LinguList
commented
9 months ago Tones should never be placed on vowels.
Relative tongue position can be ignored
Mid-long segments are an exaggeration, but if we find them to contrast with long and normal, we may want to distinguish them.
Alveolo-palatal vs. palatal consonants
I for my part do not think it is a big deal to merge them. No language has both. It is more a matter of linguistic tradition to opt for one in a certain area.
Unreleased vs. released stops
Distinction occurs in different positions of the syllable, so they would never be in opposition. I agree.
Palatalized vs. plain palatal consonants
The problem is that the series opposition in many languages, where palatalized consonants are interpreted as phonological, would be lost. But this opposition occurs in quite a few datasets, so one should discuss it.
Pre-aspirated vs. aspirated
I agree.
Voiceless vs. devoiced consonants
Yes, I agree
Bilabials vs. labiodentals
Can be ignored, I agree.
I already made an adaption to distinguish strong (fortis/tense) consonants from plain ones, but it might be a controversial one - I am following Phonotacticon which assigns [+tense] to strong consonants. I think that this choice can be motivated from an articulatory standpoint, but I am happy to discuss alternatives.
How does this surface in IPA?
arubehn
Based on the observations from the NorthEuraLex test run, there are a couple of open questions that I would like to address. I am listing all classes of sounds below that currently cannot be distinguished from another.
Tones on vowels
Do we want to support representing tones as part of a vowel (like [á])? That would technically be straightforward, but I recall a discussion where we said that we did not want to support suprasegmental features as part of segments.
Relative tongue position
e.g. [i] == [i̟] - we do not account for relative tongue positions, and I think that in the big picture, this distinction is too fine-grained to motivate introducing additional features.
Mid-long segments
e.g. [eː] == [eˑ] - mid-long vowels are currently [+long]; although looking at the NorthEuraLex data, I am not sure whether that is a good choice, since there they mostly seem to contrast with long vowels. Anyway, with only one [±long] feature, we can logically only distinguish two length qualities - I think that is justifiable, but it's a limiting factor we should keep in mind.
Alveolo-palatal vs. palatal consonants
e.g. [ç] == [ɕ] - this one honestly bugs a bit, I think we should be able to distinguish these two places from another, but I am not sure how to do that with our current feature set.
Unreleased vs. released stops
e.g. [t] == [t̚] - I think it is fair that we don't make this distinction.
Palatalized vs. plain palatal consonants
e.g. [j] == [jʲ] - I honestly didn't know that there even are palatalized palatal consonants... I currently assign palatalized sounds the same features as palatal ones, so if we want to make a distinction there, I don't know how to do that with the current logic.
Pre-aspirated vs. aspirated
e.g. [kʰ] == [ʰk] - holds true for all kinds of coarticulation that can precede the sound in question; justifiable in my opinion.
Voiceless vs. devoiced consonants
e.g. [p] == [b̥] - justifiable imo.
Bilabials vs. labiodentals
e.g. [m] == [ɱ] - in NorthEuraLex, this only affects Romance languages (Catalan, Spanish, Italian), where I know this is just an assimilation process. I don't know how many languages phonemically distinguish between bilabials and labiodental, so I don't know whether it is worthwhile to think about a way to distinguish them.
Strong consonants
I already made an adaption to distinguish strong (fortis/tense) consonants from plain ones, but it might be a controversial one - I am following Phonotacticon which assigns [+tense] to strong consonants. I think that this choice can be motivated from an articulatory standpoint, but I am happy to discuss alternatives.