Please Help in Designing a Consonant System

[HolyKnowing]

I am designing an a priori general purpose human language that is optimized for storing all human knowledge, but not in a way that is as robotic as Lojban.

I have a vowel system picked out, which was much easier to design. It is 22 vowels large, with 11 vowel qualities with distinction by length. It looks like a blend between Finnish and Korean.

i y ɯ u iː yː ɯː uː
e ø o eː øː oː
ʌ ʌː
ɛ ɛː
æ a æː aː

However, it is difficult to design a consonant system that is tasteful, because consonant inventories are more complex than vowel systems. I humbly request assistance in this endeavor. Some considerations.

Lojban has a nice feature where there are three shapes of words. One shape is for predicate-type words, which is CCVCV or CVCCV, another shape is for grammatical words, which is CV or CV"h"V. However, because there are more types than just these two, it will be necessary to design a consonant inventory to create more shapes of words. The semivowels /j/ and /w/ were reinterpreted as falling diphthongs.

Another concern is consonant clusters. The consonant system should allow for a rich system of consonant clusters that is spiritually speaking euphonious. But it is difficult to make this choice ahead of time.

Lastly, I do not want nationalists (e.g. Sanskrit nationalists) to say that I took from their language. So the consonant system should be non-political.

[bradrn]

I am designing an a priori general purpose human language that is optimized for storing all human knowledge, but not in a way that is as robotic as Lojban.

Just a warning: this board (including myself) tends to focus very strongly on naturalistic languages, rather than engelangs and auxlangs. The goal of ‘optimizing for storing all human knowledge’ is rather vague, and I (and probably most people here) would consider it probably unattainable.

Now, having got that proviso out the way:

Lojban has a nice feature where there are three shapes of words. One shape is for predicate-type words, which is CCVCV or CVCCV, another shape is for grammatical words, which is CV or CV"h"V. However, because there are more types than just these two, it will be necessary to design a consonant inventory to create more shapes of words. The semivowels /j/ and /w/ were reinterpreted as falling diphthongs.

I don’t see this as having much to do with the contents of your consonant inventory. It’s more a matter of phonotactics, in my opinion.

Another concern is consonant clusters. The consonant system should allow for a rich system of consonant clusters that is spiritually speaking euphonious. But it is difficult to make this choice ahead of time.

Lastly, I do not want nationalists (e.g. Sanskrit nationalists) to say that I took from their language. So the consonant system should be non-political.

These are rather vague requirements. Quite subjective, too: different people find different things euphonious. So, all I can do is to give the first random phoneme inventory that comes to mind and which might satisfy your requirements:

/p t c k ʔ/
/b d ɟ ɡ/
/f s ʃ x h/
/v z ʒ ɣ/
/m n ɲ ŋ/

For inspiration, I would also suggest looking at Sgai, which shares a similar goal, and which has a consonant inventory reminiscent of your criteria here.

[alice]

I'd add /l r/ and possibly /w j/ to those consonants; maybe replace the palatals with postalveolars, but that's a matter of phonetics.

[bradrn]

I'd add /l r/ and possibly /w j/ to those consonants; maybe replace the palatals with postalveolars, but that's a matter of phonetics.

They already said they’re treating /w j/ as vocalic.

[HolyKnowing]

These are rather vague requirements. Quite subjective, too: different people find different things euphonious. So, all I can do is to give the first random phoneme inventory that comes to mind and which might satisfy your requirements:

/p t c k ʔ/
/b d ɟ ɡ/
/f s ʃ x h/
/v z ʒ ɣ/
/m n ɲ ŋ/

For inspiration, I would also suggest looking at Sgai, which shares a similar goal, and which has a consonant inventory reminiscent of your criteria here.

Thank you for your splendid support!

(For future reference, my preferred pronouns are he/him/his.)

Because you have done this for a long time, your choice of inventory comes from a place of deep intuition, and therefore is extremely valuable.

Some follow-up questions:

1. What do you think of the consonants /θ ð/? Even though they're rare, a linguistics B.S. in Germanic languages told me that they're good consonants that became rare because they're soft and therefore hard to hear in noisy environments. Same reason why /h/ disappeared in the Romance languages, he elaborated.

2. A design question that I've pondered over is how to arrange the stops. Should they have voicing-only contrasts like most languages (e.g. /p, b/), a three-way contrast like Ancient Greek (/p, b, pʰ/) or Korean (/t, t͈, tʰ/) or four-way contrast like Sanskrit (/p, b, pʰ, bʱ/)?

3. What do you think about including the very common affricates /t͡s, t͡ʃ, d͡ʒ/? Will that make designing the phonotactics simpler or more complex?

[bradrn]

Because you have done this for a long time, your choice of inventory comes from a place of deep intuition, and therefore is extremely valuable.

Hah, you overrate me. It’s not anything special.

1. What do you think of the consonants /θ ð/? Even though they're rare, a linguistics B.S. in Germanic languages told me that they're good consonants that became rare because they're soft and therefore hard to hear in noisy environments. Same reason why /h/ disappeared in the Romance languages, he elaborated.

I think… well, they’re consonants. Fairly rare ones, but that shouldn’t stop you from using them if you want to.

As for the claims of that ‘linguistic BS’, an acronym has rarely been more accurate. First off, I don’t see consonants as being ‘good’ or ‘bad’. They’re just consonants. In a conlang, they might fit with your aesthetic preferences, or they might not. But no set of consonants is objectively ‘worse’ than another.

But also, /θ ð/ are not rare because they’re soft — there are many consonants which are softer (or less ‘sonorous’, to use the technical term), but more common. Rather, they’re rare because they require slightly more care to pronounce, which makes them historically unstable. In natural languages, they have a tendency to turn into /f v/ or /t d/ or /s z/, which are all much easier to pronounce.

2. A design question that I've pondered over is how to arrange the stops. Should they have voicing-only contrasts like most languages (e.g. /p, b/), a three-way contrast like Ancient Greek (/p, b, pʰ/) or Korean (/t, t͈, tʰ/) or four-way contrast like Sanskrit (/p, b, pʰ, bʱ/)?

Again: it’s up to your aesthetic preferences. Decide which one you like the best, and use it.

However, I will note that Sanskrit’s inventory is quite rare, and a Korean-style inventory more often uses ejectives: /t tʼ tʰ/, or more often /d tʼ tʰ/. There’s others which you haven’t mentioned, too — e.g. Hausa has /t (t)sʼ d ɗ/, Sindhi has /t d tʰ dʰ ɗ/, and so on.

3. What do you think about including the very common affricates /t͡s, t͡ʃ, d͡ʒ/? Will that make designing the phonotactics simpler or more complex?

Yet again, I think it’s something you can do if you want, and not do if you don’t want.

[Travis B.]

/p t k/
/b d g/
/h/

This with suprasegmental nasalization, where nasalization starts at any given vowel and regressively assimilates to /b d g/ and any other vowels, such that /b d g/ my be realized as either [b d ɡ~ɣ] or [m n ŋ], and is blocked by /p t k h/. Also, /p/ is realized as [p~f].

[foxcatdog]

Probably best is the following
*p *t *k
*s *h
*w *r/l *j
*m *n
CV syllable structure

But it probably works best with a smaller vowel system

[Darren]

/p t k/
/b d g/
/h/

This with suprasegmental nasalization, where nasalization starts at any given vowel and regressively assimilates to /b d g/ and any other vowels, such that /b d g/ my be realized as either [b d ɡ~ɣ] or [m n ŋ], and is blocked by /p t k h/. Also, /p/ is realized as [p~f].

A good start, but too many consonants. How about:

/k/
/ɣ/
/m n ŋ/

With progressive rhoticity triggered by vowels turning /ɣ m n ŋ/ into [ʀ mbʙ ndr ɴɢʀ], and blocked by /k/.

[Travis B.]

/p t k/
/b d g/
/h/

This with suprasegmental nasalization, where nasalization starts at any given vowel and regressively assimilates to /b d g/ and any other vowels, such that /b d g/ my be realized as either [b d ɡ~ɣ] or [m n ŋ], and is blocked by /p t k h/. Also, /p/ is realized as [p~f].

A good start, but too many consonants. How about:

/k/
/ɣ/
/m n ŋ/

With progressive rhoticity triggered by vowels turning /ɣ m n ŋ/ into [ʀ mbʙ ndr ɴɢʀ], and blocked by /k/.

The thing is that there are natlangs with phonologies that are similar to mine, while yours is patently absurd - lol.

[HolyKnowing]

However, I will note that Sanskrit’s inventory is quite rare, and a Korean-style inventory more often uses ejectives: /t tʼ tʰ/, or more often /d tʼ tʰ/. There’s others which you haven’t mentioned, too — e.g. Hausa has /t (t)sʼ d ɗ/, Sindhi has /t d tʰ dʰ ɗ/, and so on.

1. Why is Sanskrit's inventory rare?
2. What do you think about the inventory /d tʼ tʰ ɗ/ in relation to /d tʼ tʰ/?
3. Is pronouncing implosives like /ɗ/ challenging?

[Darren]

/p t k/
/b d g/
/h/

This with suprasegmental nasalization, where nasalization starts at any given vowel and regressively assimilates to /b d g/ and any other vowels, such that /b d g/ my be realized as either [b d ɡ~ɣ] or [m n ŋ], and is blocked by /p t k h/. Also, /p/ is realized as [p~f].

A good start, but too many consonants. How about:

/k/
/ɣ/
/m n ŋ/

With progressive rhoticity triggered by vowels turning /ɣ m n ŋ/ into [ʀ mbʙ ndr ɴɢʀ], and blocked by /k/.

The thing is that there are natlangs with phonologies that are similar to mine, while yours is patently absurd - lol.

Well, I'd argue that there are natlangs with patently absurd phonologies too

[bradrn]

However, I will note that Sanskrit’s inventory is quite rare, and a Korean-style inventory more often uses ejectives: /t tʼ tʰ/, or more often /d tʼ tʰ/. There’s others which you haven’t mentioned, too — e.g. Hausa has /t (t)sʼ d ɗ/, Sindhi has /t d tʰ dʰ ɗ/, and so on.

1. Why is Sanskrit's inventory rare?

Having an aspiration contrast on voiced consonants is very uncommon. That being said, Sanskrit’s so-called ‘voiced aspirates’ are probably better termed breathy-voiced stops — which are attested elsewhere, though still not commonly.

2. What do you think about the inventory /d tʼ tʰ ɗ/ in relation to /d tʼ tʰ/?

Combining implosives and ejectives at the same place of articulation is rare, but attested from numerous different parts of the globe (e.g. in Yucatec, Zulu, Dahalo, Lushootseed, Mazahua).

3. Is pronouncing implosives like /ɗ/ challenging?

I don’t find them too hard, but they take a bit of practice.

The thing is that there are natlangs with phonologies that are similar to mine, while yours is patently absurd - lol.

You should have a look at his scratchpad. Which honestly has been pretty interesting, actually!

[äreo]

I think the following gives you a good wide range without being unwieldy from the point of view of speakers of the most commonly spoken languages in the world:

/p b t d k g/
/f v s z ʃ ʒ h/
/m n l r/

Perhaps allow medial geminates (limiting this to sonorants, voiceless stops, and voiceless fricatives besides /h/ may be a good idea but do what you like).

I find /θ ð ʍ/ to be especially beautiful and would consider adding them as well.

[HolyKnowing]

The consonant and vowel inventories of the conlang are the most important part, because literally the rest of the language will be based off of it. Therefore it is very important to get it exactly right and to have as many eyeballs as possible looking at them.

I am struggling with choosing between the following two inventories.

m pʰ b p' f v
n tʰ d t' s z
ɲ cʰ ɟ c' ʃ ʒ
ŋ kʰ g k' x ɣ
ʔ h
r l
θ ð

m pʰ b p' ɓ f v
n tʰ d t' ɗ s z
ɲ cʰ ɟ c' ʄ ʃ ʒ
ŋ kʰ g k' ɠ x ɣ
ʔ h
r l
θ ð

The difference between the two is basically whether the stop sequence should include implosives or not. One more thing:

I think the following gives you a good wide range without being unwieldy from the point of view of speakers of the most commonly spoken languages in the world.

Don't care about the learner's feelings. Why? Because learning the sounds of a language is literally a one-time investment. You learn them once and they're stuck with you forever. So what you ought to be looking for is not how easy it is for a newbie to learn them (who might have come from Arabic, English, Chinese, etc... and so literally nothing can be known beforehand), but rather once the newbie learns them, can he grow into them? So exotic or unusual consonants are fine and dandy, provided that they don't become a nuisance once he's been speaking them for 10, 20, etc... years.

[bradrn]

The consonant and vowel inventories of the conlang are the most important part, because literally the rest of the language will be based off of it. Therefore it is very important to get it exactly right and to have as many eyeballs as possible looking at them.

I tend to disagree with this perspective. In my experience, it’s not hugely difficult to change the phoneme inventory if you dislike your existing choice. When beginning a new language, I often tend to switch between phonology, morphology and syntax, working on all three until the aesthetic settles into something I like.

[HolyKnowing]

I tend to disagree with this perspective. In my experience, it’s not hugely difficult to change the phoneme inventory if you dislike your existing choice. When beginning a new language, I often tend to switch between phonology, morphology and syntax, working on all three until the aesthetic settles into something I like.

I like verbs in the perfect tense!

[bradrn]

I tend to disagree with this perspective. In my experience, it’s not hugely difficult to change the phoneme inventory if you dislike your existing choice. When beginning a new language, I often tend to switch between phonology, morphology and syntax, working on all three until the aesthetic settles into something I like.

I like verbs in the perfect tense!

(To be pedantic, the ‘perfect’ is more of an aspect than a tense…)

But that sort of misses my point. I suggest that you stop taking specific things in isolation (like the consonant inventory or the verbal tenses). Instead, you could try making a first outline of the whole language: sketch out some key parts of the phonology, word classes, noun system, verbal system, basic word order, etc. Then try making some words and constructing some sample sentences. That should make it a bit clearer what works and doesn’t work.