(Because the bulk of this post is more about my conlang and less about specific natlangs, I reposted it here. I originally posted it in the L&L thread.)
loglorn wrote:I think it's probably just the fricatives that are syllabic. Considering the most sonorous thing to be the nucleus is usually good measure.
According to that article on Berber referred to previously;
(1) The sonority order in Berber is:
voiceless stops < voiced stops < voiceless fricatives < voiced fricatives < nasals < liquids < semivowels (they call them "high vocoids"; IIUC rounded "vocoids" would also fit here) < vowels
That's mostly pretty consistent with what many languages that have a sonority hierarchy have; but sonority hierarchies are language-specific, though there is quite a bit of cross-linguistic similarity. Some languages have sounds that aren't included above; they fit them into their sonority hierarchy if they use one. Some languages have a much coarser, less-detailed sonority hierarchy; some have a much finer, more-detailed, sonority hierarchy.
(2) A sonority peak (more sonorous than the preceding phoneme (if there is a preceding phoneme in the same word) and more sonorous than the following phoneme (if there is a following phoneme in the same word) is a syllable nucleus.
It looks to be that in that dialect of Berber only voiceless stops couldn't ever be nuclear. But maybe even they can be; I'd need to read the article more carefully to decide.
Many languages have "rules" that the phonemes in an onset must not decrease in sonority; or must increase in sonority; or must
"jump up" or "skip up" by some minimum amount; as one moves from the beginning of the syllable towards the nucleus.
Many languages have "rules" that the phonemes in a coda must not increase in sonority; or must decrease in sonority; or must
"jump down" or "skip down" by some minimum amount; as one moves from the nucleus of the syllable towards the end of the syllable.
Adpihi's arrangement is not far-fetched.
A syllable begins just before a sonority trough, or the first phoneme of the word.
Each subsequent phoneme in the syllable's body (the part that isn't the coda) must be more sonorous.
The nucleus is the sonority peak.
Each subsequent phoneme in the syllable's rime (the part that isn't the onset) must be less sonorous.
The syllable ends just before a sonority trough, or at the end of the word.
(1) if two phonemes in the same syllable are both in the same syllable margin -- both in the onset or both in the coda -- then the one closer to the nucleus must be more sonorous than the one further from the nucleus.
(2) if two phonemes in the same syllable are equidistant from the nucleus, then the one in the coda must be more sonorous than the one in the onset.
(3) if two phonemes in the same syllable are both in the same syllable margin -- both in the onset or both in the coda -- and one is voiced while the other is unvoiced, then the voiced phoneme must be closer to the nucleus than the unvoiced phoneme.
Adpihi's sonority hierarchy is:
stops, plosives, taps, flaps, etc. < fricatives, spirants, etc. < nasals < liquids < semivowels < vowels.
Let's number these in order of increasing sonority;
1. stops (T)
2. fricatives (F)
3. nasals (N)
4. liquids (L)
5. semivowels (W)
6. vowels (V)
If I encounter a consonant-cluster like T1
at the beginning of a word, then I know T1
is a syllable, and T2
is, or is the beginning of, the next syllable.
This isn't entirely impossible in Adpihi. A four-consonant root might be put into a binyan that doesn't have any vowels between the roots' consonants. So the word's root-part would be a four-consonant cluster.
Adpihi doesn't have clusters for nuclei. (In particular it doesn't have diphthongs or triphthongs or tetraphthongs.)
If we encounter a phoneme-string like TFFT or FNNF or NLLN or LWWL or WVVW, then there is a syllable boundary between the two phonemes of equal, and highest, sonority.
So Adpihi doesn't have the "no hiatus" rule that that article's dialect of Berber has. In particular, two consecutive vowels in the same word must belong to two different syllables; so instead of diphthongs, Adpihi has hiatuses.
There's also a syllable-boundary between two phonemes of equal and lowest sonority.
If we encounter a phoneme-string such as FTTF or NFFN or LNNL or WLLW or VWWV, then there is a syllable boundary between the two consecutive phonemes of like sonority that are less sonorous than the one before them both and than the one after them both.
TFFFT would be syllabized as TF F FT.
FNNNF would be syllabized as FN N NF.
NLLLN would be syllabized as NL L LN.
LWWWL would be syllabized as LW W WL.
WVVVW would be syllabized as WV V VW.
If a word begins with two phonemes of equal sonority both less than the sonority of the next phoneme after them both, then an exception might be made that allows both of those first two phonemes to be part of the onset of the word's first syllable. (The third phoneme will also be part of the body of the word's first syllable.)
If a word ends with two phonemes of equal sonority both less than the sonority of the phoneme immediately preceding them both, then an exception might be made that allows both of those last two phonemes to be part of the coda of the word's last syllable. (The third-from-last phoneme will also be part of the rime of the word's last syllable.)
I don't know what to do yet for trough-clusters of three or more consecutive phonemes in the interior of a word equally sonorous with each other, and less sonorous than the phoneme before the first of them (in the same word) and the phoneme after the last of them (in the same word).
I suppose if such a longish trough-cluster began or ended the word, it could be part of the onset of the word's first syllable, or the coda of the word's last syllable; but I don't like that idea. Maybe I'll come to like it; or maybe I'll find out it never has to happen.