I am trying to doing an exercise that requires me to reconstruct a proto language from two languages that has a difference in the [a] and [e] phonemes. I know I cannot use the majority rule because there are only 2, but I am assuming that [a] is more natural than [e], is this assumption correct?

  • 2
    Are there any conditions on the development in the languages? What is the rest of the inventory? Apr 23, 2017 at 17:33

2 Answers 2


The usual understanding of "naturalness" is either cross-linguistic statistical frequency or asymmetrical implication (e.g. "no language has X unless they also have Y"). If we take the letters a,e at IPA face value, then this could be true. The main problem is that "a" is not a particularly uniformly-used letter – it represents something low and not rounded, but the exact location is up for grabs. There actually is no IPA letter that ideally reflects the pronunciation of "a" in Spanish or German, rather, the typical pronunciation of "a" in Arabic (katab "he wrote", Levantine pronunciation) is closest to Jones cardinal vowel 4. So if for example the letter "a" is used to represent IPA [ɑ] and "e" is used to represent [æ], a "naturalness" claim is on shaker grounds. With those caveats, the most common vowel in human language is [a], and there is a widespread, probably universal implication that any language with [e] also has [a], but the reverse is not the case. There is, of course, the question of the Proto-Indo-European vowel system where "a" seems to be more marked that "e", which itself could suggest that the letters shouldn't be taken to represent those most common phonetic values.

A third factor that bears on the concept of "naturalness" is whether phonological distinctions are phonetically sensible under one theory vs. the other. If "a" causes palatalization of a preceding consonant, that is more sensible if "a" is really [e]; if "e" causes retraction of i to some back vowel, that is more sensible if "e" is really [a]. As always, you have to clear about the supposed phonetic values of the vowels in question. In Arabic, "a" has an affinity for fronting and raising to [e] or [i], which isn't particularly natural if you think of "a" as the back vowel [ɑ]. But it is less surprising if it is a front vowel, as it actually is in Arabic.

  • Also one has to keep in mind that all these are just general tendencies, which can have exceptions. E.g. French open /ε/ typically does not cause palatalisation of preceding consonant, while strangely enough, the frontal /a/ does. On the other hand, I do not recall one single language with something similar, so while such phenomenon is apparently possible, it is also very rare.
    – Eleshar
    Apr 26, 2017 at 18:09

I doubt that it is helpful to try comparing the "naturalness" of single sounds. But you you might be able to compare the naturalness of the sound systems of the protolanguage or the naturalness of sound changes, under various hypotheses of what the protolanguage was like. But you're not telling us anything else about the vowel systems of any of the languages involved.

You could try to find instances elsewhere of context free vowel changes [a] > [e] versus [e] > [a]. The latter change applied in the development of Indo-Iranian from PIE.

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.