExchange – Joseph_Marzbani Apr 10 '17 at 5:19

The letters used to represent reconstructed proto-Indo-European words are like the symbols used in algebra. They are conventional symbols for unknown entities.

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