Zompist Bboard Again

If there's a one-word joke in conlanging, it's probably phonology. The LCK says "people sometimes post phoneme inventories on my board and anxiously await comments. This mystifies me... it's like a painter showing you which colours he intends to use." At the same time, though, you'll get exchanges like this one expressing a kind of superior disappointment with how limited in scope some conlangs' "phonologies" are.

Moose-tache wrote: ↑Sat May 01, 2021 2:31 am Like a lot of amateur conlangers, Misali seems to think that “phonology” means “a chart of consonants and vowels, plus maybe some allophony.”

Raphael wrote: ↑Sun May 02, 2021 12:27 pm Sorry for the ameteur-ish newbie question, but beyond that, what else does "phonology" mean?

Creyeditor wrote: ↑Sun May 02, 2021 1:52 pm E.g. syllable structure, phonotactics of roots, affixes, and words; word stress/tone/pitch accent, phrasal stress, prosody, intonation; non-allophonic phonological processes, sandhi, etc.

Nortaneous wrote: ↑Sun May 02, 2021 5:14 pm what creyeditor said, but also meditate upon Morphological Causative Formation in Shangzhai Horpa (doi:10.1163/2405478X-90000031). here are some simplex-causative pairs:

poɣ 'be dense' > spoɣ
lùn 'get, acquire' > zlùn
rkʰù 'be cold' > skʰù
nrgə̀ 'wait' > zgə̀
tʃʌˠ 'wear (a hat)' > ɬtʃʌˠ
ʒə̀ 'accuse' > ɮdʒə̀
ɣmə 'blow' > ɣzmə
ʁlboɣ 'explode' > ʁzboɣ
fsəm 'close (mouth)' > fɬtsəm
nɣzʌv 'rub (to soften hide)' > ɣɮdzʌv
mtʂʰə̀ 'be diligent' > fɬtʂʰə̀
ʁvrdə̀ˠɣ 'sink' > ʁvzdə̀ˠɣ

as it happens, the causative affix that is applied to all these words is just |s-|, and everything else is (morpho-)phonology, due to the pressures of shangzhai horpa phonotactics. phonology can do many things

also consider any tonal system more complex than like vietnamese

The problem, at least as I see it, is there's absolutely no guidance on how to structure anything beyond a phoneme inventory and maybe some allophony, and I think the fault lies with the LCK chapter "Sounds". Making a table is easy and fits with the framing treating the phonetic space as a shopping list starting with mostly the English phonemic inventory and exploring modestly from there, but...

• Phonotactical constraints get a single page consisting of "have a syllable structure" and pointing out a couple of illegal clusters in English. No mention of the sonority hierarchy even though one English example violates it and the other doesn't, or of any other approaches since whatever is happening in Khmer or Horpa makes a nonsense of it even before you get to Miyako and the Salishan languages.
• The section on tone is less than a page and focuses heavily on contour tones rather than register tones (which I can understand - Zomp's main tonal language familiarity is Chinese); tone sandhi gets namedropped but nothing more.
• Stress- and pitch-accent are much the same - we get two simple stress rules and the single example of Latin's more complex rule interacting with vowel quality, plus a description of the accentual system of standard Japanese, but nothing that would leave the reader feeling like they can come up with their own system.

I think I've made my point. How do we fix it?

What exactly do you want to fix?

It seems like you think the LCK is too simple; I would remind you that it is written for beginners. You know lots more now; good for you.

If you think people on the ZBB need to know more about certain topics, you are welcome to provide some tutorials.

If people want to create a really outrageous sandhi system, more power to them. I would however strongly push back against suggestions that conlangers in general are doing something wrong if they don't do that.

well, two issues .... one seems like a matter of terminology .... personally i'll stick with the simpler definition of phonology, .... a table of consonants and vowels, rules of where they can go, and maybe allophony if i really need it. the stuff about root constraints etc is not really phonology, it's, as we say, morphophonology.

then there's the matter of whether it's okay to have a language where there is no sandhi, no allophony, etc .... I am saying it is, because at least with a pure CV language, which is what a lot of mine are, there is either no allophony or the allophony doesnt come into play when deriving daughter or parent languages. e.g. a three vowel language, with just /a i u/, isnt likely to have allophony so strong that /u/ turns into ⁅i⁆ or ⁅a⁆ .... at least not in stressed syllables (Moonshine has /a i u/ when stressed but it merges to /a i/ when unstressed).

I've also convinced myself that it's okay to have a tonal language with no sandhi whatsoever, so long as it is word-tone and not a distinct tone for every individual syllable. Maybe a simple tone system could get by with no sandhi even so, but I havent made such a language and likely never will.

having said all that, for my languages in particular, phoneme frequency is very important. Poswa's consonant inventory is not far off from that of English, although it has phonemic labialization, but there's no way that Poswa could ever be mistaken for English, or for a European language, or for anything spoken on the entire planet Earth, because unlike all the languages of Earth, Poswa has a very strong bias towards labial consonants, such that /p b/ are something like fifty times more common than /k g/, even though all are on an equal footing in terms of status as phonemes.

I think one of the problems with phonology is that there really are very few patterns. Stuff like allophony, sandhi and sound change don’t have any particularly clear limitations; there are patterns, but only vague ones, and the only real way to learn about them is to read a lot about sound change.

One prominent exception to this: stress patterns. These actually do have clear patterns and theory behind them, but I think that not a lot of conlangers are aware of how to utilise these.

zompist wrote: ↑Sun Aug 22, 2021 6:43 am If you think people on the ZBB need to know more about certain topics, you are welcome to provide some tutorials.

Yes please! My favourite part of the old ZBB was the tutorials, and it upsets me that we don’t seem to do them anymore.

If people want to create a really outrageous sandhi system, more power to them. I would however strongly push back against suggestions that conlangers in general are doing something wrong if they don't do that.

I don’t think this is the complaint. My understanding of the issue is that quite a lot of conlangers do really detailed morphosyntax, semantics etc., but still seem to think phonology just ‘consonants + vowels + syllable structure’, with a bit of allophony if we’re lucky. It’s rare to see a conlang with anything more involved than that: things like tone sandhi, word structure constraints and (morpho)phonological rules are omnipresent in natlangs, but rare in conlangs.

Ketsuban wrote: ↑Sun Aug 22, 2021 6:00 am...there's absolutely no guidance on how to structure anything beyond a phoneme inventory and maybe some allophony, and I think the fault lies with the LCK chapter "Sounds..."

You realize some of us have been conlanging since many years before the LCK existed? Studying natural languages provides plenty of structure on how to create a complex and realistic phonology.

My complaint about Misali was that he talks about a language "having" or "not having" a certain phoneme and leaving it at that, as if you could learn something about Maori and English by the fact that they both "have ŋ." If he doesn't care about phonological complexity, that's fine, but then he shouldn't spend a quarter of a video criticising a phonology he doesn't understand.

Is it really a problem anymore? Most of the phonology posted here is really well done!

Tutorial threads are great tho!

I suppose there's in harm in a graded set of examples, for example:

This is a simple phonology: /p t k b d g f s m n l r w j i e a o u/. It will get you started, but you will also need to think about this, this, and this.

up to

This is as complicated a phonology as anyone should ever need:
- /(about 30 phonemes)/...
- permitted clusters are: /(some clusters)/...
- /phoneme/ has allophones [a, t, m] in certain contexts...
- /a/ becomes /b/ before /c/...
- and many more!!!

the hard part is that phonology interacts with other aspects of the language. not just morphology, but also root structure constraints distinct from word structure constraints, phonological operations that morphology can exploit (e.g. word-level backing in kusunda, short and long forms in mandarin), syllable structures only licit in derived forms (-ŋθs in english), complex phonotactic restrictions with morphological implications (causative formation in shangzhai horpa), even phonosyntax (burmese developing tonal word class distinctions from phrase-final creaky voice). even a simple-looking language like gilbertese can turn out to do very bizarre things to enforce a very bizarre trimoraic foot structure

alice wrote: ↑Sun Aug 22, 2021 12:12 pm - /a/ becomes /b/ before /c/...

That's a pretty striking fortition. I'm guessing you first get a > u, then u > w and w > b. The conditioning environment is odd, but perhaps /c/ first becomes a labiovelar, which motivates the a > u raising?

(I'm guessing that's not what you meant, but I found it amusing.)

zompist wrote: ↑Sun Aug 22, 2021 5:12 pm

alice wrote: ↑Sun Aug 22, 2021 12:12 pm - /a/ becomes /b/ before /c/...

That's a pretty striking fortition. I'm guessing you first get a > u, then u > w and w > b. The conditioning environment is odd, but perhaps /c/ first becomes a labiovelar, which motivates the a > u raising?

(I'm guessing that's not what you meant, but I found it amusing.)

clearly a typo for a > b / #_C, caused by the collapse of certain pretonic vowels to schwa, followed by the merger of ə > u, reduction of word-initial vowels to semivowels as in Tocharian B, and late fortition

alice wrote: ↑Sun Aug 22, 2021 12:12 pm I suppose there's in harm in a graded set of examples, for example:

This is a simple phonology: /p t k b d g f s m n l r w j i e a o u/. It will get you started, but you will also need to think about this, this, and this.

up to

This is as complicated a phonology as anyone should ever need:
- /(about 30 phonemes)/...
- permitted clusters are: /(some clusters)/...
- /phoneme/ has allophones [a, t, m] in certain contexts...
- /a/ becomes /b/ before /c/...
- and many more!!!

I like this idea! But it feels like it could get rather long off the deep end… the phonology for my Wēchizaŋkəŋ, for instance, is ~2.5 pages, and I’m sure it’s not nearly detailed enough. (No allophony, for one thing.)

Nortaneous wrote: ↑Sun Aug 22, 2021 2:38 pm the hard part is that phonology interacts with other aspects of the language. not just morphology, but also root structure constraints distinct from word structure constraints, phonological operations that morphology can exploit (e.g. word-level backing in kusunda, short and long forms in mandarin), syllable structures only licit in derived forms (-ŋθs in english) … even a simple-looking language like gilbertese can turn out to do very bizarre things to enforce a very bizarre trimoraic foot structure

Yep, I’ve heard of most of these to some extent.

complex phonotactic restrictions with morphological implications (causative formation in shangzhai horpa)

Sounds odd, but pretty sure you posted a link at some point.

even phonosyntax (burmese developing tonal word class distinctions from phrase-final creaky voice)

Sorry, what‽

(BTW, if Nortaneous could do a tutorial on phonology, I shan’t complain.)

Also, practically every single natural language has exceptions hidden in its phonology. A good grammar makes those explicit. I'd love to see conlang grammars highlight those exceptions more. Some examples:

- The allophonic behavior of a phoneme is different than what you'd expect based on phonemes in the same POA/MOA.
- Loaned words might not be contrained by the same phonotactical rules.
- Words in specific word classes may also operate under other contraints. For instance, think about the the phonology of interjections, expressives or toponyms.
- Exceptions to stress patterns and how they may be explained.

Nortaneous wrote: ↑Sun Aug 22, 2021 5:25 pm clearly a typo for a > b / #_C, caused by the collapse of certain pretonic vowels to schwa, followed by the merger of ə > u, reduction of word-initial vowels to semivowels as in Tocharian B, and late fortition

Well done
*tips phrygian cap*

Most grammars of Austroasiatic begin with a section called "word formation," which is mostly phonology. Major syllables and minor syllables have totally different sets of rules, which are different from disyllabic words with two major syllables, and then infixes can happen on top of that that behave differently in minor syllables.

What I'm saying is, a bad phonology is just a chart, but that doesn't mean that a good phonology is a chart followed by more charts, showing allophones or whatnot. Phonology is a whole-language system. If a language is a symphony, phonology is vibrations in air. My own conlang reference grammars always either put phonology at the end or (more frequently) broken up throughout the grammar as it becomes relevant.

zompist wrote: ↑Sun Aug 22, 2021 5:12 pm

I wrote: ↑Sun Aug 22, 2021 12:12 pm - /a/ becomes /b/ before /c/...

That's a pretty striking fortition. I'm guessing you first get a > u, then u > w and w > b. The conditioning environment is odd, but perhaps /c/ first becomes a labiovelar, which motivates the a > u raising?

(I'm guessing that's not what you meant, but I found it amusing.)

Let's just say it's not quite what I meant

Another thing which is probably relevant to phonology is the relative amount of information which is packed into each syllable. Compare, for example, Goidelic, Germanic, and Finnish, in which initial syllables make many more distinctions than following syllables, with Spanish or Italian, where there are fewer distinctions in total but they can appear to much the same degree in more syllables.

Ketsuban wrote: ↑Sun Aug 22, 2021 6:00 am I think the fault lies with the LCK chapter "Sounds".

zompist wrote: ↑Sun Aug 22, 2021 6:43 am It seems like you think the LCK is too simple; I would remind you that it is written for beginners. You know lots more now; good for you.

Moose-tache wrote: ↑Sun Aug 22, 2021 9:04 am You realize some of us have been conlanging since many years before the LCK existed?

Zompist is not responsible for the quality of the conlanging hobby as a whole. Saying the LCK is at fault is a bit absurd. As Zompist mentioned, it's targeted towards beginners and does a good job of introducing linguistics to them. I referred to it when starting out, and I know many others also did, so I would suggest it does its job quite well. There is also the Advanced Language Construction Kit...

---

The issue is multifaceted, as I see it.

1) Linguistics is a MASSIVE topic.
There's tons & tons of linguistics - or even just phonetics - information to absorb to make a naturalistic conlang. At best it's difficult to put that all in 1 book, and more realistically it's impossible. To learn all of that, something like the LCK is just a starting point, and then very quickly you get into Wikipedia articles, grammars, academic articles, etc. It's spread out all over the place, and often there's conflicting information, so it's hard for people to get up to speed on all of that, if they ever do.


2) Conlanging needs are different because it's applied.
Most of the information out there describes various languages and/or language features, but (understandably) very little discusses how to apply that knowledge in conlanging. I think very often it's understanding how all of those systems interact, or even just understanding what options exist in real-world languages. Typology papers and Wikipedia articles probably come as close as anything. The papers I've seen by McGregor on Ergativity and Mithrun on Polysynthesis are good examples since they explain the phenomena, break it down into a framework of types that is easy to understand and apply, and give lots of examples from various languages where the features show up. Useful typologies like that don't seem particularly common, unfortunately. Most conlangers have to piece bits of information from here & there together to get a good idea of what's possible for various aspects of their languages.


3) Instructional design can be an issue.
Actual tutorials on conlanging are great since they at least try to address the points above. People that write them need to have a good understanding of the topic, and usually do. Sometimes the information in said tutorials is awesome, but it's not conveyed in a manner that makes it easy to digest & apply. They write it in a manner that makes sense to them, but might not to others, especially those with less linguistics/conlanging experience. Often it's an information dump. Still, writings like this are FAR better than having nothing at all, so the point here is not to disparage such attempts.


Bottom line is there's lots to learn, and it's not easy to assemble the knowledge needed. The point of conlanging (usually?) is not to learn linguistics, but that learning is often needed to pursue the activity. I'd guess most of us don't need or care to dive into the academic linguistic weeds; we just want to learn what's needed as efficiently as possible so we can move on to creating a language, which is the point.

Focusing on phonology, I've been conlanging for 11+ years, and my phonologies are finally at the point where they aren't totally noobish, but they certainly aren't great, either. I could still benefit from more information that organizes phonological topics. Take sound change, for example. I've seen lists of sound changes, but most of those are specific to one language or a closely related family. There's the Index Diachronica, which lists changes by phoneme and shows a list of languages where this has happened. However, what are the most common changes or very rare? Is there a difference diachronically vs synchronically? It's hard to tell, and so where do you begin when doing sound changes? I guess most people just list a bunch of changes, post them for feedback, and then adjust until no one "complains".

bradrn wrote: ↑Sun Aug 22, 2021 8:06 am I think one of the problems with phonology is that there really are very few patterns. Stuff like allophony, sandhi and sound change don’t have any particularly clear limitations; there are patterns, but only vague ones, and the only real way to learn about them is to read a lot about sound change.

One prominent exception to this: stress patterns. These actually do have clear patterns and theory behind them, but I think that not a lot of conlangers are aware of how to utilise these.

Yep, exactly. Knowing where to find it, getting access, and understanding it from a broad perspective are hurdles. The more the information can be organized so it's easier to locate, the easier conlangers will have in making use of that information in their languages.

Nortaneous wrote: ↑Sun Aug 22, 2021 2:38 pm the hard part is that phonology interacts with other aspects of the language. not just morphology, but also root structure constraints distinct from word structure constraints, phonological operations that morphology can exploit (e.g. word-level backing in kusunda, short and long forms in mandarin), syllable structures only licit in derived forms (-ŋθs in english), complex phonotactic restrictions with morphological implications (causative formation in shangzhai horpa), even phonosyntax (burmese developing tonal word class distinctions from phrase-final creaky voice). even a simple-looking language like gilbertese can turn out to do very bizarre things to enforce a very bizarre trimoraic foot structure

This one comment shows how much info there is to be familiar with in order to make a really high-quality phonology. Nort, you probably have the best and broadest view of anyone on the board of what real languages do phonologically. Most conlangers don't even have a particularly long list of examples, much less have the capacity to master them like an artist would with paint brushes. I'd include myself in that category.

As someone whose phonologies so far consist of phoneme inventory + syllable structure + maybe some allophony, I would definitely be interested if anyone who knows more about the topic posted some tutorials on this board. For example, I've never heard anything about stress patterns having any deeper theory behind them--does anyone have a link to an article or something about that?

Vardelm wrote: ↑Mon Aug 23, 2021 1:14 pm Zompist is not responsible for the quality of the conlanging hobby as a whole. Saying the LCK is at fault is a bit absurd. As Zompist mentioned, it's targeted towards beginners and does a good job of introducing linguistics to them. I referred to it when starting out, and I know many others also did, so I would suggest it does its job quite well. There is also the Advanced Language Construction Kit...

Yeah, I overstepped the mark and came off a bit too accusatory there, and I'm sorry about that.

I really think that letting people develop their understanding of phonology by doing might be the best approach.

alice wrote: ↑Mon Aug 23, 2021 6:21 am Another thing which is probably relevant to phonology is the relative amount of information which is packed into each syllable. Compare, for example, Goidelic, Germanic, and Finnish, in which initial syllables make many more distinctions than following syllables, with Spanish or Italian, where there are fewer distinctions in total but they can appear to much the same degree in more syllables.

yes or Taa where you have >100 initial consonants (some of which are probably clusters) and around 5 medial consonants

bradrn wrote: ↑Sun Aug 22, 2021 8:01 pm

complex phonotactic restrictions with morphological implications (causative formation in shangzhai horpa)

Sounds odd, but pretty sure you posted a link at some point.

it's quoted in OP

even phonosyntax (burmese developing tonal word class distinctions from phrase-final creaky voice)

Sorry, what‽

I'll see if I can find this

(BTW, if Nortaneous could do a tutorial on phonology, I shan’t complain.)

my keyboard's pretty broken atm but maybe in a few months once I have a new laptop. this will also give me time to procrastinate on learning autosegmental theory

Vardelm wrote: ↑Mon Aug 23, 2021 1:14 pm 1) Linguistics is a MASSIVE topic.
There's tons & tons of linguistics - or even just phonetics - information to absorb to make a naturalistic conlang. At best it's difficult to put that all in 1 book, and more realistically it's impossible. To learn all of that, something like the LCK is just a starting point, and then very quickly you get into Wikipedia articles, grammars, academic articles, etc. It's spread out all over the place, and often there's conflicting information, so it's hard for people to get up to speed on all of that, if they ever do.

My favourite linguistics quote: ‘Linguistics will become a science when linguists begin standing on one another's shoulders instead of on one another's toes’ (Anderson 1992). Indeed, most of the papers I read are just linguists doing some serious toe-standing (including Anderson 1992). Actually useful information is few and far between.

2) Conlanging needs are different because it's applied.
Most of the information out there describes various languages and/or language features, but (understandably) very little discusses how to apply that knowledge in conlanging. I think very often it's understanding how all of those systems interact, or even just understanding what options exist in real-world languages. Typology papers and Wikipedia articles probably come as close as anything. The papers I've seen by McGregor on Ergativity and Mithrun on Polysynthesis are good examples since they explain the phenomena, break it down into a framework of types that is easy to understand and apply, and give lots of examples from various languages where the features show up. Useful typologies like that don't seem particularly common, unfortunately. Most conlangers have to piece bits of information from here & there together to get a good idea of what's possible for various aspects of their languages.

One of the biggest problems is that, as I’ve mentioned before, Wikipedia is rubbish for most linguistic topics — it’s incomprehensible at best and simply wrong at worst. Luckily, there are some amazing typological books in the academic literature (Payne’s Describing Morphosyntax, Dixon’s Basic Linguistic Theory, ed. Shopen’s Language Typology and Syntactic Description) — from which I learnt most of what I know — but they tend to be paywalled and not widely advertised, so beginners don’t know about them.

3) Instructional design can be an issue.
Actual tutorials on conlanging are great since they at least try to address the points above. People that write them need to have a good understanding of the topic, and usually do. Sometimes the information in said tutorials is awesome, but it's not conveyed in a manner that makes it easy to digest & apply. They write it in a manner that makes sense to them, but might not to others, especially those with less linguistics/conlanging experience. Often it's an information dump. Still, writings like this are FAR better than having nothing at all, so the point here is not to disparage such attempts.

Now that you mention it, my ergativity tutorial probably falls into this category. (Unsurprisingly; I didn’t really intend it as a conlanging tutorial.) But I’ve been planning to write an addendum, so maybe I can try to write something about how to make an ergative conlang.

Focusing on phonology, I've been conlanging for 11+ years, and my phonologies are finally at the point where they aren't totally noobish, but they certainly aren't great, either. I could still benefit from more information that organizes phonological topics. Take sound change, for example. I've seen lists of sound changes, but most of those are specific to one language or a closely related family. There's the Index Diachronica, which lists changes by phoneme and shows a list of languages where this has happened. However, what are the most common changes or very rare? Is there a difference diachronically vs synchronically? It's hard to tell, and so where do you begin when doing sound changes? I guess most people just list a bunch of changes, post them for feedback, and then adjust until no one "complains".

Yes, I also really struggle with sound change. The ID helps somewhat, but it’s unreliable. (There was some speculation a while ago on making a new version; I hope that succeeds.) It sounds like the only real way to learn about sound change is to carefully go through paper after paper until you end up getting some sort of intuition for it. I’ve found some books with overviews of real sound changes (e.g. Blust’s The Austronesian Languages, thankfully open-access), which help, but there aren’t too many.

Nortaneous wrote: ↑Mon Aug 23, 2021 6:26 pm

bradrn wrote: ↑Sun Aug 22, 2021 8:01 pm

complex phonotactic restrictions with morphological implications (causative formation in shangzhai horpa)

Sounds odd, but pretty sure you posted a link at some point.

it's quoted in OP

Ah, so you did. (I’m pretty sure I added it to my reference manager at the time also.)

(BTW, if Nortaneous could do a tutorial on phonology, I shan’t complain.)

my keyboard's pretty broken atm but maybe in a few months once I have a new laptop. this will also give me time to procrastinate on learning autosegmental theory

Autosegmental theory itself isn’t at all difficult — I learnt it a while ago for Hlʉ̂, and there isn’t much to it. Optimality theory, on the other hand…