Conlang Random Thread

[Travis B.]

Currently working on the phonoaesthetics of my latest project. I am looking for ways to avoid too many harsh consonants in successive syllables, as this project includes not only clicks but aspirate and ejective consonants. One obvious approach is lenition with plosives becoming fricatives or approximants between vowels. However this process would clearly need some limitations to avoid the improbable situation of plosives rarely appearing within words, e.g. something like *tapakatapaka > tawağarawağa.

One way to do it is to take the metrical structure of words into account. For instance, if words are typically composed of trochees, you could only have lenition in intervocalic positions that fall between the first and second vowels of each trochee, giving, e.g. *tapakatapaka > tawakarapağa.

Also, if you have mobile stress, you can make it so that a syllable with stress is always the first syllable in a trochee, such that you can have effects like mobile stress changing which consonants undergo lenition.

[Nortaneous]

However this process would clearly need some limitations to avoid the improbable situation of plosives rarely appearing within words, e.g. something like *tapakatapaka > tawağarawağa.

‘Improbable’? Isn’t Spanish like that? (Amongst other languages.)

Spanish has voiceless plosives in word-medial position, and voiced plosives after nasals. Taking a random sample of Spanish text:

Por su textura, rica en materia seca, se presta para puré. También se consume sancochada con salsas, al horno, envuelta en papel aluminio; o en un plato típico de Perú, denominado causa a la limeña. En Colombia es conocida como papa criolla o amarilla; se come cocida, frita, en puré, asada y en sancochos, es el ingrediente principal y característico del ajiaco santafereño, plato típico de Bogotá.

However, it's possible for languages to barely have voiceless plosives at all:

After epenthetic and paragogic insertion of ɨ, the Akɨ dialect of Apalɨ lenited all intervocalic stops—that is to say, all non-word-initial stops. In combination with the word-initial lenition that took place in Apalɨ and Manat (§2.3.1.5), this change had the effect of voicing every voiceless stop in Akɨ. And indeed, Wade states that in Akɨ, voiceless stops “are so infrequent that they could have been imported into the phonology from another language” (1993: 79).

And it's common in Khoisan languages to only permit a very restricted set of consonants root-medially. This set may not contain plosives: in Nama, for example, the only root-medial consonants are w r m n. Root structure is generally overlooked in conlanging, but is an important part of the structure of natural languages.

[malloc]

And it's common in Khoisan languages to only permit a very restricted set of consonants root-medially. This set may not contain plosives: in Nama, for example, the only root-medial consonants are w r m n. Root structure is generally overlooked in conlanging, but is an important part of the structure of natural languages.

True, although the Khoisan languages are mostly isolating from what I understand, whereas my project is quite agglutinative. Nonetheless, point taken about root structure.

[bradrn]

However, it's possible for languages to barely have voiceless plosives at all:

After epenthetic and paragogic insertion of ɨ, the Akɨ dialect of Apalɨ lenited all intervocalic stops—that is to say, all non-word-initial stops. In combination with the word-initial lenition that took place in Apalɨ and Manat (§2.3.1.5), this change had the effect of voicing every voiceless stop in Akɨ. And indeed, Wade states that in Akɨ, voiceless stops “are so infrequent that they could have been imported into the phonology from another language” (1993: 79).

Oh yeah, right, forgot about Apalɨ. And I was looking at that thesis just the other day…

And it's common in Khoisan languages to only permit a very restricted set of consonants root-medially. This set may not contain plosives: in Nama, for example, the only root-medial consonants are w r m n. Root structure is generally overlooked in conlanging, but is an important part of the structure of natural languages.

I’m pleased to say that I at least try to consider root structure!

[keenir]

And it's common in Khoisan languages to only permit a very restricted set of consonants root-medially. This set may not contain plosives: in Nama, for example, the only root-medial consonants are w r m n. Root structure is generally overlooked in conlanging, but is an important part of the structure of natural languages.

True, although the Khoisan languages are mostly isolating from what I understand, whereas my project is quite agglutinative. Nonetheless, point taken about root structure.

Its perfectly legal and permitted, to use one feature of a language / language family, without using others...so its fine to ignore if its isolating.

[Ahzoh]

Are there any lists of "pre-agriculture" or "pre-metallurgy" vocabulary?

[Glass Half Baked]

Surely that is any Australian dictionary?

Also, these people have a similar project. The North American languages provide the best online resources (e.g. the online Shoshone dictionary is top notch), but you can find plenty of dictionaries from Australia and South America if you dig.

[Ahzoh]

I recently decided to simplify the paragogic "thematic" vowels of Vrkhazhian to not be based on any noun class system, but I have felt uneasy about how to best assign what nouns with what vowels.

What I have so far is to either do an echo vowel thing or base it on whether the penultimate vowel is open short or closed/open long

More: show

C₁aC₂- > C₁aC₂-a-
C₁iC₂- > C₁iC₂-a-
C₁uC₂- > C₁uC₂-a-
C₁āC₂- > C₁āC₂-i- or C₁āC₂-a-
C₁īC₂- > C₁īC₂-i- or C₁īC₂-a-
C₁ūC₂- > C₁ūC₂-i- or C₁ūC₂-a-

C₁VC₂aC₃- > C₁VC₂aC₃-a- or C₁VC₂aC₃-a-
C₁VC₂iC₃- > C₁VC₂iC₃-a- or C₁VC₂iC₃-i-
C₁VC₂uC₃- > C₁VC₂uC₃-a- or C₁VC₂uC₃-u-
C₁VC₂āC₃- > C₁VC₂āC₃-i-
C₁VC₂īC₃- > C₁VC₂īC₃-i-
C₁VC₂ūC₃- > C₁VC₂ūC₃-i-

C₁VC₂C₃aC₄- > C₁VC₂C₃aC₄-a- or C₁VC₂C₃aC₄-a-
C₁VC₂C₃iC₄- > C₁VC₂C₃iC₄-a- or C₁VC₂C₃iC₄-i-
C₁VC₂C₃uC₄- > C₁VC₂C₃uC₄-a- or C₁VC₂C₃uC₄-u-
C₁VC₂C₃āC₄- > C₁VC₂C₃āC₄-i-
C₁VC₂C₃īC₄- > C₁VC₂C₃īC₄-i-
C₁VC₂C₃ūC₄- > C₁VC₂C₃ūC₄-i-

C₁aC₂C₂- > C₁aC₂C₂-a-
C₁iC₂C₂- > C₁iC₂C₂-a-
C₁uC₂C₂- > C₁uC₂C₂-a-

C₁aC₂C₃- > C₁aC₂C₃-i-
C₁iC₂C₃- > C₁iC₂C₃-i-
C₁uC₂C₃- > C₁uC₂C₃-i-

[linguistcat]

While working on an alien species:

One of the major phonological contrasts in most languages is trilled versus untrilled, much like how many human languages differentiate voiced vs unvoiced, aspirated vs unaspirated, palatal vs velarized, etc.

[malloc]

One of the major phonological contrasts in most languages is trilled versus untrilled, much like how many human languages differentiate voiced vs unvoiced, aspirated vs unaspirated, palatal vs velarized, etc.

Cool, sounds like a nifty idea. Of course I have always liked trills and been a bit disappointed that only the alveolar trill is particularly common.

[DorotheaBrooke]

Anybody have a list of lexemes that's a bit longer than the Swadesh list? I know this isn't the optimal way to build lexicon since there are likely all sorts of polysemies, etc. but it's for a protolang so quantity is more important here than quantity.

[DorotheaBrooke]

Also, came up with a fun sound change for a language family I'm working on:

-/ŋ/ is deleted from onset clusters (except in onset clusters like /dŋ/ where it's in the more sonorous position), which makes sense given how common prohibiting onset velar nasals is. However, this rule crosses word boundaries, so words ending in vowels (or potentially also obstruents, but idk if I like that) cause the retention of the historic velar nasal since it gets syllabified as a coda.
-By analogy, speakers start inserting [ŋ] on words beginning with vowels after words ending in vowels; i.e. /ábdi/ "stone" but [rí ŋábdi] "four stones." Basically intrusive-r, but word initial.
-This could potentially go a few different ways depending on the descendants; I was thinking that one descendant could keep the pattern, another inserts [ŋ] before all initial vowels, and a third could insert [ŋ] only on words that are often preceded by vowels, like inanimate nouns (which can't serve as subjects) or verbs (in this SOV language).

[malloc]

My research into clicks has uncovered an unexpected complication. It seems that glottalized clicks are most commonly nasalized (and often voiceless as well) rather than ejective as I had assumed. Following this pattern in my own project would result in gross asymmetries, though, since the language otherwise lacks voiceless nasals while having ejective stops.

[keenir]

My research into clicks has uncovered an unexpected complication. It seems that glottalized clicks are most commonly nasalized (and often voiceless as well) rather than ejective as I had assumed. Following this pattern in my own project would result in gross asymmetries, though, since the language otherwise lacks voiceless nasals while having ejective stops.

Malloc, let me ask you: do you want clicks?

If Yes, then: what clicks do you want?

Now that you're answered that (PLEASE ANSWER THAT), use it and don't worry about asymetry.

[bradrn]

My research into clicks has uncovered an unexpected complication. It seems that glottalized clicks are most commonly nasalized (and often voiceless as well) rather than ejective as I had assumed. Following this pattern in my own project would result in gross asymmetries, though, since the language otherwise lacks voiceless nasals while having ejective stops.

It’s worth noting that click terminology is really really confusing (and confused). ‘Ejective clicks’ don’t actually exist, for one thing: the closest you can get is a click released into an ejective consonant. Similarly, the so-called ‘glottalised clicks’ are just ones where a glottal closure is held beyond the click release; they’re not ejective, implosive or creaky like other consonants called ‘glottalised’. When the glottal closure is incomplete you naturally get a voiceless nasal as the air leaks out through the nose.

What all this means is that click phonetics is hard to compare directly to non-click phonetics. I’m not sure there’s any language where the two inventories are entirely symmetric. For that matter, we don’t really have enough click languages to firmly identify any universals! So ultimately you should feel free to do what you want.

[jal]

-/ŋ/ is deleted from onset clusters (except in onset clusters like /dŋ/ where it's in the more sonorous position), which makes sense given how common prohibiting onset velar nasals is.

Onset velar nasals are relatively rare, but I'd think that clusters like the /dŋ/ don't occur as the onset (rather than as a full syllable with /d/ being the onset and /ŋ/ being the nucleus).

However, this rule crosses word boundaries, so words ending in vowels (...) cause the retention of the historic velar nasal since it gets syllabified as a coda.

"this rule crosses word boundaries" seems a bit imprecise. I get what you mean, but formulated like that it sounds like you're talking about a sandhi rule that extends to word clusters.

By analogy, speakers start inserting [ŋ] on words beginning with vowels after words ending in vowels; i.e. /ábdi/ "stone" but [rí ŋábdi] "four stones." Basically intrusive-r, but word initial.

This kinda contradicts what you wrote above. It can't be both retention and intrusive. Not saying the process itself is unlikely, but you need to better describe it imho.

This could potentially go a few different ways depending on the descendants; I was thinking that one descendant could keep the pattern, another inserts [ŋ] before all initial vowels, and a third could insert [ŋ] only on words that are often preceded by vowels, like inanimate nouns (which can't serve as subjects) or verbs (in this SOV language).

"Keeping the pattern" means the "intrusive-ŋ" right? So inter-word vowel collisions are solved by inserting a /ŋ/. "Insert /ŋ/ before all initial vowels" - that would only work if there are relatively little (I can't see a situation where 90% of words start with a /ŋ/), and I'm also not sure whether that would work given it starts off as "intrusive-ŋ". I can't easily see such an intrusive sound being recodified as a phoneme unless it's very uncommon (but based on what you write it's common). "inser /ŋ/ only on words preceeded by vowels" - same objections as the previous, once it's intrusive it's difficult to phonemize it. Otherwise this is similar to what happened in English with words like "adder" (from "nadder") and the famous French "licorne" (unicorn -> un icorn -> l'icorn -> licorn).

So I think you can save the three different patterns in daughter languages by not having the intrusive-ŋ phase in the parent language, but have some unstable system in the parent language and have the daughter languages develop the systems you described.


JAL

[DorotheaBrooke]

-/ŋ/ is deleted from onset clusters (except in onset clusters like /dŋ/ where it's in the more sonorous position), which makes sense given how common prohibiting onset velar nasals is.

Onset velar nasals are relatively rare, but I'd think that clusters like the /dŋ/ don't occur as the onset (rather than as a full syllable with /d/ being the onset and /ŋ/ being the nucleus).

To be clear, onset clusters was a typo on my part, and I generally worded that poorly; I meant just that an onset of /ŋ/ gets deleted. So for example, Protolang *ŋal *ŋwázi become [al] [wázi].

The proto-language (called Proto-Kuzitic) was pretty permissive in terms of clusters, so Stop+Nasal are totally permitted; I don't think I have an instance of *dŋV in my current corpus of words, but I have *ṭŋáz "eye," and *ḍmaʁk- "burn," in just the Swadesh list. A hypothetical form like *dŋu is totally licit, while syllabic nasals aren't permitted.

By analogy, speakers start inserting [ŋ] on words beginning with vowels after words ending in vowels; i.e. /ábdi/ "stone" but [rí ŋábdi] "four stones." Basically intrusive-r, but word initial.

This kinda contradicts what you wrote above. It can't be both retention and intrusive. Not saying the process itself is unlikely, but you need to better describe it imho.

I'm not sure if it's contradictory since I think this follows the pattern in some non-rhotic Englishes with intrusive r? I.e. historic rhotics are lost in coda position, but retained before a vowel. This retention then gets generalized as a method to prevent hiatus across morpheme boundaries. It's actually a bit stronger in this case cause Proto-Kuzitic doesn't permit hiatus within morphemes at all, unlike English.

Wrt: the plausibility of those changes, fair point; though iirc there was a language (Nenets?) which I vaguely remember inserting [ŋ] word-initially in the absence of an onset. Could be misremembering though.

[Travis B.]

-/ŋ/ is deleted from onset clusters (except in onset clusters like /dŋ/ where it's in the more sonorous position), which makes sense given how common prohibiting onset velar nasals is.

Onset velar nasals are relatively rare, but I'd think that clusters like the /dŋ/ don't occur as the onset (rather than as a full syllable with /d/ being the onset and /ŋ/ being the nucleus).

To be clear, onset clusters was a typo on my part, and I generally worded that poorly; I meant just that an onset of /ŋ/ gets deleted. So for example, Protolang *ŋal *ŋwázi become [al] [wázi].

The proto-language (called Proto-Kuzitic) was pretty permissive in terms of clusters, so Stop+Nasal are totally permitted; I don't think I have an instance of *dŋV in my current corpus of words, but I have *ṭŋáz "eye," and *ḍmaʁk- "burn," in just the Swadesh list. A hypothetical form like *dŋu is totally licit, while syllabic nasals aren't permitted.

By analogy, speakers start inserting [ŋ] on words beginning with vowels after words ending in vowels; i.e. /ábdi/ "stone" but [rí ŋábdi] "four stones." Basically intrusive-r, but word initial.

This kinda contradicts what you wrote above. It can't be both retention and intrusive. Not saying the process itself is unlikely, but you need to better describe it imho.

I'm not sure if it's contradictory since I think this follows the pattern in some non-rhotic Englishes with intrusive r? I.e. historic rhotics are lost in coda position, but retained before a vowel. This retention then gets generalized as a method to prevent hiatus across morpheme boundaries. It's actually a bit stronger in this case cause Proto-Kuzitic doesn't permit hiatus within morphemes at all, unlike English.

Wrt: the plausibility of those changes, fair point; though iirc there was a language (Nenets?) which I vaguely remember inserting [ŋ] word-initially in the absence of an onset. Could be misremembering though.

I would say that your changes here are perfectly plausible myself. As for English, if you've read some of the discussions on the forum you'd know that some people analyze their varieties of English (e.g. AusE) as not permitting hiatus at all in the first place (with all the historical long vowels other than PALM being analyzed as vowel-glide sequences, and PALM simply not being permitted in hiatus).

[Lērisama]

I would say that your changes here are perfectly plausible myself. As for English, if you've read some of the discussions on the forum you'd know that some people analyze their varieties of English (e.g. AusE) as not permitting hiatus at all in the first place (with all the historical long vowels other than PALM being analyzed as vowel-glide sequences, and PALM simply not being permitted in hiatus).

In both Australian and Southern British excepting some recessive Westcountry dialects Englishes, PALM=START, so if you are calling long vowels Vowel+glide, you also have to analyse PALM as /ɑr/, which I do, and I thing I've seen Darren do, so there really are no exceptions, except in traditional RP which is non-rhotic but preserves a non-diphthongal HAPPY that probably isn't a VC sequence.

On the preposed sound changes, a) it is Nenets, and b) yes that sounds perfectly plausible to me, although natively having intrusive r may make me a little biased.

[malloc]

It’s worth noting that click terminology is really really confusing (and confused). ‘Ejective clicks’ don’t actually exist, for one thing: the closest you can get is a click released into an ejective consonant. Similarly, the so-called ‘glottalised clicks’ are just ones where a glottal closure is held beyond the click release; they’re not ejective, implosive or creaky like other consonants called ‘glottalised’. When the glottal closure is incomplete you naturally get a voiceless nasal as the air leaks out through the nose.

Yes, the downside of taking an interest in rare linguistic features is having to wade through meager and often hazy resources. It occurs to me that this phenomenon could represent a case of rhinoglottophilia (which is much less kinky than it sounds).