Zompist Bboard Again

I find the question of "can this actually be pronounced" hangs over my head and impedes my ability to conlang. I don't want to be limited to only simple syllable structures, but I also don't want to be limited to sounds I can produce. However, my intuition for what constitutes a plausible syllable is garbage, because real languages have words like (Georgian) [ɡʷpʰɾ̥t͡skʰʷni], (Horpa) [ɣɮdzʌv] and (Nuxalk) [xɬpʼχʷɬtʰɬpʰɬːskʷʰt͡sʼ], so I feel stuck.

How do I deal with this?

I think I remember that when, a few years ago, I tried to come up with some naming languages, at first I started out from the assumption that pronunciation difficulties are entirely culturally relative and everything is equally easy to pronounce if you're used to it, but some other people here then corrected me and taught me that some sound combinations are inherently difficult to pronounce; but unfortunately, I don't really remember the details of that. (Apologies to the people who explained that stuff to me back then.)

Mostly pronounceability comes down to syllable structure, and syllable structures can often be viewed from terms of a sonority hierarchy. Generally, languages only allow consonant clusters to begin a syllable if their sonority increases, and coda clusters if their sonority decreases. Vowels, being the most sonorous, end up in the middle and are usually syllabic. This means that a syllable should steadily increase in sonority from the start, peak at the nucleus and then decrease again in the coda. The sonority hierarchy goes something like this: (Wikipedia splits voiced and voiceless stops, which is phonetically accurate but afaik never phonotactically applicable)

most sonorous: low vowels
mid vowels
high vowels (including /j w/)
flaps
laterals
nasals
voiced fricatives
voiceless fricatives
least sonorous: stops

In most languages, onset clusters will move up this chart and codas will move down it. Syllable nuclei will almost always be more sonorous; syllabic fricatives are rare and syllabic stops basically unheard of. In a grossly simplified way, a cluster which increases in sonority is an inherrently easier-to-pronounce onset than another one - i.e. /prwa/ is more pronounceable than /wrpa/. Likewise /awnt/ trumps /atnw/.
Of course, things like [ɡʷpʰɾ̥t͡skʰʷni], [ɣɮdzʌv] and [xɬpʼχʷɬtʰɬpʰɬːskʷʰt͡sʼ] don't follow that at all, as this isn't a perfect measure. Often (at least in Indo-European langs) /s/ is treated differently and gets chucked at the start or end of a word; some people have treated this as "extra-syllabic" (/ksa/ should be more commonly permitted than /ska/, but I doubt it is). Sometimes due to sound changes or some other reason, a single consonant gets treated differently - e.g. a fricative might pattern as a semi-vowel, or even as a full vowel. Sometimes morphology takes precedence, like the examples from Georgian and Horpa, which are both morphologically complex with sub-syllabic affixes. Affixes that are just made up of consonants can create consonant clusters which couldn't exist in lemma forms. Then you've got Nuxalk, which is completely batshit but can be analysed in terms of syllables; even then consonant clusters tend to clump as onset stops with syllabic fricatives which does indeed follow the sonority hierarchy.

Basically, to answer your question, pronounceability is often determined by sonority hierarchy, but sometimes it isn't. It works best in languages with simpler syllable structures, but anecdotally it seems to me that once you reach three-consonant clusters, you'll almost always find some exceptions. Exceptions to the hierarchy tend to be either fricatives (which can be syllabic or extra-syllabic), or morphologically complex forms, or motivated by historical sound changes¹. There's nothing stopping you from violating this principle as long as it's consistent within a language - even Rgyalronic languages and Nuxalk have systematic syllable structures.

---

1. Another example I just thought of - the open syllable "conspiracy" in Slavic languages. For whatever reason, at some point speakers of proto-Slavic decided that they hated syllable codas but were chill with onsets. They then chucked everything they could into the onset, making things like *bogatьstvo being split up as /bo.ga.tь.stvo/. Then the descendants mixed it all up again by dropping extrashort vowes and you get /xʂɔw̃ʂt͡ʂ bʐmi f ˈtʂt͡ɕiɲɛ/.

Ketsuban wrote: ↑Sat Jun 25, 2022 12:56 pm I find the question of "can this actually be pronounced" hangs over my head and impedes my ability to conlang. I don't want to be limited to only simple syllable structures, but I also don't want to be limited to sounds I can produce. However, my intuition for what constitutes a plausible syllable is garbage, because real languages have words like (Georgian) [ɡʷpʰɾ̥t͡skʰʷni], (Horpa) [ɣɮdzʌv] and (Nuxalk) [xɬpʼχʷɬtʰɬpʰɬːskʷʰt͡sʼ], so I feel stuck.

How do I deal with this?

I would imagine you could draw a Punnet Square for this, showing both simple & complex syllable structures, as well as -

Wait.
Maybe use simple syllables, but don't have all of the smaller elements (mora, phones, etc) be simple and easy to pronounce?

Would that work for you?

I'd add just a few nuances to Darren's excellent post:

* Other (not all) fricatives share s's ability to start a syllable: z, ś, š, ž, h. Russian has [fs]. Surely ð θ x can do it, though I can't think of examples offhand.
* Maybe worth a note that two stops can adjoin, e.g. [kt]. Greek could even have two adjoining aspirated stops.
* Nasals act like stops in some ways, like continuants in others.

zompist wrote: ↑Sun Jun 26, 2022 7:02 pm I'd add just a few nuances to Darren's excellent post:

* Other (not all) fricatives share s's ability to start a syllable: z, ś, š, ž, h. Russian has [fs]. Surely ð θ x can do it, though I can't think of examples offhand.

My "the" is basically just [ð] tacked onto the beginning of the next word if the next word begins with a consonant. It might be syllabic, though.

(Edit: I can easily sing "th' " as a nonsyllablic clitic.)

clhp'xwlhtlhplhhskwts' (which is a real Nuxalk word from wikipedia)

I fear I really need a recording, but even if you can split this monstrosity into syllables, I fail to see how it's physically sayable without vowels. I mean at some point you are surely going to have to at least insert a schwa just to be physically able to put the consonantal string together?

evmdbm wrote: ↑Tue Jun 28, 2022 6:43 am clhp'xwlhtlhplhhskwts' (which is a real Nuxalk word from wikipedia)

I fear I really need a recording, but even if you can split this monstrosity into syllables, I fail to see how it's physically sayable without vowels. I mean at some point you are surely going to have to at least insert a schwa just to be physically able to put the consonantal string together?

The only way I can see that being pronounceable is treating the sibilants and voiceless laterals as syllabic.

Travis B. wrote: ↑Tue Jun 28, 2022 11:27 am

evmdbm wrote: ↑Tue Jun 28, 2022 6:43 am clhp'xwlhtlhplhhskwts' (which is a real Nuxalk word from wikipedia)

I fear I really need a recording, but even if you can split this monstrosity into syllables, I fail to see how it's physically sayable without vowels. I mean at some point you are surely going to have to at least insert a schwa just to be physically able to put the consonantal string together?

The only way I can see that being pronounceable is treating the sibilants and voiceless laterals as syllabic.

Having worked a little bit with Salishan speakers (in the days before my hearing loss went too far), yeah, that's basically what it sounded like to me.

Ketsuban wrote: ↑Sat Jun 25, 2022 12:56 pmI also don't want to be limited to sounds I can produce.

I deliberately put this constraint on my conlangs because I want to be able to speak it – not in a fluency way but just be able to read aloud what's written. For me, creating a conlang which I cannot comfortably produce from my own mouth is pointless, but maybe that's because my conlangs have always been driven by euphony, which is subjective, and I am obviously biased.

My two pence about your problem though: I also find certain things like this difficult at times, but I try to study languages and take cues from them. My philosophy is "if X natlang does it, I can do it".

fusijui wrote: ↑Tue Jun 28, 2022 10:01 pm

Travis B. wrote: ↑Tue Jun 28, 2022 11:27 am

evmdbm wrote: ↑Tue Jun 28, 2022 6:43 am clhp'xwlhtlhplhhskwts' (which is a real Nuxalk word from wikipedia)

I fear I really need a recording, but even if you can split this monstrosity into syllables, I fail to see how it's physically sayable without vowels. I mean at some point you are surely going to have to at least insert a schwa just to be physically able to put the consonantal string together?

The only way I can see that being pronounceable is treating the sibilants and voiceless laterals as syllabic.

Having worked a little bit with Salishan speakers (in the days before my hearing loss went too far), yeah, that's basically what it sounded like to me.

See also Ridouane’s (2008) investigation of syllable structure in Tashlhiyt Berber, which has essentially the same structure: e.g. [tkkststt] is syllabified as [tk.ks.tstt], and /mqar tnt ksiʁ ar disnt nttlʢab/ is something like /m.qar.tn.tk.si.ʁar.di.sn.tnt.tl.ʢab/ — both with syllabic plosives! Notably, these do not seem to contain any schwas whatsoever, not even voiceless ones (insofar as those even exist in the first place). Even a sentence like [tsskʃftstt tftχtstt] is schwa-less.