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

[Ahzoh]

New question, so I have these sound changes (there are a lot of them)

Spoiler:

Code: Select all

Progressive assimilation of vowels:
V₁.V₂    > V₂:
v₁.V₂:   > V₂:
V₁:.V₂:  > V₂:
V₁:.V₂C  > V₂:C

Regressive assimilation of vowels:
V:₁.V₂ > V₁:

Non-High vowel harmony:
[a,a:] > [e,e:] / _.C[e,e:] or [e.e:].C_ (blocked by consonant clusters)

Simple elision:
[h,ħ]     > 0 /#_V.
[h,ħ]     > 0 /#_V:.
[j,w,h,ħ] > 0 /V_#
[j,w,h,ħ] > 0 /C._V:
[j,w,h,ħ] > 0 /V._V
[j,w,h,ħ] > 0 /V:._V
[j,w,h,ħ] > 0 /V:._V:

Diphthongization:
[a]j   > i: /_.C (root internally)   (e.g. rayb-ū > rīb-ū)
[a]j   > e: /_.C (morpheme boundary) (e.g. na-ynum-ū > nē-num-ū)
[i,u]j > i: /_.C
[a]w   > u: /_.C (root internally)
[a]w   > a: /_.C (morpheme boundary)
[i,u]w > u: /_.C

Root Internal Diphthongization:
jV > i: /C._ (e.g. na-ryub-ū > na-rīb-ū)
wV > u: /C._
hV > a: /C._
ħV > e: /C._
ji > i: /#_ (e.g. yirum "hill" > īrum) 
wu > u: /#_ (e.g. wuṣuḫtu "be ye faithful" > ūṣuḫtu)

Simple compensatory lengthening:
[h,ħ] > : /V_.C

I need ideas of what to do with the weak geminates, because simply eliding them is undesirable in most cases

Code: Select all

Weak geminates:
Vj.jV > (root-internally)   (e.g. rayyub-um > )
Vj.jV > (morpheme boundary) (e.g. ma-y-yand-um > or malay-y-ū)
Vw.wV > (root-internally)
Vw.wV > (morpheme boundary)
Vh.hV > (root-internally)
Vh.hV > (morpheme boundary)
Vħ.ħV > (root-internally)
Vħ.ħV > (morpheme boundary)

[eldin raigmore]

Root-internally do whatever you’d normally do if you can figure out what that is; otherwise, root-internally simplify the geminate to a single instance of the consonant.

(Very surprised your protospeakers didn’t already have some habit for the root-internal instance. Maybe I’m just too ignorant to see why they wouldn’t have?)

At a morpheme boundary perseverate or anticipate one of the vowels to epenthesize between the two instances of the consonant. If that would confuse things epenthesize a compromise vowel instead; same frontness as the preceding vowel and same height as the following vowel or vice-versa. (Or some other compromise?) If that’s still confusing epenthesize a schwa instead. In any case put an epenthetic vowel between the two copies of the consonant.

....

Just an idea. Might not be a good one, depending how it goes with the rest of your language.

....

Whatever your solution is and whyever you choose it I’ll be interested!

[Ahzoh]

How could I justify inalienably-possessed nouns experiencing i-ablaut in the construct state while alienably-possessed nouns do not? The distinction in alienability was lost a long time ago and it's no longer productive existing in only a few CV(:)C and CVCC monosyllabic nouns.

[Creyeditor]

Would it work to have two different construct state affixes in the proto-lang, one for inalienable nouns (containing /i/) and one for alienable nouns (not containing /i/)? You would get a generalization of the no-I-affix to all nouns with a few exception. Diachronically later, umlaut would kick in, followed by deletion of the affix.

[Ahzoh]

Would it work to have two different construct state affixes in the proto-lang, one for inalienable nouns (containing /i/) and one for alienable nouns (not containing /i/)? You would get a generalization of the no-I-affix to all nouns with a few exception. Diachronically later, umlaut would kick in, followed by deletion of the affix.

Problem is, according to various linguistic analyses, the inalienably possessed tend to be the least-marked of the two types when the matter in question would effectively mean the opposite. There was one language of the many that have it where the inalienable possession is more marked, although the language is unnamed and thus I cannot know how they came to buck the overwhelming trend.

The reasons go "it would seem redundant to overtly mark a noun that cannot be separated from its possessor; it's inherently possessed" and "because nouns like kinship are more commonly heard possessed, a hearer can hear an inalienable noun and expect it to be possessed without needing an overt marker"

[Creyeditor]

I see what you mean. Inalienable nouns do not have to be obligatorily possessed in grammar. In English you can say 'the head ', 'the brother's and 'the top' without any possessor. And possession has to be marked here overtly. Are your inalienable nouns obligatorily possessed? If yes, is it true at all stages?

If you still want to incorporate the markedness idea, you could have the inalienable nouns bearing a shorter affix if possessed, in the first stage. Many linguists would still call this less marked. A longer affix would mark alienable possessed nouns. Maybe there is even a long affix to mark unpossessed inalienable nouns. At a later stage, when the distinction is only kept in a few exceptional words, you can get rid of the unpossessed marker. Speakers can generalize the possession construction and only mark possessed nouns in general.

[Salmoneus]

I don't understand. Isn't the construct state the form of the noun for possessed things? In which case, since inalienably possessed things are by definition always possessed, wouldn't they always be in the construct state, by definition? In which case, they couldn't experience umlaut in forming the construct state, as they'd have no non-construct state to compare with?

[Creyeditor]

@sal: I really think it is important to distunguish semantic inalienability from morphosyntactic obligarory possession. Some languages just use different possessive constructions depending on alienability but still allow inalienable nouns without any possessor marking.

[Ahzoh]

I don't understand. Isn't the construct state the form of the noun for possessed things? In which case, since inalienably possessed things are by definition always possessed, wouldn't they always be in the construct state, by definition? In which case, they couldn't experience umlaut in forming the construct state, as they'd have no non-construct state to compare with?

What you're describing is obligatory possession like how it's ungrammatical in Ojibwe to just say mama "mother" but you always have to say nimama "my mother", kimama "your mother", omama (his/her mother), etc.

This is not what is occurring here.

Hmm, I suppose I could have it where the construct state was:

-wa/-w- (alienable)
-ya/-y- (inalienable)

but only the inalienable marker causes vowel mutation (singular / plural):
kap-um "water" > kap-wa / kap-w-aħ > kap / kap-aħ (> kapē)
nar-um "mother" > nar-ya / nar-y-aħ > ner / ner-aħ (> nerē)

That way they're both equally marked

[Creyeditor]

IMHO, that sounds very naturalistic and is probably attested in some natlang either for construct state affixes or more generally for possession markers.

[Salmoneus]

@sal: I really think it is important to distunguish semantic inalienability from morphosyntactic obligarory possession. Some languages just use different possessive constructions depending on alienability but still allow inalienable nouns without any possessor marking.

Fair point. However, I'm not really convinced!

If we're talking a direct/indirect possession distinction (on the basis of semantic inalienability), this is generally a fact about the marking of possessors. But what Ahzoh is talking about is the marking of possessed status. The difference is that an inalienably possessed thing can have different possessors, so it makes sense to separate out the possessive markers from the root noun. But an inalienably possessed thing is never unpossessed (even if it's not morphologically marked). So how do you distinguish the morphemes, when the unpossessed form will never occur?

There would seem to be three possibilities:

a) the base and inflected forms both occur, because sometimes the noun is possessed and sometimes it isn't. But in this case, that's not something inalienably possessed - inalienable possession is inalienable, it cannot be taken away from the noun. You may still have a direct/indirect possession distinction, but it's no longer on the basis of semantic inalienability.

b) the base and inflected forms both occur, because even though the noun is always possessed, the different forms are used in different semantic or grammatical contexts (eg, the construct may not be used for the middle noun in an X-of-Y-of-Z chain). But in that case, the umlaut is not a marker of possession, but of whatever it is that the umlaut is actually marking!

c) only the inflected form occurs, because the inflection marks possession, and the possession is inalienable. In this case, there cannot (synchronically) be said to be umlaut at all, because there is no unumlauted form for comparison.

None of this is necessarily connected to the morphological obligatoriness of possessive morphemes. It is indeed possible to have a semantically inalienable noun that is not overtly marked for a specific possessor - because although the noun must be possessed, there need not be a marker for being possessed [so, for example, the marker may be dropped in certain grammatical or semantic situations, such as indefinite or topical possession]. But it's not possible to have a genuinely semantically inalienable noun explicitly marked as unpossessed, because an inalienable noun can never be unpossessed (even if you don't know the possessor or it's clear from context). There may be times when you don't mark that it's possessed, but there shouldn't be times when you mark that it's unpossessed, which is what the ruckumlaut here would do.

To give a concrete example: if meme-mia is 'my mother', and meme-tua is 'your mother', it makes sense that in some languages you may be able to say just meme, meaning perhaps "somebody's mother" or "the mother we talked about" or the like. [whereas in a language with obligatory possession, you couldn't say this, you'd have to say *meme-qui, "unspecified person's mother" and so on]. But it doesn't make sense that you'd ever want to say **mama and mean "a mother who is not the mother of anyone or anything". It cetainly doesn't make sense that this would be the basic form of the noun from which the other forms are derived.



I suppose you could argue: it's not inalienability, but it's a property connected to inalienability, so we'll just call it that for convenience. But at the very least, that should make clear why languages don't have more elaborate marking for possession of inalienables than of alienables: the marking of possession on an alienable is semantically double-marking - it marks that, unlike most things of this class, this item is actually possessed, and THEN it indexes the possessor. Whereas with an inalienable, the possession is inherently (lexically) marked, so the overt marking is just to index the possessor. Naturally, double marking is likely to be more marked than single marking!



Although, to answer my own confusion... I guess you could have a morphological marker on the noun, not to indicate possession, but to warn the speaker of the presence of a genitive modifier, and I guess that this is what Ahzoh is doing? In which case... OK. And sorry, I should have realised that. But in that case, I wouldn't worry too much about cross-linguistic marking of possession, since you're really talking here about marking of phrase edges, which is completely different (eg some languages with a construct state also use it to mark adjectives that precede their head, iirc). Since your construct state doesn't semantically alter possession, I wouldn't think that it should interfere too much with an inalienable possession system?


In terms of how you should do it... I guess it depends how your construct state developed. It's not usually something that's fundamentally marked, AIUI - construct state is essentially just a prohibition on the double-marking of definiteness, coupled in some languages with the absence of sound changes linked to edges and stress patterns. So how this would interact with possession would depend on the details of what markers there were, and which have been dropped.



-----------------------

-------------------------------------


How I would probably do it....

Stage 1:
inalienables must have overt possession marking, in one of three ways:
a) possessive pronominal prefix - ma-mar, "my leg", tu-mar, "your leg"
b) an overt possessor - mar lago, "the idiot's leg"
c) an indefinite possessor suffix (originally a pronominal indefinite possessor, presumably) - mar-i, "someone's leg"
alienables are always marked with the b) construction

Stage 2:
The b) construction merges with the c) construction: mar-i lago:
- ma-mar, "my leg"; lon ma, "my shoe"
- mar-i lago, "the idiot's leg"; lon lago, "the idiot's shoe"
- mar-i, "someone's leg"; lon kon-ul, "someone's shoe"

Stage 3:
possession stops being obligatory; inalienables can now be marked instead with definite or indefinite markers, as can other nouns.
- mar-at, "the leg"; lon-at, "the shoe"
- mar-ul, "a leg"; lon-ul, "a shoe"
- ma-mar, "my leg"; lon-at ma, "the shoe of mine"
- mar-i-at lago-ul, "the leg of an idiot"; lon-at lago-ul, "the shoe of an idiot"
- mar-i, "someone's leg"; lon-at kon-ul, "the shoe of someone"

Stage 4:
construct state (no double-marking of definiteness):
- mar-at; lon-at
- mar-ul; lon-ul
- ma-mar; lon ma
- mar-i lago-ul; lon lago-ul
- mar-i; lon kon-ul

Stage 5: phrase-final vowels drop
- mar-at; lon-at
- mar-ul; lon-ul
- ma-mar; lonm
- mar-i lago-ul; lon lago-ul
- mar; lon kon-ul

Stage 6: umlaut, and epenthetic vowel
- mar-at; lon-at
- mar-ul; lon-ul
- ma-mar; lon-u-m
- mer-i lago-ul; lon lago-ul
- mar; lon kon-ul

Stage 7: word-final vowels drop.

This gives us two paradigms:

alienable nouns:
a shoe: lonat
the shoe: lonul
someone's shoe: lon konul
an idiot's shoe: lon lagol
my shoe: lonum


but inalienable nouns:
a leg: marat
the leg: marul
someone's leg: mar
an idiot's leg: mer lagol
my leg: mamar

A construct state, and umlaut only in the construct state of inalienable nouns. Plus a distinct paradigm for pronominal possession of inalienable nouns, which could be dropped for most nouns but retained for a small number of irregulars.


In any case, I'd definitely think about the general path of former obligatory possession, with an umlauting suffix as an indefinite possession marker that was then generalised in many cases. You could even generalise it to all instances of the noun, but have it dropped in phrase-edge conditions (i.e. only retained in the construct state).

[Creyeditor]

Sorry, I tthink I was using imprecise terminology again. A marker of unpossessed inalienable noun cannot refer to the absence of a (semantic) possessor, as Sal correctly points out. I was thinking of contexts where English (and probably other languages, Indonesian comes to mind) use inalienable nouns with definite/indefinte/-out articles, e.g. 'The foot is an important bodypart.', 'I hate fingernails', 'A head has a larger volume than a cherry'. These often have an implicit indefinite possessor, IINM. Many languages that obligatorily mark possession on these nouns have a specific marker for these cases. I was referring to this kind of marker, improperly as an 'unpossessed' marker.

[Ahzoh]

To give a concrete example: if meme-mia is 'my mother', and meme-tua is 'your mother', it makes sense that in some languages you may be able to say just meme, meaning perhaps "somebody's mother" or "the mother we talked about" or the like. [whereas in a language with obligatory possession, you couldn't say this, you'd have to say *meme-qui, "unspecified person's mother" and so on]. But it doesn't make sense that you'd ever want to say **mama and mean "a mother who is not the mother of anyone or anything". It certainly doesn't make sense that this would be the basic form of the noun from which the other forms are derived.

Well from reading this, alienability distinction is restricted only to attributive possession and not predicative possession, so there would require an instance where the non-possessed form must exist.

Also there is the notion that the inalienable has fewer morphological features which in practice just means that the possessed and possessor would simply be in juxtaposition ("zero-marked") while the alienable would require more overt marking like an affix or pronoun.

Anyways, the *wV and *yV suffixes are the construct state markers; it is not a construct state form on top of alienability markers. But as I said, the nonpossessed form would have to exist when expressing predicative possession.

Frankly, this all is just becoming more trouble than it's worth. I just wanted to produce some irregular ablaut in a small set of "biliteral-root" nouns. Now if I want such abalut it would be more sensible to do it on alienable nouns, which would exist in far greater numbers than inalienables, even for biliteral-roots.

[Khemehekis]

- mar-i lago, "the idiot's leg"; lon lago, "the idiot's shoe"

I see what you did there.

[Ahzoh]

- mar-i lago, "the idiot's leg"; lon lago, "the idiot's shoe"

I see what you did there.

As did I. Because of it, I made a root L-G-W "be stupid, be foolish"

[Khemehekis]

- mar-i lago, "the idiot's leg"; lon lago, "the idiot's shoe"

I see what you did there.

As did I. Because of it, I made a root L-G-W "be stupid, be foolish"

[Salmoneus]

Serious question: does any real language actually have a "construct state suffix"? (or indeed prefix?)

[Creyeditor]

If you want to research on this topic (and specifically possessed nouns), it might make sense to google antigenitive instead. At least that's the term I am familiar with from Nilotic languages. A quick google search turn up this pdf on construct states in African languages in general. After a quick scanning, It seems to mention Wolof as a "pure" antigenitive, marked by -u in the singular and -i in the plural.

[Ahzoh]

Serious question: does any real language actually have a "construct state suffix"? (or indeed prefix?)

Yes. In languages like Syriac and Akkadian there is a suffix -ay and -i respectively and in Ge’ez it is the suffix -a

Furthermore, Akkadian has the construct state suffixes as distinct from the absolute state which requires the removal of all suffixes in order to give the noun a predicative meaning.
Also, Akkadian lacks any definite marking, so the construct state isn’t simply a matter of making it incapable of taking a definite marker.

[Sequor]

Serious question: does any real language actually have a "construct state suffix"? (or indeed prefix?)

In modern Arabic(s), feminine nouns ending in -a take -t as the construct state suffix. Something typical:

madiina 'a city'
madiinat ʕomar 'Omar's city' (i.e. the city Omar comes from)
madiinat-o 'his city' (where -o = 3.SG.MASC enclitic pronoun, here functioning as a possessor)

Considering the Classical Arabic form of these would be madiinatun, madiinatu ʕumara, madiinatu-hu, it looks like the -t failed to drop when followed by a possessor (reasonable in terms of prosody really...), and now it's basically a construct state suffix.

Classical Nahuatl also has a construct state, usually called the "possessed" or "possessive" form. Nouns ending in a vowel mark it with the suffix -uh [w] in the singular, and all nouns mark it with -huān [wa:n] in the plural.

absolutive state:
chichi 'pet dog'
chichi-meh 'pet dogs'

construct state:
chichi-uh 'pet dog of...'
chichi-huān 'pet dogs of...'
to=chichi-uh 'our pet dog' (1PL=dog-CONST.SG)