(Conlangs) Q&A Thread - Quick questions go here

A forum for all topics related to constructed languages
Davush
greek
greek
Posts: 672
Joined: 10 Jan 2015 14:10

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Davush »

Salmoneus wrote: 12 Dec 2020 21:19
Thank you for the informative reply! That helps a lot.
Salmoneus
MVP
MVP
Posts: 3033
Joined: 19 Sep 2011 19:37

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Salmoneus »

I'm glad. This issue actually leads into some interesting and (by conlangers) I think underexplored areas!
Axthieb
rupestrian
rupestrian
Posts: 10
Joined: 08 Dec 2020 22:25
Location: Austria

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Axthieb »

Hi, I've spent a lot of time working on my conlang's vocabulary lately. Formerly I had used a method that derived 6 rather specific words from one usually very general root word. Eventually I realized, that, with this method, there was going to be a lot of roots with meanings that would have to share, at least in part, meanings that already exist. I also knew I could increase the number of derived words and spin a bigger etymological net.
My solution was a pattern by which I created a further 15 words, which were usually based on the words from which they branched off of. Sometimes one of those farther words would be derived from the root itself, or in other systematically inconsistent ways. That usually happened when it either seemed to suit the endeavor better or when I saw no other way.

Image

My question now is whether this is actually a good way to get words that seem believably real, and not forcedly artificial.

I will give an example:

[zɛn] movement, course, outcome
[zɛn.cʰyɸ] milipede, centipede, isopod
[zɛncʰ.d͜ða] limb, body part
[nɛx.zɛn] monster, monstrosity
[çy.zɛn] scurry, drift

[zɛnx] (n.) liquid
[zɛn.ʍɔx] honey
[d͜ðe.zɛnx] unstable
[zɛnx.d͜ða] hydraulic
[d͜ðe.zɛnx.d͜ða] pump

[zɛnˈza] later, afterwards, (arch.) next
[ˌzɛn.zaˈçyːx] delayed, unexpected
[ˈzɹɛɹnˌsa] unreliable

[sɹɛsˌɛnˈzaç] to let something rest, give a break
[ɹɮɛn.zɛn] wheat
[zɛnɮ] (coll.) wheat, flour
[ɹɮɛn.zɛn.ʍɔx] mart, market, stall, booth
[ɹɮɛns.ʍɔx] (coll.) mart, market, stall, booth

[zɛn.cʰe] pursuue, listen
[zɛnˈcʰex] ask
[iθ.sɹɛ.zɛn.cʰe] (impolite) (v.) gossip, slack, babble time away
[ˌθɹɛˌsɛnˈcʰe] (rare)(impolite) (v.) gossip, slack, babble time away
[iθ.sɛn.cʰe] (coll.)(impolite) (v.) gossip, slack, babble time away

[zɛn.d͜ðiɮ] wagon, cart
[zɛn.d͜ðiɮ.d͜ða] boat
[zɛn.d͜ði.ɮɛx] river, rails

The phonology itself is fine the way it is, so please don't mind it. The syllable structure is CCVCC where the pattern of allowed onset and coda clusters differ from eachother and are highly irregular. (Also V includes the 3 diphthongs)

Please tell me what you think of the process so far, and what you think of the derivations.
(edit: missing square brackets)
Last edited by Axthieb on 22 Dec 2020 07:32, edited 1 time in total.
User avatar
Creyeditor
MVP
MVP
Posts: 5091
Joined: 14 Aug 2012 19:32

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Creyeditor »

This looks like a good idea for complex derivational morphology. Maybe the next step could be a list of the prefixes and suffixes used (e.g. /d͜ðe/-), each with a list of more specific meanings. Instead of action, it could be a list that includes for a given root X, e.g. 'to create X', 'to become X', 'to turn something into an X'. You could look at natlangs for inspiration at wiktionary.
Oh and having overlapping meaning is actually pretty realistic.
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Axthieb
rupestrian
rupestrian
Posts: 10
Joined: 08 Dec 2020 22:25
Location: Austria

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Axthieb »

maybe i was too quick in my earlier post, but I already do have a list of derivational pre and suffixes.
d͜ðes is a prefix that means "whatever is being put effort in for" of which I'm sometimes not sure what i was thinking when wrote that. On occasion I mix adjective and verb suffixes. My other affixes are similarly specific.I guess that on second thought the whole thing seems fine. I believe my doubts were basically feelings of inadequacy.
User avatar
eldin raigmore
korean
korean
Posts: 6353
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by eldin raigmore »

“Axthieb” wrote:.... My solution was a pattern by which I created a further 15 words, which were usually based on the words from which they branched off of. Sometimes one of those farther words would be derived from the root itself, or in other systematically inconsistent ways. That usually happened when it either seemed to suit the endeavor better or when I saw no other way.
....
My question now is whether this is actually a good way to get words that seem believably real, and not forcedly artificial....
I don’t know, but I love it so far!
Salmoneus
MVP
MVP
Posts: 3033
Joined: 19 Sep 2011 19:37

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Salmoneus »

The system is not, in my opinion, very realistic - for human speakers, at least.

Unrealistic elements in my opinion include:

- vague, conceptual roots. In real languages, 'roots' are almost just ordinary words in the protolanguage. Therefore, they should have the same sort of meanings as ordinary words. The idea of the conceptual 'root' seems to be a confusion due to the efforts of early philologists working with PIE: in effect, because it's often hard to work out what the actual meaning of a PIE word was, they're given vague labels - we may know a certain word probably meant something to do with so-and-so, but we don't know exactly what it meant. In particular, lots of PIE words themselves seem like they may be derivations from more basic roots (I'm thinking here of 'extended' roots, where two roots differ only in the final consonant and seem semantically related or even interchangeable, so we guess that the consonant was once an affix), but we have absolutely no idea what significance the derivational affixes have. So those primordial roots may have to be described in very vague terms. But there's two things to note about this: first, that most of this vagueness/abstractness is the result of dealing not with a living language, the evidence for which comes only from later descendents - in most real language, you don't get 'roots' with anything like the vagueness you get in descriptions of PIE; and second, that this is abstraction in later analysis, not abstraction in origin. You don't start out with abstract meanings, and then derive from them to get concrete meanings. Instead, you start with ordinary words, derive ordinary words, and THEN, retrospectively, notice a pattern that suggests an abstract meaning. And often the 'roots' you notice will actually not have a historical basis anyway. [imagine English with a future simplified spelling. People would be saying "well, clearly 'terer' causes fear (-er is an agent suffix), and a 'terarium' must be a 'place of fear' because of its horrible frightening inhabitants (in the same way that a 'solarium' is a 'place of loneliness'), and a 'terasor' is some kind of frightening monster (we don't know what -sor means but it's also in 'dynasor' so it must be something really bad!)"...
[sometimes, when people think there is a root, whether that's a real etymological connection or just a folk etymology, the abstract meaning they imagine for that root can further shape the nuanced meaning of words sharing that root... but this is very much a secondary phenomenon]

- roots that don't overlap in meaning. In real language, there are lots of synonyms. Since roots are just (or were just) words, roots will also have lots of synonyms. LOTS of them.

- lots of derivational affixes. Having a few derivational affixes is something almost languages do. Having four or five or six is not uncommon. But having fifteen fully productive derivational affixes is unusual. English has a huge number of affixes - but largely because we've taken all the productive affixes of Latin (which already had an unusually large number of affixes) and added them to all our (many) inherited affixes and a fair few from other languages as well. But English's promiscuity is unusual. And when there's lots of derivational affixes, it's unusual for all of them to be fully productive - you'd expect many to be old and no longer productive, or only productive for small subsets of words. I'm certainly not saying having 15 derivational affixes is unrealistic, it's not. But having 15 fully productive derivational affixes is unusual, I'd say.

[wiktionary lists 62 derivational affixes in Proto-Germanic. However, this is misleading - many are simply gender or conjugation variants of one another, or are transparent compounds of one another, or only appear in one branch of the family, or were already no longer productive and only appear in a handful of words]

- full derivational paradigms. Just because you have derivational affixes doesn't mean you need to use them. In real languages, few if any 'roots' will be found in all possible derived forms - most will only appear in a small fraction of the possible forms.

- distinct derivational affixes. The reason some languages have many derivational affixes isn't usually due to precise distinctions of meaning, but simply duplication. Actual derivational affixes are usually very vague in sense, and unpredictable (that's why they're derivation, not inflexion!). So, say you want to make a noun into a verb in Proto-Germanic. Wiktionary says you have 9 suffixes to choose from! But that doesn't mean you can derive 9 verbs with different meanings. Not at all! Most derived verbs will simply have either -jana or -o:na. There's no difference in meaning between these two - mostly it's just that the former is older and more irregular than the latter. Some roots appear with either - usually the -jana verb is older, and the -o:na verb is a more recent and hence more transparent derivation from a noun or adjective. It's possible that -o:na is actually from -ojana or the like originally - i.e. it may be the same affix originally, but one version has reanalysed some material from the end of the root as part of the affix (which is very common). From -o:na, we get -iso:na, which is even more transparently denominative, but it's also very rare and not productive in any daughter language. We also get -ino:na, which is again rarely found; it sort of has a meaning, in that it's often semi-productively associated with jobs and duties, but not very firmly. There's -lo:na, which tends to form verbs of continuing action, or diminutives (drip>dribble), but it's indistinguishable from -ro:na (wend>wander, spit>spatter). -itjana and -atjana have the same meaning - both vaguely intensive - and are just applied to different conjugations. And so on.
Likewise, think of English: we have -ise, -ate and -ify, and compounds of these, and compounds of these with -ic, but they don't really have clear and distinct functions. You couldn't, from dictionary meanings of roots and affixes, work out what the difference in meaning between 'placate' and 'placify' is.

- general regularity. The unifying theme in these points is: your system looks like a system. Real languages are rarely systematic. Words are derived in haphazard ways as they are needed - they're not designed to fit into a coherent, efficient system. If, say, we have the word 'plastic', and want to derive a verb from it - in English, across time and space, different people will try different things. Some will say 'plasticate', some will say 'plasticise', some will say 'plasticify', some will say 'plastic' (zero-derivation), some will say 'plastify' (taking the -ic as a derivative affix and choosing derive their new word directly from the root), some will say 'plasticitise' (prefering to first derive an abstract noun, plasticity, and then derive from that). A bunch of words will arise, and be broadly synonymous, and then over time some of them will develop a more specialised function - sometimes in ways vaguely predictable from the affix, but often not. So in English we do in fact have the words 'plasticise', 'plastify' and 'plasticate', and they all mean almost exactly the same thing, with only small differences in nuance - 'plasticate' usually means only making fully plastic, 'plasticise' often implies the use of a plasticising agent, while 'plastify' is more generic.

Some real languages are unusually systematic in their derivation. But I'm not sure any real language is THIS regular.



Of course, none of these issues are relevant is your language isn't intendeded to be naturalistic.
User avatar
Pabappa
greek
greek
Posts: 577
Joined: 18 Nov 2017 02:41

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Pabappa »

One of my languages has a large number of noun classes and has mostly lost the ability to use its noun roots in bare form .... that is, apart from a few very basic concepts, all nouns must have a noun class prefix, and the same root can appear with a wide variety of noun classes. Becuase of radical sound changes, in some cases originally distinct roots have merged and come to be associated semantically. This means the roots have very wide semantic ranges, which even native speakers would have difficulty pinning down.

For example I have huuku "to recognize, remember" and tiuku "blossom, flower bloom" ... because we wait for the flowers to blossom every year and we remember what they look like before they appear. But then an unrelated root meaning "thirsty" merged with this, creating a new verb giuku, and flowers are also thirsty, so the semantics makes sense with this meaning as well. So what does uku by itself mean? Native speakers would have a difficult time answering this, because it never occurs by itself, and the range of meaning covered by its derivations is very broad.

My language has a whole lot of empty space in its paradigms, .... e.g. for example it has a whole noun class just for turtles, which isn't used much, ... but even the more commonly used noun classes, such as i- "handheld object", have lots of gaps. I dont have a word *i-uku for example. However, if I wanted one, it would be quite easy to come up with a meaning. Glass or cup maybe? Those are thirsty objects. Or maybe it's a mirror, because mirrors remember what we look like. (The problem of having *too much choice* can be handled in various ways, so this doesnt present a problem for me or for you.)

So I think your system can and will work. How it could arise naturally I'm not sure. But I think having roots with highly abstract meanings and wide semantic ranges is perfectly sound.

If you decide to stay with this as it is, and then run into other problems .... for example if you have too many possible meanings for a single root+affix combination, you may want to look at having more than just this one method of deriving new words .... but I think this is a very good base to build on.


Also, it's worth asking .... isnt Semitic also known for having roots with highly abstract meanings, and wide semantic ranges of the words derived from those roots?
Kavunupupis, šiŋuputata.
When I see you pointing at me, I know I'm in trouble. (Play)
Axthieb
rupestrian
rupestrian
Posts: 10
Joined: 08 Dec 2020 22:25
Location: Austria

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Axthieb »

Well, first of all let me thank you for all your inputso far. I'd always wondered what kind of impression the insides of my conlang would make.

The first thing I want to make clear is, that while this is a system, I'm not always using the affixes in predictable ways and it is supposed to be open to derivations that are not obviously or conceivably related. That actually brings me to my next point.

I value the criticism, especially this honest kind. I've read through it a couple times an I've got the impression, that, back when I started working on this project and conlang, the word making was maybe more realistic. A small example would be the first root I set up: [θɛn] body, form | [θɛn.my] tree | where [my] is what I've kind of come to understand as a nonsense suffix, because I pulled it from thin air, and never gave it any other meanin than being the second syllable of the word for tree.
But having fifteen fully productive derivational affixes is unusual.
I'm not sure I unrstand what that means. What exactly does productive mean?
I guess I didn't sew the derivation in tightly enough, but if a productive affix is one that is accessible to speakers as opposed to obsolete and archaic except for set uses, then I guess I'm fine, because most suffixes aren't supposed to be accessible.

Actually, that my vocabulary has become so systematic is because I felt it would be more unreasonabe having some irregular scribble that sometimes merges the first consonant into a cluster, or making the vowel merge with the root's vowel if they were to form an allowed diphthong. This is assuming no rules or hierarchy of rules for this seemingly inconsistent use apply. In other words: Is it really more reasonable to use affixes, or rather consistent derivation, in unpredictable, inconsistent and possibly erratic ways?
I guess that would make sense as well, I'm just so damn unsure.

Also thank you very much for that bit about IE. I'd never really known how that worked.
For example I have huuku "to recognize, remember" and tiuku "blossom, flower bloom" ... because we wait for the flowers to blossom every year and we remember what they look like before they appear.
That is very simlar to how I think when I make up words.

Edit: I want to add that I occasionally combine two roots as a sort of pseudo compound.
Also I knew that duplicate meanings or synonyms are common, I just avoid that as a general rule to get more words of different meanings more quickly.
User avatar
eldin raigmore
korean
korean
Posts: 6353
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by eldin raigmore »

“Axthieb” wrote:
But having fifteen fully productive derivational affixes is unusual.
I'm not sure I unrstand what that means. What exactly does productive mean?
I guess I didn't sew the derivation in tightly enough, but if a productive affix is one that is accessible to speakers as opposed to obsolete and archaic except for set uses, then I guess I'm fine, because most suffixes aren't supposed to be accessible.
English has (about) 800 affixes (I have heard). Only 8 are inflexional; the others are all derivational.

The distinction between derivational and inflectional is a bit schizophrenic.
It means one of two things:
1. Inflection is completely transparent semantically, while derivation is not.
2. Inflection never changes part of speech, while derivation does.

If you try to apply them both you wind up with the boundary between inflection and derivation being kind of fuzzy.
* inflection usually doesn’t change part of speech while derivation often does;
* inflection is usually semantically transparent while derivation frequently isn’t, quite;
* inflection is usually fully productive while derivation often isn’t.

“Fully productive” means it applies to any wordbase of a given part-of-speech.
“Semantically transparent” means the meaning of the full word is completely predictable from the meanings of the affix and of the word base; and conversely the meaning of the base word is fully predictable from the meanings of the affix and of the full word.

I think your critic was saying that it would be unusual to have that many derivational affixes that were all fully productive..

Not all derivational affixes can be applied to all base words of the part(s)-of-speech to which they can be applied, even if they’re “semantically transparent” when so applied. But I’m not familiar with how many there are even for English.

Some Algic languages have some derivational affixes that can be applied to every root with full semantic transparency yet are still called derivational because they change the part-of-speech. Others are not fully transparent; others are not fully productive. The result of putting them altogether is called a word-stem. The stem may have inflection done to it to make the surface word.

....

I only recently found out about English having around 800 affixes of which only about 1% are inflection.
I don’t know how many of the derivational affixes are semantically transparent and how many aren’t;
nor how many are fully productive and how many aren’t;
nor how many change part-of-speech and how many don’t.

I know even less about other languages.

....

I do know that, cross-linguistically, there is a statistical correlation between transparency and productivity.
The more transparent an affix is the more productive it is likely to be and the likelier it is to be productive;
and at the same time
the more productive an affix is the more transparent it is likely to be and the likelier it is to be transparent.

....

I hope that was both to the point and helpful and informative.

....

IMHO if you’re doing a lot of deriving words of one part-of-speech from words of some other part-of-speech, a whole bunch of regular, transparent, almost-fully productive derivational mechanisms — whether affixes or something else — might very well be the way to go. It would make conlanging easier and make your conlang easier to learn, and would resemble many natural languages. Add in irregularities and non-transparencies and lowered-productivity if you feel your conlang needs to seem diachronically aged.
Most natlangs have had diachronics happen to them but this particular piece of some of them has not yet become glaringly obvious, as far as I know; so whether it is part of your conlang’s goal is up to you.
User avatar
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Sequor »

Salmoneus wrote: 22 Dec 2020 14:04- lots of derivational affixes. Having a few derivational affixes is something almost languages do. Having four or five or six is not uncommon. But having fifteen fully productive derivational affixes is unusual. English has a huge number of affixes - but largely because we've taken all the productive affixes of Latin (which already had an unusually large number of affixes) and added them to all our (many) inherited affixes and a fair few from other languages as well. But English's promiscuity is unusual. And when there's lots of derivational affixes, it's unusual for all of them to be fully productive - you'd expect many to be old and no longer productive, or only productive for small subsets of words. I'm certainly not saying having 15 derivational affixes is unrealistic, it's not. But having 15 fully productive derivational affixes is unusual, I'd say.
Interesting claim. I think Latin's highly productive derivational affixes might be limited to just about eight actually (not sure if I might be forgetting or wrongly discounting any):

[supine stem]-iō (forms verbal nouns)
[supine stem]-us (-ūs m., 4th declension, forms verbal nouns)
[supine stem]-or/rīx (forms agent nouns)
-ul/ol-us/a/um (forms diminutives)
-iē(n)s (forms "X times" adverbs, e.g. sexiēs 'six times', derived from numerals)
-ē, -iter (forms adverbs of manner from adjectives)
-Vbilis (forms "can be X-ed" adjectives i.e. -able/-ible)
-ēnsis (forms "from X city" adjectives i.e. demonyms)

To these you could add comparative -ior, superlative -issimus, and the passive participle ending -Vt-us/a/um, but these are traditionally considered inflections and not derivations (correctly so, IMO). And even within this list, a bunch of cities and regions have irregular demonym adjectives (Rōma > rōmānus), and the [supine stem]-iō and [supine stem]-us suffixes basically compete over the same meaning (both were largely acceptable in Classical Latin though). Also, -or can attach to anything, but I think feminine -rīx needs to be attached to a supine stem ending in -t-, so amātor ~ amātrīx 'love-er', corruptor ~ corruptrīx 'corrupter', but only caesor 'cut-er, hewer' and not *caesīx/*caestrīx? (EDIT: just found persuāsōr ~ persuāstrīx 'persuader' and ēsor ~ ēstrīx 'eater' are attested in Classical Latin!)

In Late Latin and more so later, -Vtiō becomes tremendously productive to form abstract nouns (hence English -tion), and same goes for -itās whence English -ity, while the Classical productivity of [supine stem]-us is not exploited as much... So basically still eight suffixes. (I think -ntia and Greek -ía, hence English -ncy and -y, were used for the same thing but to a lower extent?)

I find it amusing how varied the derivation of nouns from adjectives is in particular, considering how this language often needs such adjectives due to them being preferred over genitives when describing something rather than expressing possession. Star Wars is most idiomatically bella sīderea or bella stēllāria, the starry (star-ish) wars, using the derived adjectives of sīdus or stēlla 'star'. And so also Julius Caesar's bellum gallicum 'the war in Gallia, the Gallia war' (traditionally translated 'the Gallic wars'). But Latin has -eus/-ius, -ārius, -īvus, -Vlis(/-Vris), -icus to form these, and it's a huge tossup which is conventionally used: āēr āer- 'air' > āerius, cella 'storeroom' > cellārius, aestās 'summer' > aestīvus/a/um, cunīculus 'rabbit' > cunīculāris... Why not "cellāris", for example? Well, -icus is used with words of Greek origin, but that's about it?

And then of course there's the long parade of non-productive affixes: -ger ("X-wearing"), -fer ("X-bearing"), -igāre/-īgāre (denominal verbs), -urīre (desiderative verbs), -ō (-ōnis m., feminine -ōnia, in names and nicknames), -ius (names), -ō (manner adverbs), -tim (manner adverbs), -men (instruments), -mentum (instruments), -trum (instruments), -ūra (verbal nouns), -tūdō (abstract nouns), -itās (abstract nouns, productive post-classically), -met (emphatic personal pronouns), -ce (emphatic demonstratives in pre-Classical Latin), -nam (emphatic interrogatives), -pte (emphatic possessive adjectives), ...

Maybe this post shows how a natlang does this derivational stuff.
hīc sunt linguificēs. hēr bēoþ tungemakeras.
User avatar
qwed117
mongolian
mongolian
Posts: 4094
Joined: 20 Nov 2014 02:27

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by qwed117 »

Honestly, if I looked at the words you gave, the derivational system is opaque enough that the roots more like an extremely hypothetical proto-word in a language set maybe 4000+ years ago which obviously makes the phonetic changes not exactly distant enough for the relationship. Like the semantic difference between 'millipede' and 'wagon' and 'wheat' is so different that I have trouble believing that they could be maintained as a single seme, which would imply that over time the effect of diachronic phonetic change should easily outweigh that of any analogical forces keeping them similar
Spoiler:
My minicity is [http://zyphrazia.myminicity.com/xml]Zyphrazia and [http://novland.myminicity.com/xml]Novland.

Minicity has fallen :(
The SqwedgePad
User avatar
Omzinesý
mongolian
mongolian
Posts: 4082
Joined: 27 Aug 2010 08:17
Location: nowhere [naʊhɪɚ]

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Omzinesý »

Slavic has this s => š => x change under some conditions.
Could there also be t => c => k change under some conditions in a lang? It could create interesting morphological alternations. Does it even appear somewhere?
My meta-thread: viewtopic.php?f=6&t=5760
Axthieb
rupestrian
rupestrian
Posts: 10
Joined: 08 Dec 2020 22:25
Location: Austria

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Axthieb »

Hi, again. Thank you for all the further input, I'm learning a lot through this discussion.

So, first I want to say that through this new understanding I've decided that roots are no longer words on their own in most cases. I will, in my own writings, list roots that are also actual words as well among other words.

My improved understanding of morphology led me to the conclusion that words need to be less predictably derived if the resulting meanings are similarly unpredictable.
Like the semantic difference between 'millipede' and 'wagon' and 'wheat' is so different that I have trouble believing that they could be maintained as a single seme, which would imply that over time the effect of diachronic phonetic change should easily outweigh that of any analogical forces keeping them similar
I used to think that my hypothetical peoples would have mindsets accustomed through culture and different beliefs, that such forces as the analogical would actually keep them together. That is, they were supposed to be different and see things as wildly different that it would make little sense to us.
That was before I strayed farther from this past idea.
I do want vocabulary and semantics to work in rather naturalistic ways.
That's why I'm changing small parts of the conlang in accordance.
User avatar
Pabappa
greek
greek
Posts: 577
Joined: 18 Nov 2017 02:41

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Pabappa »

Omzinesý wrote: 25 Dec 2020 18:17 Slavic has this s => š => x change under some conditions.
Could there also be t => c => k change under some conditions in a lang? It could create interesting morphological alternations. Does it even appear somewhere?
Probably. Possibly in Hawaiian .... but there's no real way to know, since there are so few consonants phonemes to keep track of ... it was entirely allophonic and thus at the phonemic level it's just a simple /t/ > /k/ shift.

In a conlang, I once did [tʰ] > [tx] > [kx] > [kʰ], meaning that the /t/ phoneme became /k/ in every position in which it was aspirated, but when it was not aspirated it remained as /t/. I dont know of any precedent for that shift in the wild, but it made sense to me and that's the strategy I usually go by.
Kavunupupis, šiŋuputata.
When I see you pointing at me, I know I'm in trouble. (Play)
User avatar
qwed117
mongolian
mongolian
Posts: 4094
Joined: 20 Nov 2014 02:27

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by qwed117 »

Pabappa wrote: 26 Dec 2020 00:59
Omzinesý wrote: 25 Dec 2020 18:17 Slavic has this s => š => x change under some conditions.
Could there also be t => c => k change under some conditions in a lang? It could create interesting morphological alternations. Does it even appear somewhere?
Probably. Possibly in Hawaiian .... but there's no real way to know, since there are so few consonants phonemes to keep track of ... it was entirely allophonic and thus at the phonemic level it's just a simple /t/ > /k/ shift.
For what it's worth, a look at the Ni'ihau Hawai'ian page on Wikipedia suggests that /t/ > /k/ only presents near back consonants, so a change like /tB/ ->/tʷB/ -> /kʷB/ ->/kB/ could have happened. It's worth noting that before this, the Marquesic /*k/ had already become Hawai'ian /ʔ/, so /t/ >/k/.
Spoiler:
My minicity is [http://zyphrazia.myminicity.com/xml]Zyphrazia and [http://novland.myminicity.com/xml]Novland.

Minicity has fallen :(
The SqwedgePad
Salmoneus
MVP
MVP
Posts: 3033
Joined: 19 Sep 2011 19:37

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Salmoneus »

qwed117 wrote: 27 Dec 2020 03:10
Pabappa wrote: 26 Dec 2020 00:59
Omzinesý wrote: 25 Dec 2020 18:17 Slavic has this s => š => x change under some conditions.
Could there also be t => c => k change under some conditions in a lang? It could create interesting morphological alternations. Does it even appear somewhere?
Probably. Possibly in Hawaiian .... but there's no real way to know, since there are so few consonants phonemes to keep track of ... it was entirely allophonic and thus at the phonemic level it's just a simple /t/ > /k/ shift.
For what it's worth, a look at the Ni'ihau Hawai'ian page on Wikipedia suggests that /t/ > /k/ only presents near back consonants, so a change like /tB/ ->/tʷB/ -> /kʷB/ ->/kB/ could have happened. It's worth noting that before this, the Marquesic /*k/ had already become Hawai'ian /ʔ/, so /t/ >/k/.
This (the conditioned change with intemediates) is possible, but unlikely. It seems more likely that conditioned /k/ > [t] is a secondary change in Ni'ihauan!

The change /t/ > /k/ has happened at least 20 times independently in Austronesian. In a few of these cases, we've managed to catch it in the act. In Samoan, /t/ > /k/ is a sociolinguistic marker: [t] is retained in formal speech, while [k] is found in informal speech (to simplify). It's not seemingly phonologically conditioned in Samoan. Likewise, in the Hawai'ian of about two centuries ago, early Western linguists, in writing down the speech of the natives, were inconsistent in whether they wrote 'k' or 't', without an obvious phonological basis.

The assumption is that, in the absence of /k/, /t/ is often realised as [t~k]. In some languages, the backer realisation came to be generally preferred, leading to /t/ > /k/. It is likely that during the period of free variation (or even after it), the Ni'ihau dialect developed an additional allophonic rule preferring the front realisation before front vowels.

The idea of /t/ > /c/ > /k/ has the problems that:
a) /c/ is rare
b) /c/, despite being more back than /t/, is often associated with high or front vowels, rather than with backing
c) /t/ > /c/ is no more motivated than /t/ > /k/ would be, and then you're left with also needing /c/ > /k/ - so too weird changes instead of one
d) why are there over 40 Austronesian languages that show /k/ for inherited /t/, but zero that show /c/? If it were a chain shift, you'd expect the chain to at least sometimes be slow or interrupted!


It's also worth remembering that an even more extreme, direct, no-intermediaries, no-influence-from-vowels POA swap is well-attested closer to home: English /t/ > (future) /?/.
User avatar
Pabappa
greek
greek
Posts: 577
Joined: 18 Nov 2017 02:41

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Pabappa »

Marshallese has /tʲ/ as its reflex of proto-Oceanic /t/, so spontaneous palatalization is in fact possible. However I dont think this can really help explain Hawaiian because Marshallese did the shift as part of a global reshaping of its whole phonology into pairs of palatalized versus velarized consonants, so it had to be one way or the other. (And the velarised counterpart came from earlier /s/.)

Another good reason why /t/ > /tʲ/ isnt really useful here is that, as you point out, it is most likely to happen before a front vowel, which means we'd be stuck at /c/ with no way to get to /k/, since Polynesian's vowels are very stable.

I avoided a direct shift from /t/ > /k/ in my work because the language already had /k/ and there was no gap crying out to be filled. But with Polynesian languages, the consonant inventories are so small that almost every shift immediately creates a yawning gap because a major phoneme is suddenly absent.

Even so,
Omzinesy wrote: Slavic has this s => š => x change under some conditions.
Could there also be t => c => k change under some conditions in a lang? It could create interesting morphological alternations. Does it even appear somewhere?
A conditional /t/ > /k/ is certainly within reach ... I just dont think Polynesian is a good model to follow.

What are the phonotactics of your language? Do you have a lot of consonant clusters? If not, are there preexisting vowel alternations you could pull on?
Kavunupupis, šiŋuputata.
When I see you pointing at me, I know I'm in trouble. (Play)
User avatar
Sequor
sinic
sinic
Posts: 438
Joined: 30 Jun 2012 06:13

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Sequor »

Salmoneus wrote: 27 Dec 2020 13:21The idea of /t/ > /c/ > /k/ has the problems that:
a) /c/ is rare
b) /c/, despite being more back than /t/, is often associated with high or front vowels, rather than with backing
c) /t/ > /c/ is no more motivated than /t/ > /k/ would be, and then you're left with also needing /c/ > /k/ - so too weird changes instead of one
d) why are there over 40 Austronesian languages that show /k/ for inherited /t/, but zero that show /c/? If it were a chain shift, you'd expect the chain to at least sometimes be slow or interrupted!
Another thing I notice is that Spanish-speaking toddlers acquiring the language very often replace /k/ with /t/, saying effectively adult "la taza" 'the cup' [laˈtasa] for "la casa" 'the house', and pato 'duck' for Paco (a personal male name, = Francisco). But you don't hear [c]. Somehow, the opposite (t>k) doesn't seem to happen among these L1 toddlers, not much at least.
hīc sunt linguificēs. hēr bēoþ tungemakeras.
Khemehekis
mongolian
mongolian
Posts: 3885
Joined: 14 Aug 2010 09:36
Location: California über alles

Re: (Conlangs) Q&A Thread - Quick questions go here

Post by Khemehekis »

Sequor wrote: 28 Dec 2020 00:04
Salmoneus wrote: 27 Dec 2020 13:21The idea of /t/ > /c/ > /k/ has the problems that:
a) /c/ is rare
b) /c/, despite being more back than /t/, is often associated with high or front vowels, rather than with backing
c) /t/ > /c/ is no more motivated than /t/ > /k/ would be, and then you're left with also needing /c/ > /k/ - so too weird changes instead of one
d) why are there over 40 Austronesian languages that show /k/ for inherited /t/, but zero that show /c/? If it were a chain shift, you'd expect the chain to at least sometimes be slow or interrupted!
Another thing I notice is that Spanish-speaking toddlers acquiring the language very often replace /k/ with /t/, saying effectively adult "la taza" 'the cup' [laˈtasa] for "la casa" 'the house', and pato 'duck' for Paco (a personal male name, = Francisco). But you don't hear [c]. Somehow, the opposite (t>k) doesn't seem to happen among these L1 toddlers, not much at least.
My (English-speaking) brother was unable to pronounce /k/ as a small child, and he'd replace it with /t/. He also used /d/ instead of /g/.
♂♥♂♀

Squirrels chase koi . . . chase squirrels

My Kankonian-English dictionary: 87,413 words and counting

31,416: The number of the conlanging beast!
Post Reply