Zompist Bboard Again

Scratch intro

It's about time I had one of these.

Things I post here will mostly be a step or two past raw notes and brainstorming, but I won't try to keep things either focused or cumulative.

Vædty Qyṣ (phonology)

vædty qyṣ [væˑˈθtɨˑ ˈqɨʂ] is Vædty Qyṣ for unnamed conlang. (Not really.)

I've got no particular plans for this language. Maybe it'll be a source for vocabulary borrowed into Akiatu's ancestors.

Consonants

m	n
	t	ʈ <ṭ>		k	kʷ <kw>	q	qʷ <qw>
	s	ʂ <ṣ>						h
v	ð <d>			ɣ <g>
	ɾ <r>
	r~l <rr, l>
			(j)		(w)

The voiced fricatives are supposed to look like they derive from voiced plosives, though that's not going to be apparent in the phonotactics.

l is [l] in coda and otherwise [r]. I'll distinguish these allophones orthographically; note though that rr never represents a genuine geminate, and can follow a coda consonant in the previous syllable.

Retroflex ṭ ṣ are genuine subapical retroflexes. I wish I knew more about how such things tend to behave phonologically/phonetically.

So far it looks like j w might be best thought of as positional allophones of the high vowels. They occur only in coda position, unlike the consonants.

Every syllable has an onset, and there's no particular reason to think of any of these onsets of epenthetic.

Other than the plosives, all consonants can also occur in coda position. Coda voiced fricatives are devoiced. There's no difference in which consonants can occur in coda medially and word-finally.

Vædty Qyṣ has a somewhat odd collection of possible onset clusters. They all consist of an oral fricative followed by a noncontinuant; there's regressive voicing assimilation in all clusters.

	m	n	t	ʈ	k	kʷ	q	qʷ
s	[zm]	[zn]	[st]	*	[sk]	[skʷ]	*	*
ʂ	[ʐm]	[ʐɳ]	*	[ʐʈ]	*	*	[ʂq]	[ʂqʷ]
v	[vm]	[vn]	[ft]	[fʈ]	[fk]	[fkʷ]	[fq]	[fqʷ]
d	[ðm]	[ðn]	[θt]	*	[θk]	[θkʷ]	[θq]	[θqʷ]
g	[ɣm]	[ɣn]	[xt]	*	[xk]	[xkʷ]	[χq]	[χqʷ]

(Actually I'm not going to promise that all of these will end up attested, but they're ones I'll consider legal.)

Vowels

i	u
e	ɔ <o>
æ	ɑ <a>

ɨ <y> is also very common, but I'll consider it epenthetic.

I'm thinking that maybe this derives from a system with i ɨ u ɛ ɔ ɐ, via a chain shift ɐ→ɑ ɛ→æ i→e ɨ→i, fwiw. (Or maybe it started with y instead of ɨ.)

The distinction between e and æ is lost following q and qʷ. (Also preceding them?) I'll write the resulting vowel as æ. In the same position, i u shift to ɪ ʊ, but no contrasts are lost.

The distinction between i and epenthetic y is lost following retroflex ṭ ṣ. I'll write the result as y.

A syllable has at most one vowel. Many syllables have no underlying vowel; that's when you get epenthetic y.

Actually, epenthetic y is common enough that I'll be explit about the rules. Starting from the end of the word, y gets inserted in the following contexts:

• C_#, when C is a plosive (and thus can't be coda)
• C_C# and C_CC, no matter what
• C_CV, when CC do not form a legal onset cluster

Stress

The simple rule is that the word-final syllable gets stress.

Suffixes that add a syllable also shift stress. E.g., ˈskæd dog → skæˈdæ dogs.

Some clitics, though, don't shift stress: skæˈdæ=ve for dogs.

There are other clitics that do attract stress. When they attach outside a non-stress-attracting clitic, the existing stress becomes secondary. Thus skæˌdæ=ve=ˈdi only for dogs.

But without an intervening non-stress-attracting clitics, the stress-attracting ones behave the same as suffixes: skæ.dæ=ˈdi only dogs.

Actually it's entirely possible that it's best not to try to understand these patterns in terms of a distinction between clitics and affixes. Maybe I'll just gloss all bound morphemes as suffixes.

I've mentioned one way in which you can get secondary stresses in morphologically complex forms. That can also happen with compounds (though I'm not actually sure how much those stress patterns will differ from what you'd ordinarily get in phrases).

I doubt there'll be any evidence of rhythm. If you want, you can think of the stressed syllables as heading syllabic-trochaic feet, but I'm not sure what that'd get you.

Stressed syllables are generally louder than unstressed ones, but not longer, and I don't think there'll be any significant vowel reduction in unstressed syllables.

Conversely, it'll often be an epenthetic y that gets stressed.

Stressed syllables will host pitch accents, and if I ever understand intonation, maybe I'll tell you more about that.

Processes

I've already mentioned that vowels in open syllables get a bit lengthened. I'll transcribe them as half-long. (This does not interact with stress.)

There'll be other things, but I'll mention them as I come up with introduce particular bits of morphology.

Vædty Qyṣ (nouns)

This is just about morphology: plural, definiteness, case, and focus markers.

This morphology is entirely suffixing/encliticising.

Eventually I'll also need possessor agreement (I think it'll probably go in the same `slot' as the definiteness marker), but I'm not even a little bit in the mood to put together an agreement paradigm right now.

Plurals

I'll start with plurals.

Plural-marking is pretty much never obligatory. (Alternatively, you could say that all nouns can be used as mass nouns.) It's most common with nouns referring to human beings, as you might expect, but there's no semantically-definable class of nouns that can't occur with plural marking. (You could say that there's no class of nouns that can only be used as mass nouns.)

Most nouns fall into one of two broad classes, depending on how they form their plural: nouns of one class use an echo vowel, and the other nouns use the suffix -kaj.

Plurals of both sorts add syllables, and the new syllables attract stress.

Echo vowels

In the simplest cases, nouns of the first type really do just append a copy of the stem's rightmost vowel:

• skæd dog → skædæ dogs
• qygum shrine → qygumu shrines
• vor → tree rat voro tree rats

With some nouns, the stem vowel changes before being copied. This is most common when the stem vowel is y, which is never simply copied; it can become either i or u (there's no way to predict which):

• datys star → datisi stars
• tyn bead → tunu beads

Some common nouns preserve other ablaut patterns, roughly unwinding the chain shift I mentioned in the last post: e→i æ→e a→æ.

• ṭyṣmel fool → ṭyṣmirri fools
• væd rain → vede rains

For the purposes of vowel copying, stem-final (epenthetic) y is usually not treated as a vowel, instead dropping in the plural:

• moqy flower → moqo flowers

This pattern can co-occur with ablaut:

• vkyty tree → vkiti trees

Stem-final nonlow vowels can get copied as j or w, depending on frontness/rounding:

• ko branch → kow branches
• qi water → qij waters

There are sporadic cases of ablaut feeding this sort of plural, including with stem-final y:

• myrraty stream → myrratij streams
• ṭaxkyvæ ritual mask → ṭaxkyvej ritual masks

Alternation of a with æ occurs in the common nominaliser -da:

• gymda life → gymdæj lives (< gym to give birth)
• ṣṭujda campsite → ṣṭujdæj campsites (< ṣṭuj to lie down)

-kaj

The other main way to form plurals is with variants of the suffixe -kaj.

• gyl person → gylkaj people
• mon lake → monkaj lakes

-kaj becomes -qaj when immediately following a low vowel or ṣ, or if the most recent consonant was uvular or retroflex:

• vaqwy vine → vaqwyqaj vines
• da spirit → daqaj spirits

(q and qw lose their influence if a coda alveolar or velar intervenes.)

The vowel in the suffix becomes æ with a front vowel in the previous syllable:

• tyqæṣ campfire → tyqæṣqæj campfires
• madimadi butterfly → madimadikæj butterflies

Definiteness

Maybe eventually I'll say something about when nouns must be marked as definite. For now I'll just stick with how to mark them.

It's easier than marking plurals.

With a singular noun, the suffix is just -s; except that it becomes -ṣ if the most recent consonant was uvular or retroflex. The suffix can trigger insertion of epenthetic y.

• skæd dog → skædys the dog
• vkyty tree → vkytys the tree
• tyqæṣ campfire → tyqæṣyṣ the campfire

With plurals, the definiteness marker differs depending on the noun's plural class.

With copy-vowel plurals, the copy vowel deletes (though stem changes are retained) and the suffix -as (or -aṣ or -æs or -æṣ) is appended.

• skæd dog → skædæ dogs → skædæs the dogs
• tyn bead → tunu beads → tunas the beads
• vkyty tree → vkiti trees → vkitæs the trees

With vowel-final stems that form their plural with a `copy' glide, the glide is retained; vowel harmony is determined not by the glide but by the nuclear vowel in the previous syllable:

• ko branch → kow branches → kowas the branches
• qi water → qij waters → qijæs the waters

Er, these forms are actually inconsistent with the phonotactic constraints I listed last time. I think I want to say that the glides can end up as onsets only at a morpheme boundary, as here.

With -kaj plurals, the suffix-final j is replaced with s or ṣ.

• gyl person → gylkaj people → gylkas the people
• da spirit → daqaj spirits → daqaṣ the spirits

When a definiteness marker adds a syllable, that syllable attracts stress.

(As I mentioned above, definiteness marking will in some cases get disrupted or replaced by possessor agreement.)

Case

I'm going to distinguish between the core cases and the other ones (which might not really be cases).

The core cases are nominative, accusative, dative, and genitive.

Nominative is unmarked.

Accusative gets marked with -n. (Vædty Qyṣ is basically an Esperanto relex.) This will sometimes require an epenthetic y.

Dative gets -ti or -ṭi.

Genitive gets -v, which will become vy after a consonant.

When the marker of a core case adds a syllable, that syllable attracts stress.

None of the other case markers attract stress, and you might prefer to think of them as cliticising postpositions.

...Er, but I've run out of steam for the night, and am just going to stop here.

Vædty Qyṣ (nouns, cont'd)

Possessor agreement

I've decided that possessor agreement does go in the same slot as the definiteness marker, and also that the definiteness marker(s) I covered in the last post are (also) used with singular third-person possessors.

I'll generalise what I said last time about definite plurals: it's also true that before a possessor agreement marker, the copy vowel (but not a copy glide) in copy vowel plurals is replaced by a, and the -kaj suffix loses its j.

I'm also going to assume that the fronting of a to æ and the velar/uvular and alveolar/retroflex alternations are governed by a general rule, and I'll stop mentioning those. (The general rule will cover at least all suffixes. I think it won't cover clitics.)

Here's the full paradigm. The examples use qod footstep (pl. qodo). It might help in understanding some of the examples that plural-marking is generally optional.

	Marker	Examples
1s	-oj, -wi	qodoj my footstep, qodawi my footsteps
1p	-rm	qodrym our (excl) footsteps, qodarym our (excl) footsteps
1+2	-mki	qodmyki our (incl) footstep, qodamki our (incl) footsteps
2s	-d	qodyd your (s) footstep, qodad your (s) footsteps
2p	-rri	qodrri your (pl) footstep, qodarri your (pl) footsteps
3s	-s	qodys their (s) footstep, qodas their (s) footsteps
3p	-t	qodty their (pl) footstep, qodaty their (pl) footsteps

(Incidentally, I've just realised I screwed up the rules for y-insertion. It's entirely possible I've also been screwing up actual y-insertion. Eek.)

Case

Here's a full paradigm of the cases you might consider morphological:

	Marker
nominative	--
accusative	-n
dative	-ti
genitive/locative	-v
benefactive	=ve
instrumental	=myh
locative	-v=hyr
allative	-v=ty
ablative	-v=æv
comitative	-v=ky

I've described -v as marking both genitive and locative. It directly forms a locative expression only with nouns that inherently refer to places; with other nouns, the `long' locative -v=hyr must be used.

Remember that the markers I've classed as clitics do not attract stress.

There are a couple of other devices that you might think of as marking cases. They involve true postpositions or relational nouns, though; a diagnostic is that they occur outside the focus markers, which I'll introduce next.

Vædty Qyṣ has no nominal and distinct from the comitative marker. Generally all but the last in a list of conjuncts take that marker; and any case marking that governs the whole conjunction appears explicitly only on the final conjunct. (This is true as much for the cases that are marked with suffixes as it is for the other ones.) Thus ṭyṣmirriv=ky maryṣqaj=myh by fools and sages, for example. Conjuncts must be individually marked for number and definiteness; though often the final conjunct will be marked plural to reflect the plurality of the list, even if its own reference is actually singular.

Focus markers

There are three focus markers that cliticise onto the end of a noun phrase. Like the case markers, they take scope over the whole list when they follow a conjunction.

Here they are:

• =di only, precisely
• =gy all, also, even
• =(a)ra which, any

Coda

That's all for now. Maybe I'll be able to give some real examples once I've started work on verbs.

Vædty Qyṣ (verbs, first pass)

I seem to be enjoying this thing enough that I want to have a go at verbs and simple clauses.

I've managed to avoid mentioning so far that Vædty Qyṣ is pretty consistently head-final, though maybe that was obvious enough in the treatment of nouns. Anyway, it's consistently head-final, in its verb phrase as well.

And its verbs also tend to host a fair bit of associated functional material in various suffixes and enclitics. To be honest it'll probably mostly be fairly run-of-the-mill head-final agglutination, and that's certainly going to be how it starts.

The verbal template is going to look something like this:

stem - valency - TAM - negation - agreement = clitic

But I'm not going to approach this by filling in a template. I'd rather approach it bit by bit, preferably with some fun interactions between the bits. But I'm never going to come up with that sort of thing without playing around for a while, so I'm just going to start with some basics: a past/nonpast distinction, and things to the right of that. Hopefully that's enough for some playing.

The clitic

In fact I'll start at the end, with the clitic. It's obligatory whenever the verb occurs before pause, which might be always. The requirement is prosodic: intonatianal phrases will get a final boundary tone, but the verb itself can't host that tone because its stressed final syllable will always attract a phrasal pitch accent; so a non-stress-attracting clitic must step up.

I'm so far resisting the urge to let myself be inspired by Cantonese and allow a whole host of these sentence-final particles (which is another way of thinking about what they are). So far there are just seven of them.

The first three are shared with nouns; they represent focus of different sorts:

• =di only, precisely
• =gy all, also, even
• =(a)ra any

(With nouns (a)ra can also mean something like which or what, but I don't see how that would work with verbs. I don't actually remember what exactly I was thinking with that that initial a; I guess it'll surface after a consonant.)

Three of the particles represent the speech act:

• =ko, for assertions
• =nul, for polar questions
• =saj, expressing speaker authority

I'm going to leave these underdescribed here, but =saj could use a bit of clarification. A while back I was trying to come up with a simple evidential system, and thought it would be cool to have an evidential that you'd use when the very act of making a statement guarantees that the sentence of true. I was thinking of classic speech act cases like "I now pronounce you married," spoken by an appropriate authority in an appropriate situation. My initial reaction was that this was too much me being a speech act geek and too little something you'd expect to find in an actual language. Then it occurred to me that all I wanted was a particle that means hereby that gets treated syntactically as an evidential.

Anyway, that's roughly what =saj is doing here (though it's no longer obviously clasesd as an evidential). It expresses the speaker's claim to authority---not as an authoritative witness, but as an authority over the matters that the sentence is about. So it's a marker of certain sorts of illocutionary force.

If you're wondering about imperatives, I'm thinking that they'll usually get one of the focus particles. I suppose though that if there are imperatives of the "George will do the dishes" sort, they could take the hereby particle =saj.

The last of the verbal clitics marks indirect evidentiality (details another time):

• =næx they say

(I can already see though that the temptation will be strong to go the Cantonese route and just multiple these particles. But best hold off on that until I've got a language where I want to allow sentence-final particles to stack.)

Past, negation, agreement

So I said that for now I wouldn't go further with TAM than allowing a past/nonpast distinction. The past marker is kyr There's a nonpast marker that's just m, but so far it only surfaces when a buffer is required between vowels. But maybe it'll end up with a role in complex tenses or something.

The negation marker is vi.

So far it looks like past tense kyr and negative vi will be invariant. They certainly combine in the simplest way, yielding kyrvi. Since nonpast m so far only occurs before vowels, it doesn't combine with vi.

The verb is inflected to agree with its subject in person and number; unlike possessor agreement, there's no clusivity distinction in the first person. Some of the agreement markers take slightly different forms after the past and negation markers, and otherwise can differ depending on whether they follow a consonant or a vowel. Here's the system so far:

	Negative	Past	other C-stem	other V-stem
1s	-visu	-kyru	-u	-mu
1p	-visæm	-kyræm	-æm	-mæm
2s	-vid	-kyryd	-yd	-d
2p	-virri	-kyrri	-irri	-mirri
3s	-vi	-kyr	--	--
3p	-viti	-kyri	-ti	-ti

The m that shows up in the 1s, 2p, and 3p forms after a vowel represents the nonpast morpheme. I don't know what the s that occurs in the first person after negating vi represents.

Coda

I want to start putting together actual sentences, but I'll save that for the next post.

akam chinjir wrote: ↑Mon Aug 19, 2019 1:23 am I'm going to leave these underdescribed here, but =saj could use a bit of clarification. A while back I was trying to come up with a simple evidential system, and thought it would be cool to have an evidential that you'd use when the very act of making a statement guarantees that the sentence of true. I was thinking of classic speech act cases like "I now pronounce you married," spoken by an appropriate authority in an appropriate situation. My initial reaction was that this was too much me being a speech act geek and too little something you'd expect to find in an actual language. Then it occurred to me that all I wanted was a particle that means hereby that gets treated syntactically as an evidential.

Anyway, that's roughly what =saj is doing here (though it's no longer obviously clasesd as an evidential). It expresses the speaker's claim to authority---not as an authoritative witness, but as an authority over the matters that the sentence is about. So it's a marker of certain sorts of illocutionary force.

If you're wondering about imperatives, I'm thinking that they'll usually get one of the focus particles. I suppose though that if there are imperatives of the "George will do the dishes" sort, they could take the hereby particle =saj.

That's interesting, the =saj clitic is approaching then the idea of an egophoric but in the end follows a different principle. To clarify, are you stating that the clitic always track the authority of the present speaker? Based on your description, I'd guess that you don't have any anticipation of authority (unlike with prototypical egophorics) and so =saj would never appear in questions. What about =saj and reported speech then?

I'd also enjoy some example use cases for the focus clitics since it would be nice to know how narrow or wide focus duty they carry out. I'm guessing you are thinking more or less along the lines of contrastive focus, i.e. new information that's contrary to a previous supposition.

If you don't allow any particle stacking, what happens in sentences where more than one clitic would be applicable? Such a sentence would, for example, be a contrastive polar question:

"Did he even clean up his mess afterwards?"

Would it be =gy or =nul that wins in this case and how will the grammatical content of the omitted clitic be handled?

Thanks for the questions!

I think with egophorics, the issue is the epistemic authority of the speaker, and the issue is whether the listener will believe what the speaker says.

With =saj, I'm imagining it used in speech acts that the speaker needs authority to perform, and =saj is affirming that the speaker does in fact have that authority. And I think indirect reports would focus on what the speaker did, not on what they said.

Suppose X says: I now pronounce you married=saj.

That's an act of marrying, or some kind of pretense. The =saj indicates that the speaker has the authority to perform that action; and if they do have that authority, and the circumstances are otherwise right, the addressees end up married. (Or anyway that's how it goes in speech-act-example-land.)

If you wanted to report that, you could say, "And then X said that they were married." Here it wouldn't make sense to include the =saj within the scope of "said," I think. (I guess if your authority to utter assertions were in doubt, it might make sense to include =saj with wider scope.)

But actually if you were going to report this speech act, probably you'd say someting more like "And then X married them"---you'd report what they did, not what they said. (Or maybe, "And then X pretended to marry them," if X didn't really have the authority, or the circumstances weren't really right, or whatever.)

As for the focus particles...

I think I've somehow managed to avoid saying so far that Vædty Qyṣ is consistently head-final, which resolves one sort of scope issue: the focus particles always take scope over the clause at whose right edge they appear. And I'm hoping that the freedom of the verb's arguments to scramble will make it possible to distinguish between sentence-focus, predicate-focus, and verb-focus.

I try to avoid the expression "contrastive focus," since I personally have misunderstood it in a host of different ways. But, yeah, I'm thinking of focus as involving some sort of contrast with a contextually-salient set of alternatives; but I don't want any of them just to amount to a marker of new information. (Maybe sometimes they'll verge on miratives?)

Roughly speaking, I think you'd use =di only, precisely when you're indicating that the salient alternatives are false. E.g.:

mava-s   vor     -yn  hyrtag=di
Mava-DEF tree.rat-ACC hunt  =only
"Mava is only hunting tree rats"

Here the idea is: that's the only thing she's doing. You might say this, for example, to contradict someone who's just said that Mava is gathering firewood. Or if someone's concerned about the noise, and you want to assure them that not very much is happening.

=di can also attach to nouns, though, so if what you wanted to emphasise is that it's only tree rats that she's hunting, you'd say something like this instead:

mava-s   vor     -yn =di   hyrtag=ko
Mava-DEF tree.rat-ACC=only hunt  =DECL
"Mava is only hunting tree rats"

Contrasting with =di, you'd use =gy when to indicate that the salient alternatives are true, so that what you're not saying is true in addition to those other things.

mava-s   vor     -yn  hyrtag=ygy
Mava-DEF tree.rat-ACC hunt  =also
"Mava is also hunting tree rats"

As in: not only is Mava collecting firewood, she's also hunting tree rats. (And =gy can also attach directly to a noun phrase, with similar results as with =di.)

I'm actually less sure how =ana any will play out. I'm imagining it showing up in negative sentences and maybe in constituent questions, but I'm not entirely sure yet.

As for stacking---well, as I said, the temptation towards a system inspired by Cantonese is strong. But for now I think I'm going to try to hold my ground. That means just one of these clitics, the minimum you need to host a boundary tone.

As for which one you get, I think it's probably up to the speaker, but it'll depend on what else ends up going on in the language. Like, if there's a full adverb meaning also, then forgoing =gy in polar questions won't seem so bad; though if polar questions can also be unambiguously signalled by intonation, you could instead give up =nul.

(It may be trickier if nul also ends up as the marker for embedded questions, which seems likely.)

Thanks again!

akam chinjir wrote: ↑Mon Aug 19, 2019 9:18 am Thanks for the questions!

No problem. It's good fun thinking about the consequences of seemingly simple grammar design choices.

Yeah, gach, I definitely agree with that.

And here's me playing with some grammar choices.

Vaedty Qyṣ, some sentences

Okay, let's try some simple sentences.

mava-s   vor     -yn  hyrtag=ko
Mava-DEF tree.rat-ACC hunt  =DECL
"Mava is hunting tree rats"

This is SOV, which I take to be the basic constituent order. The direct object takes accusative case; this case-marking is not sensitive to specificity or definiteness or anything. The subject is third person singular, so there's no overt agreement marker on the verb. I take this to represent a lack of person and number features rather than the presence of a null morpheme, and gloss accordingly.

For the time being I'll assume that proper names require an explicit definiteness marker.

The verb is not marked for tense, so is interpreted as nonpast. There's nothing to indicate aspect. I think the main interpretive options are present progressive and immediate-future inceptive, but that could easily change. I think habituals will look like futures, morphologically speaking; anyway this wouldn't get a habitual interpretation.

I'll throw in ṣyn now to help probe possible constituent orders:

• ṣyn mavas voryn hyrtag=ko
• mavas ṣyn voryn hyrtag=ko (topicalised subject, higher than ṣyn)
• mavas voryn ṣyn hyrtag=ko (focused ṣyn, next to the verb)
• voryn ṣyn mavas hyrtag=ko (topicalised object; possibly also focused subject)
• voryn mavas ṣyn hyrtag=ko (topicalised object; focused ṣyn)
• *ṣyn voryn mavas hyrtag=ko (the object can't move anywhere higher than the the subject without also moving higher than ṣyn)

This is supposed to be scrambling of a sort that's utterly unremarkable in an SOV language. One unremarkable thing is the use of the position right before the verb for focus.

I guess you might think of the first of the orders as being basic, in the sense that it involves no discourse-driven movement, but the second is probably more common. Maybe it depends on whether you think of whole-sentence focus or predicate-focus as being more `neutral'. If you think that sort of issue matters in the first place.

An indirect object is I guess most neutrally placed before a direct object, if both are represented by full noun phrases, and a VP adjunct will go before an indirect object:

mava-s   kwety  -v  =t  =ara   mon -ys -ti  kero-n   ṭæjke-kyr =ko
Mava-DEF purpose-LOC=ALL=INDEF lake-DEF-DAT bone-ACC throw-PAST=DECL
"Mava threw a bone into the lake for no reason"

(Or: Mava was throwing bones into the lake... and other variants. kwetyvtara for for no reason is supposed to be a bit idiomatic; the same word presumably could mean for what purpose, why, though for an actual question I suppose it would have to occupy the focus position right before the verb: mavas monysti kerosyn kwetyvtara ṭæjkekyrgy? why did Mava throw the bone into the lake?)

More or less any permutation of direct object, indirect object, and VP adjunct can be achieved by scrambling, however. I think.

An indirect object will always take dative case. This is an easy way to distinguish selected indirect objects from adjuncts.

With an argument:

mava-s   gagul qygum -ys -ti  my-kyr =ko
Mava-DEF Gagur shrine-DEF-DAT go-PAST=DECL
"Mava went to Gagur shrine"

With an adjunct (taking allative case marking):

mava-s   gagul qygum -s  -yv -ty  my-kyr =ko
Mava-DEF Gagur shrine-DEF-LOC-ALL go-PAST=DECL
"Mava went towards Gagur shrine"

More or less, there'll be subtle distinctions of the sleep in the bed vs sleep in Toronto sort. (Except that the Vædty don't have beds or cities.) But I don't have more to say about that now.

The general rule is that a verb can take at most one object that'll get accusative case, and all other selected objects will get dative. Probably there'll be aa applicative or antipassive, maybe unproductive, to mix this up a bit.

Unsurprisingly a verb's arguments often aren't all represented by full phrases. First- and second-person subjects are fine represented only by agreement on the verb; a third person subject needs a pronoun. The third person forms that are used in this context come from the paradigm of emphatic pronouns:

1s	kwoj
1p	kwyrym
1+2	kwymki
2s	kwyd
2p	kwyrri
3s	kwys
3p	kwyt

This is just a hypothetical noun kwy marked for possessor agreement. Maybe kwy grammaticalised from something meaning body or nose or something.

Objects (both direct or indirect) can be represented by clitics immediately before the verb; probably these can also co-occur with topicalised objects. Here are those clitics:

1s	ho=
1p	m=
2s	d=
2p	rr=
3s	hi=
3p	ti=

These can double up if there's both a direct and an indirect object, with the direct object clitic closer to the verb; but if I ever feel like I understand person/case constraints there'll probably be restrictions on non-third-person direct object clitics in the presence of an indirect object.

There'll be two inflecting copula-like verbs, one that's used with nominal predicates (including any adjectives), and another that's used with locative predicates and in existential and possession clauses.

ṣy goes with nominal predicates (sorry, Mandarin speakers); it's obligatory in all tenses and so on. Normally it attaches to the predicate as a suffix:

mava-s   hyrtag-yda -gin  -ṣy =ko
Mava-DEF hunt  -NMLZ-ADJLZ-COP=DECL
"Mava is skilled at hunting"

Predicate adjectives but not predicate nouns agree in number with the subject:

gædi-s  -v  =ky  mava-tol   hyrtag-yda -gin  -i -ṣy -ti =ko
Gadi-DEF-GEN=COM Mava-PN.PL hunt  -NMLZ-ADJLZ-PL=COP-3pS=DECL
"Gadi and Mava are skilled at hunting"

(tol is an associative plural marker that can be used with proper names; it precludes the need for a definiteness marker. This is an example of the construction I mentioned in an earlier post, where the head noun of a conjunction gets plural marking to represent the plurality of the whole conjunction.)

sæw is the existential verb:

mava mon -ys -v  =yhyr sæw  =ko
Mava lake-DEF-LOC=LOC  exist=DECL
"Mava is at the lake"

As you can see, this is a full verb.

Naturally you can topicalise the location, especially with an indefinite subject, resulting in an existential sentence proper:

mon -ys -v  =yhyr tyqæṣ    sæw  =ko
lake-DEF-LOC=LOC  campfire exist=DECL
"There is a campfire by the lake"

It's not obvious here, but the verb still agrees with the subject, which is still unmarked for case, which is presumably to say that it's nominative.

With an animate `location', the location argument does not need the long locative case, and can simply take the -v ending, which in this context I'll gloss as marker of genitive case:

mava-s  -yv  hyne -j  kero-w  sæw  -ti =ko
Mava-DEF-GEN INDEF-PL bone-PL exist-3pS=DECL
"Mava has some bones"

Notice that the verb agrees with kerow bones, and that kerow is not marked to agree with a possessor---mavasyv hynej kerow of Mava some bones does not constitute a single noun (or determiner) phrase or a single argument of the verb.

The correct analysis of these constructions might end up being fairly subtle. Right now I'm thinking of the possessor as a topicalised complement of the verb, not a subject, and presumably that'll affect things like its position relative to certain adverbs and maybe its availability to questioning, relativisation and so on. But I can't now say whether I'll want to play with that sort of thing.

A possible contributing factor is that Vædty Qyṣ might end up with other verbs that take apparent genitive (or dative) subjects, and they might also fail to trigger agreement on the verb, and they might or might not have restricted possibilities in questions, relative clauses, and so on. We'll see. (If I continue enjoying this language!)

I'm actually going back-and-forth on the question of whether the ṣy copula should be a suffix or a clitic. I've actually already edited the post on just this point; for the time being, it's a suffix. The relevant issues involve interaction with stress and final clitics, and the use of focus clitics especially with nominal predicates. For the moment it looks like having it be a suffix avoids some mess, so I'll stick with that.

One consequence of this, I'll say, is that a noun functioning as a predicate can't take a focus enclitic; those can only occur predicate-finally.

So for it's only a tree rat, you'll get:

kwys vor     -ṣy =di
3s   tree.rat-COP=only
"It's only a tree rat"

And a polar question will look (unsurprisingly) like this:

kwys vor     -ṣy =nul
3s   tree.rat-COP=Q
"Is it a tree rat?"

Vædty Qyṣ, the future

Oops, first some housekeeping. I was taking it for granted that the y in the copula ṣy would always surface. I think the best way to guarantee that is to say it's underlyingly ṣi, with conflation of i and y after retroflexes. Then the copula will have a si allomorph after alveolar consonants.

But, the future.

I'll start with a marker, -sæni (just -æni after obstruents). The subject agreement markers will treat this just like a regular vowel-final stem, which is to say that the 1s, 1p, and 2s markers will include the -m- that I said represents nonpast. I'll assume that this marker is semantically inert here: nonpast + future = future.

So it's going to look like this:

hykwad-ys  va  -ris      -mas-kas-yv  gym       -do   -sæni=ko
sun   -DEF come-RSLT.PTCP-NEG-day-LOC give.birth-ACAUS-FUT =DECL
"The sun'll come out tomorrow"

This is meant to be a straight prediction, so the declarative clitic =ko is appropriate. To hedge a bit, you instead could use indirect evidential =næx.

In first person futures, =saj will often be appropriate. Remember that in effect it expresses the claim that the speaker has the authority to perform the action they're performing with the sentence in question. In first person futures, it might suggest that the speaker is undertaking a commitment of some sort:

hi =qorr-yl  -sæni-mu =saj
3sO=die -CAUS-FUT -1sS=ILLOC
"I'll kill it"

=saj can also go with second-person plurals, often giving the utterance imperative force:

vor     -s  -yn  dmys-mas -tu-sæni-d  =saj
tree.rat-DEF-ACC skin-PRIV-TR-FUT -2sS=ILLOC
"You will skin the tree rat"

Non-imperative second-person futures will often get =næx, officially an indirect evidentiality marker, but I can already feel it broadening in the direction of Mandarin ba 吧 or Canadian eh.

vor     -s  -yn  dmys-mas -tu-sæni-d  =næx
tree.rat-DEF-ACC skin-PRIV-TR-FUT -2sS=IND
"You'll skin the tree rat, eh?"

I said earlier that habitual forms will look like futures. I've decided I'll use a participle for that, and participles are a big topic that I won't go into here.

I think in principle it should be possible to put the future marker inside the scope of the past marker, for a future-in-the-past meaning:

gædi qwy -va  -sæni-kyr =gy
Gadi back-come-FUT -PAST=FOC
"Gadi would not return"

The core use of this combination is for events that are now past, but are future relative to the topic time.

Vædty Qyṣ, some diachrony

It looks like Vædty Qyṣ is going to end up at laest substantial enough for Akiatu's ancestors to acquire vocabulary from it. And since Akiatu's been in search of a verb-final language to give it some verbalising morphology, it's looking like Vædty Qyṣ is the donor. So maybe Vædty Qyṣ should have its own historical depth.

Anyway that's what this post is about. Mostly it's sound changes, though I'll also show how the sound changes give me about the plurals I want.

The sound changes are kind of dull, mostly unconditional, but they result in fairly drastic superficial changes (and a few not so superficial ones). My current thinking is that grammar will mostly not seem to change so much. (A factor is that the sound changes are boring enough that even without much analogy they wouldn't substantially disrupt the languages's agglutinative tendencies.)

Let's say the time depth is something like 500 years.

Qɨsə basics

Qɨsə (that's what I'll call the parent language) has a somewhat simpler inventory than does its Vædty descendent:

m	n
p	t	k	q
b	d	g
	s
	ɾ
	r

That's to say: no retroflex or labialised consonants, and there's a genuine series of voiced plosives; and there's p but no h.

i	ɨ	u
ɛ		ɔ
	ɐ

There's contrastive length on all vowels. Epenthetic ə is also common, it's inserted to preserve a simple syllable structure.

In transcribing historical forms I'm going to follow the IPA strictly, forgetting the conventions I've been using with Vædty Qyṣ itself. I'm also not going to worry about when r acquires its [r~l] allophony, maybe it's already there in Qɨsə.

Syllables can be CV(C) or CrV(C); the first consonant in a complex onset must be a voiceless obstruent, and a coda consonant must be resonant (including ɾ). I'll write the rhotic in complex onsets as r, but there might be morphophonology that erases the ɾ/r distinction in this context.

Phonotactics are enforced by insertion of ə, scanning from the right, much as happens in Vædty Qyṣ itself, though the rules are a bit different. (The differences: allowing Cr onsets but disallowing FC ones, and disallowing coda fricatives.)

I don't think there'll be major differences in prosody: final stress, unstressable clitics, boundary tones, and pitch accents. (Which is not to say I've worked out the actual melodies, or that they couldn't change.)

Sound changes

First are a bunch of changes that upset the consonant system somewhat:

• p → b / V_V (including between epenthetic vowels)
• p → f elsewhere
• fr → fʷ
• tr → ʈ
• sr → ʂ
• kr → kʷ
• qr → qʷ

At this point we start getting interactions between the retroflex consonants and the uvulars. I won't try to figure out the details, but there's one change that'll probably be fully general (since it's needed to derive the name of the language!):

• t s → ʈ ʂ / qə_, qʷə_

The ə in the context will always be epenthetic, so you could take it to represent the lack of an underlying vowel between the consonants.

Possibly overlapping with that first set of changes is a long pull chain shift among the short vowels.

• ɐ → ɑ
• ɛ → æ
• i → e
• ɨ → i (except after retroflex consonants)
• u → ɨ / _C[coda], C[labialised]_, C[retroflex]_, q_, _q

At this point, vowel-harmonising shifts of suffixed ɑ to æ can get started. Accompanying this, you'll get sporadic æ → ɑ in suffixes. (Which is to say, you'll increasingly get æ in suffixes _only_ as a result of harmony.)

Next you get merger of the nonlow central vowels:

• ɨː → ɨ
• ə → ɨ

The most significant part of this change is that, like former ə, ɨ and former ɨː now get reanalysed as epenthetic. This means that where they are not phonotactically required, they get dropped. Note that this takes place after obstruent+rhotic clusters have finished changing into no phonemes: Qɨsə phonotactics no longer allows onset clusters (but it does still allow resonant coda consonants).

Now it's time for the voiced fricatives.

• fʷ → v
• b → v
• d → ð
• g → ɣ

Starting with these changes, syllable structure can begin to shift to allow both coda fricatives and fricative + plosive onsets. Given how many vowels are epenthetic, this can quickly affect the syllabification of many words. E.g., with this change, there's no longer any need for an epenthetic vowel in (a hypothetical word) vɨka, so it will become vka.

The last change among the consonants is the loss of remaining f (which should only occur morpheme-initially):

• f → h

Finally, vowel length gets lost:

• iː → i
• uː → u
• ɛː → e
• ɔː → ɔ
• ɐː → ɑ / q_
• ɐː → æ elsewhere

Here's how these changes apply to prɛdɨtə qɨsə:

• → prɛdətə qəsə (merger of short ɨ with ə)
• → frɛdətə qəsə (p→f)
• → fʷɛdətə qəsə (fr→fʷ)
• → fʷɛdətə qəʂə (s→ʂ/qə_)
• → fʷædətə qəʂə (ɛ→æ, the vowel chain shift)
• → fʷædɨtɨ qɨʂɨ (raising of epenthetic vowels)
• → væðɨtɨ qɨʂɨ (fʷ→v, d→ð)
• → væθtɨ qɨʂ (more complex syllables now allowed, some epenthetic vowel are lost)
• → Vædty Qyṣ (Vædty Qyṣ orthographic conventions)

Plurals

The description I gave of plural formation was meant to hint at historical depth. Now I'm in a position to fill in those hints a bit.

In the parent language Qɨsə, there are two main strategies for forming plurals.

The first lengthens the final vowel in the stem, and appends a short echo vowel if the stem is consonant-final, and otherwise appends nonsyllabic u after a back vowel and nonsyllabic i otherwise. On the way to Vædty Qyṣ, analogy keeps the echo vowels in harmony with the lengthened stem vowels, which in turn do not undergo the chain shift, resulting in alternations.

For example:

• prɨkɨtə tree → prəkətə → frəkətə → fʷəkətə → fʷɨkɨtɨ → vɨkɨtɨ → vkɨtɨ <vkyty>
• prɨkɨːtɨ trees → prəkɨːtɨ → prəkɨtɨ → frəkɨtɨ → fʷəkɨtɨ → fʷəkiti → fʷɨkiti → vɨkiti → vkiti <vkiti>

The result are a good number of words in which the vowel alternates in the plural, much as I described in the earlier post.

The way analogy affects these plurals is going to be complicated. There'll definitely cases in which nouns with alternating stems settle on a single vowel quality for both singular and plural forms. But the alternations will also be common enough to be the basis for analogy, so there'll also be nouns that acquire a change in vowel quality in the plural. It'll be a bit messy, with room for lots of variation amoung speakers/bands, but nothing terrible.

The second way to form plurals in Qɨsə is by means of certain augments. These may go back to a noun class system or something, but in Qɨsə itself mostly play a role in noun→noun derivation. Some of these have a collective or plural or indefinite sense, and some nouns form their regular plural in part by using one of these augments. Moreover, the augments themselves are mostly of a CV shape, and mostly add nonsyllabic u or i as a plural marker, without any lengthening of vowels.

I think in Qɨsə itself there were a few of these augments that showed up semiregularly in plurals, but the system got simplified a bit, and only -kɐ survives into Vædty Qyṣ. In actual plurals, it would be -kɐi̯, which in Vædty Qyṣ ends up as kɑi̯ <kaj>. Meanwhile, whatever non-plural uses -kɐ had in Qɨsə have mostly been lost, so kaj as a whole comes to be thought of as the plural marker.

A consequence. As I work on Qɨsə more, I'll probably end up with some semantic classes within which nouns tend to get the kaj suffix. We'll see.

Coda

That's mostly it for now. Some little thoughts about how this affects other things I've covered:

• I designed the case system with the thought that different cases and postpositions were grammaticalised to different extents, and now I've got more of a framework in which to think that through. One possibility is that one or more of the noun augments I mentioned above could have gotten incorporated into the case system, though I don't know how well that would play with my decision to make the noncore case markers clitics.
• I have no plans to derive the agreement markers from full pronouns, at least until such time as I go further back in the history of the language. Maybe though the subject agreement suffixes, the possessor agreement suffixes, and the object proclitics will look more similar to one another in Qɨsə.
• The Vædty Qyṣ copula looks a lot like its definiteness marker. This is actually a coincidence, but it looks like it might be possible to give them a common origin, with the definiteness marker being sə and the copula being sri, with the shared s maybe having a demonstrative sense.
• I don't currently have any useful ideas about the past tense marker, negation, or the various clitics I've discussed. Or the future. I think at least the past tense marker and the negator are very old. But my plan is to focus next on participles and other somewhat deverbalised forms, and there'll be room for Vædty-specific innovations there.
• I guess Vædty itself is a name for one of the regions where Qɨsə was spoken or to which its speakers migrated. There'll be other such regions, and thus other Qɨsə languages. One of these days I'll have some fun doing comparative work in this little language family.

Edit. I ended up doing a bunch of edits. Some were just to fix dumb mistakes. The biggest problem, though, was that I'd ordered various changes so that Vædty Qyṣ i could only derive from a Qɨsə long vowel, and I decided I didn't like that, so I reordered things a bit drastically.

Also, the more I play with things in a diachronic context, the more it looks like I'm going to want to adjust things. No surprise there, I guess.

Vædty Qyṣ, the resultative participle

Yeah, this is going to be one of those SOV languages with a host of participles and converbs and nominalisations and so on. Here I'll just talk about one of them, the resultative participle.

In case it matters, I'm assuming that a participle is any verbal form that functions as an adjective, particularly as an attributive adjective, which is to say, heading a sort of relative clauses.

Morphology

The Qɨsə form of the resultative suffix is riːs, often requiring an epenthetic ə before or after it or both. This mostly becomes rris is Vædty Qyṣ.

The big issue with this suffix is that regular sound changes will have this merge with a preceding plosive or s to form v ʈ ʂ kʷ qʷ. As far as I can tell these alternations will be completely regular, so I'll say it's preserved into Vædty Qyṣ. Further, when this results in ʈ ʂ, the s of the suffix will harmonise to ʂ.

Meanwhile the long iː will end up as i.

I'm also going to assume a sporadic change that otherwise deletes the suffix-initial r(r) whenever it does not follow a non-epenthetic vowel. That's to say, the suffix will become is rather than trigger vowel epenthesis before it.

Usage

Here's an example:

va  -rris     -i  gyl   -ka-s
come-RSLT.PTCP-PL person-PL-DEF
"the people who've come"

You can see that the resultative participle, like adjectives in general, agrees in number with the modified noun; it does so with an echo vowel, yielding -isi or -rrisi.

In the example, gyl person is modified by va come. gyl corresponds to the implied patient or theme of va, and the meaning is that the va-ing has reached its natural result: the people have come.

This sort of participle forms most naturally from unaccusative verbs, verbs that take a patient or theme argument but not an agent. Here are some examples:

• hækwis sweaty (< hæky to sweat)
• qorris dead (< qol to die)
• ṣṭujrris lying down (< ṣṭuj to lie down)

Some of the verbs that can form these participles have uses where you might think of them as unergatives. For example, ṣṭuj lie down can describe someone intentionally lying down from a standing position. What matters is that the verb clearly picks out a sort of result state. The result state picked out by ṣṭuj is one in which the person has lain down completely, for example.

Transitive verbs normally cannot form this participle directly, but need a valency-reducing operation first. This will be an anticausative rather than a passive:

• hyrtagysæṭyṣ hunted (< hyrtagysæty to be hunted < hyrtag to hunt)
• qwyṭymtynis cooked (< qwyṭymtyn to cook (intr) < qwyṭym to cook (tr))

Note that hyrtagysæṭyṣ hunted describes something that has been hunted and caught or killed, not something that is being hunted; this nuance is probably clearer with qwyṭymtynis cooked. Also, I don't know how to convey the nuance of hyrtagysæty to be hunted in English: unlike a passive, it doesn't imply an agent, and it could not occur with the Vædty Qyṣ analog of a by-phrase.

Some transitive verbs can directly form resultative participles, though this isn't easy to predict and the resulting sense can be idiosyncratic. Here's an example:

• gymis having begun, alive (< gym to give birth to)

Unsurprisingly, maybe, consumption verbs behave a bit distinctively in these participles:

• tærris having eaten (< tæ to eat)

This participle is oriented towards the person who ate, the affected agent, rather than the thing eaten, which is what you'd normally think of as the patient. (These verbs also have distinctive behaviour with causatives, tæl is to feed, to make eat not to make be eaten, though the latter is what you'd expect given how causatives usually work in Vædty Qyṣ.)

Resultative participles, and in fact participles in general, form their negation with -mas rather than the regular verbal -vi:

• qwyṭymtynismas uncooked, raw
• tærrismas unfed, not having eaten, hungry

Additionally, from va come and my go, together with the bound form kas day:

• varriskasys today, the day that has come
• varrismaskasys tomorrow, the day that has not come
• miskasys yesterday, the day that has gone

Adding in the nominaliser -myn:

• varrismasmynys the future, what has not come
• mismynys the past, what has gone

In fact -myn occurs productively with resultative participles:

• gymismyn the living, those who have been born (< gym to give birth to)
• qorrismyn the dead (< qol to die)

Resultative participles can take dative complements, but not locative or other adjuncts, and only a very limited set of adverbs.

With a source argument (which can take dative case rather than ablative as a complement of the verb va come):

gagul qygum -ys -ti  va  -rris     -i  gyl   -ka-s
Gagur shrine-DEF-DAT come-RSLT.PTCP-PL person-PL-DEF
"the people who've come from Gagur shrine"

With the adverb væw well, thoroughly:

væw  qwyṭym-tyn  -is        vor     -ys
well cook  -ACAUS-RSLT.PTCP tree.rat-DEF
"the well-cooked tree rat"

These participles can occur predicatively as well as attributively, using the regular copula ṣi (which will always have its si allophone in this context):

gædi   qorr-is       -ysi=næx
Ghathi die -RSLT.PTCP-COP=IND
"I guess Ghathi is dead"

(In previous posts I romanised the name gædi as Gadi in my translations, but it's really something like [ɣæˈði], and I suppose Ghathi is better.)

In such sentences, normally you'll negate the copula rather than the participle:

mig  vor     -ys  væw  qwyṭym-tyn  -is       -ysi-vi =ra
PROX tree.rat-DEF well cook  -ACAUS-RSLT.PTCP-COP-NEG=INDEF
"This tree rat is not well-cooked"

And that's as good a thought to end with as any

Coda

This participle is meant to resemble patient-oriented participles in many languages, such as English's past participle; crosslinguistically, these have a strong tendency to have resultative senses. There are two distinctive points that I want to highlight.

First is the fact that these participles can't be formed directly from most verbs that can assign a theta role to an external argument (that is, transitive verbs, unergatives, and, I'll assume, passives).

The second point is how `small' they are, particularly the fact that they can't occur with most adverbs. I'm assuming that the participle marker selects a VP complement (as opposed to vP), and is thus incompatible with adverbs that are not strictly VP-internal. Now if only I had a clearer sense of which adverbs those are; I assume that væw well, thoroughly will count, as might other adverbs that clearly characterise the result rather than manner; but I'm hoping it's reasonable to say that these participles can't occur with manner adverbs, for example, or with place/time adverbs.

Vædty Qyṣ, verbal updates

This is meant to give Vædty Qyṣ verbs a bit more depth. It substantially revises verb conjugations in the present, past, and future tenses.

1. The simple tenses

There are two tenses that I'll call simple: the present and the past. Both go back to Qɨsə (though in Qɨsə it was a nonpast rather than a present), and in Qɨsə they used the same markers of subject agreement:

	S	Pl
1	-u	-ɛm
2	-bə	-bəsi
3	--	-si

Meanwhile, the nonpast was marked by -m and the past by -krɛ.

We need the following bits of Qɨsə morphophonology:

• The nonpast tense -m assimilates to -n before the 3p marker -si. (There is no general process of nasal assimilation in Qɨsə, though.)
• Past tense -krɛ loses its final vowel before the vowel-initial first person forms, in which the vowel lengthens, yielding -kruː and -krɛːm.

There are also a few irregularities on the way to Vædty Qyṣ:

• Nonpast -m survives only in the present tense forms, and only after a vowel-final stem; that's to say, it survives only as a buffer between (nonepenthetic) vowels.
• In the 3p nonpast, the s of the agreement marker fortifies to t, -nsi → nti. (Maybe this should be a regular sound change, or even a regular process within Qɨsə.) The n subsequently drops, leaving -ti.
• The s also fortifies in the 2p nonpast: -mbəsi → -bəsi → -bəse → -vəse → -vse → -vte. This fortification definitely isn't a regular sound change, since it doesn't happen in the past tense, whose marker remains -kwavse. I'll assume that in both tenses the 2p forms are influenced by the 3p forms.
• I have kru → kwu → kwɨ as a regular change, but so far kruː just ends up as kwuː, which seems wrong; I'll assume that rounding comes to nothing before long u, and the result is just kuː.
• I'll assume æ → a / _ v[coda], either as a late sound change or as a process within Vædty Qyṣ. To reconcile this with Vædty Qyṣ's sporadic vowel harmony, I'll say this doesn't happen with a front vowel in the next syllable.

I'll demonstrate all this with the conjugations of two verbs, prɐ → va to come and mə → my to go, giving both Qɨsə (Q) and Vædty Qyṣ (VQ) forms. (It might help to remember that VQ y represents epenthetic ɨ.)

The nonpast/present:

	prɐ to come (Q)	va to come (VQ)	mə to go (Q)	my to go (VQ)
1s	prɐmu	vamu	məmu	mu
1p	prɐmɛm	vamæm	məmɛm	mæm
2s	prɐmbə	vav	məmbə	myv
2p	prɐmbəsi	vavte	məmbəsi	myvte
3s	prɐm	va	məm	my
3p	prɐmsi	vate	məmsi	myte

The past:

	prɐ to come (Q)	va to come (VQ)	mə to go (Q)	my to go (VQ)
1s	prɐkruː	vaku	məkruː	myku
1p	prɐkrɛːm	vakwem	məkrɛːm	mykwem
2s	prɐkrɛbə	vakwav	məkrɛbə	mykwav
2p	prɐkrɛbəsi	vakwavse	məkrɛbəsi	mykwavse
3s	prɐkrɛ	vakwæ	məkrɛ	mykwæ
3p	prɐkrɛsi	vakwæse	məkrɛsi	mykwæse

In the two simple tenses, Vædty Qyṣ thus ends up with the following paradigm:

	Nonpast	Past
1s	-(m)u	-ku
1p	-(m)æm	-kwem
2s	-v	-kwav
2p	-vte	-kwavse
3s	--	-kwæ
3p	-te	-kwæse

It's gotten a little bit more complicated, and I hate it a little bit less.

2. -niː forms

Besides the simple tenses, the Qɨsə verb had at least three forms built using the formative -niː. The one that I'm most concerned with uses -sɛːn+niː, because that's the one that turns into a future tense in Vædty Qyṣ.

(It's not a future tense in Qɨsə. In fact Qɨsə has no dedicated future tense, for future time reference you'd use either the nonpast or a modal construction.)

-niː combines directly with the agreement markers, resulting in the following forms:

	S	Pl
1	-niju	-niːmɐ
2	-niːbə	-niːbəsi
3	--	-niːsi

Unsurprisingly, -sɛːn + -niː results in -sɛːniː, dropping an n. I also want this to become sɛniː, with shortening of the first vowel, in all forms in which the i remains long (that is, all forms except for 1s); and this needs to happen before the chain shift among the short vowels. For now I'll assume this occurs at a point when sɛːniː is reanalysed as monomorphemic, assuming a rule banning multiple long vowels in a single morpheme.

There's a further complication. I've decided that this will end up as one of a small number of bisyllabic suffixes that leave secondary stress on the preceding syllable; primary stress will fall as usual on the word's final syllable. (So I'm not quite cloning Turkish stress, so far.)

Anyway, here's prɐ → va to come in this conjugation (showing stress in the Vædty Qyṣ forms):

	prɐ (Q)	va (VQ)
1s	prɐsɛːniju	vàsenejú
1p	prɐsɛːniːmɐ	vàsænimá
2s	prɐsɛːniːbə	vàsænív
2p	prɐsɛːniːbəsi	vàsænivsé
3s	prɐsɛːniː	vàsæní
3p	prɐsɛːniːsi	vàsænisé

I'll reiterate that the Qɨsə forms don't constitute a future tense. The -niː verbs forms had primarily modal significance; you could think of them as different types of subjuntives. I'll talk about this more when I post about the `modal' participle in Vædty Qyṣ, which is built using -sɛn but without -niː.

There's another -niː form, -po + -niː, that's the source of a Vædty Qyṣ imperative.

And verbs just in -niː could be used in the complements of some prepositions, such as rroqə after. There might still be relics of such constructions, but for the most part they've been taken over in Vædty Qyṣ by converbs.

One last thing. In the post about the future tense, I mentioned the possibility of combining future tense -sæni with the past tense marker to get a future-in-the-past sense. I kind of want to do this instead by deriving from a Qɨsə modal form that joins -sɛːn with the past tense marker; the result in Vædty Qyṣ would be -senkwæ (-senku, -senkwem...). But so far I can't see how to do that and at the same time have -sɛːn derive a participle.

Vædty Qyṣ, the modal participles

Alongside its future tense in -sæni, deriving from Qɨsə sɛːn+niː, Vædty Qyṣ has a participle in -sen, from Qɨsə sɛːn. I'm going to assume that Qɨsə -sɛːn already produced a participle, and that the forms with and without niː were near equivalent in the range of modal senses they could convey. But the two diverged on the way to Vædty Qyṣ.

The form with niː (which I guess was some sort of copula) ended up as the regular future tense, bleached of all specifically modal meaning (unless you think that futures are inherently modal, but that's a conversation for another day).

The participle ended up with three main uses.

First, habitual:

mava-s   vor     -yn  hyrtag-sen -si =ko
Mava-DEF tree.rat-ACC hunt  -PTCP-COP=DECL
"Mava hunts tree rats"

(Compare mavas voryn hyrtag=ko Mava is hunting tree rats.)

The syntax here is just what we saw with the resultative participle, the copula suffixed to the participle; in the present tense and with a third-person subject, there's no further inflection.

Here's a past tense example with a plural subject:

gædi  -s  -v  =ky  mava-tol   vor     -yn  hyrtag-sen -e -ṣy -qwæ -se =ko
Ghathi-DEF-GEN=COM Mava-PN.PL tree.rat-ACC hunt  -PTCP-PL-COP-PAST-3pS=DECL
"Mava and Ghathi would hunt tree rats"

There's plural agreement both on the participle itself and in subject agreement on the copula. The translation uses "would" for a past habitual, it seems the best choice in English. The meaning is not the same as "used to", since it implies no contrast with the present.

The second sense these participles can have is something like an immediate future, with meanings like about to, on the verge of, ready to. Despite the label, the sen participle is best interpreted as describing something's current state: it's more about current readiness or potential than about future fact. (Whereas the future tense is supposed to be entirely about future fact.)

In theory the sentences I've already given as examples could be translated with this immediate-future meaning, though I think it's unlikely they'd be used that way. With a definite object or specific time reference, though, the habitual reading wouldn't be available, and the immediate future is what you'd get.

Specific time reference:

ṣyn mava-s   vor     -yn  tæ -sen -si =ko
now Mava-DEF tree.rat-ACC eat-PTCP-COP=DECL
"Now Mava is about to eat tree rat"

Definite object:

mava-s   vor     -ys -yn  tæ -sen -si =ko
Mava-DEF tree.rat-DEF-ACC eat-PTCP-COP=DECL
"Mava is about to eat the tree rat"

Heck, even a nonspecific object with an imprecise specification of amount:

mava-s   hynæ vor     -yn  tæ -sen -si =ko
Mava-DEF some tree.rat-ACC eat-PTCP-COP=DECL
"Mava is about to eat some tree rat"

More or less: these sentences are anchored in a way that rules out a generic, habitual interpretation. Or anyway that's the idea.

There's a third use of this participle, in the consequent clause of conditional statements. I'll talk about that another time.

Of course this is a participle, so it can also be used attributively:

vor     -yn  hyrtag-sen -e  gyl   -kaj
tree.rat-ACC hunt  -PTCP-PL person-PL
"people who hunt tree rats"

And it can also take the -myn nominaliser:

qol-sen -myn -kaj
die-PTCP-NMLZ-PL
"those about to die"

The modal -sen participle is `bigger' than the resultative -rris participle, syntactically speaking. As my examples have shown, with transitive verbs it licenses an accusative object, and it can occur with manner adverbs. However, I think it's limited to VP/vP adverbs, and it definitely can't take speaker-oriented adverbs with meanings like certainly or unfortunately. Also, it forms its negation with mas rather than with vi. (But of course when it's used predicatively, the clause as a whole can have adverbs of all sorts, and will be negated with vi.)

Vædty Qyṣ's other modal participle, the -vo/-ho participle, has about the same syntax. It has meanings that get expressed in English with "can," "should", or "must," but unlike these English auxiliaries it can express only root and not epistemic modality.

For example:

mava tæ -vo  -ṣy =qo
Mava eat-PTCP-COP=DECL
"Mava must eat"

The meaning here must be that Mava has a need or obligation to eat, not that it must be the case that Mava is eating.

The choice of the -vo or the -ho allomorph is straightforward: it'll be -ho after a nasal or liquid, otherwise -vo. The Qɨsə form was -po, the p becoming b→v between vowels and f→h after a coda, which could only be m n ɾ r, and nothing funny happens to these codas. (Hmm, though maybe I should change that. Maybe something like r+po → rfo → rho → so? I'll think on it.)

I don't know enough yet about modals that don't distinguish can, should, and must, so I can't go into detail about how this participle will get used with exactly what sense. But let me have a go at it.

Just now I translated mava tævoṣyqo as Mava must eat---with must rather than can or should. This was mostly because I thought her need to eat is much more likely to be salient than her ability to eat; but of course there might be a situation in which it's precisely her ability to eat that's at issue, and in such a situation a different translation would be warranted.

One thing. Mava must eat but she won't eat is a bit strange, but the following is not at all strange:

mava tæ -vo  -ṣy =qo,   kwys tæ -sæni-vi =di
Mava eat-PTCP-COP=DECL, 3s   eat-FUT -NEG=only
"Mava should eat, but she won't eat"

One sort occasion in which you'd often find this participle is in statements assigning or reporting responsibilities:

gadi   gagul qygum -ys -ti  my-vo  -ṣy =saj
Ghathi Gagur shrine-DEF-DAT go-PTCP-COP=ILLOC
"Ghathi will go to Gagur shrine"

The =saj here indicates that this statement actually assigns the responsibility, and does not merely report it.

The English translation there could have been Ghathi is to go to Gagur shrine, and I think it'll quite often be true that where English allows a modal infinitive of that sort, Vædty Qyṣ will allow the -vo participle. One difference is that English allows these infinitives to take an unmarked passive sense: Mava is the one to ask. Vædty Qyṣ would require an explicit valency change here:

mava=di   gæ -ma  -so  -si =ko
Mava=only ask-PASS-PTCP-COP=DECL
"Mava is to be asked"

(Apparently I have in fact decided that rh → s, since the passive marker here is -mar. I'll say this is a regular sound change. Note that this is just the alveolar tap, not the trill, which I transcribe as rr.)

Or you could do something structurally maybe more similar to the English using a myn nominalisation:

mava=di   gæ -ma  -so  -myn -si =ko
Mava=only ask-PASS-PTCP-NMLZ-COP=DECL
"Mava is the one to ask"

Another place you'll find the vo participle is in the antecedents of (at least) counterfactual conditionals.

Oh, and in the complements of nouns like harro obligation:

mava-s  -yv  hyrtag-yrr-ho   harro      sæw  =ko
Mava-DEF-GEN hunt  -AP -PTCP obligation exist=DECL
"Mava has an obligation to hunt"

And no doubt in lots of other places.

Er, and for purposes of agreement it forms its plural with -w.

Vædty Qyṣ, focus and aspect

It's just occurred to me that maybe the main way that perfective aspect will get marked in Vædty Qyṣ is with the focus particle =gy and, also, even.

To be frank, the idea that perfectivity has to do with thinking about the event as a whole, without concern for its internal temporal structure, has never seemed particularly helpful to me. In part because it's often false: with resultative constructions, which are often perfective, you're very much concerned with temporal structure.

I find it a lot more helpful to think of it in terms of sequencing and foregrounding; right now especially foregrounding. The basic idea is that if you've got two conjoined clauses, one imperfective and one perfective, probably the imperfective clause provides the background for the perfective one. (I'm pretty sure this way of thinking about it first became clear to me from something William Annis once said on Conlangery, fwiw.)

Anyway it seems to follow that a lot of what you do with perfectivity you could do instead with focus marking, and it's just occurred to me that that's a natural way to use Vædty Qyṣ's postverbal clitics.

In particular the =gy clitic seems like it would work here. It can carry over into translations as "and, also, even." Roughly, it indicates that the speaker intends to add to a conversationally-salient body of presuppositions, whose truth the speaker is taking for granted.

I think I'm going to help this along by introducing another postverbal particle, =wo, which I might translate as "as" or "of course," and which marks the sentence as something the speaker intends to take for granted.

So something like this could have a background-then-foreground interpretation:

madi -s   vor     -yn  hyrtag-kwæ =wo, gadi  -s   tyqæs   -yn  vadyl -kwæ =gy
Mathi-DEF tree.rat-ACC hunt  -PAST=as  Ghathi-DEF campfire-ACC handle-PAST=also
"While Mathi was hunting tree rats, Ghathi set a campfire"

Of course other interpretations are be possible. But for the time being I'm going to say that it's in this part of the grammar that Vædty Qyṣ comes close to grammaticalising perfectivity.

I seem to have run out of steam with Vædty Qyṣ.

Here's an inventory I'm liking at the moment. Let's say for now that the language is called Nðɑħɑɑ̯.

lab	dent	alv	alv+phar	postalv	vel	uv	phar	glot
m	n̪ð	n	nˤ	ɲ	ŋ	ɴ
p	t̪θ	t	tˤ	t̠ɕ	k	q		ʔ
b	d̪ð	d	dˤ	d̠ʑ	g
	θ	s	sˤ	ɕ		χ	ħ	h
β~ʋ	ð~l	z~ɹ	xˤ~ɹˤ	ʑ~j		ʁ~ʁ̞	ʕ~ʕ̞

The Akiatu river basin needs some early settlers to seed a linguistic area in which some of my favourite things can be overrepresented, maybe this is it (at least for ʋ θ q).

A while back (not on this thread) I posted a couple of things about a language (or family) called Iqę́hhǫ, but backed off because I decided almost immediately that I needed to work out the history a bit more. Maybe Nðɑħɑɑ̯ will turn out to be the consonant inventory of some fairly distant ancestor of Iqę́hhǫ's. It has to lose quite a bit, and gain some things, but nothing too dramatic, given that there's probably a few thousand years to play with.

Okay, some things about the inventory.

I'm only just starting to do serious reading about pharyngeals and pharyngealisation, but I think what I have here is fairly unexceptional. My main question is whether there should be another pharyngealised (emphatic) series.

n̪ð is pretty weird, I think unattested, and maybe I'll start to hate it once I've live with it a bit longer. For the moment, though, I'm a bit infatuated. As you can see, Nðɑħɑɑ̯ uses affrication to reinforce place contrasts among the coronal plosives: t̪ð t t̠ɕ and d̪ð d d̠ʑ. The idea with n̪ð is to reinforce the contrast with alveolar n in the same way. But maybe it doesn't really call for reinforcement; it's a reasonably common contrast, after all, at least in Australia. Anyway the affrication is a phonetic detail, and shouldn't play any role in the phonology. Which is to say that if I do end up deciding I hate it, I should be able to dispense with it at no cost.

I haven't made up my mind about vowels. I've put ɑ in the name of the language, thinking that it could be a retracted allophone of something like ɐ, but I probably want to differentiate how pharyngealised consonants (including q ɴ) affect vowels from how true pharyngeals (including χ ʁ and for some purposes maybe ʔ h) do; and maybe I'll only end up with ɑ next to the pharyngealised consonants. (I've also written ɑ̯ for ʕ̞.)

There's an interesting phenomenon here, if I've learned the right lessons from recent reading. I'm used to thinking of vowels as falling into ±ATR (advanced tongue root) pairs: e ɛ, o ɔ, and so on. And I guess I'd have thought that pharyngeals, especially when they're described as +RTR (retracted tongue root), would tend to make +ATR vowels -ATR, within an inventory that includes both sorts of vowels underlyingly. Well, and I'm pretty sure that can happen in languages that have ATR harmony. But my impression now is that the more common pattern is for the retraction to be allophonic and not to follow a neat +ATR → -ATR pattern. Like, you might get i→e u→o, in which all the vowels are +ATR, or you might get ɛ→ə, which if anything takes -ATR ɛ to +ATR (or -RTR) ə. (Those particular examples are from the Athabaskan language Chilcotin, as reported in Sharon Rose, Variable laryngeals and vowel lowering.)

Anyway, it feels likely that Nðɑħɑɑ̯ will end up with four underlying vowels---Iqę́hhǫ has i e a o---with various retracted allophones.

I haven't thought much about the rest of the language. Really this is just me playing with an inventory, and trying to understand a few things that are new to me. But the Iqę́hhǫ languages are verb-first and have ergative agreement patterns, so that's a place to start (or at least a place to end up!).

Well, I've thought a bit about codas. I think the plosives will all become fricatives in coda position, and that there'll be some place assimilation before a following consonant---not across-the-board, but I guess subplace distinctions (among coronals, and among dorsals) will get lost. So you'd get things like t̪θ+n → θ+n → sn.

You'll still have voicing contrasts in coda, helped by the fact that the voiced fricatives can be realised as approximants. (Though maybe they will still devoice before pause?)

I think -ɴ is going to be pretty weak, fully assimilating in place to a following consonant, or being realised only as nasalisation on a preceding vowel. Having a -ɴ suffix or two could be fun, with results like qat̪θ+ɴ → qaθ+ɴ (coda fricativisation) → qað+ɴ (voicing before ɴ) → qɑ̃ð (nasalisation and retraction before ɴ, which drops) → qɑ̃l (allophonic ð→l).

Well, we'll see.

I'm not convinced that +ATR exists in the sense you're assuming it exists. It seems like a neat hack to get vowel height contrasts to fit nicely into a binary feature system, although it's real in some languages. Either way, assuming features inhere in characters of the IPA is sort of a type error.

Yeah, that paragraph was there largely because I was struck by the fact (if it's a fact) that pharyngeal vowel lowering doesn't pattern nicely with ATR distinctions. Not really surprised, I so far don't know enough that any of this stuff can be really surprising, but I did find it striking.

(Fun fact: it's just barely possible that the very first bit of phonology that I read specifically for conlanging research was about ATR harmony in Mongolian, where you actually do have uvulars that occur with only the -ATR vowels---assuming I've remembered and understood that stuff correctly.)

I have to say Aqiatu is going to have some serious depth if all of this is intended to provide some vocabulary and ancestors of neighbouring languages. But some historical context would be good - where are these settlors going to come from and why did they migrate? What happened to the original inhabitants?

Vaedty Qys has given me some ideas too - not that they'll be of much use for now. Cheyadeneen and Vedreki are rather different in structure.