Page 2 of 4

Re: Akiatu scratchpad (subject control; "mwi")

Posted: Sun Oct 07, 2018 11:39 am
by Xwtek
I have suggestion. If you're going to have Austronesian-esque conlang, you should have /ŋ/. And particle conparative (Yes, like English than)

Re: Akiatu scratchpad (subject control; "mwi")

Posted: Sun Oct 07, 2018 7:02 pm
by akam chinjir
Not aiming for it to seem Austronesian, though. (Don't know enough about Austronesian languages even to try it. Seems like an unreasonably broad target, anyway.)

On particulars, I decide for /ɲ/ over /ŋ/ (and didn't want both). Comparatives will be parasitic on motion/distribution constructions; locative constructions will be possible but probably not hugely common. Currently you can't form comparatives with true adjectives, only with verbs, though that might change.

It should look more or less like this:

Code: Select all

hjaci kinaisu   iwasu     a   itamu
Hjaci surprised go.beyond LOC Itamu
Hjaci is more surprised than Itamu

Code: Select all

hjaci kinaisu   a   itamu aiwa
Hjaci surprised LOC Itamu beyond
Hjaci is more surprised than Itamu

Code: Select all

hjaci pahupahu maita a   itamu
Hjaci angry    reach LOC itamu
Hjaci is as angry as Itamu

Code: Select all

hjaci pahupahu a   itamu=wati
Hjaci angry    LOC Itamu=DEIC
Hjaci is as angry as Itamu

Akiatu scratchpad (Other nonfinite complements; "na")

Posted: Sun Oct 07, 2018 9:12 pm
by akam chinjir
Other nonfinite complements; "na"

In the last big post I talked about sentences like this one:

Code: Select all

itamu₁ kamaisu mwi ____₁ pahupahu papa
Itamu  begin   SS        be.angry REDUP(INC)
Itamu was beginning to get angry
Here, the semantic subject of the subordinate clause shows up (only) as the syntactic subject of the matrix clause. This pattern requires that the subordinate clause be headed by mwi.

This time I'll start with sentences like this one:

Code: Select all

kipaja hjaci₁ wakaisu nai na ____₁ hakjawi acitau
Kipaja Hjaci  help    PFV DS       fire    bless
Kipaja helped Hjaci bless the fire
The semantic subject of the subordinate clause is still controlled, but now it's controlled by the object in the matrix clause, here "Hjaci": Hjaci is both the one who is helped and the one who blesses the fire. In this construction, the subordinate clause is headed by na, which I'll gloss DS (for different subject).

Object control verbs often have a broadly causative sense, as here:

Code: Select all

ki wapanaiwi itamu₁ ahjai ka  na ____₁ kitikwa  wañi aja
KI elders    Itamu  let   PFV DS       REFL(AP) say  out(PFV)
The elders let Itamu speak
The two examples so far involve verbs that take both a direct object and a clausal complement. This is actually not a common pattern. More often, if a verb takes a clausal complement, any additional argument will appear as an indirect object:

Code: Select all

itamu wañi aja      i   hjaci kja  tamwipaku kunasi     acitau jaku
Itamu say  out(PFV) DAT Hjaci COMP canoe     must(PASS) bless  PFV
Itamu said to Hjaci that the canoe must be blessed
na complements are possible if the indirect object is raised to focus position (as discussed in a previous post):

Code: Select all

itamu hjaci₁ wañi aja      na ____₁ tamwipaku acitau jaku
Itamu Hjaci  say  out(PFV) DS       canoe     bless  PFV
Itamu told Hjaci to bless the canoe
As in this example, the sense of the verb can differ depending on whether it gets a kja complement or a na complement. The contrast here is quite typical: the kja complement picks out a proposition, and the na complement picks out a course of action.

Here is another example that requires raising of the indirect object:

Code: Select all

kijapa hjaci₁ hwati tapikau   takau      na ____₁ wamau ka    a   mikuwi hatau
Kijapa Hjaci  CAUS  have.plan REDUP(PFV) DS       go    TRANS LOC waters great
Kijapa persuaded Hjaci₁ ____₁ to go to the ocean
(Incidentally, the fact that hjaci is a raised indirect object here is signalled by the used of hwati, otherwise give: it's used in causative constructions with transitive verbs, with an intransitive verb you'd find ahjai, otherwise make, let.)

I'd also like this construction to be possible with oblique arguments in general, something like this:

Code: Select all

?itamu sati tamwipaku₁ wañi aja      na ____₁ kunasi     acitau jaku
 Itamu COM  canoe      say  out(PFV) DS       must(PASS) bless  PFV
 Itamu said of the canoe that it must be blessed
(The comitative preposition sati can be used with wañi and some other verbs to give the subject matter.)

This might not work: since (unlike dative i) sati does not delete in preverbal position, the raised tamwipaku is buried in a preposition phrase, and (if I understand all this right) can't be related by c-command or movement to the subject position in the subordinate clause, which in turn means (still assuming I understand all this) that there can't be a control relationship here. But I'll worry about that another time.

You can also have explicit subjects in na clauses; they must be licensed by ki:

Code: Select all

itamu ucini na hjaci ki  tamwipaku acitau jaku
Itamu want  DS Hjaci DET canoe     bless  PFV
Itamu wants Hjaci to bless the canoe
In English, this sort of case has the embedded subject rise to become the object of the matrix verb; in Akiatu, it can stay in the subordinate clause, marked in effect as a possessor.

Here is another example:

Code: Select all

itamu wañi aja      na tamwipaku ki  kunasi     acitau jaku
Itamu say  out(PFV) DS canoe     DET must(PASS) bless  PFV
Itamu said that the canoe must be blessed
The difference between this na clause and a (finite) kja clause is that in the latter but not the former the implied tense of the subordinate clause is independent of that of the matrix clause. For example, if you wañi aja kja P say that P, the reported P could be past, present, or future relative to the saying; if instead you wañi aja na S ki F, the implied tense of S's F-ing is dependent on that of the matrix clause (typically it is near-future). In general Akiatu prefers na ...ki over kja when both are semantically appropriate.

The subject in this construction can be dropped, as is generally true verb arguments in Akiatu, but the ki remains:

Code: Select all

itamu wañi aja      na ki  kunasi     acitau jaku
Itamu say  out(PFV) DS DET must(PASS) bless  PFV
Itamu said that it must be blessed
It may be counter-intuitive, but you can also also get an explicit subject in mwi clauses, even though they are marked as having the same subject as the matrix clause. In such cases the mwi implies a close relationship between the matrix and embedded subjects, most often possession, especially inalienable possession:

Code: Select all

itamu ucisu mwi hjakiwani ki  tawaru
Itamu want  SS  brother   DET sing
Itamu wants her brother to sing

Re: Akiatu scratchpad (subject control; "mwi")

Posted: Mon Oct 08, 2018 4:32 am
by Frislander
Akangka wrote: Sun Oct 07, 2018 11:39 am I have suggestion. If you're going to have Austronesian-esque conlang, you should have /ŋ/. And particle conparative (Yes, like English than)
No, there's no need a for either of these. Firstly Austronesian is a very broad family, encompassing over a thousand languages of wildly differing structures, but most people are only aware of a few of them. It appears from these two features you posit that you are thinking in particular of central-Malayo-Polynesian (i.e. Malay and Javanese), but "Austronesian" could also easily refer to languages as diverse as those of Polynesia, Vanuatu and Taiwan. Furthermore while many languages in the family do have /ŋ/, it is by no means universal, and the particle comparative is probably restricted entirely to those Central Malayo-Polynesian languages (see the WALS Map, where the particle comparatives among the Austronesian languages on the map are restricted to the western half of the Malayo-Polynesian family).

Akiatu scratchpad (secondary predicates)

Posted: Mon Oct 15, 2018 9:11 am
by akam chinjir
Secondary predicates

To make this whole thread a little easier to navigate, I've put a table of contents in the first post, as well as a link---which I'll continue to update---to the most recent main post.

In the last couple of posts I talked about nonfinite clausal complements. Using mwi for same subject:

Code: Select all

hjaci₁ tapikau   takau      mwi ____₁ wamau a   mikuwi hatau ka
Hjaci  have.plan REDUP(PFV) SS        go    LOC waters great TRANS
Hjaci₁ decided ____₁ to go to the ocean
And with na for different subject:

Code: Select all

kijapa hjaci₁ hwati tapikau   takau      na ____₁ wamau ka    a   mikuwi hatau
Kijapa Hjaci  CAUS  have.plan REDUP(PFV) DS       go    TRANS LOC waters great
Kijapa persuaded Hjaci₁ ____₁ to go to the ocean
These mwi- and na- clauses are complements because the verb in the matrix clause selects for a clausal argument. But similar clauses can also be used as adjuncts, and fill a sort of adverbial function.

There are a number of kinds of case to consider, but today I'll focus on just one: secondary predication. This is a construction that applies a subordinate predicate to one of the arguments of the matrix verb; you use mwi when that argument is the subject, and na when it is the object (or some other argument raised to focus position). In both cases, the secondary predicate normally follows the verb (Secondary predication is not possible with finite kja clauses.)

Cross-linguistically, secondary predicates come in two flavours, resultative and descriptive. In Akiatu, resultatives are handled with serial verb constructions, discussed previously (and no doubt to be discussed again). The secondary predicates I'll discuss here are purely descriptive.

One more thing: nothing really turns on whether the English sentences I use to translate my examples make use of secondary predicates, but I'll do my best to choose translations that are fairly good guides to the structure of the Akiatu.

Here is an example:

Code: Select all

kipaja wamau hja         ka    a   ikjamii mwi papija
Kipaja go    arrive(PFV) TRANS LOC river   SS  jump
Kipaja went to the river, running (= Kipaja ran to the river)
The verb wamau does not take a clausal complement, it selects only for a (locative-marked) destination. mwi papija is instead an adjunct, an adverbial clause. Its interpretation is as you might expect: mwi indicates same subject, so the clause offers a description of the subject of the main clause that fills out the statement's meaning. Here it does so by indicating the manner of Kipaja's movement, namely that he was running. (There's a running = jumping metaphor here, as also sometimes seen with the English "bound.")

In this example, the use of na indicates that it is the matrix verb's object rather than its subject that gets the further description:

Code: Select all

kipaja isau     taki wamau wa  na aitapi
Kipaja medicine hold come  CIS DS raw
Kipaja brought the medicine unprepared
Lexical aside: bringing is holding (taki) plus a path with a destination (wamau) plus cislocative motion (wa).

The tricky question is when exactly to use secondary predicates. The easy answer (so far) is that I haven't yet decided.

But there are still some details I can go through.

One thing that I didn't mention last time, but which is also true with clausal complements, is that these clauses can use nominal predicates. As in main clauses, these predicates can be marked with the affirmative particle iti, but this is not necessary:

Code: Select all

hau wamau wa  mwi (iti) wajari
1s  come  CIS SS   AFF  ally
I come as a friend
Locative predicates are also possible. In a finite clause, these would require a posture verb such as ijau to sit, but those verbs cannot appear (in this use) in a nonfinite clause; and without a verb, the locative preposition a is also not possible. As a result, the secondary predicate is just a locational noun phrase:

Code: Select all

kipaja isau     taki wamau     wa  na akitawa icai
Kipaja medicine hold come(PFV) CIS DS head    top
Kipaja brought the medicine on his head
The secondary predicate can also be a phrase with an explicit preposition:

Code: Select all

itamu jisaka piwa aja       na sati urakja
Itamu fish   eat  away(PFV) DS COM  yam
Itamu ate the fish with yams
It was not Itamu or the eating that was accompanied by yams, it was the fish, and this is indicated with the different-subject complementiser na.

Using the secondary predicates rather than a simple preposition phrase changes the meaning significantly. The PP analogues of the previous two examples both have meanings that are questionable at best:

Code: Select all

?kipaja isau     taki wamau     wa  akitawa icai
 Kipaja medicine hold come(PFV) CIS head    top
 Kipaja brought the medicine on his head (?)
This says that Kipaja's walking took place on his head, which is presumably nonsense.

Code: Select all

?itamu jisaka piwa aja       sati urakja
 Itamu fish   eat  away(PFV) COM  yam
 Itamu ate the fish accompanied by yams
Here again the preposition phrase characterises the event as a whole, not a particular argument: this says that the yams accompanied Itamu in eating the fish, not the sort of thing you are likely to want to say.

As with clausal complements, the matrix argument that controls the subject in a na subordinate clause must be in the focused position. Consequenttly, object control can operate across indefinite objects:

Code: Select all

kipaja apatu hwati mawa      i   itamu na ɲikatiwi
Kipaja spear give  find(PFV) DAT Itamu DS offering
Kipaja gave Itamu the spear as an offering of friendship
This calls the spear, and not Kipaja or Itamu, an offering.

Again as with clausal complements, object control can be fed by argument raising:

Code: Select all

kipaja itamu hwati mawa      apatu na ijau a   ikjamii kura
Kipaja Itamu give  find(PFV) spear DS sit  LOC river   bank
Kipaja gave Itamu the spear (as she was) sitting by the river
The use of na (rather than mwi) makes it unambiguous that it was Itamu, not Kipaja, who was sitting by the river; to provide the na clause with a subject, itamu must rise to the focus position, leaving the underlying direct object stranded after the verb.

I said last time that I wasn't sure whether to allow object control when the raised object has an explicit preposition---I was worried about technical matters involving movement and c-command. I've decided to hand-wave those worries away, so this sort of thing is now officially allowed:

Code: Select all

itamu niwa ki  apatu makjai   jisaka na tawaru
Itamu INST DET spear to.spear fish   DS sing
Itamu was fishing, her spear singing
The Akiatu construction here is signinficantly different from that of the English translation. In the English, "her spear" occurs in the subordinate clause, but the Akiatu ki apatu is in the matrix clause, complement to the instrumental preposition niwa.

Akiatu does allow an explicit subject in a secondary predicate, but only one that is related by inalienable possession to a controller in the matrix clause:

Code: Select all

itamu wamau ka    a   ikjamii mwi akitawa ki  saimuhi
Itamu go    TRANS LOC river   SS  head    DET hurt
Itamu is going to the river, her head hurt
The ki in the subordinate clause marks akitawa head as the possessor (= semantic subject) of saimuhi to be hurt, injured; Itamu's (inalienable) possession of the head in question need not be similarly marked.

Secondary predications can sometimes be fronted. Doing this backgrounds them, typically indicating that they serve a scene-setting purpose. This does not mean in any precise sense that such fronted clauses give old information, but in that position they are not part of the sentence's focus.

Code: Select all

mwi papija wai kja  kipaja wamau hja         ka    a   ikjamii
SS  jump   TOP COMP Kipaja go    arrive(PFV) TRANS LOC river
Running, Kipaja went to the river
Here, running is given as an additional background detail, but the focus is on Kipaja's destination. Note that the particle wai flags the fronted clause as backgrounded material, and that the main clause often takes an explicit complementiser, here kja, when there is a fronted subordinate clause.

It won't often be easy to think of a context in which it would make sense to front a secondary predicate in this way:

Code: Select all

?mwi wajari wai hau wamau wa
 SS  ally   TOP 1s  come  CIS
 As a friend I come (?)
Here my sense is that the natural English translation focuses "as a friend," but the Akiatu would normally do the opposite. One possibility (I think) would be to use this sentence for contrastive focus:

Code: Select all

mwi wajari wai hau wamau hja         wa, mwi taiñu wai ikihwa
SS  ally   TOP 1s  come  arrive(PFV) CIS SS  enemy TOP leave
As a friend I came, as an enemy I am leaving
Maybe not the best note to end on, but there you go.

Akiatu scratchpad (tija "now" and mikwa "already")

Posted: Sat Oct 20, 2018 12:23 pm
by akam chinjir
tija now, mikwa already, and acuta soon

This post will be about the three two title adverbs and their combinations. There are two main complications: interactions with aspect, and the use of these words in adverbial clauses. (Aside: the post was already too long even without acuta, so I'm holding off on that one for now.)

The normal position of these words is before the verb, between the two positions in which you can find an object, the focus position and the incorporated position. (And there can be significant differences in interpretation depending on whether an object precedes or follows these adverbs.)

tija now

tija now refers to the reference or narrative time---by default the time of speech, but this easily shifts, especially in narrative contexts. It never has the sequencing sense of and then.

You could be forgiven for thinking that tija always refers to the time you're already talking about, and therefore can never contribute anything to a sentence. Actually it's not that bad, even if you're only concerned with truth conditions.

This is most obvious when tija is used in a perfective clause; the result is an immediate-past perfect:

Code: Select all

amwikitu tija urasu jaku a   hjaci kiwa
Amwikitu now  enter PFV  LOC Hjaci inside
Amwikitu just now entered Hjaci     (← Amwikitu is an ancestral spirit)
The logic here is that a perfective treats an event as a whole, but a whole event cannot be wholly contained in the present moment (we can talk about the metaphysics there if you want :) ), so the event gets shunted back to the immediate past.

tija can also be used to jump out of a temporal digression, either from a narrative to the current moment, or out of a flashback or something and back into the main narrative sequence. In this use, tija often gets moved to the front of the sentence and marked with wai:

Code: Select all

tija wai hawi hakwai saka        cuwatai sai  ikwakawi hatau utami aja
now  TOP 1p   know   manage(PFV) why     COMP giants   great cease away(PFV)
And now we know why the great giants are no more
(tija... hakwai saka is actually also another recent past construction, something like we just learned, but that seems less idiomatic in English.)

tija also interacts with information structure. It tends to suggest a contrast with another time, usually past. How this plays out can depend on the position of an object, among other factors:

Code: Select all

kipaja isau    =su  tija ahjai japikuwa
Kipaja medicine=FOC now  CAUS  potent
Now Kipaja is preparing the medicine (before, he was preparing something else)

Code: Select all

kipaja tija isau     ahjai japikuwa
Kipaja now  medicine CAUS  potent
Now Kipaja is preparing medicine (before, he was doing something else)

Code: Select all

isau     wai kipaja ahjai japikuwa
medicine TOP Kipaja CAUS  potent
Kipaja is now preparing the medicine (before, he was doing something else with
    the medicine)
Finally, adverbial when or while clauses can be formed with tija:

Code: Select all

itamu papija tija  kja  hjakiwani tawaru
Itamu jump   while COMP brother   sing
Itamu danced while her brother sang
Such adverbial clauses will often be topicalised:

Code: Select all

tija  kja  hjakiwani tawaru wai, itamu papija
while TOMP brother   sing   TOP  Itamu jump
Itamu danced while her brother sang
A nonfinite adverbial clause is also possible:

Code: Select all

tija  na   hjakiwani ki  tawaru wai, itamu papija
while TOMP brother   DET sing   TOP  Itamu jump
Itamu danced while her brother sang
The nonfinite clause is more clearly marked as old information than its finite counterpart; a translation like Itamu danced during her brother's singing would maybe convey this better.

One last comment about this: in these constructions, tija isn't a preposition taking a clausal complements, it's still an adverb despite being fronted. (The proof of this is that tija can't take a nominal complement.)

mikwa already

English "already" tends to locate an expected event in the (possibly recent) past, often with the implication that it happened earlier than expected and often with focus on the resulting state of affairs. Akiatu mikwa overlaps with all this, especially in combination with tija now.

On its own, though, and in a perfective clause, mikwa resembles already only in locating the reported event in the past; there is no implication that the event was expected or predictable, much less that it took place earlier than expected, and mikwa actually moves the narrative present to the time of the reported event. In English the best translation in this sort of context might be "once."

Code: Select all

kipaja kiwa   mikwa   anatu      urasu jaku hu  jakwanai
Kipaja inside already meet(PASS) enter PFV  ABL ancestor
Kipaja was once entered by an ancestor
The English "Kipaja was already entered by an ancestor" would make sense only given the expectation that Kipaja was going to be entered (= possessed, sort of) by an ancestor at some time or another; the Akiatu has no such presupposition. Instead, this sentence could be used to introduce a narrative about that time when Kipaja was entered.

In an imperfective clause, mikwa operates a bit more like "already," indicating a contrast with a counterfactual state of affairs in which something expected or predictable has not yet come about:

Code: Select all

hjaci mikwa   suwasu
Hjaci already asleep
Hjaci is (already) asleep (that is, is not still awake)
There's still no implication that something has happened earlier than expected (or, for that matter, later than expected).

Maybe it's worth pointing out the difference here with tija: tija can imply a contrast with another (usually past) time, mikwa can imply a contrast with a counterfactual present.

You can get the implication that the event is earlier than expected by using tija mikwa now already, in that order:

Code: Select all

hjaci tija mikwa   suwasu
Hjaci now  already asleep
Hjaci is already asleep (so early!)
(There's probably a finally adverb too, but I haven't thought about that yet.)

tija mikwa is not a constituent in sentences like this one. For example, topicalisation can only front one of them (normally tija), never the pair.

The two words can also occur in the opposite order, and in this case they do constitute a constituent. mikwa tija already now produces a perfect of resulting state:

Code: Select all

kipaja mikwa   tija     ihjatu wa
Kipaja already now(PRF) arrive CIS
Kipaja has arrived (→ and is still here)
The verb phrase that occurs under mikwa tija need not be marked as perfective, but it will always be interpreted as describing something that took place in the past (relative to the narrative present at least): a sentence in mikwa tija is about a resulting state, not a continuing one. For example:

Code: Select all

itamu mikwa   tija     suwasu
Itamu already now(PRF) sleep
Itamu has already slept (→ and is now awake and refreshed)
(Contrast itamu (mikwa) suwasu Itamu is (already) sleeping.)

A mikwa tija predicate is stative, which means that it is negated with miwa rather than hwai:

Code: Select all

kipaja miwa mikwa   tija     ihjatu wa
Kipaja NEG  already now(PRF) arrive CIS
Kipaja has not already arrived
(Neither mikwa on its own nor tija mikwa affect the choice of negator.)

Like tija, mikwa can be used to form adverbial clauses; in this contexts it'll mean after:

Code: Select all

itamu kitikwa  wañi kihwa        mikwa kja  hjaci hakjawi acitau jaku
Itamu REFL(AP) say  set.out(PFV) after COMP Hjaci fire    bless  PFV
Itamu started speaking after Hjaci blessed the fire
These adverbial clauses can also be topicalised:

Code: Select all

mikwa kja  hjaci hakjawi acitau jaku wai, itamu kitikwa  wañi kihwa        
after COMP Hjaci fire    bless  PFV  TOP  Itamu REFL(AP) say  set.out(PFV) 
After Hjaci blessed the fire, Itamu started speaking
tija mikwa adverbial clauses normally move the tija, not the mikwa (remember that they can't both move, since they don't form a constituent):

Code: Select all

tija kja  wapanaiwi mikwa   ijau      ihjatu wai, 
now  COMP elders    already sit(IPFV) arrive TOP  
    hjaci ikau       hakjawi acitau jaku
    Hjaci right.then fire    bless  PFV
With the elders already arriving, Hjaci blessed the fire
(The topicaliser aka, rather than wai, together with ikau right then in the main clause, helps indicate that the main clause action was spurred by the topicalised event.) The bit here about aka is wrong, and I've updated the examples with wai to correct the mistake.

By contrast, since mikwa tija is a constituent, it moves as a unit to the front of an adverbial clause:

Code: Select all

mikwa   tija     kja  hjaci ikihwa aja       wai,
already now(PRF) COMP Hjaci leave  away(PFV) TOP
    hakjawi=wati ikau miwa ijau mainaki na acitau
    fire   =LOC  then NEG  sit  no.one  DS bless
With Hjaci having left, there was no one to bless the fire
Coda

This is already far too long, so no detail yet about acuta soon, except that I'll mention that it's sort of a mirror-image of mikwa already, and in particular that it can mean before when it starts an adverbial clause.

There's also a chance I'll work an evidential meaning into some use of mikwa, not sure yet though.

Akiatu scratchpad (acuta "soon")

Posted: Wed Oct 24, 2018 1:38 pm
by akam chinjir
acuta soon, &c

So here's what would've been the rest of the last post, on acuta, sort of a future-oriented counterpart of mikwa already. I've added a few odds and ends at the end.

acuta locates a reported event or state of affairs in the future. Its precise sense tends to vary with the kind of speech act.

In predictions (mostly with third person subjects), it picks out the near future. In this case, it also shifts the narrative present to the time of the reported event or state of affairs:

Code: Select all

kamiru acuta ihjatu jaku
hunter soon  arrive PFV
The hunters are going to arrive
(Likely, the subsequent sentence would continue to describe the hunters' arrival or make other predictions about that time.)

In commitments, assurances, or statements of intention (mostly with first person subjects), it also picks out the near future, though without the shift in narrative present:

Code: Select all

hau acuta isau     taki wamau wa
1s  soon  medicine hold come  CIS
I'll bring the medicine soon
In consolations (mostly in first or second person), it picks out an indefinite time in the future, again without moving the narrative present. In this use, it can be reinforced with ituwani or wanitu sometime, anytime; one of these adverbs is all but required to get this sense with a third person subject.

Code: Select all

hau/sama acuta wamau hja         a   mikuwi hatau (ituwani)
1s /2s   soon  go    arrive(PFV) LOC waters great  sometime
Someday I/you will go to the ocean
All of these senses are also possible in imperfective clauses. Here's an example:

Code: Select all

wapanaiwi acuta timarikau
elders    soon  chant
The elders are going to chant (soon)
acuta gets a more emphatic sense when it occurs along with tija now or mikwa already, both of which it follows.

With tija, two senses are possible: either that the reported event will take place sooner than expected, or that action must be taken now to prevent the predicted event.

Unexpectedly soon:

Code: Select all

tija acuta cautuka
now  soon  rain
It's about to rain (← sooner than expected)
Requiring prevention:

Code: Select all

hakjawi tija acuta utami aja
fire    now  soon  cease away(PFV)
The fire is about to go out (← you'd better do something about that)
With mikwa already, acuta implies that the predicted event is impending and that action must be taken in anticipation of it; you could use mikwa acuta in a warning, for example.

Code: Select all

wamikawi mikwa   acuta cautuka
airs     already soon  rain
It's about to rain (← warning about a storm)
Neither tija acuta nor mikwa acuta are constituents.

acuta can also precede tija now, and in this case the combination is a constituent. In this and two other respects acuta tija resembles mikwa tija already now, which, recall, results in a perfect of resulting state. The two other similarities are that acuta tija also neutralises the perfective/imperfective distinction in the embedded phrase, and that it selects the stative negator miwa (rather than hwai). The meaning of acuta tija is not yet: an expected event has not yet taken place.

Code: Select all

jaikawi acuta tija kitikwa  kamaisu
yams    soon  now  REFL(AP) begin
The feast has not begun yet
All these expressions can move to the front of a subordinate clause, with a resulting meaning related to before. (The claims I've made about constituency are supported by the fact that acuta tija can but tija acuta and mikwa acuta cannot move like this.)

An adverbial clause with acuta before:

Code: Select all

acuta na kamuri ki  ihjatu wai, hjaci urakja acitau jaku
soon  DS hunter DET arrive TOP  Hjaci yam    bless  PFV
Before the hunters arrived, Hjaci blessed the yams
With tija... acuta, implying an imminent event that maybe needs to be prevented:

Code: Select all

tija kja  hakjawi acuta utami aja       wai,
now  COMP fire    soon  cease away(PFV) TOP
  hjaci ikau witamwi ahjai ijau jau
  Hjaci then wood    CAUS  sit  REDUP(PFV)
The fire was about to go out, so Hjaci added some wood
(Note the ikau in the main clause, implying that the main clause action was spurred by the possibility reported in the subordinate clause. Last time I mistakenly said that the topicaliser aka could be used instead of wai with the same implication; oops, that was just wrong.)

An example with mikwa... acuta, implying something impending that must be prepared for:

Code: Select all

mikwa   na jaikawi ki  acuta kitikwa  kamaisu kihwa        wai, 
already DS yams    DET soon  REFL(AP) begin   set.out(PFV) TOP
    hikunai   ikau wamau a   hakjawi=wati wa
    everybody then come  LOC fire   =LOC  CIS
When the feast was about to begin (= impending),
    everyone was coming to the fire
And with acuta tija, meaning not yet:

Code: Select all

acuta tija na  kamuri ki  ihjatu wai,
soon  now  DS  hunter DET arrive TOP
    itamu ijau jisaka makjai
    Itamu PROG fish   spear
With the hunters not yet having arrived,
    Itamu was fishing
Some odds and ends
  • tija now can be reduplicated; the resulting tijatija means now and then, sometimes. Its meaning can be reinforced by ituwani sometime.
  • There's another adverb tikai again, still that maybe should be grouped with the ones I've been discussing. (It'll mean still when the main verb takes the continuative complement wamau, also come, go.)
  • tikai also has a reduplicated form; tikaitikai means over and over, on and on.
  • For whatever it's worth, when more than one of these adverbs occur, their normal order (in which they remain separate constituents) conforms to the following ranking:
    tija now < tikai again < mikwa already < tikai still < acuta soon
    Note the two different positions for tikai. mikja tija and acuta tija both occur in the same position as tija on its own.

Akiatu scratchpad (numbers)

Posted: Sat Oct 27, 2018 4:34 am
by akam chinjir
Numbers

Cardinal numbers are base five up to twenty and thereupon base twenty. Here they are, up to nineteen:

1itu
2ami
3pai
4cita
5haku
6haku (sati) itu
7haku (sati) ami
8haku (sati) pai
9haku (sati) cita
10amiku
11amiku (sati) itu
12amiku (sati) ami
13amiku (sati) pai
14amiku (sati) cita
15paiku
16paiku (sati) itu
17paiku (sati) ami
18paiku (sati) pai
19paiku (sati) cita
20harati

The system is simple and perfectly regular. The comitative preposition sati is optional but uncommon here.

The ensuing numbers up to 39 are formed with harati 20 followed by the number given above for the remainder; sati is possible, but only directly following harati. Thus, 32 can be either harati amiku ami or harati sati amiku ami.

40 is haratiwi ami two twenties. Ensuing numbers are formed as with harati, except that now sati is required; for example, 42 is haratiwi ami sati ami.

The same system can be followed up to 399, which is haratiwi paiku cita sati paiku cita.

400 is amirati, 800 is amiratiwi ami, and so on. Higher numbers are formed as you might expect. Precise numbers this high aren't often (ever?) necessary, but they're formed about how you'd expect. For example, 6552 (= 16*400 + 7*20 + 12 → 16*400 sati 7*20 sati 12) would be amiratiwi paiku itu sati haratiwi haku ami sati amiku ami. Note that sati is required both after the number of 400s and the number of 20s. This means that above 8000 this system can in theory lead to ambiguities.

8000 can be papairati, with the syntax you might guess, but most often it's used on its own and just indicates an indefinitely high number.

Numbers follow the noun, possibly separated by one of Akiatu's few true adjective; relative clauses and deictic clitics follow:

janaki hakufive people
janaki amaki hakufive good people
janaki haku na suwasufive sleeping people
ki janaki haku=watithose five people

There's a complication: the cardinal numbers itu one and ami two are never used in this way. Instead we find the adjectives ahiwa one, solitary, alone, whole, unique and iruwa two, paired. These can actually co-occur with each other and with the (other) cardinal numbers; that they are syntactically distinct from the (other) cardinal numbers can be shown by their position relative to the adjectives inisa same and tañuci other.

janaki ahiwaone person
janaki iruwatwo people
janaki ahiwa iruwatwo solitary people
janaki iruwa ahiwaone pair of people
janaki ahiwa hakufive solitary people
janaki iruwa hakufive pairs of people
janaki ahiwa inisathe same one person
janaki ahiwa tañucione other person
janaki iruwa inisathe same two people
janaki iruwa tañucitwo other people
janaki inisa hakuthe same five people
janaki tañuci hakufive other people

One-word numbers up to at least harati 20 can be reduplicated to get a distributive sense; in this usage, the numbers itu one and ami two are used rather than their corresponding adjectives.

Code: Select all

itamu sati hjaci apatu ami ami   taki  wamau wa
Itamu COM  Hjaci spear two REDUP carry come  CIS
Itamu and Hjaci brought two spears each
OR: Itamu and Hjaci brought spears two at a time
Some funky syntax is possible here:

Code: Select all

itamu sati hjaci apatu taki  wamau ami ami  =wati wa
Itamu COM  Hjaci spear carry come  two REDUP=DEIC CIS
Itamu and Hjaci brought two spears each
OR: Itamu and Hjaci brought spears two at a time
Here the object is in incorporated position directly before the verb, where only the head noun can appear; the remainder of the noun phrase is left stranded after the verb, where a final deictic element is obligatory. (Cf. the post on direct objects.)

If the subject is singular, only the at a time meaning is available:

Code: Select all

itamu apatu ami ami   taki  wamau wa
Itamu spear two REDUP carry come  CIS
Itamu brought spears two at a time
Reduplicated numbers can also be used as adverb phrases, again with only the at a time meaning available. A few constructions are possible, each somehow making it clear that the reduplicated number is not part of the object noun phrase:
  • itamu sati hjaci ami ami apatu taki wamau wa---the object is in incorporated position, after the number phrase
  • itamu sati hjaci apatu=su ami ami taki wamau wa---the object takes the focusing su, which must be the final element in the noun phrase
  • ami ami wai itamu sati hjaci apatu taki wamau wa---the number phrase is topicalised, which is not in general possible for a constituent of the object
Ordinal numbers use a relative clause. I haven't really posted about those yet, but for now I'll use nonfinite relative clauses headed by na, and hope I don't change my mind about these yet again. For first, use hjasi (otherwise nose). For other ordinals, use tai followed by the appropriate cardinal number (using tai ami rather than tai iruwa for second):

cucu na hjasifirst child
cucu na tai amisecond child
cucu na tai paithird child

Relatedly, one can use kasu to follow for the next item in a sequence and the adjective tañuci other for the last item:

cucu na kasunext child
cucu tañucilast child

Update: fixed some stupid errors, all minor.

Re: Akiatu scratchpad (relative clauses, I)

Posted: Wed Oct 31, 2018 11:37 pm
by akam chinjir
Relative clauses, I

I guess I'm going to commit on this. It'll require a couple of posts.

It might help if I start with a review of the process I call argument raising:
  • Preposition phrases, whether giving indirect objects or adjuncts, normally occur after the verb. Direct objects normally precede the verb.
  • Preposition phrases can however be moved to a position before the verb---to what I've called the focus position. If the verb is transitive, this will leave any overt direct object after the verb.
  • When raised in this way, the prepositions i (dative) and a (locative) delete, leaving just a nominal phrase.
Anyway the basic typological facts are these. Relative clauses follow the noun and are obligatorily headed by either kja (for finite clauses) or na (for nonfinite ones). A relativised subject or object is simply gapped; anything else is either argument-dropped or replaced by a resumptive pronoun. Work similar to that done by relative clauses can also be done with other constructions, such as topicalised clauses, including coordinate structures, secondary predicates (discussed earlier), and simple juxtaposition.

I'll start with the simplest cases: kja (finite) relative clauses that relativise the subject or object of the embedded clause. I'll re-use the strategy of indicating gaps explicitly and using subscripts to indicate coreference.

Code: Select all

janaki₁ kja  ____₁ jisaka makjai mawa
person  COMP fish  spear  find(PFV)
the person₁ that ____₁ caught the fish

Code: Select all

jisaka₁ kja  janaki ____₁ makjai mawa
fish    COMP person spear find(PFV)
the fish₁ that the person caught ____₁
You'll maybe notice that the two relative clauses here have exactly the same apparent structure: nothing overt indicates that jisaka is object in the first one but janaki is subject in the second. And that's to say that both are ambiguous, strictly speaking: the first could mean the person that the fish caught, and the second could mean the fish that caught the person.

The issue here is somewhat general. Subjects and objects both precede the verb, but there is no case-marking, so when a transitive verb has only one of them overt, it might not be obvious whether that's the subject or the object. Normally this will be obvious just from context and common sense, and when it's not obvious main clauses allow both arguments to be made explicit---so they can be distinguished by their relative order. But that route is not available in relative clauses, since a gapped subject or object cannot be made overt.

There are three things that could be done to disambiguate relative clauses like the ones above.
  • If it's the subject that's getting relativised (and therefore gapped), include an overt object marked with the focus clitic su (which won't have its usual informational significance in this context).
  • If it's the object that's being relativised, make the verb passive.
  • In either case, instead use a na relative clause, on which more below.
So on its intended meaning the first example above could be adjusted in either of these ways:

Code: Select all

janaki₁ kja  ____₁ jisaka=su  makjai mawa
person  COMP       fish  =TOP spear  find(PFV)
the person₁ that ____₁ caught the fish

Code: Select all

janaki₁ na ____₁ jisaka makjai mawa
person  DS       fish   spear  find(PFV)
the person₁ that ____₁ caught the fish
And the second in these ways:

Code: Select all

jisaka₁ kja  ____₁ anatu      makjai mawa      hu  janaki
fish    COMP       meet(PASS) spear  find(PFV) ABL person
the fish₁ that ____₁ was speared by the person

Code: Select all

jisaka₁ na janaki ki ____₁  makjai mawa
fish    DS person DET       spear  find(PFV)
the fish₁ that the person caught ____₁
Indirect objects introduce some further complications. First, here's an example of a sentence with the indirect object raised to a preverbal position:

Code: Select all

kipaja itamu=su  hwati mawa      apatu
Kipaja Itamu=FOC give  find(PFV) spear
Kipaja gave Itamu a spear
This allows the indirect object to be relativised just like a regular object:

Code: Select all

janaki₁ kja  kipaja ____₁ hwati mawa      apatu
person  COMP Kipaja give  find(PFV) spear
the person₁ that Kipaja gave ____₁ a spear
(This could also mean the person who gave Kipaja a spear, and could be disambiguated either via passivisation or by using instead a na relative clause.)

English allows a couple of other constructions, and it's maybe worth mentioning why you can't do analogous things in Akiatu.
  • "the person that Kipaja gave a spear to." This gaps the complement of a preposition. In Akiatu (like in many other languages) the analogous construction would violate an island constraint and is ungrammatical.
  • "the person to whom Kipaja gave a spear." Here a relative pronoun has pied-piped the preposition to the front of the clause. Akiatu doesn't have relative pronouns, however, so pied piping is impossible here.
Akiatu can however use a resumptive pronoun, yielding something like this:

Code: Select all

janaki₁ kja  apatu hwati mawa      i   kinaki₁
person  COMP spear give  find(PFV) DAT RESUME
the person₁ that Kipaja gave a spear to ____₁
There are three main resumptive pronouns, differing by animacy (a bit roughly speaking):
  • kinaki, for human beings
  • kimuki, for animals and forces of nature, including ancestors
  • kimija, for everything else
(I'll discuss the distinctions here in more detail sometime soon.)

Resumptive pronouns can be used with other prepositions as well:

Code: Select all

janaki₁ kja  itamu papija tau           sati kinaki₁
person  COMP Itamu jump   together(PFV) COM  RESUME
the people₁ that Itamu danced with ____₁

Code: Select all

apatu₁ kja  Itamu jisaka makjai mawa      niwa kimija₁
spear  COMP Itamu fish   spear  find(PFV) INST RESUME
the spear₁ that Itamu caught the fish with ____₁
Note that argument raising wouldn't obviate the need for a resumptive pronoun; unlike dative i (and locative a), the prepositions sati and niwa are retained in preverbal position.

They're also required for possessors, whether inalienable or alienable:

Code: Select all

janaki₁ kja  hau kinaki₁ ki  apatu kwaipitu wukau
person  COMP 1s  RESUME  DET spear break    broken
the person₁ whose₁ spear I broke

Code: Select all

hau cucu₁ kja  kinaki₁ takauni    ahwaicu  hu  citaisita
1s  child COMP RESUME  sex.father lie.down ABL Citaisita (← a village)
my child₁ whose₁ father comes from Citaisita
(A brief aside on family structure. A child's takauni is the man as a result of sex with whom the child's mother became pregnant with the child. The woman doesn't normally have trouble knowing who this is---maybe because of mores, but definitely because of magic (roughly speaking). But the takauni is not one of the child's primary caregivers and there's nothing really analogous to marriage; his role in the child's life is maybe analogous to that of an uncle or a godfather. There's usually a primary male caregiver, but it's usually one of the woman's brothers or cousins, not a sex partner, and not necessarily the same one for different children.)

(And an aside on crazy relative clauses. "The man as a result of sex with whom the child's mother became pregnant" is gread pied piping, but Akiatu would probably go with something more like "the man who the child's mother had sex with (him) and thereby became pregnant." I don't yet have this vocabulary---or the associated mores---worked out, but if "had sex with" ends up a transitive verb then the object will just be gapped; a resumptive pronoun will be necessary only if the sex partner is given in a preposition phrase.)

A full discussion here of islands and such could be very complicated. I'll skip that for now.

All the above examples have finite relative clauses headed by kja. There are also nonfinite ones headed by na.

(Yet another aside. I was going back and forth on the question of whether to make something of the mwi/na distinction in relative clauses. Maybe you'll remember that clauses headed by mwi inherit their semantic subject from the subject in the matrix clause. I had a vague idea that a mwi relative clause could be one that relativises its subject, but couldn't make sense of that idea without making mwi a relative pronoun, which I didn't want---both because the whole idea was to use the same complementisers that show up in other complement clauses, and because relative pronouns are about as SAE a feature as a language could hope to have. Anyway the result of all this was that all nonfinite relative clauses will use na.)

Here are the initial examples from above in their versions with na:

Code: Select all

janaki₁ na ____₁ jisaka makjai mawa
person  DS       fish   spear  find(PFV)
the person₁ that ____₁ caught the fish

Code: Select all

jisaka₁ na janaki ki  ____₁  makjai mawa
fish    DS        DET person spear  find(PFV)
the fish₁ that the person caught ____₁
As you can see, when the embedded clause relativises its subject, as in the first example, the subject is gapped, and, importantly, there is no ki. ki is required, though, when the subject is not gapped, as in the second example. Note that there must still be a ki if the subject is pro-dropped:

Code: Select all

jisaka₁ na ki  ____₁ makjai mawa
fish    DS DET       spear  find(PFV)
the fish₁ that she caught ____₁
You can also used na relative clauses with raised objects and resumptive pronouns, as in the examples with kja above.

I should acknowledge that I'm still mulling the question of what exactly distinguishes nonfinite from finite clauses, semantically speaking. I hope it has something to do with a lack of an independent time reference, and I've tended to assume that the perfective/imperfective distinction gets neutralised in nonfinite clauses. On the other hand, it's clear that mikwa already and acuta soon can be used to indicate relative time. But there's a lot that's not nailed down.

Anyway that what I've got to say about the main structures that look like relative clauses and act like relative clauses. It leaves some pretty major loose ends though: nonrestrictive and headless relative clauses, relativising on adverbials like watiwi there, and other structures that do similar things. Probably my next post will cover some of that.

Akiatu scratchpad (relative clauses, II)

Posted: Sun Nov 04, 2018 6:56 pm
by akam chinjir
Relative clauses, II

This post will follow up on some points that I didn't address last time. Not everything though: there'll have to be at least one more post, on correlative structures.

Little relative clauses

(This really should have had a mention last time.)

Akiatu mostly has stative verbs where English has adjectives, and it mostly has little relative clauses where English has attributive adjectives. So little relative clauses are very common. Here's an example:

Code: Select all

wipakija  na jasijasu
butterfly DS be.colourful
a/the colourful butterfly
A reminder of some details. In a relative clause, na without ki implies that the clause is nonfinite (that's the na) with a gapped subject (that's the lack of ki). That'll always be what you want in these little relative clauses.

Another detail: a noun phrase on its own can function as a predicate, so you can also have little relative clauses consisting of na + a noun:

Code: Select all

kipaja na isaunaki
Kipaja DS healer
Kipaja, who is a healer
OR: Kipaja the healer
As the last example maybe implies, little relative clauses like these can be restrictive or nonrestrictive with no overt signalling of the difference. They can also be used for epithets, as in the second translation given of the last example, or as in this:

Code: Select all

itamu na ipatu
Itamu DS spear
Itamu the spear
Naturally this construction allows a fair bit of metaphor and metonymy.

It also allows nesting, with ordering restricted if at all by semantics.

Headless relative clauses

It's maybe worth a mention that there are no headless relative clauses. But you can come close.

Relative clauses can be headed by third person pronouns, giving meanings like English "the one who/that":

Code: Select all

kai kja  hakjawi acitau jaku
3s  COMP fire    bless  PFV
the one who blessed the fire
Various constructions with indefinite pronouns and generic nouns are also possible:

Code: Select all

cunai  na itamu kjaitiwa nai
anyone DS Itamu insult   PFV
anyone who insults Itamu

Code: Select all

janaki na kasu   a   pumuki
person DS follow LOC current
people who go with the current
(The correlative constructions that I'll talk about in another post can also often be used where you think you want a headless relative clause, like a "whoever" clause, for example.)

Relativising adverbials

Here's a sentence with a locative adjunct:

Code: Select all

cucuwi   papija parawara a   witamwi karau
children jump   wander   LOC tree    foot
The children are playing at the foot of the tree
Based on the last post, you could construct relative clauses of the following sort:

Code: Select all

witamwi kja  cucuwi   papija parawara a   kimija karau
tree    COMP children jump   wander   LOC RESUME foot
the tree at the foot of which the children are playing
You might also expect to be able to say something more like the tree where the children are playing, and you can, but it requires the resumptive adverbial watiwi:

Code: Select all

witamwi kja  cucuwi   papija parawara watiwi
tree    COMP children jump   wander   there
the tree where children are playing
The watiwi can be omitted, but mostly only when the head of the relative clause is (also) the pronoun watiwi, or a noun with a meaning close to place, especially wakwai. (With wakwai you could interpret a subsequent clause as a noun complement rather than a relative clause, noun complements having exactly that syntax, equivalent to a relative clause with no gap and no resumptive pronoun.)

That's how it works with locative adverbials. Similar constructions are also possible with kiwani then and kimata for that purpose, for example.

Nonrestrictive relative clauses

The little relative clauses I discussed above don't have to differ according to whether they're restrictive or nonrestrictive. Longer relative clauses, especially ones with kja, are likely to be prosodically marked as parenthetical, maybe by bracketing pauses, a lowered pitch, and a faster pace. Something like this:

Code: Select all

kipaja, kja  ijau      ijaisa ijaisa tikwa kiji  aja,     
Kipaja  COMP sit(PROG) bat    REDUP  face  smile out(PFV)
    apatu hwati mawa      i   itamu
    spear give  find(PFV) DAT Itamu
Kipaja, who was smiling like a bat, gave the spear to Itamu
(Yeah, that's a very English-y way to distinguish nonrestrictive relative clauses. Maybe I'll eventually come up with something more interesting.)

There's a strong tendency to make the additional comment parenthetical not just prosodically and semantically but also as it were syntactically. That is, rather than a nonrestrictive relative clause, you'll often get a syntactically independent sentence.

Code: Select all

kipaja, kai  ijau      ijaisa ijaisa tikwa kiji  aja,     
Kipaja  COMP sit(PROG) bat    REDUP  face  smile out(PFV)
    apatu hwati mawa      i   itamu
    spear give  find(PFV) DAT Itamu
Kipaja, he was smiling like a bat, gave the spear to Itamu
(My inclination is to say that the coreferring pronoun---here kai---cannot be dropped from the embedded sentence, as if it were also as it were pragmatically parenthetical.)

That embedded sentence can also be shunted to the end of the sentence, like so:

Code: Select all

kipaja apatu hwati mawa      i   itamu,
Kipaja spear give  find(PFV) DAT Itamu,
  kai ijau      ijaisa ijaisa tikwa kiji  aja
  1s  sit(PROG) bat    REDUP  face  smile out(PFV)
Kipaja gave the spear to Itamu, he was smiling like a bat
In these examples, kai 3s isn't a relative pronoun, it's just the subject of another sentence that embedded in but not syntactically integrated with the main statement.

You'll also often find secondary predication where you might expect a nonrestrictive relative clause:

Code: Select all

kipaja apatu hwati mawa      i   itamu
Kipaja spear give  find(PFV) DAT Itamu
    mwi  ijaisa ijaisa tikwa kiji  aja
    SS   bat    REDUP  face  smile out(PFV)
Kipaja gave the spear to Itamu, smiling like a bat
The use of mwi shows that this must be a secondary predicate rather than a shifted relative clause.

One more thing

As you might expect, you can use an independent prefatory statement rather than a restrictive relative clause, something like this:

Code: Select all

hawi kimuri=su  kiwaita mawa      saraikuti, kiwi ijau a   hakjawi=wati
1p   hunter=FOC see     find(PFV) yesterday  3p   sit  LOC fire   =LOC
Yesterday we saw some hunters, they are sitting by the fire
But Akiatu loves its subordinate clauses, and isn't above subordinating them by putting them in topics, so something like this might actually be preferable:

Code: Select all

saraikuti kja  hawi kimuri=su  kiwaita mawa      wai,
yesterday COMP 1p   hunter=FOC see     find(PFV) TOP
    kiwi ijau a   hakjawi=wati
    3p   sit  LOC fire   =LOC
Yesterday we saw some hunters, they are sitting by the fire
But in that direction lie correlative clauses, which need a post all their own; so that's it for now.

Re: Akiatu scratchpad (subject control; "mwi")

Posted: Sun Nov 04, 2018 9:38 pm
by Xwtek
akam chinjir wrote: Sun Oct 07, 2018 7:02 pm Not aiming for it to seem Austronesian, though. (Don't know enough about Austronesian languages even to try it. Seems like an unreasonably broad target, anyway.)

On particulars, I decide for /ɲ/ over /ŋ/ (and didn't want both). Comparatives will be parasitic on motion/distribution constructions; locative constructions will be possible but probably not hugely common. Currently you can't form comparatives with true adjectives, only with verbs, though that might change.

It should look more or less like this:

Code: Select all

hjaci kinaisu   iwasu     a   itamu
Hjaci surprised go.beyond LOC Itamu
Hjaci is more surprised than Itamu

Code: Select all

hjaci kinaisu   a   itamu aiwa
Hjaci surprised LOC Itamu beyond
Hjaci is more surprised than Itamu

Code: Select all

hjaci pahupahu maita a   itamu
Hjaci angry    reach LOC itamu
Hjaci is as angry as Itamu

Code: Select all

hjaci pahupahu a   itamu=wati
Hjaci angry    LOC Itamu=DEIC
Hjaci is as angry as Itamu
your 'iwasu', 'aiwa', looks like how Indonesian works, though.

Dua itu lebih besar dari tiga.

Two DEM more big from three.

Your iwasu could have a meaning shift to exclusively comparative

Re: Akiatu scratchpad (subject control; "mwi")

Posted: Sun Nov 04, 2018 10:14 pm
by akam chinjir
Akangka wrote: Sun Nov 04, 2018 9:38 pm your 'iwasu', 'aiwa', looks like how Indonesian works, though.
Oh, interesting. I guess that sort of pattern isn't too unusual? (I'm trying to do a fair bit with path verbs like iwasu and locational nouns like aiwa, though it's not very apparent in what I've posted so far, I think.) Fwiw, the most immediate inspiration for iwasu is gwo 過 in Cantonese, like here:

Code: Select all

佢靚過我
kéui leng   gwo     óh
3s   pretty go.past 1s
S/he's prettier than me
Dua itu lebih besar dari tiga.
Two DEM more big from three.
Hmm, seems like I have a determiner in common with Indonesian as well, or near enough. itu janaki is some/any person (or itunai is someone, anyone.
Your iwasu could have a meaning shift to exclusively comparative
Yeah, things like that will happen---maybe not this particular case (not sure yet), but there end up being lots of ways things can combine in the main verb complex, and the plan is for those to grammaticalise in different ways with different daughter languages.

Re: Akiatu scratchpad (relative clauses, I and II)

Posted: Mon Nov 05, 2018 9:24 am
by Vlürch
This seems really cool, I like how minimalistic the phonemic inventory is. Makes me think of some South American indigenous language for some reason, maybe Warao or something. Then again, I could just be thinking of Warao because Warao...🤤 Also seconding the Austronesian vibe. Also, a kind of Japanesey vibe somehow. Basically, it's great how your language is unique enough to point to so many (vaguely similar but still very different) directions.

Re: Akiatu scratchpad (relative clauses, I and II)

Posted: Tue Nov 06, 2018 5:40 pm
by akam chinjir
Hey, thanks! I definitely have lots to learn, about many languages. (I do know just enough about Japanese to occasionally avoid word forms that strike me as sounding somehow too Japanese.)

Akiatu scratchpad (correlative clauses)

Posted: Fri Nov 16, 2018 2:30 am
by akam chinjir
Relative clauses, III: Correlatives

This post is about what I'll call correlative clauses. These are often typed as a sort of relative clause, and that's why I introduce them in this context. They do often correspond to restrictive or headless relative clauses, especially ones headed by free choice indefinites such as "whoever" or "anything."

Here's a simple example, to help fix ideas:

Code: Select all

hau cunai  =su  kiwaita mawa      wai, kinai ikau amau.amau tikwa
1s  whoever=FOC see     find(PFV) TOP, RESUM then IDEO      face
Whoever I see, I will embrace
(Okay, there's an orthogonal complication here, an ideophone. This one describes someone as overflowing with life and contentment. The tikwa face lets it function as a predicate. Embracing is likely implied, but it's not really the meaning.)

Some things to notice:
  • The correlative clause is subordinate and topicalised (here with wai).
  • The construction makes use of a pair of pronouns, the indefinite cunai in the correlative clause and the resumptive kinai in the main clause. The indefinite is being used here as a free-choice pronoun, though it's the same in form as the question word meaning who?.
  • The indefinite pronoun is focused; here there's an explicit focus particle (the su), though that's not strictly necessary.
  • The resumptive pronoun is topicalised. Here I'm trying out a new idea, and just putting it in the usual subject slot (it's before the ikau). The alternative would be to make it an explicit topic, presumably with wai; Akiatu does allow nested topics, but I think not here. (I've started thinking that Akiatu will allow morphologically unmarked valency shifts at least in certain special contexts. This'll require some revision to things I've said before, though maybe not a tremendous amount.)
  • The locations of the pronouns in the two clauses is otherwise nothing special: the indefinite pronoun in the correlative clause goes where focused phrases go, and the resumptive pronoun in the main clause goes in a topic position (sort of). The main thing: this construction doesn't involve wh-movement, everything is in situ. (Actual questions do usually involve wh-movement.)
Here are the available pronoun pairs, with glosses corresponding to the indefinites:

IndefiniteResumptiveGloss of indefinite
cunaki, cunaikinaiwhoever, anybody, who?
cumukikimukiwhatever, anything, what? (of animates)
cumija, cumiikimijawhatever, anything, what? (of inanimates)
cutiwakitiwawhatever kind, any kind, what kind?
cuwatisuwi, kuwi, watiwiwherever, anywhere, where?
cuwakwaisuwi, kuwi, watiwiwherever, anywhere, where?
cuwanikiwaniwhenever, ever, when?
cumukikimukihowever, somehow, how?
cuwataikiwataifor whatever cause, why?
cumatakimatafor whatever purpose, why?

This is all too regular, and might get adjusted at some point. Some subtleties:
  • Except for kinai, the resumptive pronouns are also just the pronouns (or proforms, for the adverbials).
  • cunaki and cunai whoever do not differ in meaning at all. There may be isoglosses, when I get around do doing those, but for now these are just in free variation. Note that they correspond to the same resumptive pronoun, kinai.
  • And the same for cumija and cumii whatever.
  • There are two classes of nonhumans, the -muki animates and the -mija inanimates. The animate class actually leaves out teeny things like individual insects but includes forces of nature such as rivers, fires, and storms (and swarms of insects). (The -muki actually relates to pumuki current.)
  • The -muki forms have another use corresponding to manner adverbs (however). (Besides current, pumuki also means manner, way.)
  • The suwi/kuwi/watiwi distinction is three-way proximal, with suwi for places near the speaker, kuwi for places near the listener, and watiwi, which is proximity-neutral. (Note that that means watiwi can be used of places near either discourse participant, it just doesn't specify.) (ku used to be kisa, I decided I didn't like that form.)
  • cuwati and cuwakwai wherever differ subtly in meaning, the latter tending to imply that it's discrete locations or sites at issue, the former more neutral.
  • cuwatai from whatever cause an cuwatai for whatever purpose correspond to efficient and final becauses, respectively.
Rather than an indefinite proform, you can also use the determiner cuwa whichever, any with an arbitrary noun.

Here's an example with kuwa:

Code: Select all

hawi cuwa kamuri=su  kiwaita mawa      saraikuti wai,
1p   any  hunter=FOC see     find(PFV) yesterday TOP
    kinai  ijau a   hakjawi=wati
    RESUME sit  LOC fire   =LOC
Which(ever) hunters we saw yesterday, those ones are sitting by the fire.
OR: The hunters (that) we saw yesterday are sitting by the fire.
Here the English translation with the relative clause sounds more natural. The reason is that this sentence isn't about any arbitrary hunters that we happen to see---we've already seen them, it's a definite group. The previous sentence differs because it's implied time reference is future, so it really is about whatever arbitrary people I happen to see, not a particular definite group of people, and in that case English is much more friendly to a whoever correlative clause. Akiatu likes correlative clauses in both cases.

(One issue: subjects and objects go before the verb, and relative clauses go after the noun, so Akiatu correlatives are a way to avoid a sort of center embedding; not really a troublesome sort, but correlative constructions will still often be preferred.)

The rest of this post will take up two complications: first subjects and definiteness, and second the difference between the wai and aka topicalising particles.

Akiatu subjects, you might remember, have to be definite---in that way they remain close to topics. But of course indefinite pronouns cannot be definite. So what do you do when your semantic subject is indefinite?

In an early post on clause types (here), I treated this issue by giving a structure like the following for "presentative" sentences:

Code: Select all

ikjamii kura wamau hja         itunaki wa
river   bank come  arrive(PFV) someone CIS
Someone came to the river
In the earlier post I said that the indefinite subject gets put after the verb, and the locative expression moves to the front to fill the resulting gap. Now I want to allow the same strings, but change the analysis.

What I want to say now is that with an indefinite semantic subject, it's possible to do a valency shift so that the semantic subject occurs as an object, and any existing object is demoted to an oblique (taking the locative preposition a). No subject can then surface; I suppose you could say there's a pro-dropped expletive subject, but I'm just going to say that there's no subject at all. (I mentioned above that I've started thinking Akiatu will have morphologically unmarked valency shifts, this is another example.)

The above construction results when the valency shift is followed by argument raising, with the oblique locative argument occurring in focus position. Another possibility, though, is for the demoted semantic subject to be focused. Then you'd get something like this:

Code: Select all

itunaki=su  wamau hja         a   ikjamii wa
someone=FOC come  arrive(PFV) LOC river   CIS
Someone came to the river
That the semantic subject is actually an object here can be shown in a number of ways. One is that it follows the negation particle:

Code: Select all

hwi mainaki=su  wamau hja         a   ikjamii wa
NEG  no.one =FOC come  arrive(PFV) LOC river   CIS
No one came to the river
(You need mainaki rather than itunaki in a negative clause.)

Here's how that'll look in a correlative clause:

Code: Select all

cunai  =su  piwa aja       a   ki  jisaka=wati wai,
whoever=FOC eat  away(PFV) LOC DET fish  =DIST TOP
    kinai wai atausa tija pawa      mikwa   mija  taima
    RESUM TOP belly  now  certainly already thing bad
Whoever ate those fish must be sick now
This has some details worth mentioning:
  • The verb in the correlative clause is transitive, but because of the lowering of the (indefinite) semantic subject its semantic object (ki jisaka=wati those fish) must surface as an oblique.
  • The resumptive pronoun actually is topicalised; the resulting use of a topic+subject pair to express inalienable possession is common.
  • Though there's a simple verb nukiwa be sick, this uses a more complicated expression to say that the person's belly is bad; one complication is that the true adjective taima bad, ugly, sick needs the support of the dummy noun mija thing before it can serve as a predicate.
Okay, that's it for indefinite semantic subjects. The other complication I want to take up is the wai/aka distinction.

wai is used for topics of three general sorts:
  • Topics that (re)center discourse on a new topic. Of course this new topic must be one that it's possible to refer to definitely, since topics must be definite. But it won't be the main topic you're already talking about; that'll most often be referred to with a subject. Note that this use of wai can be used for what's often called contrastive focus (which in information-theoretic terms isn't really a sort of focus, since it's not new information).
  • Topics that set the scene. These will often be locative phrases or clausal. They describe the background against which the event described in the main clause takes place.
  • Topics that give the general subject matter of the ensuing statement or provide a context within which it should be interpreted. (Like the example above: kinai wai atausa as for that one, her belly.)
Topics of the first sort can usually be interpreted as raised constituents of the embedded clause, topics of the other sorts often cannot.

aka has as its central use the flagging of topics that provide premises or presuppositions; we'll see a lot of it if I ever get around to conditionals. It gives a correlative clause a hypothetical sense: if anyone does rather than whoever did, for example:

Code: Select all

cunai  =su  piwa aja       a   ki  jisaka=wati aka,
whoever=FOC eat  away(PFV) LOC DET fish  =DIST TOP
    kinai wai atausa pawa      ahwaicu        hwaicu     mija  taima
    RESUM TOP belly  certainly lie.down(STAT) REDUP(INC) thing bad   
If anyone eats those fish, they'll get sick
OR: Anyone who eats those fish will get sick
Along with the shift to aka, this replaces the earlier already (is) a sick thing with (will) become a sick thing, roughly speaking.

There'll be plenty of cases where the wai/aka distinction is very subtle, and when it's fairly clear it'll often involve other differences (as in the above pair of examples). But the correct use of aka will end up being pretty important for fluent Akiatu.

Re: Akiatu scratchpad (correlative clauses)

Posted: Fri Nov 16, 2018 1:25 pm
by chris_notts
What about non-specific relatives? Did I miss how it works? Example:

I want to marry a woman who speaks Spanish (no specific woman in mind)

The topic construction doesn't seem well suited to a non-specific non-generic referent, but maybe it could do it by extension from the specific relative case.

Re: Akiatu scratchpad (correlative clauses)

Posted: Sat Nov 17, 2018 2:31 am
by akam chinjir
Oh, good point!

First, a confession: I've done a really meagre amount of research on languages that use correlative clauses of a roughly similar sort---I looked at some examples from Hindi, that's about it. So it might be weird to put them in a topicalised position, and it certainly does seem to rule out using them for nonspecific cases like the ones you mention.

One thing is that Akiatu (Hindi too, for that matter) also allows post-nominal relative clauses, and this could easily be a case where that's the preferred strategy. Here's what that would look like (changing the example a bit because Akiatu society doesn't really have marriage):

Code: Select all

hau ucisu mwi cunai   na  kwamuri jiraci   añiki kiwaita mawa
1s  want  SS  someone REL hunter  language know  see     find
I want to find someone who knows Kwa̰ Mɨɨrts
(cunai someone is also one of the correlative pronouns and a question word, here it's used as a nonspecific indefinite pronoun. I can't right now remember what little prosody I've worked out for Akiatu, but I expect that after the relative clause---between añiki know and kiwaita see---there's a jump back to phrase-initial pitch, so the verb sequence won't be as confusing as maybe it looks in writing.)

(Also: the Akiatu word for hunters was always meant to be derived from the ethnonym Kwa̰ Mɨɨrts, but for some reason I gave it in previous posts as kamuri even though kwamuri is legal and seems more likely. As of now that's changed.)

But there's actually a way to use a correlative construction here, too. It's just that you have to treat I want to find someone as the topic and someone knows Kwa̰ Mɨɨrts as the comment, like this:

Code: Select all

hau ucisu mwi cunai  kiwaita mawa aka, kinai ikau kwamuri jiraci   añiki
1s  want  SS  anyone see     find TOP  RESUM then hunter  language know
Anyone I want to find, they know Kwa̰ Mɨɨrts
OR: I want to find someone who knows Kwa̰ Mɨɨrts
Maybe that works?

There's another, unrelated thing that I forgot to mention in the post, which is that you can have two correlative/resumptive pairs, like this:

Code: Select all

cunai  =su  kausitu  a   cunai   wai, kinai acuta ikau ikjamii kura kiwaita mawa kinai
whoever=FOC look.for LOC whoever TOP  RESUM soon  then river   bank see     find RESUM
Whoever is looking for someone can find them by the river
Thanks for the question!

Re: Akiatu scratchpad (correlative clauses)

Posted: Sun Nov 18, 2018 2:08 am
by chris_notts
akam chinjir wrote: Sat Nov 17, 2018 2:31 am But there's actually a way to use a correlative construction here, too. It's just that you have to treat I want to find someone as the topic and someone knows Kwa̰ Mɨɨrts as the comment, like this:

Code: Select all

hau ucisu mwi cunai  kiwaita mawa aka, kinai ikau kwamuri jiraci   añiki
1s  want  SS  anyone see     find TOP  RESUM then hunter  language know
Anyone I want to find, they know Kwa̰ Mɨɨrts
OR: I want to find someone who knows Kwa̰ Mɨɨrts
Maybe that works?
Maybe that is tge best way. I think that is pragmatically similatpr to the English example, since in the nonspecific relative the relative clause itself is not really presupposed, it's asserted, and the rest of the main clause is non-focal.

I wonder if all languages allow non-specific relatives? We need a new chapter on WALS...

Re: Akiatu scratchpad (correlative clauses)

Posted: Sun Nov 18, 2018 2:25 am
by akam chinjir
There are supposed to be languages without relative clauses at all. I think Pirahã is supposed to be one? (I don't remember exactly what Everett's claim that Pirahã lacks recursion is supposed to mean, but surely it entails that there aren't any relative clauses.)

You're right that it'd be interesting to know about any languages that have relative clauses but don't allow nonspecific heads, in this context particular if it involved correlatives. Maybe there's something in the relative clause chapter of Language Typology and Syntactic Description? It's been a long time since I last looked at it, and I don't have time to check now, though.

Re: Akiatu scratchpad (correlative clauses)

Posted: Sun Nov 18, 2018 3:03 am
by chris_notts
akam chinjir wrote: Sun Nov 18, 2018 2:25 am Maybe there's something in the relative clause chapter of Language Typology and Syntactic Description? It's been a long time since I last looked at it, and I don't have time to check now, though.
I read those books many years ago at uni and loved them. I thought about buying them on kindle, but unfortunately two of the three volumes don't work on traditional kindles and need devices with better screens. I think it's because they are "print replica", i.e the publisher couldn't be bothered to reformat them for variable screen size devices.