Akam's scratchpad (two speedlangs)

[akam chinjir]

Thanks!

Akiatu itself was actually supposed to be fairly peripheral to the main goings-on in the conworld. (It's still a bit peripheral, though a lot less so, both geographically and world-historically.) The project as a whole has a fairly broad geographic spread (I generally say "subcontinent," which is about as vague as I want it to be), and the events and characters that got it started are still more than 2000 years in Akiatu's future.

Part of what this means is that if an inventory or something catches my fancy, but doesn't hold my attention for more than a few days, there's still probably a place for it. And part of what that means is that it doesn't seem like a waste of time to post on something that I might not end up taking very far.

(I hope I do take it pretty far! I'd really like it if Nðaḥaa̯ does turn out to be an ancestor of the Iqę́hhǫ languages, which I'm really enthusiastic about but don't want to work on more without more historical perspective---and a deeper understanding of all things laryngeal. But we'll see.)

As for the settlers I mentioned---I was thinking of them as the region's first (human) inhabitants, foragers who gradually settled the region for no particular reason. But I don't have any very concrete ideas about that part of the history.

I've actually wondered if it might make sense to put up some worldbuilding posts, maybe I'll give that a try.

[akam chinjir]

Okay, forget about ɴ. It was silly to try contrasting it with ŋ.

And I used the symbol ʕ̞, which doesn't make any sense (whatever a pharyngeal approximant is, it's not a lowered pharyngeal fricative). When I want to transcribe a pharyngeal approximant, I'll just write it as a̯ or ɑ̯.

I'm still fond of nð. Maybe that's here to stay.

Maybe q is somewhat affricated, but I'll still write it just as q.

In other orthography: I'm not going to distinguish dental and postalveolar articulations with diacritics, it's just nð tθ dð (with no ̪) and tɕ dʑ (with no ̠). And I'll use a subdot for pharyngealisation (ṇ ṭ ḍ ṣ ẓ ṛ) and in ḥ And I think β~ʋ will be v~w.

Like this:

lab	dent	alv	alv+phar	postalv	vel	uv	phar	glot
m	nð	n	ṇ	ɲ	ŋ
p	tθ	t	ṭ	tɕ	k	q		ʔ
b	dð	d	ḍ	dʑ	g
(f)	θ	s	ṣ	ɕ	(x)	χ	ḥ	h
v~w	ð~l	z~r	ẓ~ṛ	ʑ~j	(ğ)	ɣ	ʕ, ɑ̯

I've included the fricative allophones of p k g.

I still don't know about vowels. The main open question is whether any of the changes conditioned by pharyngeals should be neutralising. Like, maybe it's i e a o u, but i e can merge to ẹ and u o to ọ, where ẹ ọ could be ɛ ɔ, or eˤ oˤ, or ɛˤ ɔˤ, or something. For the time being, I'll assume that at least i e a o occur underlyingly.

I'd like there to be a lexical high tone, but I'm not going to make up my mind for sure until I know more about laryngeal stuff.

The maximum syllable is CVCC. Consonant clusters are limited to CC both finally and intervocalically.

Roots are mostly monosyllabic and always consonant-initial, with CVC and CVCC being common and CV fairly rare. Bisyllabic roots are mostly of the shape CVCVC(C), with the medial C either a glottal or a voiced plosive/approximant; the vowels in such a root are often identical, and always are across h ḥ ʕ (which have no oral constriction to block assimilation); CVCCV roots do also occur, and probably there are some trisyllabic roots, too.

Phonetically, VʕV VhV VʔV could easily end up as something like Vːˤ V̤ː V̰ː.

Hmm, I'm going to end up with both aʕa and aʕ, and both are going to end up sounding an awful lot like ɑːˤ. Maybe there's a way for aʕ (aa̯) to have a clear contour, more a diphthong than a long vowel. That'll require that ʕ not back a all the way to ɑ, but that seems reasonable enough.

Plosives are fricativised whenever they occur in postvocalic coda position---both word-finally and in CC clusters. It follows that a plosive can never surface as such when it is the first consonant in a cluster, and, conversely, that a fricative in that position could always in principle represent an underlying plosive. I'll just assume when possible that it's an underlying fricative. (That's not possible with f x ğ, which can only ever be allophones of p k g.)

The voiced fricatives (including voiced plosives that have fricativised) will get their approximant allophones at least optionally whenever they follow a vowel, and maybe always when they precede a consonant. They'll always stay fricatives whenever following a consonant.

The fricativisation rule made an exception for root-initial plosives. The exception is plausibly stress-related, since stress will consistently fall on the first syllable of the root. But this might turn out to be the only significant stress-related pattern or process.

(It may turn out that only stressed syllables can have branching codas, but if so, that'll be because you only get branching codas in monosyllabic roots, not directly because of stress.)

There'll be some restrictions on CC clusters, probably with some additional restrictions enforced across morpheme boundaries. Here are some high points:

• If both are coronals, one will assimilate to the place of the other. The usual pattern will be for the first to assimilate to the second, but if both are alveolars and one is pharyngealised, then the result will be two pharyngealised alveolars.
• If one is velar and the other is uvular, the velar will become uvular. (So ɴ actually can occur, as an allophone of ŋ.)
• If the second is voiced and the first is a voiceless obstruent other than q, then the first will become voiced.
• There'll be some resistance to letting postalveolars occur next to uvulars or pharyngeals, but I may allow some exceptions, at least within roots.
• Maybe I should reign in sequences of voiceless fricatives, but I'm not sure how.
• These adjustments can result in geminates, which are allowed to surface, though geminates are rare in roots.

One more phonotactic. I think you'll often get an epenthetic vowel after word-final consonants, especially after word-final clusters. The quality of the vowel will depend on the preceding consonants; obvious things like echo vowels after consonants with no oral closure, a after uvulars and pharyngealised alveolars, i after postalveolars, o after labials...

I want to close with some roots. Not very many, because I don't yet have a handle on phoneme frequency, and anyway this is a terrible way to approach lexicon design, but maybe I can give a taste. For now I'll assume there aren't any high tones.

• bajk eat; meal, food
• ɕi water
• dðoṛ anger; thunder
• dʑaʔam person
• geṛ (geḍ) rock, stone; hard, rigid; unyielding, stubborn; courageous
• heχṭ steam; smell
• ḥaṛol man
• kolos wild boar
• mamk mud
• poa̯ṇ woman
• qeɣeχ giant
• qow sand, paste; crush
• taw third gender
• tem drink, breathe
• vaɕ (vatɕ) child; young; give birth
• ʔijk medicine; heal; healer

Oh, and the name of the language: nðaḥaa̯. I don't yet know what it means, but it's a perfectly well-formed root.

Edit: I corrected a couple of vowels based on the next post.

[akam chinjir]

Vowels

Okay, I've decided. Nðaḥaa̯ has four vowel phonemes, which I'll mostly transcribe as i e a o. (Maybe ɐ would be more accurate for a.)

o has u as a fairly common allophone, especially word-finally and next to velars.

Pharyngealised alveolars, along with q, lower a preceding high vowel, merging i with e and ruling out o's high allophone. That means I got one root wrong in the last post: *giṛ should have been geṛ rock. Note though that the uvular fricatives and the pharyngeals proper do not trigger this lowering---so iχ can contrast with eχ and uχ is a possible variant of oχ. This effect is not blocked by an intervening coda consonant---so in the last post *hiχṭ was another mistake, because even though the χ doesn't trigger lowering the ṭ does, and what you get is heχṭ steam.

This lowering can actually spread to a previous syllable, but only if all intervening consonants are pharyngeal or glottal.

Each of the vowels has an allophone that I'll refer to as retracted. I'll sometimes transcribe these as ɪ ɛ ɑ ɔ~ʊ. The vowel lowering process I just described always results in one of these allophones, and they also surface whenever a vowel is adjacent to any pharyngeal, pharyngealised, or uvular consonant.

The situation with a is actually a bit more complicated. aʕ (that is, ɑɑ̯) remains a phonetic diphthong, audibly distinct from a long vowel. And retraction of a is blocked when it's also adjacent to a postalveolar consonant; a hypothetical ḥaɕ would be ḥ[ɐ]ɕ, with the flanking consonants battling to a draw. Meanwhile, when a is under postalveolar but not pharyngeal influence, it fronts to æ. (Under the same circumstances, o can also front somewhat, to ɵ or ʉ.)

[evmdbm]

I've actually wondered if it might make sense to put up some worldbuilding posts, maybe I'll give that a try.

Yes please!

[akam chinjir]

I'll give it a try! (Soon, I promise.) But first you get this post about nouns. Maybe it includes some implied word-building?

Oops, and also: I've made up my mind about tone, there's a marked high tone, which I'll transcribe with acute accents. There's some funny business, which I'll be trying out in this post, but this thread doesn't right now need more phonology, so for now I won't go into the rules. (Also I'm still looking into some things.)

Intro to noun stems

The great majority of underived nouns consist of a root preceded by either a gender prefix or a plural marker:

• ó-dðoṛ O-thunder
• e-ɕi E-water
• a-heχṭ A-steam
• íj-geṛ PL-rock

I'll just gloss the genders as the O, E, and A genders, as above.

These markers can only attach directly to roots.

It's easy enough to associate the three grammatical genders with the three social genders that the Nðaḥaa̯ people distinguish: words referring specifically to people classed as íjḥáṛól men are in the A gender, words referring specifically to people classed as íjpoa̯ṇ women are in the E gender, and words referring specifically to people classed as íjtáw are in the O gender.

Words with with a semantic gender that clearly aligns with their grammatical gender are one of the main sorts of exception to the rule that nouns need a gender prefix of some sort: ḥaṛol man, póa̯ṇ woman, and taw, for example, can occur on their own as noun stems. Other examples are of two main sorts: kin terms often have inherent gender; and words for spirits and certain figures in folklore or myth are often considered inherently O (like dej ancestral spirit and qeɣeχ giant). (I'm not settled enough on the details of the kinship system to risk examples of those terms, however.)

The other main exception to the rule that nouns require a gender or number prefix is that some O nouns that begin with v and E nouns that begin with ʑ do not require a gender prefix:

• viʔ tree; wood
• ʑea̯ṣ shadow, shade; protection, care

A gender nouns whose roots begin with ʕ still normally require the a- prefix, however.

Most Nðaḥaa̯ words lack semantic gender. I kind of hope that there end up being some symbolic patterns, but for now the main patterns are phonological:

• Words beginning in an uvular, pharyngeal, glottal, or pharyngialised consonant tend to be A gender: . aheχṭ steam, smell . aqów sand, paste
• Words beginning in a labial consonant tend to be O gender. . óbajk meal . ómámk mud
• Words beginning in a postalveolar consonant tend to be E gender.
  • edʑaʔám person
  • eɕi water
• Words beginning in any other consonant tend to get the gender implied by their (first) vowel; both e and i imply E gender.
  • etem breath, air, mood
  • ókolos wild boar

I expect that borrowed words whose source is vowel-initial will take on the gender implied by the initial vowel, but for now I plan to rigorously avoid thinking about the languages from which Nðaḥaa̯ might borrow words.

There are also a number of derivational prefixes that are specified for gender. vi- (O), éθ (E), and aχ (A) derive nouns referring to humans and (occasionally) animals. One common use is to derive agent nouns:

• viʔíjk healer < ʔíjk medicine, to heal; contrast aʔíjk medicine
• éθháɕ forager < haɕ to forage, to gather, to collect
• aɣváhár hunter < váhár to run, to hunt (with χ → ɣ before a voiced consonant)

In general, if one of these prefixes can be used, then so can the other two; though some combinations will be socially more probably than others. (And I'd say more about that if I'd worked out the gendered division of labour in any detail.)

These prefixes can also supply semantic gender when that somehow seems appropriate, e.g. éθkolos female boar vs aχqólós male boar. (Animals other than monkeys are rarely conceived of as having a taw gender.)

I'll also mention the postural prefixes:

• bó- (hanging, O gender)
• éʔ- (standing, A gender)
• iχ- (sitting, E gender)
• ṣaḥ- (lying down, A gender)
• táj- (squatting, E gender)

You might suppose that the (phonologically unexpected) assignment of éʔ- standing to the A (`masculine') and of táj squatting to the E (`feminine') gender reflects crosslinguistically common pissing symbolism. Though maybe the coda consonants also have something to do with it.

The use of these is often unsurprising; viʔ wood provides some nice examples:

• éʔvéʔ standing wood → tree
• ṣaa̯viʔ lying down wood → log
• iɣviʔ sitting wood → wood
• bovíʔ hanging wood → branch

Sometimes the contrast is maybe not semantically transparent:

• ṣaḥqów lying down sand → sand
• iχqów sitting sand → paste

There's also the fun contrast between éʔdow, bódow, and ódow, all based on the root dow penis (if that sort of thing is fun for you).

The postural prefixes are often omitted when the resulting imprecision isn't a problem.

The prefix qaw- derives collective nouns and associative plurals. Collective nouns are treated as either A gender or plural, depending on context; associative plurals are always treated as plural.

• qavváɕ cohort < vaɕ child
• qavviʔ forest < viʔ tree
• qawdej the ancestors < dej ancestral spirit
• qawzejka Zejka's group

The presumably related qa- forms the plurals of derived stems:

• qaviʔíjk healers
• qaqavváɕ cohors

Diminutives are formed by prefixal CV reduplication, with some phonological adjustments of the V; unlike other derived stems, these generally take the regular gender prefixes:

• egegeṛ pebble < geṛ rock, stone
• óvóvíʔ shrub, bush < viʔ tree
• óvovaɕ young child, kid < váɕ child

Augmentatives require a prefix omo-, always resulting in an O gender noun:

• omodej great ancestral spirit; the ancestral spirit of a giant
• omogeṛ boulder

Okay, one more prefix. ɕa- most often attaches to roots with intuitively verbal senses, producing an E gender noun referring to some oblique argument of that verb, most often an instrument, though locations are also fairly common, and idiosyncratic meanings aren't exactly rare.

• ɕabájk food < bájk eat (contrast óbajk meal)
• ɕahaɕ companion < haɕ gather (folk etymology, at least among women: someone with whom you go foraging)
• ɕakáttá dance ground < káttá dance

I feel like I'm forgetting something, but also like that's enough for now.

[akam chinjir]

I seem to like Nðaḥaa enough that it might be a while before I have it sorted out enough for more posting, so much for a little sketch.

One thing is that I'm fairly set on having most noun and verb stems include nominalising and verbalising morphology, the nominalisers specifying gender and the verbalisers valency. And I'd like the whole system to feel organic and complicated---a lot more so that it did in the nouns post.

One idea I'm toying with is having the genders sometimes marked not by prefixes but by (labial, palatal, and pharyngeal) prosodies. So, for example, you might have a root foʢṇ woman, female whose noun stem is formed by palatalisation/fronting, yielding fejɲ (foʢn itself could occur as an adjectives---adjectives don't require stem formatives).

These prosodies aren't supposed to be active in Nðaḥaa itself, and I don't yet know exactly how they'll work or where exactly they'll show up.

It would also be nice if I had some idea of how Nðaḥaa's phonology might have come from something less ornate, at enough to derive some worthwhile distributional patterns, but so far I'm nowhere with that.

Meanwhile, I might as well report some adjustments:

• The high tone's ended up feeling wrong to me, and it's gone. Maybe pharyngealisation will turn into a low tone in some descendents.
• The pharyngealised coronals are now also retroflex (so ṭ represents ʈˤ), meaning that the four coronal series (tθ t ṭ tɕ) all differ in (sub)place, as well as being distinguished by secondary gestures.
• The sonorant allophone of the voiced alveolar fricative will now be trilled. I think I'll also allow the uvular and pharyngeal approximants to be in free variation with trills; probably not ʋ, though. This'll have orthographic consequences: the uvulars will be ʁ ʀ and the pharyngeals ʕ ʢ. (But I'll write the name of the language just as Nðaḥaa.)
• I had a p that predictably became f in some positions; now it's an f that in some positions varies freely with p.
• I had i merging with e under the influence of pharyngeals. I'm leaning towards making it i → e alongside e → ɛ.
• Restrictions on CC clusters, and some of the related allophony, seems to be changing.

[akam chinjir]

Pharyngeals, again, and tone, again

My head's really not been in conlanging for the last month or so.

One issue is that every time I come back to this I run up against my own ignorance about how pharyngeals work. I'm slowly working on that, but it'll be a while, and in the meantime I'd really like to be working out other parts of the language.

Here's what I've got so far. (Or what I think I've got---please let me know if you spot anything that seems wrong!) I guess the main single thing I'm drawing on is Moisik's The epylarynx in speech, but I'll warn you that I've so far read it neither completely nor well.

Anyway, pharyngeal(ised) consonants can be associated both with a distinctive tongue shape and with distinctive epilaryngeal activity.

There are at least two quite different sorts of tongue shape that could be involved. There's the backed/retracted sort, which is what you might associate with an RTR feature, and with vowel lowering in general. And there's the double-bunched tongue that you might associate with one kind of American /r/ (but it's also found in (some?) Caucasian pharyngeals, for example), and doesn't tend to lower vowels so much as centralise and even front them (so you get cases of a → æ next to pharyngeals, for example).

Anyway the way I've so far pieced things together, it's the tongue shape that's mostly significant with pharyngealised consonants (and I think also pharyngealised vowels?)---which is a reason why some people prefer to describe Arabic emphatics (for example) as uvularised, rather than pharyngealised.

Whereas it's the epilaryngeal activity that's key with full-on pharyngeal (or epiglottal) consonants like ħ ʕ. These might also involve tongue backing or bunching, but it's less significant (and they often have less of a lowering effect on nearby vowels than do pharyngealised consonants).

With uvulars I think you often get a split (I'm drawing here on Rose, Variably laryngeals and vowel lowering). When pharyngealised consonants pattern differently from full-on pharygeal ones, q tends to pattern with the pharyngealised ones, but χ ʁ ʀ tend to pattern with the full-on pharyngeals. (iirc, Rose says that ɴ is too rare to draw any conclusions; I can't remember what she says about ɢ.)

Glottals can also pattern as pharyngealised in some sense, though obviously they don't have the distinctive tongue shape. (So they won't lower vowels in general; but there are languages in contexts in which all vowels merge to ɑ under the influence of pharyngeals, and in those languages h ʔ tend to have the same effect.)

(Sylak-Glassman, Deriving natural classes, complicates things a bit by considering a much broader range of languages---e.g., it's hard to make sense of Ubykh's qʲ q qˤ contrasts on Rose's approach---so if this sort of thing really interesting to you then you might want to check out that as well.)

I mentioned epilaryngeal activity. I so far have no head for back-of-the-throat anatomy, and have retained very little of what I've read. The main thing that matters here is that I've convinced myself that it can result in a lower pitch on (the beginning of) a following vowel, perhaps in part by triggering some breathy phonation.

Anyway that's about the context in which I'm now thinking about these things in Nðaḥaa.

The main thing this gives me (it also helps that I've got a better handle on Nðaḥaa morphology) is the confidence to reintroduce a high tone.

I'll sketch the whole system (using nonce roots and affixes).

You might have a root téáw. (Insofar as it matters, that'll end up getting treated as a single syllable, btw.) And you might add a suffix -es. The result will be téáwés---the high tone will spread to the right, onto the suffix. (There's no analogous spreading to the left, onto prefixes.)

A detail: a high tone cannot spread onto a syllable that already has a high tone; and if it spreads to the point where it ends up adjacent to another high tone, the second one will end up deleting. (Like, qá+ma+téá would result in qámátea.)

There are certain consonants that block this spreading. These are called depressor consonants, and they include all voiced plosives. So, if you started with téád, you'd end up with téádes.

Well, it's trickier than that, since non-root-initial plosives tend to fricativise or even sonorant-ise after vowels, so actually you'd get téázes or even téáres. (Which produces some complication, because underlying z~r isn't a depressor consonant, and won't block tone spreading.)

Anyway, what I've got worked out for now is that χ ʁ~ʀ ħ ʕ~ʢ h are also depressor consonants.

Depressor consonants can also result in contour tones, at least phonetically. Voiced consonants tend to reduce the pitch at the beginning of a following high-toned vowel; pharyngeals and h tend to lower the pitch at the end of a preceding high-toned vowel. I'm not going to worry about representing this orthographically, though.

One thing though: it seems reasonable to stipulate that there can't be a high tone on a vowel that's preceded by a voiced consonant and followed by a pharyngeal or h. E.g., you could have dá and táh, but you can't have dáh.

A bit more about vowels

One thing: I've decided that the syllable template allows for two vowel slows. This allows for both long vowels and opening diphthongs.

This means that I've talked myself into a four-way contrast between i ij ii iij. I can live with that, partly because that j is actually ʑ~j, and I could assume it's somewhat fricated in this context.

A consequence: nðaḥaa will contrast with nðaḥaʕ, so I have to decide once and for all which it is.

Anyway there are four vowels, which I've been thinking of as i e a u~o. But there are a bunch of consonants that affect vowel quality.

• The palatals ɲ tɕ dʑ ɕ ʑ~j tend to front preceding vowels.
• The pharyngeals χ ʁ~ʀ ħ ʕ~ʢ tend to lower preceding vowels.
• The pharyngealised consonants ṇ ṭ ḍ ṣ ẓ~ṛ tend to lower both preceding and following vowels.

Note that a vowel that follows a pharyngealised consonant and precedes a palatal one will be both fronted and lowered.

The result is the following four systems of vowel contrasts:

	i	e	a	u~o
neutral	i	e	ɐ	u~o
fronted	i	e	æ	y~ø
lowered	e	æ	ɑ	o
fronted+lowered	e	æ	æ	ø

Phonologically the main point of interest is that the e a contrast is lost between a pharyngealised consonant and a palatal: there's no distinction between ṭaj and ṭej, for example, both would end up as ṭæj.

I'm going to keep the orthography simple, though. (I'll continue to write u~o as o.)

Coda

This is, at least, I hope, enough to go on with.

The next post will be about something non-phonological, I promise.

[akam chinjir]

Pronoun table

I'm going to post this so it's all in one place. I won't be talking about all of it right away, but it shouldn't be too long.

	erg/poss	erg+pfv	abs	free	+ŋ	+di	+tal	+ges
1s	ŋ	ŋi	wo	bów	ŋoŋ	nzi	ntal	ŋges
1+2p	éjé(s)	éjés i-	heŋ	ikáŋ	éjéŋ	éjézi	éjétál	éjégés
1p	al~að	al i-, aθ	je	jaj	alb	ardi	artal	alges
2s	ne	ni	sa	weṛo	ħoŋ	ħezi	ħetal	ħeges
3sF	e, s	eh	--	wíí	eɲ	ezi	etal	eges
3sM	a, r	ah	--	wáá	aŋ	azi	atal	ages
3sT	o, w	oh	--	wóó	om	ozi	otal	oges
2p,3p	ij	ih	(e)m	ḥam	ijɲ	irdi	irtal	ijges
LOC	ke, k	ki, keh	jaṛ	jaaṛ	koŋ	kezi	ketal	keges

I haven't figured out a nice way to standardise the orthography for the voiced continuants (which for now are β~w ð~l z~r ẓ~ṛ ʑ~j ʁ~ʀ ʕ~ʢ), that could easily be confusing.

As you can see, across the pronoun system there's an inclusive first person plural, and there's no distinction in the plural between second- and third person.

The 1+2p ergative/possessive form éjés is a so-called footed prefix that can actually surface as an independent phonological word (this happens obligatorily before a perfective verb).

LOC is for locative nouns; as you can maybe infer, they're getting their own gender. (I say "locative," but certain abstractions, including at least one sort of deverbal noun, will take this gender.)

-ŋ -di -tal -ges are used in possessive constructions. You can think of them as classifiers, maybe.

Edit: for reasons that'll come up later, the 1p erg+pfv form now has a footed variant. Also I fixed an error in the 2s erg+pfv form.

[akam chinjir]

Some morphosyntax and some prosody

(Spamming today, sorry! But this is the last one.)

So Nðaḥaa is a verb-initial language with no case-marking and a somewhat peculiar ergative agreement system. In this post I'll set out some basics.

The verb template

The full verb template will look something like this:

AGREE(ERG) - PERFECTIVE - ANDATIVEetc - CAUSATIVEetc - ROOT - VALENCY

• I'll talk about ergative agreement below. As you might expect, it surfaces only in certain valency frames.
• The semantics of the perfective will need a separate post. It's marked by i ~ ʑ ~ h, or by stem changes, under various morphophonological conditions.
• There'll be andative and venitive prefixes, and one with a stay-like meaning (restative?). They'll have uses related to both motion and aspect and I'll talk about them another time. Maybe I'll decide they should (also?) be able to go after the causatives and such.
• The productive causative is wá ~ wáá ~ wááwá, the passive is ḍe ~ ḍej (perfective dʑe(j)). These can occur together, in either order, and maybe you could even get more than two (e.g. ḍe-wá-ḍe PASS-CAUS-PASS). There's some prosodic trickery that I'll talk about below: all allomorphs of the causative prefix, and the ḍej allomorph of the passive prefix, must head a (moraic trochaic) foot, and can head a phonological word distinct from the one headed by the verb root. I'll occasionally call these "footed" prefixes.
  
  I probably also want some kind of reflexive (maybe it'll also be a reciprocal), and it'll also go here, presumably.
• The root could actually be a compound; at least V+V and V+N compounds will be possible. Most simple roots will just be CVC or CVVC, but there'll be other possibilities, including, I think, some vowel-initial roots (the morphophonology will treat the initial vowel as a prefix).
• The valency suffixes will need their own post. (I'm actually still changing my mind about them, so the forms I use below might change or disappear.) For now I'm just going to gloss these as TR (for transitive), ITR (for intransitive), and IMPS (for impersonal); though the system distinguishes different sorts of transitive, and (e.g.) unergatives from unaccusatives from middles.
  
  One thing: many roots will be able to occur with more than one of these suffixes, and often it'll look like one of the resulting stems relates to another as its passive or its causative or its reciprocal. (Like, verbs in -a will be unaccusative, and verbs in -aχ---or -a-χ, maybe--- will mostly look like the causatives of the corresponding -a forms.)

Cross-referencing

A verb can cross-reference zero, one, or two arguments. How many arguments should be cross-referenced is normally plain from the verb's valency suffix (except that nonverbal predicates don't take such a suffix; they always cross-reference a single argument).

ˈʑé.fól
jéf -ol
rain-IMPS
"It's raining"

When there's just one argument getting cross-referenced, you use an absolutive clitic:

ˈβá.ha.ri.βo
wáhar-i  =wo
hunt -ITR=1s.ABS
"I am running"

One thing, though: a nonlocative third-person singular absolutive argument does not trigger overt agreement. I'll gloss this as if a phonologically null clitic were present, like so:

ˈβá.ha.ri
wáhar-i  =∅
hunt -ITR=3s.ABS
"They (sing) are running"

So far I've got no reason to care whether that ∅ actually represents anything.

When two arguments are cross-referenced, one of them gets an ergative prefix. This is the argument that's structurally higher in the clause---usually, it'll be the argument you intuitively think of as the subject. The other argument gets an absolutive clitic (or nothing).

iʑˈβá.haɾ.ni.heŋ
ij    -wáhar-ni=heŋ
3p.ERG-hunt -TR=1+2p.ABS
"They (pl) are hunting us (incl)"

Simple!

Placement of the absolutive clitic

The absolutive clitics are second-position clitics, and what counts as second-position is determined prosodically: they follow the first prosodic (or phonological) word within their domain. This will often put them right after the verb, as in the examples, above, but most things that end up before the verb can host them instead. (The exceptions I know about so far are clause-level conjunctions, and left-dislocated topics.)

One word that often hosts absolutive clitics is ʔaṛ, which you can think of as a grammaticalised adverb meaning already or now; I'll gloss it as IAM for iamitive (< Latin iam already), and often translate with some sort of perfect or with "already" or "now." (If you know some Mandarin, ʔaṛ has a lot in common with sentence-final le 了.)

ˈʔɑṛ.ħæ̃ iʑˈβá.haɾ.ni óó
ʔaṛ=heŋ      ij    -wáhar-ni óó
IAM=1+2p.ABS 3p.ERG-hunt -TR PRT
"They are hunting us now!"

(The translation is meant to suggest a change-of-situation meaning. The sentence-final particle conveys urgency. Oh, and even in the `phonetic' line of glosses I'll write long vowels by doubling.)

A preverbal focused element can also host the clitic:

iʑˈkó.lós.séŋ iʑˈβá.haɾ.ni
ij-kólós    =heŋ      ij    -wáhar-ni
PL-wild.boar=1+2p.ABS 3p.ERG-chase-TR
"It is wild boars chasing us"

(Aside: I haven't decided what if any bits of syntactic ergativity there'll be in Nðaḥaa, but here's a sentence where that sort of thing might come into play---since it's the ergative argument getting focused. In Mayan languages, for example, you might expect to see an agent focus construction here.)

If the focused element is more than one phonological word, the clitic will interrupt it:

ˈħɑm.meŋ iʑˈkólós iʑˈβá.haɾ.ni
ḥam=heŋ      ij-kólós     ij    -wáhar-ni
3p =1+2p.ABS PL-wild.boar 3p.ERG-chase-TR
"It's those wild boars chasing us"

(Here you see the free 3p pronoun ħam used as a sort of definite determiner. Note the morpheme order, something like them us wild.boars they chase.)

Absolutive clitic placement works the same way in intransitive clauses, and clauses with nonverbal predicates.

ˈʔɑṛ.βo ˈβá.ha.ri
ʔaṛ=wo     wáhar-i
IAM=1s.ABS run  -ITR
"I am running now"

iʑˈdʑam.je ˈnðɑ.ħɑʢ
ij-dʑam  =je     nðaħaʕ
PL-person=1p.ABS Nðaḥaa
"We are Nðaḥaa people"

The trickiest point I know about so far concerns the so-called footed prefixes, which can but mostly do not have to constitute independent phonological words. (Whether they do depends on the speed of talking, for example.) Here's a pair of examples with the 1+2p.ERG form éjé(s), first as a true prefix:

ˌé.jéˈwa.har.nim
éjé     -wáhar-ni=m
1+2p.ERG-chase-TR=3p.ABS
"We are chasing them"

And here it is with independent éjés:

ˈé.jé.sém ˈwá.har.ni
éjés    =m      wáhar-ni
1+2p.ERG=3p.ABS chase-TR
"We are chasing them"

There's no difference in meaning here, except insofar as talking speed can reflect differences in register.

Default ergative agreement (gender)

This bit is somewhat provisional, because I'm not sure how this interacts with possession, but maybe it's worth mentioning.

When a verb has a third person singular ergative argument, normally it agrees with in gender as well as in person and number:

aˈβá.har.ni ˈḥɑ.ṛɔl aˈfíí
a      -wáhar-ni=∅      ḥaṛoð a-fíí
3sM.ERG-chase-TR=3s.ABS man.M M-snake
"A man is chasing a snake"

But there's a complication when the (singular, third person) ergative argument is definite. There are two ways to mark definiteness on common nouns. One of them uses a prefix, and when it is used, the noun can only trigger default feminine ergative agreement:

eˈβá.har.ni βɑˈḥɑ.ṛɔl aˈfíí
e      -wáhar-ni=∅      wa   -ḥaṛoð a-fíí
3sF.ERG-chase-TR=3s.ABS DEF.M-man.M M-snake
"The man is chasing a snake"

Definiteness can also be signaled using the appropriate third person pronoun, and in that case the verb does agree as expected:

aˈwá.har.ni ˈwáá ˈḥɑ.ṛɔl aˈfíí
a      -wáhar-ni=∅      wáá ḥaṛoð a-fíí
3sM.ERG-chase-TR=3s.ABS 3sM man.M M-snake
"The man is chasing a snake"

Coda

Finally some sentences!

There are a couple of points on which I've not tried to be careful---so the example sentences might turn out questionable in the lights of near-future posting.

First, only one of them has a sentence-final particle. With Vædty Qyṣ I refrained from letting myself be too inspired by Cantonese on this, but with Nðaḥaa I'm not holding back: there'll be a whole host of sentence-final particles. As in Vædty Qyṣ they'll have the job of hosting boundary tones; though they'll be a bit less integrated into the grammar, and won't interact with focus at all (or at least that's the current plan). Instead, they'll be speech-act typing, and express the speaker's attitude towards what they're saying or their expectations wrt the listener, things like that. Anyway, they're sort of ostentatiously missing from most of the examples above.

Second, transitivity is a bit complicated in Nðaḥaa, and (e.g.) the use of awáharnim ijfíí he is chasing snakes rather than wáhari oŋ ijfíí he is chasing/running after snakes (with the verb marked as unergative rather than transitive and absolutive cross-referencing of the chaser, not the chasee) could use more discussion than I'm giving it here.

[akam chinjir]

Some worlding

So I'm going to start trying to set out some of what I know about the world in which these languages are spoken. Blanket disclaimer: I know vanishingly little about most of the things you need to know to do this well (e.g., physical geography). Sometimes that means I leave things pretty sketchy, but undoubtedly it'll also mean that some things don't make sense.

Please let me know when something doesn't make sense.

Some basics. For a world that's not earth or counter-earth or whatever, it's pretty earthlike. But it's not supposed to be a reimagining of any particular region or period or culture. In terms of broad categories, the areas I'm so far mostly interested in are tropical, neolithic (I guess), and subsist on foraging and (in some cases) horticulture, though there are also agrarian societies to the south (the so-called Dog People), and I'll get to them eventually.

Most of my posting has been about a variant of the language Akiatu spoken in a village called Ihjatikai by the northern bank of the Akiatu River, in maybe 120 BA or so. (It's the river that gives the language its name. I'll give dates BA and AA---Before/After the Ancestor---but that's not an in-world system.)

At this point the Akiatu is flowing roughly east by northeast, with the mountains at least a few days' walk to the north. I tend to picture the river as being a few hundred metres across, with the (rain)forest coming pretty close to the banks; though I suppose there's probably substantial seasonal variation on both points. (Maybe someday I'll know more about rivers.)

The mountains to the north are volcanic, though currently dormant. They form the southeastern part of an enormous ring of volcanoes and volcanic islands that circle a region I call the Sea of Giants. (It's called that because of a widespread belief that in ancient times volcanoes were inhabited by giants; more on this when I talk about ancestral and other powers.)

The Akiatiwi tend to refer to the peoples to their north as Hunters (kwamuri, which I think is actually an ethynonym) or, especially once you reach the mountains, as the Bird People. (Neither group is linguistically or culturally homogeneous.)

Starting from Ihjatikai it's about ten days by canoe to the enormous waterfall that the Akiatiwi call the Thunder, which is immediately above the point where the Akiatu reaches the ocean.

The Thunder is meant to be ridiculously magnificent, maybe a kilometre across and hundreds of metres high (though not necessarily in a single drop). The falls are split by a sizeable island, on whose ocean-facing cliff there's an important shrine to the Great Ancestor.

The Great Ancestor is a massive volcanic island roughly 50 km out to sea, part of the ring of volcanoes I mentioned earlier. It's a massive eruption of the Great Ancestor that anchors the dating system I mentioned earlier.

The Akiatiwi refer to the people who live below the Thunder, around the mouth of the Akiatu River and on the islands stretching out to and around the Great Ancestor, as the Fish People. (These people are also neither linguistically nor culturally homogeneous, though most speak languages derived from Nðaḥaa---I think with retention of voicing contrasts in plosives, loss of tone and pharyngeals but gain of some sort of ATR harmony in vowels, and I'm not sure what else.)

The year zero eruption of the Great Ancestor isn't Krakatoa-level, but it's world-ending for the Fish People, who survived only in two refugee populations. Some managed their way inland, others set out to see, eventually, over the course of the next 2000 or so years, to settle most of the islands (mentioned earlier) that encircle the Sea of Giants. (This migration is supposed to remind you of, but not really to rival, the Austronesian migrations across the Pacific.)

(Incidentally, on the last of these islands, not hugely distant from the mainland though quite far to the west from where we started, something like 2000 years after the eruption, there's a shaman named Čepes speaking a language named Hamụz, and it's with Čepes and Hamụz that this whole project got started.)

But that's enough about the future, I want to talk about the past.

Quite far to the south of the Akiatu River, though near a significant tributary, and about 1500 years BA, there's a hilltop shrine called Gagur. I hope it's not too confusing that there's also a language called Gagur. The river nearby swells to fill a shallow valley, and you'd probably call the result a lake; the Qɨsə languages, including Vætdy Qyṣ, are mostly spoken around this lake.

I tend to think of this region as more grassy than forested, but still hot and humid. There's little or nothing of what you'd call settlement. Gagur shrine probably does always have a handful of people there (there's a fire that needs to be kept burning), but for most Gagur speakers it's a place they go to just for particular special occasions (births, deaths, festivals).

The shrine and the lake and the surrounding region is actually supposed to compose a little linguistic area, with various features getting shared by ancestrally unrelated languages. One dynamic is that increasing numbers of people are arriving in the region from the south, motivated by the increasing power of certain great chiefdoms---the Dog People---including eventually the rise of agriculture. (But I don't yet have anything like a timeline for the development of agriculture, this might end up being too early for that to be a real factor.)

The Gagur language is ancestral to Akiatu, by the way. (I think I haven't posted about it here, but I did an initial sketch over on reddit, in case you're interested.) So one of the reasons for the linguistic mix around the Gagur shrine in to supply Akiatu with some early influences (and some adult learners!---given that that seems to be a tried and true way to simplify a language's morphology).

Gagur speakers themselves eventually migrated, heading north, and reaching a place they called Aadwi, on the southern bank of the Akiatu proper. (And by this point the language has changed enough that I start calling it Aadwi instead of (Late) Gagur.) I'm pretty sure that canoe-building becomes an economically significant activity in Aadwi---maybe there are especially appropriate trees in the area. But I've hardly worked out any of the social or technological context of this yet. One thing: I'm pretty sure it implies that Aadwi is actually a settlement, not just a ritual centre. Though I think few of the Aadwi people would be truly sedentary.

Also the fishing is really good at Aadwi, and progressively so as you go downriver, which may be a big part of the reason why the descendents of the Gagur eventually settled as far along as Ihjatikai, and then even further.

More about that another time.

But the next post, I think, will have to be about magic.

[akam chinjir]

(So much for magic.)

A while back dewrad posted a 48-hour speedlang challenge, and I sketched a language I called Bááru. (It had a short-lived thread here, 48 Hours of Bááru.) There were some things I liked about it, but I ended up not taking it anywhere, and some of the things I liked about it have ended up elsewhere.

But last month there was a week-long speedlang challenge on the discord server associated with the Conlongs subreddit, and for that I resuscitated Bááru---well, I reused the name, and took up the idea of having inflecting focus particles that interfered with verb agreement (an idea that owes maybe a bit too much to Lavukaleve, via the Terrill grammar), and Wedi is still eating snakes. I'd been planning to wait till I'd done revisions, but that looks like it'll take some time, so I'll just link to the original speedgrammar (I've made a handful of minor corrections):

• Bááru the Language (pdf)

Maybe that'll be interesting to someone. For me, one of its high points is that it's so far the closest I've ever come to writing something with the look and feel of a grammar. (I don't mean it's especially close!)

I had a week, and spent a day or so of that redoing my LaTeX glossing setup, and probably two of the remaining days on pronouns and agreement paradigms, because I'm terrible at those. I think the main thing that suffered from the time constraints was the weird business about postpositions, which is really undigested.

It's verb-last, with a fair bit of agglutination and polysynthetic orthography (no spaces) in the verb. Here's a sample sentence:

wériɗe íbí ꜜémiiwalíjuusuwáʔ
wéri -ɗe íbí   é-        ɴ- pii-   walí -uu    -su       -áʔ
Wedi ERG snake 3PL>3SING TR INC(1) eat  INC(2) III.S.ABS PFV
"Wedi started eating the snake"

The requirements of the challenge were to include some form of harmony; some marginal phonemes; phonemic prenasalisation; morphosyntax sensitive to information structure; at least five grammaticalised aspects; some two-part morphemes; and some translation targets.

More recently I took a flight across the Pacific and ended doing about a 12-hour speedlang sketch of a language called (for very clever reasons) Totǫhk. I guess I'll share that too:

• Totǫhk the Language (pdf)

I had some fun with sound changes, and produced some consonant clusters I really liked (current favourite, onset kθv), and I definitely concluded I want to spend some time with an SVO language, but I'm not sure if I'll try to take this anywhere.