Page 3 of 4

Re: Akiatu scratchpad (correlative clauses)

Posted: Sun Nov 18, 2018 3:10 am
by akam chinjir
Yeah, they're great.

I use the Kindle app on a tablet, they show up nicely on that.

Akiatu scratchpad (ideophones and manner adverbs)

Posted: Fri Nov 30, 2018 12:48 am
by akam chinjir
Ideophones and manner adverbs

I'll start with manner adverbs because they're easy.

Except: I'm a bit bogged down in sorting out some word-order stuff, and the position of manner adverbs might end up changing (again).

For now, manner adverbs go before the verb, right before the lower of the two preverbal positions where you might find a direct object. (So normally a focused object would precede and an unfocused object would follow a manner adverb.)

Manner adverbs are formed by reduplicating an appropriate verb. (For now this is full reduplication.)

Code: Select all

itamu aikja aikja jisaka piwa aja
Itamu quick REDUP fish   eat  away(PFV)
Itame ate the fish quickly.
It's about as simple as that, really.

Though there's something to be careful about with this example: "Itamu quickly ate the fish" has an interpretation in which "quickly" isn't a manner adverb (the meaning is something like Itamu immediately ate the fish), and so far Akiatu does not have a way to say things like that.

There's a sort of adverbial clause that I haven't posted about yet that can also do the work (and occupy the position) of manner adverbs, consisting of (maybe just a) VP followed by the subordinating particle (conjunction?) ma. This allows sentences like the following:

Code: Select all

itamu aikja iwasu   a   hjaci ma  jisaka piwa aja
Itamu quick surpass LOC Hjaci SUB fish   eat  away(PFV)
Itamu ate the fish more quickly than Hjaci
As you can see, you'll use a ma adverbial if you need something more complex than just the manner verb.

You could actually just use ma in place of reduplication, come to think of it:

Code: Select all

itamu aikja ma  jisaka piwa aja
Itamu quick SUB fish   eat  away(PFV)
Itamu ate the fish quickly
This is a bit staid, and I suppose emphasises the quickness more (since it gets its own clause).

Ideophones are similar in certain ways to manner adverbs: they often characterise the subject's behaviour or manner, they can occupy the same position as manner adverbs, and they often look like they result from a process of reduplication.

But there are differences: ideophones can also have non-manner meanings, they can also occupy different positions in the sentence, and even when they look like reduplications (which is common) they are not actually derived that way (the apparently reduplicated base is generally not itself a word---so amau amau overflowing with contentment does not derive from an independently meaningful amau, it's as it were irreducibly reduplicated.)

Ideophones are also more vivid---both in their meaning and their typical enunciation---than manner adverbs. They'll normally receive prosodic emphasis, and rarely allow another sentence element to be focused. They're rare outside of narrative, oratorial, and performative contexts (though quite common in those contexts); you tend to use ideophones when you've got an audience and not just an interlocutor, I guess. (And I guess you also tend to be moving your hands.)

It's often difficult to translate ideophones in even a minimally satisfying way. (So my translations below will sometimes be a bit weird.)

They can also be phonologically a bit unusual. So far I'm pretty sure of these points:
  • There'll be nonstandard stress patterns, quite often with iambic rather than trochaic feet, and with consecutive heavy (and therefore stressed) syllables acceptable.
  • There'll also be distinctive tonal patterns. At least: there'll be some ideophones that are entirely high-toned, and some characterised by a low tone.
  • A broader range of geminates will occur, at least the plosives. There'll also be some word-final coda plosives.
  • There'll be overlong, even arbitrarily long, vowels and consonants, unusual length generally having an intensifying or onomatopoetic significance.
  • I'm leaning towards syllabic nasals, possibly only word-initially and before a homorganic plosive. (Mostly I want something like ɲɲɲcacaca, with an arbitrarily long initial ɲ and an arbitrarily reduplicated ca, to express the sounds of insects.)
  • I'm also leaning towards allowing x. (There might be some varieties of Akiatu in which k → x before high vowels, but mostly I seem to have concluded that x is sound-symbollically appropriate for fire.)
There'll also be some distinctive reduplication patterns:
  • Three or even more repetitions are possible, at least with short (bisyllabic) bases.
  • You can have a short ult or penult that can repeat any number of times (ɲɲɲcacaca, mentioned above, is an example.) I guess I'll write these syllables three times in citation forms.
  • Reduplication with an i/u alternation, as in màuti màutu struggling under a load (the grave accent is marking a low tone).
Often there'll be a small family of ideophones with little or no difference in meaning. For example, you might have something like piti piti, piti pitu, pitititi, pitik, and pituk. Maybe these all describe the sound of something small---like a small stone---falling in water, with some different nuances. (Maybe pititititi would be used if several stones fell in the water at the same time, piti piti if you were skipping stones, and so forth.)

Both in these families and generally there'll be a lot of sound symbolism.

I mentioned that ideophones can go in places where you wouldn't find a regular manner adverb.

First, when followed by tikwa face, they can be used as predicates:

Code: Select all

hau amau.amau tikwa
1s  IDEO      face
I'm really happy!
These sentences can't get very complex, but an object is possible:

Code: Select all

hau itamu amau.amau tikwa
1s  Itamu IDEO      face
So happy! was I with Itamu.
(This describes Itamu as the object of my happy behaviour, not the cause of my happy feeling. Maybe I ran up to her and embraced her.)

You might think of the object in this construction as a "towards" dative, though being before the verb (the only place it can occur) it manages without a preposition.

These ideophone predicates do not distinguish perfective from imperfective senses.

Another possible position for an ideophone is clause-initially; the clause must then have an explicit complementiser, most often exclamatory sai.

Code: Select all

tarauhu.auhu sai  hjaci kahawa hja         i   hakjawi
IDEO         EXCL Hjaci move   arrive(PFV) DAT fire
Injured! Hjaci stumbled to the fire
The predicate in a sentence like this can just be a noun:

Code: Select all

paurrrak sai  ki  wamikawi
IDEO     EXCL DET storm
Flash! went the lightning
(I don't know how to justify the use of the ki determiner there, but it definitely feels right.)

wamikawi is from wamika air, wind, with the pluralising or collectivising suffix wi. I'm leaning towards the view that it can be used definitely without specific supporting context, the way we can use "the sky" or "the sun" definitely. (That's not meant to explain the ki, though.)

You can also just have an exclamation:

Code: Select all

amau.amau sai
IDEO      EXCL
So happy!
The construction with sai is potentially recursive:

Code: Select all

panai.panau sai  paurrrak sai  ki  wamikawi
IDEO        EXCL IDEO     EXCL DET storm
Bang! went the thunder and flash! went the lightning
This can be done with different members of the same ideophone family or with the two parts of a reduplication:

Code: Select all

pitik sai  pituk sai  mitana urasu kjai rasu        a   ikjamii ka
IDEO  EXCL IDEO  EXCL stone  enter down REDUP(COMP) LOC river   TRANS
Plink! Plonk! The (two) stones fell in the river

Code: Select all

tarauhu sai  auhu sai  hjaci kahawa hja         i   hakjawi
IDEO    EXCL IDEO EXCL Hjaci move   arrive(PFV) DAT fire
Injured! Stumbling! Hjaci went to the fire
Some ideophones can show up as intensifiers, immediately preceding a verb or adjective. These are mostly very lexically-specific. For example, ukura can only go with the adjective sakija red, bright.

Code: Select all

tiwana wai tija mija  ukura    sakija
sun    TOP now  thing very.red red
The sun is so bright today!
These intensifying ideophones can also be used on their own and clause-initially:

Code: Select all

ukura    sai  tiwana
very.red EXCL sun
So bright is the sun!
(Above I wanted ki before wamikawi, but here I don't want it before tiwana. This has to be largely a prosodic issue: wamikawi does and tiwana doesn't have (secondary) stress on its first syllable, so ki puts an unstressed syllable after sai. Huh.)

Ideophones are also quite free to just go where you want to put them. (I think that's the same as English "bang" and "plonk" and such---are these ideophones?)

One place ideophones tend not to show up is in isolated sentences of the sort you pick out of the air for examples, one symptom of the artificiality of such sentences. (Obviously I ought to be doing more to generate an actual corpus of "real" Akiatu. Anyone know any good examples of bat folklore?)

Maybe I should mention: there are languages with thousands of ideophones, and I don't think Akiatu is one of those, but it'll have at least hundreds.

Re: Akiatu scratchpad (ideophones and manner adverbs)

Posted: Wed Dec 05, 2018 12:07 am
by akam chinjir
Locative phrases

I've given up trying to remember to write /ɲ/ as ñ---from now on it'll just be ɲ.

Here's a locative phrase:

Code: Select all

a   jakikwai  kiwa
LOC clan.hall inside
in  the clan hall
The primary use of a locative phrase is as a circumstantial adverbial phrase, which normally follows the verb complex, though they can also be topicalised or raised.

The example above follows the most common pattern, which has three parts:
  1. A preposition, here a. i (dative/allative), hu (ablative), and niwa (instrumental/perlative) are also possible. (There's one more preposition, comitative sati, maybe it too will someday gain a use in locative phrases.)
  2. A nominal phrase, which I'll call the landmark phrase. Here it's just the noun jakikwai. (Strictly I guess this is a DP rather than an NP: it can include determiners such as ki.) In this construction, the landmark phrase must be definite.
  3. A locational noun, here kiwa inside; this picks out a location relative to the landmark phrase. These nouns cannot be used independently with their locative meaning (though some also have uses as names for body parts). A locational noun is normally required when a preposition is used with a spatial sense (the main exception that I won't mention below is the use of a with path verbs).
Here's a list of some locational nouns.
  • aiwa beyond
  • hjata previous, before (in a sequence, in time
  • kihamii under the water (e.g., akjamii kihamii, under the river)
  • kiwa inside, during (cf. ukiwa belly)
  • kura bank (of a river); shore (of a lake or ocean); edge, boundary
  • maihi next, after (in a sequence, in time)
  • mau above, on top of, the upper part of
  • ɲuwani under, below, the lower part of (also legs, feet)
  • paika side, edge, path (also arm, hand)
  • tai rank, height, level
  • taku outside
  • tikwa front, in front of, front part of (also face)
  • tuwaika back, behind, back part of (also bum)
Here are a few more examples:

Code: Select all

itamu parawara a   ikjamii aiwa   wara
Itamu wander   LOC river   beyond about
Itamu is going for a walk across the river

Code: Select all

hawi papija tau           a   piwawi maihi
1p   jump   together(PFV) LOC feast  after
We danced after the feast

Code: Select all

i   niwa ki  pumuki =su   paika wamau maka  a   apwaki mau ka
IMP PER  DET current=PROX side  go    go.up LOC hill   top TRANS
Go along this path up the hill
One distinctive use of this construction is in comparatives:

Code: Select all

hjaci kitikwa  piwa aja       a   itamu tai
Hjaci REFL(AP) eat  away(PFV) LOC Itamu rank
Hjaci ate as much as Itamu

Code: Select all

itamu naki   ahwita a   hjaci aiwa
Itamu person tall   LOC Hjaci beyond
Itamu is taller than Hjaci
Locative phrases do not always require a locational noun. Place names and names for events (such as festivals) can occur without them, for example.

Code: Select all

itamu ahwaicu  a   ihjatikai
Itamu lie.down LOC Ihjatikai
Itamu lives in Ihjatikai (a village)

Code: Select all

kwamuri wamau hja         a   jakuɲi hjata  wa
hunter  come  arrive(PFV) LOC Jakuɲi before CIS
The hunters came before Jakuɲi (an annual feast)
Any old noun can also gain a locational sense with the addition of one of the deictic clitics, usually wati, though su and ku are also possible.

Code: Select all

hjaci ijau a   itamu=wati
Hjaci sit  LOC Itamu=DEIC
Hjaci is sitting by Itamu
Note that these clitics have a specifically spatial sense, they cannot be used in temporal locative phrases.

Locational nouns can also take wati (but not su or ku) as an affix; the resulting noun has two main uses (in both of which both spatial and temporal senses are possible).

First, it can be used in locative phrases without an explicit landmark:

Code: Select all

ijaisa pijatu    a   kiwa  -wati
bat    hang(LOC) LOC inside-WATI
The bat is inside

Code: Select all

itamu ikihwa aja       a   kiwa  -wati
Itamu leave  away(PFV) LOC inside-WATI
Itamu left during (the event)
Second, it can be used in a manner that superficially parallels the locative phrases introduced above, but with an indefinite landmark phrase:

Code: Select all

itamu aku     mawa      a   piwawi kiwawati
Itamu be.born find(PFV) LOC feast  during
Itamu was born during a feast
This isn't really the same construction though. With a regular locational noun, as in piwawi kiwa during the feast, piwawi feast is an inalienable possessor, a position that allows a full DP. In piwawi kiwawati during a feast, piwawi is instead a noun modifier, a position where you'll rarely find more than the head noun and which excludes determiners such as ki.

One complication here: though both constructions allow a relative clause, the relative clause cannot come between the landmark phrase and the locational noun. There are at least two possibilities.

First, the relative clause can be postposed:

Code: Select all

hawi ijau a   witamwi ɲuwani    kja hjaci aku     mawa      watiwi
1p   sit  LOC tree    foot(LOC) REL Hjaci be.born find(PFV) there
We were sitting under the tree where Hjaci was born
Second, you can use a correlative clause:

Code: Select all

hjaci aku     mawa      a   cuwa  witamwi ɲuwani    wai, hawi ijau watiwi
Hjaci be.born find(PFV) LOC which tree    foot(LOC) TOP  1p   sit  there
Which tree Hjaci was born under, we were sitting there.
I'm torn whether or not to allow a construction like this, with the locative raised and the relative clause stranded after the verb:

Code: Select all

hawi witamwi ɲuwani    ijau kja  hjaci aku     mawa      watiwi
1p   tree    foot(LOC) sit  COMP Hjaci be.born find(PFV) there
We were sitting under the tree where Hjaci was born
This looks like secondary predication, except the semantics are different and it uses a finite subordinate clause (with kja) rather than a nonfinite one (with mwi or na). I quite like this construction, so its chances are good. Naturally, if I allow it in this case it'll also be possible for objects in general.

I a little bit want to put a ki before witamwi: hawi ki witamwi ɲuwani ijau kja... we were sitting under that tree where.... Maybe.

Locative phrases can be modified, and I'll mention two ways.

First, ikau right then, just, exactly can go right before the preposition, suggesting precision:

Code: Select all

hjaci akimau ikau  a   hakjawi paika
Hjaci stand  right LOC fire    side
Hjaci is standing right by the fire
Second, you can insert a measure of (spatial or temporal) distance along with the ablative preposition hu:

Code: Select all

hawi akimau a   hikuti haku pai   hu  mikuwi hatau shore
1p   stand  LOC day    five three ABL waters great bank
We are eight days from the ocean

Code: Select all

kai akimau (ikau) a   wai  cita hu  jakikwai  tikwa
3s  stand   right LOC pace four ABL clan.hall face(LOC)
She is standing (exactly) four paces in front of the clan hall
(In the last example, ikau could also go right before the number. Other modifiers will be possible in these positions, but I haven't got them yet.)

Re: Akiatu scratchpad (locative phrases)

Posted: Wed Dec 05, 2018 5:48 am
by Bob
Wow. Really great.

Some comments:

I'm a big fan of glossing. Using standard grammatical abbreviations is a big plus.

I have this book by Yallop, Australian Aboriginal Languages, you might like it.

Akiatu scratchpad (path verbs)

Posted: Sun Dec 09, 2018 11:33 pm
by akam chinjir
Path verbs

Path verbs get used especially in descriptions of motion and spatial distribution.

The most common such verb is wamau come, go. It can be used like this:

Code: Select all

itamu wamau hja         a   jakikwai  ka
Itamu go    arrive(PFV) LOC clan.hall TRANS
Itamu went to the clan hall
Some points to notice:
  • wamau is in most respects a normal verb. Here for example it gains a perfective sense from the complement hja arrive.
  • The difference between come and go is marked not by the choice of the verb but by a particle, which can either directly follow the verb complex or (as above) follow its locative complement. By far the most common such particles are cislocative wa and translocative ka, but at least wara back and forth, around, about and wi passing by are also possible. Descriptions of motion must include one such particle, at least when it's the subject that moves, and I'm going to call these motion particles.
  • The path verb's complement looks like a locative phrase (the topic of the last post), but there are two differences.
    • The preposition is always locative a, regardless of the specific semantics. Here the complement gives a destination, so you might expect dative i instead. But the complement of wamau will always be a destination, so nothing is lost if this isn't also signalled by the preposition.
    • The preposition's object is just jakikwai clan hall, where a locative phrase would require a specifically locational element---like the tikwa front in jakikwai tikwa in front of the clan hall or the =wati there in jakikwai=wati at the clan hall, for example. These locational elements become optional in the complements of path verbs, typically included only in the interests of semantic precision.
  • The complement of a path verb can be omitted if it can be recovered from context; it will always be semantically present, however.
  • I'll say that the complement gives the point or reference, for the path verb. A path verb, then, defines a path relative to a point of reference. The point of reference of wamau is the path's destination, but other path verbs will relate differently to their points of reference.
Here is another example, using ikihwa depart, leave, whose point of reference is the path's starting point.

Code: Select all

itamu ikihwa aja       a   awajakwai ka
Itamu leave  away(PFV) LOC village   TRANS
Itamu left the village
In this case the semantically appropriate preposition would be ablative hu, but again we get a. Also note that though "leave" marks no come/go distinction, Akiatu still has the wa/ka distinction. Also, there's a subtlety here: the complement aja has a centrifugal sense here (away), so it wouldn't make sense with the cislocative motion particle wa (the example has translocative ka).

kahawa move is a special case: it's a path verb with no point of reference. It implies motion or distribution along a path but tells you nothing about the nature of that path. It's most often used for small local movements, often as a complement (more on that below), though kahawa i can mean about the same as wamau a and kahawa hu as ikihwa a.

You can use path verbs without motion particles to describe something's distribution in space:

Code: Select all

mitana pumuki  wamau a   kjatikwai
stone  current go    LOC village.hall
The stone path went to the village hall

Code: Select all

mitana wikjacasu a   hakjawi
stone  circle    LOC fire
The stones went around the fire
With this sort of sense, though, path verbs most often occur as complements. One relatively neutral possibility is to use a posture verb as the main verb:

Code: Select all

mitana ahwaicu  wikjacasu a   hakjawi
stone  lie.down circle    LOC fire
The stones lay around the fire
In the interests of simplicity I've used definite subjects in these examples, but indefinite ones will actually be more common. Like this:

Code: Select all

pumuki kura aku     niwasu   mapi   na  jasijasu
path   side be.born go.along flower REL colourful
Colourful flowers grew along the path
Notice that as usual the locative preposition a cannot occur before the verb, so pumuki kura by itself serves as the complement to the path verb niwasu go along.

There'll be some lexicalised verb+path collocations, but I don't have a list to offer now.

Three path verbs can indicate aspect when used as complements.

wamau and kahawa can have a continuative sense, wamau tending to emphasise that an activity is still taking place in the present (or the narrative present, anyway), kahawa that someone is continuing an activity into the future; kahawa does and wamau does not also imply a perfective sense; kahawa is especially common with the manage to complement saka and its negation kuu; and it's normally kahawa that corresponds to the English "keep (on)."

Code: Select all

itamu hjakiwani tikai kitikwa  tawaru wamau
Itamu brother   still REFL(AP) sing   CONT
Itamu's brother is still singing

Code: Select all

hjaci kitikwa  piwa saka           kahawa
Hjaci REFL(AP) eat  manage.to(PFV) CONT
Hjaci managed to keep eating
ikihwa leave can have an inceptive sense:

A path complement can also be used together with a motion particle. Easy cases involve the main verb taki hold, which can be used for either bring or take, depending on the choice of motion particle.

Code: Select all

itamu ki  apatu taki wamau   hja         wa
Itamu DET spear hold come/go arrive(PFV) CIS
Itamu brought her spear

Code: Select all

itamu ki  apatu taki wamau   hja         ka
Itamu DET spear hold come/go arrive(PFV) TRANS
Itamu took her spear
Path complements, especially kahawa move, are frequently used with a resultative sense:

Code: Select all

hjaci apatu supi  kahawa i   itamu
Hjaci spear touch move   DAT Itamu TRANS
Hjaci nudged the spear towards Itamu

Code: Select all

hjaci witamwi ajamija urasu a   hakjawi
Hjaci wood    throw   enter LOC fire
Hjaci threw the wood on the fire
Two things: these sentences do not require motion particles because it's not the subject who's moving; and path complements used as resultative complements imply a perfective sense, but when used to describe a distribution in space they do not.

An explicit causative is also possible:

Code: Select all

itamu itai ahjai wamakasu jaku
itamu rope CAUS  coil     settle(PFV)
Itamu coiled the rope
One other thing to note is that it's not common to use path verbs as complements to manner-of-motion verbs:

Code: Select all

?itamu pijatu wamau hja         a   ikjamii ka
 Itamu jump   go    arrive(PFV) LOC river   TRANS
 Itamu ran/jumped/danced to the  river
(A secondary predicate or some sort of manner adverbial would normally be better here.)

Here are some of the more common path verbs:
  • akaɲa go over, across, onto. Point of reference: the top of something.
  • akjasu go past, to pass by. This will often be used with the motion particle wi passing by.
  • amawaitu go around, to the side of.
  • caɲi to cross, go across. Point of reference: a boundary or linear obstacle.
  • ikihwa leave. Point of reference: the starting point.
  • iwasu go beyond. You'll use this rather than (say) akjasu go past, pass by in order to emphasise the distance traveled or when the point of reference is given just as a distance:

    Code: Select all

    hjaci iwasu     wasu       a   hikuti haku itu (hu  suwi)
    Hjaci go.beyond REDUP(PFV) LOC day    five one  ABL here
    Hjaci went more than six days (from here)
    
    As a complement, iwasu can be used in comparatives:

    Code: Select all

    itamu suwasu iwasu     wasu       a   hjaci
    Itamu sleep  go.beyond REDUP(PFV) LOC Hjaci
    Itamu slept longer than Hjaci
    
  • jima go between, among, through. The point of reference will be somehow plural, or at least heterogeneous. The implied path will usually pass through this heterogeneity to the other side, though the motion particle wara around, back and forth will imply a less linear or goal-oriented path.
  • kahawa move. No point of reference.
  • kaja ascend. The contrast with akaɲa go over, onto is significant: kaja implies a path up the side of something, akaɲa a path either terminating at or passing above the top of something.
  • kasu to follow, accompany. The point of reference will be something that provides guidance, maybe a distant (even celestial) landmark, or maybe a person who's leading the way. The contrast with niwasu go along can be subtle. Both can take a trail as a point of reference, but kasu emphasises the need to take guidance from the trail: the trail isn't just a convenience, it's actually your way of knowing which way to go. kasu also shows up in two other path verbs, wamakasu to spiral, coil (< wama shell, spiral, coil) and wakjacasu to circle (< wakjai circle, loop).
  • makwaja go home. The point of reference (the destination) is often left implicit.
  • matai go as far as. This verb tends to replace wamau come, go in order to emphasise the distance traveled or when the destination is given as a distance from the starting point:

    Code: Select all

    hjaci matai  tai        a   hikuti ami (hu  suwi)
    Hjaci go.far REDUP(PFV) LOC day    two  ABL here
    Hjaci went two days away
    
    matai can also be used as a complement to form comparative statements:

    Code: Select all

    hjaci kitikwa  piwa matai  tai        a   itamu
    Hjaci REFL(AP) eat  go.far REDUP(PFV) LOC Itamu
    Hjaci ate as much as Itamu
    
    (Compare hjaci kitikwa piwa aja a itamu tai from the last post.)
  • niwasu go along, through. Here the point of reference is itself a path, maybe an actual trail or a stream you can follow or somesuch, or possibly instead a region that the path traverses from one end to another. When the point of reference is a region you can get a subtle distiction with jima go between, among, through. Something like this: you'd niwasu a forest, but jima the trees. (But with no plural marking in Akiatu the distinction might be marked only by the choice of verb.)
  • parawara wander. The point of reference, when present, gives a region. The usual motion particle is wara around, and in its presence you often get just para. (The result can be a subtle prosodic distinction between parawara and para wara.)
  • piraa exit.
  • tamapai go under, beneathe.
  • urasu enter. Water and fire optionally get medium-specific alternatives, respectively miku enter water (also just water) and hajasu enter fire.
  • wamau come, go. Point of reference: the destination.
  • wijasu descend. This implies a path down the side of something, whereas tamapai go under, beneathe implies a path terminating at or passing under the bottom of something.
  • wamakasu spiral, coil, circle repeatedly.
  • wikjacasu go around, circle. This verb implies no goal other than circling the point of reference. By contrast, when amawaitu means something like go around, this implies a path that goes around some obstacle but then continues on towards a further destination. The usual motion particle with wikjacasu is wara around.
There are some characteristic ways to indicate perfective aspect with path verbs.
  • hja arrive will often be used when the point of reference is a destination.
  • Partial reduplication (of the verb's final foot) or the complement siwa will indicate that the full path was traversed. This can yield a different sense from hja arrive with akaɲa go over, onto: akaɲa hja is went onto, whereas akaɲa kaɲa or akaɲa siwa is went over, to the other side of; and similarly with amawaitu to around, beside and tamapai to under, beneathe.
  • hana return and maru return home are common when semantically appropriate.
  • jaku fix in place can be used with urasu enter and kasu follow.
  • saka succeed can indicate that completing the path required overcoming some obstacle, kuu fail that the obstacle was not in fact overcome.
I suppose that's it for now.

Akiatu scratchpad (causatives)

Posted: Tue Dec 18, 2018 11:12 pm
by akam chinjir
Causatives

I just realised that I've never written an overview of how to make causatives, and I've been inconsistent in my own practice.

I don't just mean I've never posted about it, I mean even in all my notes, there's nothing more than brief comments in the entries for ahjai and hwati in my dictionary.

So I'll fix that here.

I'll focus on monoclausal and biclausal constructions with the verbs ahjai make, do, cause, let and hwati give, make, let, ignoring various other things you can do with complements and so on.

First, the monoclausal constructions. These imply direct causation---what you generally get in English with "make."

A digression: there's an old idea that we can semantically analyse "kill" as "cause to die," and a classic counterargument that if you cause someone to die, you can do the causing on Tuesday even if they don't die till Thursday, but you can't kill them on Tuesday unless they die on Tuesday. That's because English "cause" expresses indirect causation; if you analysed "kill" instead as "make dead," the analysis wouldn't be vulnerable to the same sort of counterargument (I'm not saying it would be correct).

Anyway, in monoclausal causative constructions, the difference between ahjai and hwati is purely syntactic: you used ahjai with intransitive verbs and hwati with transitive ones. (There's no way to make a monoclausal causative if you start with a ditransitive verb.)

Here's how it looks with ahjai:

Code: Select all

itamu rawu    tima        →  jisaka itamu ahjai    rawu    tima
Itamu content ready(PFV)     fish   itamu do(CAUS) content ready(PFV)
Itamu was satisfied          The fish satisfied Itamu
You'll notice that the original subject ends up as a direct object, and that the use of ahjai does not affect the choice of perfective complement.

One other thing: here the plain verb corresponds to an English past participle, whereas the Akiatu causative corresponds to a plain verb in English. That's pretty common. In general, Akiatu uses explicit causatives a lot more than English does. (Some important examples with hwati are hwati hakwai make know (that), tell, hwati aɲiki make know (of), introduce, and hwati kiwaita make see, show.)

With hwati, the result is a normal ditransitive construction (remember that hwati also means give), with the original subject ending up as an indirect object. It can look like this:

Code: Select all

hau jisaka piwa aja        →  itamu jisaka hwati      piwa aja       i   hau
1s  fish   eat  away(PFV)     Itamu fish   give(CAUS) eat  away(PFV) DAT 1s
I ate fish                    Itamu made me eat fish
You'll notice that the object goes before hwati.

However: in the monoclausal construction with hwati, it's actually more common for the indirect object to raise to a preverbal position, stranding the actual direct object after the verb. For example:

Code: Select all

itamu hau hwati      piwa aja       jisaka
Itamu 1s  give(CAUS) eat  away(PFV) fish
Itamu made me eat fish
This doesn't imply any particular focus on the indirect object. (My thinking on this sort of thing has been shifting, I'll have to do a post on it before too long. It may be that the preference for a raised indirect object is completely general, or at least very common with animate indirect objects.)

A monoclausal causative can be passivised, with kiwaita see:

Code: Select all

hau kiwaita   hwati      piwa aja       jisaka (a   itamu)
1s  see(PASS) give(CAUS) eat  away(PFV) fish    LOC Itamu
I was made to eat fish (by Itamu)
The opposite order---a (monoclausally) causativised passive---is not possible.

In the post about passives, I said that certain adverbs (I gave the example of ikijiku often) can come between the passivising morpheme and the main verb. That sort of thing is kind of up in the air at the moment, but if I keep it, the same adverbs will also come between the causative morpheme and the main verb. That would allow sentences like this:

Code: Select all

ki  kaɲi  kipaja hwati      ikijiku     kjaitiwa kaiku     itamu
DET pride Kipaja give(CAUS) often(IPFV) poke     down(PFV) Itamu
His pride often made Kipaja insult Itamu
And that's it (for now) about monoclausal causatives.

Biclausal causatives also use ahjai and hwati, though in this case there is no (overt) syntactic difference.

(Maybe there's a covert syntactic difference, if with hwati the causee is an underlying indirect object; but it always shows up before the verb and with no preposition, so who can say?)

What you get instead is (usually) a semantic difference: ahjai usually implies compulsion (like "make"), whereas hwati implies permission, enabling, or even help (more like "let").

In both cases, the causee comes before the verb, like a direct object, and the caused behaviour gets descibed in a na complement. Like this:

Code: Select all

itamu hau ahjai    jaku na jisaka piwa
Itamu 1s  do(CAUS) PFV  DS fish   eat
Itamu made me eat fish
The use of jaku be fixed in place as a perfective complement for ahjai is pretty standard (even if the caused behaviour involves plenty of motion). You'll notice that the embedded verb has no complement; aja away wouldn't be wrong, but the perfective/imperfective distinction often doesn't get marked in nonfinite clauses, so you'd usually add a complement only for it's non-aspectual semantic content.

Or we could have this, with hwati:

Code: Select all

itamu hau hwati      aja       na jisaka piwa
Itamu 1s  give(CAUS) away(PFV) DS fish   eat
Itamu let me eat fish
(aja is a common perfective complement for causative hwati.)

There's an idiosyncrasy. Remember the translocative particle ka, which occurs in motion descriptions? It's got one use as a perfective complement, with causative ahjai, and the result is more of a let causative.

Code: Select all

itamu akjanai  ahjai    ka         na ikihwa aja
Itamu stranger do(CAUS) TRANS(PFV) DS leave  away
Itamu let the stranger go
There's a pretty strong implication that someone or something (quite likely Itamu herself) had previously been preventing the stranger from leaving: ahjai ka implies a withdrawal of compulsion, maybe.

There you go.

Akiatu scratchpad (saying and thinking)

Posted: Tue Dec 25, 2018 11:43 am
by akam chinjir
Saying and thinking

I'm going to talk about three ways to go about quoting someone, not for any very particular reason. And I'm going to do it at great length and with great pedantry.

Direct quotation will pretty much always use kwasu, which I'm officially just calling a quotation particle (I've been glossing it as QUOT). On the one hand, it kind of has to be a verb, even if an odd one. On the other hand, often enough it might as well just be a colon, a bit of punctuation with phonological content.

Quotation using kwasu will look like this:

Code: Select all

hjaci kwasu, kwamuri tija wamau hja        wa
Hjaci QUOT  hunter   now  come  arrive(PFV) CIS
Hjaci said, the hunters have come
I'm putting a comma there to represent a pause. That's one of the oddities of this word, that you get a pause between it and its apparent complement. Another is that the complement, though a clause, doesn't have an overt complementiser. (Could I get away with saying that the pause means that?) And a third is that kwasu, if understood as a verb at all, gets a perfective interpretation despite not having a resultative or anything like that; no other verb can be used on its own in a main clause with a perfective sense.

The example sentence would be appropriate if you were interested in Hjaci's words, or in Hjaci's act of uttering those words. So it would make sense in a narrative involving Hjaci, for example, but not if you just wanted to pass on the information about the hunters.

Besides the normal leeway for translation and paraphrase that you get with direct quotation, kwasu usually has embedded pronouns shift in accordance with the perspective of the current speaker. For example, you might have this:

Code: Select all

hjaci kwasu, kai tija wamau hana        wa  hu  mikuwi hatau
Hjaci QUOT   3s  now  come  return(PFV) CIS ABL waters great
Hjaci said, I have (just) returned from the ocean
Note that word-for-word direct quotation would have had to use the first person pronoun hau (it's Hjaci talking about herself), but what we find is the third person kai (it's not Hjaci talking about talking about herself). hau would normally get used in the reported statement only when someone is reporting on their own prior words.

kwasu has one more peculiarity: when used in this way, there's no grammatical way to mention a listener. For example this is ungrammatical;

Code: Select all

*hjaci itamu kwasu, kai tija wamau hana        wa
 Hjaci Itamu QUOT   3s  now  come  return(PFV) CIS
 Hjaci said to Itamu, I have (just) returned
To convey the sense intended here, you need to use kwasu as a sort of complement to another verb:

Code: Select all

hjaci itamu tami  kwasu, kai tija wamau hana        wa
Hjaci Itamu speak QUOT   3s  now  come  return(PFV) CIS
Hjaci said to Itamu, I have (just) returned
Something to notice here: kwasu seems to give the verb a perfective sense, as if functioning as a regular verbal complement. But the subsequent clause doesn't require an overt complementiser, as if kwasu were still operating sort of like a colon.

(tami speak, say on its own can't be used for quotation. It can take a direct object describing someone's words: jiraci amaki tami speak beautiful words. It's the normal verb used to say that someone can speak a particular language: akiatu jiraci tami speak Akiatu. And with a comitative complement it can have a sense like chat, converse: tami sati Itamu talk with Itamu.)

I'm going to mention a couple of slightly tangential uses of kwasu. First, it can be used in enumerations. For example, if you wanted to say, First, it can be used in enumerations, you'd start with itu kwasu, using the regular cardinal number itu one. (And then for the second thing you'd say ami kwasu two:, and so on.)

Second, it shows up in a fairly common sort of topicalising construction. Here's an example:

Code: Select all

sati hau ma  kwasu, jisaka wakiru piwa
COM  1s  SUB QUOT   fish   easy   eat
As far as I'm concerned, fish are easy (=good) to eat
The meaning here is something like: when it comes to me or if you ask me. The comitative preposition sati here has a meaning more like about than like with; more on this usage in a bit.

(Aside: for what it's worth, this is meant to be a fairly direct relex of Mandarin 對我而言 or 對我來說, though kwasu otherwise owes more to the classical 曰. Maybe amusingly, the ma relates to both the 而 and the 來 come: it's got a lot in common with subordinating uses of 而, and is supposed to be related etymologically to wamau come. Though---I've started some work on diachronics, and I'm not at all sure I'm going to be able to pull that off. Right now wamau looks like it'll go back to something like guemur, with the ur → au shift being quite recent.)

That's probably enough for now about kwasu.

Next there's waɲi think, say. This goes easily with a (finite) kja complement:

Code: Select all

kipaja waɲi aja      kja  apatu aku     tima
Kipaja say  out(PFV) COMP spear be.born ready(PFV)
Kipaja said that the spear was ready
Notice the choice of resultative: aja here has a centrifugal sense, and implies speech rather than thought.

Instead of a clausal complement, waɲi can take a nominal direct object, which typically characterises the content of the person's words, not the words themselves (for which you'd more likely use tami, mentioned above). Something like this:

Code: Select all

hjaci amawaini amaki waɲi aja
Hjaci plan     good  say  out(PFV)
Hjaci proposed a good plan
The listener can be mentioned using the dative preposition i:

Code: Select all

hjaci amawaini amaki waɲi aja      i   itamu
Hjaci plan     good  say  out(PFV) DAT Itamu
Hjaci proposed a good plan to Itamu
The subject matter can also be mentioned, generally using comitative sati for animates and perlocative/instrumental niwa for inanimates:

Code: Select all

hjaci amawaini amaki waɲi aja      i   itamu niwa piwawi
Hjaci plan     good  say  out(PFV) DAT Itamu PER  feast
Hjaci proposed a good plan for the feast to Itamu
Now, there's a great deal of resistance to putting both a clausal complement and a preposition phrase after the verb. With a clausal complement, then, an indirect object will almost always get raised to a preverbal position, though generally without the focus-marker su:

Code: Select all

kipaja hjaci waɲi aja      kja  apatu aku     tima
Kipaja Hjaci say  out(PFV) COMP spear be.born ready
Kipaja said to Hjaci that the spear was ready
(Remember that the dative preposition gets dropped before the verb. Also, mentioning a listener is another way to make it explicit that you're talking about talk rather than thought.)

It's less common to include separate mention of a subject matter when there's also a clausal complement. When it happens, you usually get a (non-finite, object-controlled) na complement, like this:

Code: Select all

kipaja niwa apatu waɲi aja      na aku     tima
Kipaja PER  spear say  out(PFV) DS be.born ready
Kipaja said of the spear that it was ready
You can also use a na complement controlled by an indirect object:

Code: Select all

itamu hjaci waɲi aja      na wamau a   mikuwi hatau ka
Itamu Hjaci say  out(PFV) DS go    LOC waters great TRANS
Itamu told Hjaci to go to the ocean
mwi (nonfinite, same-subject) complements are also possible, generally with a future orientation:

Code: Select all

hjaci waɲi aja      mwi wamau a   mikuwi hatau ka
Hjaci say  out(PFV) SS  go    LOC waters great TRANS
Hjaci said she would go to the ocean
waɲi can also take an interrogative complement, and mean something like wonder, ask:

Code: Select all

hjaci waɲi  ha sai kwamuri wamau hja    wa
Hjaci think Q  IRR hunter  come  arrive CIS
Hjaci was wondering whether the hunters had come
There's more. In principle, a clausal complement can be topicalised, with the topic particle wai; in this case a recipient will usually remain postverbal, with i. But doing this is rarely appropriate.

Clausal complements are far more frequently focused. This also puts them before the verb, but in this case they'll follow any overt subject; they'll also be marked by the focus clitics su. You get sentences like this:

Code: Select all

itamu kja kwamuri wamau wa =su  waɲi aja
Itamu kja hunter  come  CIS=FOC say  out(PFV)
Itamu said the hunters are coming
Overt recipients are rare in this construction, but when given they must follow the verb:

Code: Select all

itamu kwamuri wamau wa =su  waɲi aja      i   hjaci
Itamu hunter  come  CIS=FOC say  out(PFV) DAT Hjaci
Itamu said to Hjaci that the hunters are coming
Note though that you can't topicalise or focus when using kwasu.

Okay, this is already an absurdly long post given that it's so far just about two words, but I promised three ways to quote, so here's the third: using the hwati hakwai make know, tell.

This looks like a causative construction, and maybe that's all it is. I went over the causative syntax in the last post---in fact I did that so I'd be sure to get the syntax of hwati hakwai right. Here's how it works.

hakwai know is a transitive verb, often taking a clause for its object. (I actually don't know how universal this is, but as far as Akiatu is concerned (niwa akiatu jiraci ma kwasu...) clausal complements are direct objects.) That's why it forms its causative using hwati (otherwise give). A nice non-coincidence: the semantic subject of hakwai know becomes the indirect object of the causative construction. It will almost always end up before the verb (therefore with no preposition):

Code: Select all

itamu hjaci hwati hakwai jaku kja  kwamuri wamau wa
Itamu Hjaci CAUS  know   PFV  COMP hunter  come  CIS
Itamu told Hjaci that the hunters are coming
There's an important difference from waɲi: the listener or audience will always be semantically implied even if not explicitly mentioned. I don't think you can even get around this with an antipassive. The whole point of this verb is that you let someone know something; this makes no sense if there's no someone that you're letting know. I guess you could say that a phrase referring to an audience is a complement of hwati hakwai, but just an adjunct of waɲi.

There's another difference: just as with waɲi, you can use a (same-subject, nonfinite) mwi complement, but here the result is oriented towards the past rather than the future:

Code: Select all

hjaci itamu hwati hakwai mwi wamau a   mikuwi hatau ka
Hjaci Itamu CAUS  know   SS  go    LOC waters great TRANS
Hjaci told Itamu that she had gone to the ocean
(But with a na complement the meaning is pretty much the same as with waɲi: tell (someone) to....)

And yet another difference. When hwati hakwai takes an interrogative complement, the result isn't an indirect question:

Code: Select all

itamu hjaci hwati hakwai ha sai kwamuri wamau wa
Itamu Hjaci CAUS  know   Q  IRR hunter  come  CIS
Itamu told Hjaci whether the hunters are coming
(NOT: *Itamu asked Hjaci whether the hunters are coming)
I'll also mention an important contrast with "tell." You can tell someone falsehoods, but you cannot hwati hakwai them (or let them know) falsehoods, because people can only know truths. Though I think I'll say this phrase is lexicalised at least to the extent that you can hwati hakwai someone something even if they weren't convinced of what you said (so they didn't end up knowing it) and even if they already knew it (so it wasn't you letting them know it).

And that surely is enough about that (for now).

Akiatu scratchpad (focus, I: the why)

Posted: Mon Jan 07, 2019 1:48 pm
by akam chinjir
Focus, I: The why

Ugh, I already think I got causatives wrong. Or at least incomplete. But there's no time for that now.

This time I'm talking about focus---specifically the uses to which it can be put. I'll say only the bare minimum about how you actually signal focus. (The next post will be all about that.)

Here's that bare minimum. To focus the direct object of a transitive verb:
  • append the clitic =su, as well as a high boundary tone
  • move the object a bit higher (lefter) in the clause (it'll come before some adverbs that normally precede the direct object)
You thus get pairs such as these:

Code: Select all

itamu ihjaisa piwa
Itamu bat     eat
Itamu is eating bat

Code: Select all

itamu ihjaisa=sú  piwa
Itamu bat    =FOC eat
Itamu is eating *bat*
As here, I'll flag focused expressions using asterisks. (This isn't meant to imply that the English sentence should be spoken with any particular emphasis.) But that's the last time I explicitly mark the boundary tone.

So why would you flag some element of a sentence in this way?

(Nothing in the next few paragraphs is meant to be Akiatu-specific.)

If someone says that Itamu is eating *bat*, the point is likely to draw a contrast with other things she might be eating. The focused element indicates how these alternative possibilities differ from the one that's explicitly described. Itamu's eating *bat*---contrast that with Itamu eating fish, or boar, or chicken. *Itamu*'s eating bat---contrast that with Hjaci eating bat, or Kipaja eating bat.

The particular alternatives that are salient, and the point of drawing a contrast with them, will depend on context. Here are five possibilities (I don't claim the list is exhaustive):
  • If you've just said that Itamu is eating fish, I might counter that Itamu's eating *bat* in order to contradict you. The only relevant alternative is the one you've just asserted, and I'm focusing ihjaisa bat to indicate the point of disagreement.
  • Or maybe you've just asked me what Itamu is eating, and I focus ihjaisa because that's the part of my statement that answers your question.
  • A bit more generally, I might be indicating that all the alternatives are false: Itamu is eating fish, and is not eating anything other than fish.
  • Or my point might be that it's more surprising or noteworthy that Itamu is eating bat than it would be if she were just eating (say) fish, boar, or chicken. Note that in this case (unlike the previous ones) I need not be denying that she's (also) eating those other things.
  • Finally, I might be commenting on a difference between another state of affairs that's also been mentioned or is in some other way salient. Yesterday Itamu ate *fish*, today she's eating *bat*. Hjaci is eating *fish*, but Itamu is eating *bat*. And so on.
So far so good.

But a sentence can also include elements whose interpretation is sensitive to focus. Like negation:

Code: Select all

itamu hi  ihjaisa=su  piwa
Itamu NEG bat    =FOC eat
Itamu is not eating *bat*
Someone saying this would most likely be conceding that Itamu is eating something, while denying that it's bat she's eating. In this way, focus can be used to signal which part of a sentence is getting negated.

Let me rub this in. Here's an affirmative sentence with a bunch of arguments:

Code: Select all

kipaja apatu hwati mawa i   itamu a   ikjamii kura
Kipaja spear give  PFV  DAT Itamu LOC river   bank
Kipaja gave a spear to Itamu by the river
By focusing different arguments to the verb, it's possible to generate all the following negations (don't worry too much about the syntax for now):
  • kipaja hi apatu=su hwati mawa i itamu a ikjamii kura Kipaja didn't give *a spear* to Itamu by the river
  • kipaja hi itamu=su hwati mawa apatu a ikjamii kura Kipaja didn't give *Itamu* a spear by the river
  • kipaja hi ikjamii kura=su hwati mawa apatu i itamu Kipaja didn't give a spear to Itamu *by the river*
  • ikjamii kura wai hi kipaja=su hwati mawa i itamu a apatu *Kipaja* didn't give a spear to Itamu by the river
Even (maybe; more on this next time):
  • kipaja hi hwati=su jai nai a apatu i itamu a ikjamii kura Kipaja didn't *give* a spear to Itamu by the river
Anyway, I hope it's clear enough how focus is interacting with negation in these examples.

It's not just negation that works this way. So do polar questions:

Code: Select all

ha sai itamu ihjaisa=su  piwa
Q  IRR Itamu bat    =FOC eat
Is Itamu eating *bat*?
Most likely someone using this sentence would be taking it as given that Itamu is eating something; what they want to know is whether it's bat that she's eating.

Here's an example involving a conditional:

Code: Select all

ita   sai itamu ihjaisa=su  piwa wai...
maybe IRR Itamu bat    =FOC eat  TOP
If Itamu is eating *bat*...
The adverb ita maybe is sensitive to focus even when not flagging the antecedent of a conditional:

Code: Select all

itamu ihjaisa=su  ita   piwa
Itamu bat    =FOC maybe eat
Maybe Itamu is eating *bat*
And there are a good number of adverbs that work the same way. Here's hwika just, merely:

Code: Select all

itamu jisaka=su  hwika piwa
Itamu fish  =FOC just  eat
Itamu is just eating *fish*
That's to say, all she's eating is fish.

Here's wija even:

Code: Select all

itamu ihjaisa=su  wija piwa
Itamu bat    =FOC even eat
Itamu is even eating *bat*
And hiku all:

Code: Select all

itamu jisaka=su  hiku piwa aja
Itamu fish  =FOC all  eat  away(PFV)
Itamu ate all the fish
It's maybe worth emphasising that hwika just, wija even, and hiku all (as well as other adverbs like them) cannot occur as constituents of the noun phrases with which they are associated. They are adverbs, and get associated with noun phrases by focus, not by constituency.

Three more things about these adverbs.

First, they can occur with no explicit indication of focus:

Code: Select all

itamu hi  ihjaisa piwa
Itamu NEG bat     eat
Itamu is not eating bat
The most likely interpretation here is something like it's not eating fish that Itamu is doing, with focus on the predicate as a whole. But what's happening is not that Itamu is eating fish, with whole-sentence focus, is also possible.

Second, these adverbs can also get hosted by an otherwise sentence-initial complementiser. Above we saw an example with ita maybe and sai (a complementiser that selects a non-indicative clause)---the result was the antecedent of a conditional. With indicative clauses, the complementiser will normally be kja.

If an element of the clause is explicitly focused, fronting the adverb does not really affect interpretation:

Code: Select all

hwika kja  itamu jisaka=su  piwa
just  COMP Itamu fish  =FOC eat
Itamu is just eating *fish*
With no explicit focus, the result is whole-sentence focus:

Code: Select all

hwika kja  itamu jisaka piwa
just  COMP Itamu fish   eat
Itamu is just eating fish (→ and that's all that's happening)
Third and finally, these adverbs can rise out of a subordinate clause that's a complement of the matrix verb:

Code: Select all

hau hwika waɲi aja kja  itamu jisaka=su  piwa
1s  just  say  PFV COMP Itamu fish  =FOC eat
I only said that Itamu is eating *fish* (→ that's the only thing I said she's eating)

Code: Select all

hau hi  waɲi aja kja  itamu ihjaisa=su  piwa
1s  NEG say  PFV COMP Itamu bat    =FOC eat
I didn't say that Itamu is eating *bat* (→ that's not what I said she's eating)
There's a lot more to say, but it mostly requires that I go over a fair bit of syntax, and that will have to wait till next time.

Re: Akiatu scratchpad (focus, II: the how)

Posted: Wed Jan 09, 2019 11:18 pm
by akam chinjir
Focus, II: The how

Just to repeat the bare minimum: to focus the direct object of a verb, append the clitic =su along with a high boundary tone, and move it to a position higher in the clause.

Code: Select all

itamu ihjaisa=su  piwa
Itamu bat    =FOC eat
Itamu is eating *bat*
(I'll use asterisks to signal focus, they're not meant to indicate any particular prosody in English. I won't do anything to indicate the boundary tone.)

That a focused object is higher than an unfocused one can be shown by the position of various adverbs, which follow a focused one but precede an unfocused one. (Most of the adverbs mentioned in the previous post are examples.)

Any other argument of the main verb other than the subject can be focused more or less by pretending that it's a direct object. Here's an example with an instrument:

Code: Select all

itamu niwa apatu=su  jai jaku      ihjaisa
Itamu INST spear=FOC do  stay(PFV) bat
Itamu caught a bat with a *spear*
ihjaisa bat is still a direct object in this example---if it had been demoted to an oblique role it would require the locative preposition a. But it can only occur after the verb; this would be ungrammatical:

Code: Select all

*itamu niwa apatu=su  ihjaisa jai jaku
 Itamu INST spear=FOC bat     do  stay(PFV)
 Itamu caught a bat with a *spear*
Even though a focused object and an unfocused one occupy different positions in a sentence's structure, only one of those positions can actually be occupied. Thus, when some other arguments moves into the focus position, the object ends up after the verb.

Here's my current story about what's going on. On its way to the higher focus position, a noun has to stop off in the position right before verb where you expect to find a direct object, and the trace or copy or whatever that it leaves there prevents the actual object from moving there, and that leaves the object stranded after the verb.

(It's also possible for that noun just to stop in the lower position. I'll come back to that possibility in another post.)

One more detail. Two of the prepositions---dative i and locative a---can only appear after the verb. Thus, when a dative or locative argument gets focused, it appears without a preposition:

Code: Select all

kipaja itamu=su  hwati mawa      apatu
Kipaja Itamu=FOC give  find(PFV) spear
Kipaja gave *Itamu* a spear
I'm toying with the idea of making it possible to focus, not a full NP or DP, but just some part of it, leaving a remnant in place after the verb.

One thought is that a sentence like this could be grammatical:

Code: Select all

itamu jisaka=su  piwa aja pai
Itamu fish  =FOC eat  PFV three
Itamu ate three *fish*
Grammatical, but maybe not very useful. (I guess it would be appropriate if someone has just said, mistakenly, that Itamu ate three bats or asked, perversely, what Itamu ate three of.)

The other way around is maybe better:

Code: Select all

itamu pai  =su  piwa aja jisaka
Itamu three=FOC eat  PFV fish
Itamu ate *three* fish
(Normally a number would require a noun to host it, but there'll be exceptions besides this anyway. In particular, I know it will be possible to topicalise an NP's head noun, stranding a number.)

Here's an example with a focused noun modifier:

Code: Select all

hau sama waɲi na tamwi=su  taki wamau jisaka wa
1s  2s   say  DS wood =FOC hold come  fish   CIS
I told you to bring the *wooden* fish
And a possessor:

Code: Select all

hjaci itamu=su  taki wamau apatu wa
Hjaci Itamu=FOC hold come  spear CIS
Hjaci brought *Itamu*'s spear
(itamu could instead be interpreted here as a raised benefactive argument.)

Clefts can also be used to focus any of a verb's arguments, with occasionally subtle effects on meaning.

Code: Select all

(iti) ihjaisa na   itamu ki  hi  piwa
 AFF  bat     COMP Itamu DET NEG eat
It was bat that Itamu didn't eat
The cleft is formed by using the focused NP as a predicate nominal, and following it with a relative clause, typically a na nonfinite clause. The positive polarity item iti is common but not obligatory in this construction.

The sentence with a cleft differs from the simpler itamu hi ihjaisa=su piwa aja Itamu didn't eat *fish* because itamu is focused not in relation to the focus particle hi NEG but (in the absence of anything focus-sensitive in the matrix clause) in relation to the conversational context. (Maybe it's part of the story that Itamu refused to eat bat, but someone else has just said that Itamu didn't eat boar, and I have to set them straight.)

With negation in the matrix clause:

Code: Select all

miwa ihjaisa na   itamu ki  piwa
NEG  bat     COMP Itamu DET eat
It wasn't bat that Itamu ate
This differs only subtly from the monoclausal construction. Perhaps it more clearly implies that Itamu ate something, just not bat.

Clefts make double negation possible:

Code: Select all

miwa ihjaisa na   itamu ki  hi  piwa
NEG  bat     COMP Itamu DET NEG eat
It wasn't bat that Itamu didn't eat
(Different negation particles are needed in the two clauses: stative miwa for the cleft itself, hi with the nonstative embedded verb.)

More generally, clefts let you use multiple focus-sensitive elements without linking them to the same argument:

Code: Select all

iti wija Itamu na  hi  jisaka=su  piwa
AFF even Itamu REL NEG fish  =FOC eat
Even Itamu didn't eat *fish*
Now, as you can see in the last example, clefts allow you to focus the subject, something we haven't seen before. But this isn't the only way to focus the subject. There's also a morphologically-unmarked valency alternation that turns the subject into a direct object and demotes any existing object to a locative oblique. (I tentatively introduced this alternation when I discussed correlative structures. I'm no longer tentative about it.) This leaves the clause with no subject. (I suppose you could instead say that it's got a covert expletive subject, if you're into that kind of thing.) Often something will get topicalised to make up for this, but that's not strictly necessary. (If it's the demoted object that's topicalised, then as usual it'll lose the locative preposition when moving before the verb.)

Here's an example:

Code: Select all

ihjaisa wai hi  itamu=su  piwa aja
bat     TOP NEG Itamu=FOC eat  away(PFV)
*Itamu* didn't eat bat
That itamu isn't just marked with the focus clitic =su but is actually in the usual focus position, which is to say that it's lower in the clause than you expect to find a subject, is shown here by the fact that it follows the negator hi.

You can also focus the verb. The simplest way to do this is, in effect, to unfocus everything else:

Code: Select all

jisaka wai itamu piwa aja
fish   TOP Itamu eat  away(PFV)
Itamu *ate* fish
Of course piwa eat isn't really focused here, it's just that everything else is marked as old information (since both subjects and topics must be definite). But for some purposes that's enough. In particular, it's enough for focus-sensitive particles:

Code: Select all

jisaka wai itamu hwika piwa aja
fish   TOP Itamu just  eat  away(PFV)
Itamu only ate the fish (→ that's all she did to it)
If you want stronger focus---maybe if you're contradicting someone---you can usejai do as an auxiliary verb.

Code: Select all

itamu piwa=su  jai aja       a   jisaka
Itamu eat =FOC do  away(PFV) LOC fish
Itamu *ate* the fish
As you can see, the original object is now a locative oblique. piwa, normally a verb, seems to be a noun here, the (focused) direct object of jai do.

(I'm a bit torn about this construction, but it gives me something I desperately need---a way to nominalise verbs---so its odds of survival are pretty good.)

A few more particulars:
  • Adverbials giving spatial or temporal location, even when they don't need a preposition, can be treated just like nouns, and raise no special issues.
  • Elements in a subordinate clause can be focused in the normal way, within that clause. That's true even in the clauses I've been calling nonfinite.
  • Many adverbials, including adverbial clauses, will normally be in focus just in virtue of being in the sentence, and need no special marking.
  • Those adverbs that you're most likely to think of just as particles---including the focus-sensitive ones I mentioned in the last post---can themselves be focused, if at all, only by intonation.
  • I'm pretty sure that a subordinate clause as a whole can be focused, but I haven't thought much about this yet.
Coda on quantifiers

In the last post I gave hiku all as an example of a focus-sensitive particle, and I now think that was a mistake. hiku will get interpreted relative to an explicitly focused argument, but in the absence of such an argument it will get linked not to the predicate as a whole but to the object (if the verb is transitive) or the subject (otherwise). But that's a topic that needs its own treatment.

Akiatu scratchpad (partial reduplication)

Posted: Mon Jan 14, 2019 3:34 pm
by akam chinjir
Partial reduplication

Just a quick note to say I'm changing how partial reduplication works.

The existing system simply copies a word's final foot, so far just to form inchoatives from stative verbs. (For example, suwasu be asleepsuwasu wasu fall asleep.)

But---if I've been learning the right lessons from some recent reading, that's not really how partial reduplication tends to work. Instead of just copying a syllable or a foot, you tend to use the segments from the base to build a new one---and the one you build won't necessarily have the same shape as the one in the base.

Like, a particular reduplication process might be constrained to always produce a heavy syllable from the base's initial segments. If the base has the form CV.CVC, what you'll get is CVC-CV.CVC---copying the second consonant so you'll end up with a heavy syllable, even though it's not part of the base's first syllable.

Meanwhile, the reduplicant might end up phonologically simpler than the base in various ways---consonant clusters eliminated, maybe, or a default vowel used regardless of what's in the base.

So here's the latest version of how this is going to work in Akiatu. Partial reduplication will always result in two light syllables with simple onsets: CVCV. How it gets there will depend on the shape of the base.
  • If the base ends in two light syllables, they get copied, with any medial glides deleted.
  • If the base ends in a heavy syllable, the initial consonant gets copied between the two vowels (e.g., taitati). Except: if the syllable is jau, instead you get jaku, because of the ban on ju sequences.
  • If the base ends in a heavy syllable followed by a light syllable, CVVCV (maybe with medial glides as well, but those will delete), then the initial CV is ignored and you get kVCV (e.g., maihwikihi.
Fall asleep is still suwasu wasu. But now sit down is ijau jaku instead of ijau jau; I'm counting that as an improvement.

And also as a nice coincidence: I've been using jaku as a reduplicative complement meaning something like to be fixed in place, settled, and here I have it resulting from the reduplication of ijau sit. Pretty cool!

(And it's actually not unheard of to use a similarly derived form in compounds, so I think I'm going to let myself say that the complement jaku really is reduplicated ijau.)

Re: Akiatu scratchpad (partial reduplication)

Posted: Thu Jan 17, 2019 6:21 pm
by Xwtek
Can the reduplication be separated? If not, I suggest you to write it in a single word, likeː

ijaujaku, suwasuwasu, maihwikihi.

To be fair, This is not how reduplication work in Indonesian either. But I personally don't like a orthographical word that doesn't coincide with grammatical word.

Re: Akiatu scratchpad (partial reduplication)

Posted: Thu Jan 17, 2019 10:06 pm
by akam chinjir
They'll be independent phonological words. Some of them will also be able to show up as complements to other verbs, if that makes a difference.

(There are languages with a reduplication pattern that's also used in some compounds---like, instead of using the whole word for the second element of the compound, you reduce it the way you would for reduplication. So I'll be doing something like that. E.g., jaku derives from ijau sit, but shows up not just in ijau jaku sit down but also in, say, paja jaku tie in place.)

I'm not sure that fully justifies separating them from the base word, but that's my rationale.

Re: Akiatu scratchpad (partial reduplication)

Posted: Thu Jan 17, 2019 10:18 pm
by Xwtek
akam chinjir wrote: Thu Jan 17, 2019 10:06 pm They'll be independent phonological words. Some of them will also be able to show up as complements to other verbs, if that makes a difference.

(There are languages with a reduplication pattern that's also used in some compounds---like, instead of using the whole word for the second element of the compound, you reduce it the way you would for reduplication. So I'll be doing something like that. E.g., jaku derives from ijau sit, but shows up not just in ijau jaku sit down but also in, say, paja jaku tie in place.)

I'm not sure that fully justifies separating them from the base word, but that's my rationale.
Is it possible with other verbs like paja kihi, ijau wasu, etc? Otherwise, it's just jaku that is special.

Re: Akiatu scratchpad (partial reduplication)

Posted: Fri Jan 18, 2019 3:36 am
by akam chinjir
It's not fully mapped out yet, but yeah, the idea is that it'll happen with other verbs too---when used as resultative complements, it'll be their reduced (reduplication) form that's used. I think it'll be especially common when using stative verbs as resultative complements to yield a sort of causative.

Here's an example of something I think will work, with wasu being the reduplication form of suwasu be asleep:

Code: Select all

atai cacija tawaru wasu
Atai baby   sing   asleep
Atai sang the baby to sleep
...But this particular example require a surreptitious valency switch, because normally the direct object of tawaru sing is the song, not the target, and I'm not yet sure how much of this sort of valency switching I want to allow.

So---I don't think this'll be fully productive, but there'll be examples other than ijau/jaku.

(And that example is actually just a happy coincidence. I was using jaku as a resultative meaning fix or settle in place well before I came up with reduplication rules that happened to make it derivable from a verb with a suitable meaning.)

Re: Akiatu scratchpad (partial reduplication)

Posted: Fri Jan 18, 2019 3:53 am
by Xwtek
akam chinjir wrote: Fri Jan 18, 2019 3:36 am It's not fully mapped out yet, but yeah, the idea is that it'll happen with other verbs too---when used as resultative complements, it'll be their reduced (reduplication) form that's used. I think it'll be especially common when using stative verbs as resultative complements to yield a sort of causative.

Here's an example of something I think will work, with wasu being the reduplication form of suwasu be asleep:

Code: Select all

atai cacija tawaru wasu
Atai baby   sing   asleep
Atai sang the baby to sleep
...But this particular example require a surreptitious valency switch, because normally the direct object of tawaru sing is the song, not the target, and I'm not yet sure how much of this sort of valency switching I want to allow.

So---I don't think this'll be fully productive, but there'll be examples other than ijau/jaku.

(And that example is actually just a happy coincidence. I was using jaku as a resultative meaning fix or settle in place well before I came up with reduplication rules that happened to make it derivable from a verb with a suitable meaning.)
Then:
  1. You shouldn't call it partial reduplication, call it resultative form.
  2. You really shouldn't make it productive, (except if Akiatu's verbs is closed class). For example if there is language A with verb tekiraki to count using abacus (literally use a abacus), patsuraki to have a sword duel (literally use a sword), photraki to cut a tree (literally use a axe), their resultative forms is all raki, However raki is actually a denominal suffix that mean to use X in language A.

Re: Akiatu scratchpad (partial reduplication)

Posted: Fri Jan 18, 2019 10:30 pm
by akam chinjir
It's tricky, because those forms do in some ways behave like independent words. But, as far as I know that sort of reduction is always associated with reduplication, even in languages where it also shows up in compounds.

The main alternative, I think, is to treat it as a particular sort of compounding.

Synchronically, there aren't (so far) any verbal or verb-forming suffixes, though there are some old ones that are usually still recognisable, and I've wondered about making them extra-metrical, in which case they'd be left out of reduplication (and stress assignment). E.g., one of them is -ru, which shows up (or example) in hakjaru to burn (intransitive) (from hakja fire) and kamaru to feel pain (from kama pain). These might end up reduplicating as hakjaru haka and kamaru kama (and getting stress on the first syllable).

Re: Akiatu scratchpad (partial reduplication)

Posted: Sat Jan 19, 2019 4:47 am
by Xwtek
akam chinjir wrote: Fri Jan 18, 2019 10:30 pm The main alternative, I think, is to treat it as a particular sort of compounding.
That's what I mean. Make your partial reduplication form as resultative form, and make a rule that it's common to compound a verb and the resultative form of itselve
akam chinjir wrote: Fri Jan 18, 2019 10:30 pm Synchronically, there aren't (so far) any verbal or verb-forming suffixes, though there are some old ones that are usually still recognisable, and I've wondered about making them extra-metrical, in which case they'd be left out of reduplication (and stress assignment). E.g., one of them is -ru, which shows up (or example) in hakjaru to burn (intransitive) (from hakja fire) and kamaru to feel pain (from kama pain). These might end up reduplicating as hakjaru haka and kamaru kama (and getting stress on the first syllable).
Sorry if my message is not clear. I mean if your language has a loanword from language A. Your language may not have verb-forming suffixes. However, the language might borrow a word from may have it. And since your language has no verb forming suffixes, it will be analyzed as part of root. However, your rule of reduplication means those root with identical final 2 syllables will be confusable in resultative form.

Re: Akiatu scratchpad (partial reduplication)

Posted: Sat Jan 19, 2019 9:17 am
by akam chinjir
Ah, sorry for misunderstanding, it's a good point. Unfortunately, I think my answer's going to have to be that I'll deal with ambiguities like that on a case-by-case basis. (And they'll no doubt arise even without borrowing, given how simple Akiatu phonology is.)

(And thanks for the questions and suggestions, very helpful.)

Re: Akiatu scratchpad (partial reduplication)

Posted: Sat Jan 19, 2019 11:07 am
by Salmoneus
I know this is a trivial and self-centred comment, given the great breadth and detail of your posts...


...but thanks! Your discussion of partial reduplication and how reduplicants can become autonomous words has both given me a good idea for one language, and reassured me about something I did in another!

Re: Akiatu scratchpad (partial reduplication)

Posted: Sat Jan 19, 2019 2:22 pm
by akam chinjir
I'll take it :)

Reduplicants can definitely become independent phonologically, with their own stress and so on, and you can get similar reductions in compounds (I think especially synonym compounds, I've been trying and failing to dig up the examples I thought I remembered). I don't know that they often end up getting used outside of reduplication constructions, though.

There's one sort of context in which similar sorts of reduction are common without reduplication, nicknames and such. Like: you get "Dave" by taking enough segments off the front of "David" to form a single closed syllable, even though those segments don't actually form a syllable in "David" itself.