Akiatu scratchpad (questions)

[akam chinjir]

(Aside: I'll be changing the title of this post to reflect the latest entry, you can use the links below to find the promised content.)

Latest posts: questions.

Table of Contents

1. Phonology
2. Word formation
3. Pronouns and ki
4. Clause types (matrix clauses)
5. Resultatives
6. More about resultatives perfectives
7. Direct objects
8. Passives
9. Argument raising and impersonal passives
10. Subject control; mwi
11. Other nonfinite complements; na
12. Secondary predicates
13. tija "now" and mikwa "already"
14. acuta "soon," &c
15. Numbers
16. Relative clauses, I
17. Relative clauses, II
18. Relative clauses, III: Correlative clauses
19. Ideophones and manner adverbs
20. Locative phrases
21. Path verbs
22. Causatives
23. Saying and thinking
24. Focus, I: The why
25. Focus, II: The how
26. Partial reduplication
27. ki, the definite determiner
28. Locative subjects
29. Raising and control, again
30. Resultatives, again
31. Stress basics
32. Some morphophonology
33. Phonological do-over
34. Clitics
35. Compounds
36. Telicity, aspect, applicatives, focus...
37. Questions

Introduction

Hi everybody! Here's a thread I'll be using to sort out ideas for a language I've been working on; hopefully it'll be interesting to someone other than me.

First, a quick aside: I've been lurking hereabouts for quite a while now, but only recently started commenting, after the move to the new board. I'll post a proper introduction over on the census thread once life settles down a bit; for now I guess you can call me Akam.

Anyway, the language I'll be posting about here is called Akiatu (/akjatu/, something like [əˈkˣjäː.tʊ]).

Long overdue update: actually that should be /akijatu/ [ˌä.kˣɪˈjäː.tʊ]. I can't have thought it should actually be /akjatu/ for more than a day or two, to be honest.

The Akiatu people live near the mouth of a major river in an imaginary world. I'm a bit hesitant to say much about their society and situation, because my background is so far pretty thin on relevant geological, environmental, technological, and anthropological issues. But for concreteness I'll be assuming that the Akiatu live in about a dozen permanent settlements, mostly surrounded by rainforest; they practice shifting cultivation, and fish with stone-tipped spears. Sample sentences will mention yams, canoes, ancestral spirits, and maybe ancient giants. The implied social structure will be dominated by clan mothers, male village heads, and complicatedly gendered shamans. The Akiatu will (I'll assume) be linked to neighbouring (and not-so-neighbouring) peoples by trade and ritual networks of various sorts.

The broader context is an unnamed subcontinent together with a long looping chain of volcanic islands off to the west. The Akiatu were among the first human settlers of this subcontinent, and as far as my larger project is concerned their language is a diachronic starting point. So: hopefully I can at least some of the time make Akiatu seem like a language with a history, but telling that history is not one of my goals here; and I'll make some design decisions with the aim of opening up interesting and divergent possibilities for daughter languages.

One such design decision that I made early on was that Akiatu would be ruthlessly analytic, and we'll see how that plays out. As it happens, three of the languages I know best are English, Mandarin, and classical Chinese, and those are nicely analytic---so those are likely to be the main natlang inspirations for this language.

Here are some other broad features of the language. It mostly patterns as a VO language, with the big exception that objects will precede verbs (admittedly I'm wavering on this one, have to sort out information structure before I'll be sure). Aspect distinctions are drawn in a variety of periphrastic ways, many of them not really grammaticalised; tense is expressed, if at all, by standard temporal adverbs. Animacy, definiteness, and specificity all play important roles in the syntax, though there are no markers specifically for those categories. The language is pro-drop (in fact argument drop). There are few ambitransitive verbs, at least two ways to form passives, and also an antipassive, which serves syntactic as well as pragmatic purposes, though the language's alignment is nom/acc. Verb chains are common. There's a closed class of about 15 pure adjectives. There's a lot of topicalisation and subordination (and a lot of topicalisation of subordinate clauses). Explicit complementisers do a lot of work, even in matrix clauses, a bit like taking Mandarin's sentence-final particles and putting them at the start of the clause.

Here goes...

[akam chinjir]

Phonology

One early decision was to make the phonology simple, largely so I wouldn't get bogged down in it (which is something that I do). But it couldn't be so simple that nothing interesting could come of it. I settled on the inventory that I'll give in a moment, and a tendency towards longish words made up of CV syllables---figuring that those words could reduce and squash in a variety of interesting ways.

Here are the consonants:

Code: Select all

p t c k 
  s     h 
m n ñ 
  r 
    j w

(ñ represents IPA ɲ.)

Those consonants are accompanied by the most boring of three-vowel systems, a i u.

A word can start with a vowel, but otherwise all syllables have onsets. The legal onset clusters are pw, mw, kj, kw, hj, and hw. A syllable with one of these clusters cannot contain a u; further, kj and hj cannot be followed by i. The only underlying vowel sequences are ai, au, and (in word-final syllables) aa, ii, and uu. All such sequences are tautosyllabic, yielding phonetic diphthongs and long vowels. All syllables are underlyingly open. j cannot occur either before or after u.

A syllable with two vowels is heavy, all other syllables are light. Form feet in the following way, starting at the right edge of the word:

• If the rightmost unfooted syllable is heavy, and it's not the word's rightmost syllable, treat it as light for the purposes of this algorithm (this only happens in reduced forms, see below).

• If one of the three rightmost unfooted syllables is heavy, form a foot starting with that syllable; otherwise form a bisyllabic foot of two light syllables.

• Repeat as necessary to reach the left edge of the word, possibly leaving one light syllable unfooted.

The leftmost syllable in each foot gets stress, the rightmost stress is primary. Stress affects vowel quality (see below) and volume; the syllable with primary stress also counts as prominent for intonational purposes (but I don't as yet have any idea what that's going to mean).

The syllable with primary stress must be heavy. If it's not underlyingly heavy, two things can happen:

• If the next syllable starts with j or w, that glide geminates.

• Otherwise, the stressed syllable's vowel becomes long.

Orthographically I'll distinguish between (for example) aija and ajja. They are phonetically identical, but the first has an underlying vowel sequence ai while the second is underlyingly /aja/ but has had j geminate to avoid giving primary stress to a light syllable.

(In fact I'll probably be very inconsistent whether I indicate this heavying orthographically at all; so far my habit has been to do it in example sentences but not otherwise. I'll also (try to remember to) indicate unstressed heavy syllables with a grave accent on the first vowel.)

Function words often occur unstressed, and some (the monomoraic ones) cannot be stressed. Some function words have different forms depending on whether or not they will be stressed (e.g. the singular personal pronouns, and hwi / hwai "not"). A word with no stress is a clitic and must find a phonological host. No surprises here: a clitic always attaches in the direction you'd expect given constituency relations, and the clitic does not affect its host's stress (e.g., ki=mijja=wati, "that (thing)," with a clitic on each end). (I'll probably forget to mark clitics as such a lot of the time, and just write them as separate words.)

There are some patterns in phoneme distribution that I hope hint at some interesting diachronics.

• r is rare word-initially (though rawu "feel good, be satisfied" is an important exception).

• The palatals (c and ñ) rarely occur before a.

• Alveolar + u syllables are rare, except word-finally; a disproportionate number of verbs end in tu, su, or ru.

I haven't worked out the phonetics in any great detail, but here's a bit:

• Stops aren't significantly aspirated, obstruents aren't ever significantly voiced.

• There's no significant palatalisation. Instead k is affricated before high vowels and sometimes before ai; c and k are always clearly distinct.

• p and m are not rounded unless followed by w; kw can just be [kʷ].

• r is a trill, though not usually an especially assertive one.

• Vowels centralise in unstressed syllables: a → [ə], i → [ɪ], u → [ʊ], ai → [ɛ], au → [ɔ]. Update: I originally had i → [e] and u → [o], that's changed.

...This is getting long, I'll put word formation and morphophonology (such as it is) in a separate post.

[akam chinjir]

Word formation

Akiatu might have no active, productive processes of derivation; I'm not sure yet. But there are various signs that things were more fluid in the past. Here are some examples.

• -kwa, -kuwa seem to have a possessive significance in many stative verbs: japikuwa "be charismatic, powerful," mihikwa "be soaking wet"

• -kwai shows up in words for places: awajakwai "camp, settlement," janakwai "clan hall," kjatikwai "men's house"

• -miku-, -mihi-, and -mii show up in a lot of words relating to liquids: ajamii "to weep," ikjamii "river," pumuki "current, way"

• -tami forms speaker-oriented adverbs from true adjectives (a small closed class in Akiatu): siwitami "insignificantly," amakitami "beautifully, fortunately"

• -wi, which marks plural pronouns, also shows up in a good number of words for groups, mostly groups of animates: ataiwi "eyes," janakiwi "clan," tamwipakuwi "raiding party" (< tamwipaku "canoe")

• A disproportionate number of verbs end in -tu, -su, or -ru. There are often state or event nouns corresponding to these verbs and ending in -ni, as well as patient nouns ending in -nai or -naki. For example: suwasu "to sleep," suwani "sleep," suwanai "sleeper"; jakwasu "to die," jakwani "death," jakwanai "ancestor." (In fact nouns for people ending in -naki or -nai are fairly common, e.g. isaunaki "healer" < isau "medicine," akimaunaki "person of standing (e.g. in a ritual)" < akimau "to be standing.")

Compounding will be quite productive. There will be at least (head-final) N+N compounds. There will also be various V+V constructions that you might interpret as compounds. And backgrounded or nonspecific direct objects will get treated somehow distinctively, and that might end up looking like noun incorporation, which you might think of as a sort of compounding. (I'm currently a bit up in the air about what to do with direct objects, I'll probably be posting about this before too long.)

Full reduplication, especially of verbs, produces manner adverbs: tawaru "laugh, sing" → tawaru tawaru "laughingly, in a laughing manner," mutañi "fall over, stumble" → mutañi mutañi "clumsily."

I don't know how I'll handle compounding and reduplication, prosodically speaking, but I'm pretty sure I'll ban consecutive heavy syllables: a word-final heavy syllable that ends up before a word-initial heavy syllable will have to break or shorten (ai → aji, au → awu, aa → a, and so on).

There's also a process I'll call reduction. (This is a bit weird and if you don't buy it at all then please let me know.) Here's how reduction works:

• If the source word has just one foot, then drop any unfooted syllable, and if the foot consists of three syllables then also drop the third one. For example, janaki "person" → naki, mitana "stone" → tana, aijiki "island" → aiji.

• Otherwise, drop all unstressed syllables. For example, mainutawa "goal, target, purpose" → maita, mitajasa "thing, object" → mija

These are used, for example, to get an inchoative from a stative verb (suwasu "be asleep" → suwasu wasu "fall asleep"); you could think of this as a second form of reduplication, maybe. Reduced forms can also be used with ki to form pronouns of a sort, especially for inanimates; for example, kitana "it, they (of things made of stone)" or kimija "it, they (for inanimates in general)." Reduced forms will probably also show up in lots of compounds. (E.g., tamwipaku "canoe" seems to start with the reduced form of witamwi "tree.")

Reduction can run afoul of Akiatu phonotactics in two ways. It can put a u next to a j, which will trigger j → w. It can also put two heavy syllables side-by-side; the first of the two will be phonetically lightened, with ai → [ɛ] and au → [ɔ]. (This is different from what will happen if compounding or reduplication puts two heavy syllables side-by-side.)

I feel a bit like I'm forgetting something, but that's all I'll say for now about word formation and repair strategies.

[akam chinjir]

Pronouns and ki

Before going on there are a couple of things I can mention quickly that will give useful background.

Here are the personal pronouns; the singular forms have reduced variants that occur when the pronouns are not stressed (e.g., most of the time):

Code: Select all

  Sing  S.Red.  Plural 
1 hau   àu      hawi
2 sama  sàu     sawi 
3 kai   hài     kiwi

In the third person, especially for inanimates, it's also possible to use ki + the reduced form of some appropriate noun, usually a fairly generic one. (In the last post I gave kitana as an example---it would be used for things made of rock---and also kimija, which can be used for anything inanimate.)

Demonstratives use the same ki + reduced noun base, adding a deictic element. Here's how it works using naki < janaki "person":

Code: Select all

kinaki=su   this/these (PROX, by me) (person/people)
kinaki=kisa that/those (MED, by you) (person/people)
kinaki=wati that/those (DIST, elsewhere) (person/people)

hawi "we" and naki < janaki "people" can also be used as generic pronouns, like certain uses of "one" or "you" or "they" in English.

However---pronouns are freely dropped, as are whole arguments, so you don't necessarily see these forms all that often.

A related point: there aren't many ambitransitive verbs in Akiatu, so when an otherwise transitive verb occurs without an object, normally that's because the object is understood, not because the verb is being used intransitively. For example: hau piwwa is "I am eating it" rather than "I am eating (something)."

The morpheme ki that shows up in demonstratives and pronouns has a couple of other uses in which it licenses subjects (to be a bit fancy about it). This is how alienable possession is handled, for example:

Code: Select all

itaamu ki apaatu 
Itamu  KI spear
Itamu's spear

(Yeah, ki is a lot like English 's here, though phonologically it's hosted by the following noun and pro-dropping means it can occur without an overt preceding noun: ki apaatu "her spear.")

(Inalienable possession, by contrast, is unmarked: itaamu hjakiwaani "Itamu's (younger) brother.")

You also use ki if a nonfinite subordinate clause needs a subject:

Code: Select all

hau kiwaita tau           na itaamu ki jisaaka makjai
I   see     together(PFV) DS Itamu  KI fish    to.spear 
I saw Itamu fishing

itaamu here is the subject of jisaaka makjai "fishing"; because this is a nonfinite subordinate clause, the subject must be licensed by ki. (tau is a resultative complement that implies that not only did I see Itamu, but also she might have seen me; its main effect is to render the clause perfective. na is a complementiser for nonfinite subordinate clauses whose subject is not controlled by the subject of the matrix clauses: DS different subject contrasts with SS same subject. More on all these points coming up eventually.)

That should be enough for now, and for tonight. Tomorrow I might be able to put something up about resultatives and aspect.

[akam chinjir]

Updates

I forgot some things. I've edited the previous posts (and also added titles), but here are the new things:

• Unstressed heavy syllables get a grave accent on their first vowel (when I remember)
• Function words often occur with no stress; monomoraic ones cannot be stressed.
• Some words have distinct stressable and unstressable forms, for example the singular personal pronouns and also hwi/hwai "not."
• The reduced unstressable forms of the singular personal pronouns are àu "1s," sàu "2s," and hài "3s."
• A word with no stress must cliticise. It will do so in the direction you'd expect, given constituency relations, and it will not affect its host's stress. (For example, ki=mijja=wati "that (thing)," with clitics fore and aft.)
• A disproportionate number of verbs end in -tu, -su, or -ru. Many of these verbs have corresponding state or event nouns ending in -ni, and patient nouns ending in -nai or -naki. For example: suwasu "to sleep," suwani "sleep," suwanai "sleeper"; jakwasu "to die," jakwani "death," jakwanai "ancestor." (In fact nouns for people ending in -naki or -nai are fairly common in general, e.g. isaunaki "healer" < isau "medicine," akimaunaki "person of standing (e.g. in a ritual)" < akimau "to be standing.")

[akam chinjir]

Clause types (matrix clauses)

Oops, not aspect after all.

Before anything else I'll mention that pre-verbal subjects and topics must be definite.

Noun predicates

Noun predicates need not be marked at all:

Code: Select all

kipajja isaunaki
Kipajja healer
Kipajja is a healer

The particle iti can be used, however:

Code: Select all

kipajja iti isaunaki
Kipajja AFF healer
Kipajja is a healer

iti is a polarity item, not a verb; it occupies the same position as the negative particle miwa:

Code: Select all

itaamu miwa isaunaki
Itamu NEG  healer
Itamu is not a healer

iti (or miwa) is all but required if there is no explit subject:

Code: Select all

iti isaunaki
AFF healer
He is a healer

Identity statements can have the same syntax as other noun predications, so kipajja isaunaki could also be "Kipajja is the healer," given a helpful context. But it would be normal for isaunaki to be topicalised here:

Code: Select all

isaunaki wài kipajja iti
healer   TOP Kipajja AFF
Kipajja is the healer

Adjectives

True adjectives are a small closed class in Akiatu, and anyway they can't be used as predicates. You can use a noun to support them, however.

Code: Select all

itaamu janaaki ahwiita
Itamu  person  tall
Itamu is a tall person

Most often, meanings that we would express using adjectives in English will call for a stative verb in Akiatu:

Code: Select all

itaamu ataurika
Itamu  be.happy
Itamu  is happy

Nouns are also possible:

Code: Select all

tamwipaaku witaamwi
canoe      wood
The canoe is wooden

What all this means is that there's actually no such thing as adjectival predication in Akiatu.

Locative statements

There is no copula for use in locative sentences: instead of saying (say) that Itamu is by the river, you'd have to say that she is sitting by the river, or fishing by the river, or whatever---that is, you can't get away from using an appropriate contentful verb.

Though there's really no limit to what verb you would use, five posture verbs are especially common:

• ahwaicu "to be lying down"
• akimau "to be standing"
• ijau "to be sitting"
• kaiku "to be squatting"
• pijatu "to be hanging"

With human subjects, you'd use the posture verb that best suits the person's actual posture, if known---defaulting to ijau "to be sitting," but also sometimes using akimau "to be standing" with an honourific significance. With non-human subjects the choice of posture verb is somewhat grammaticalised, but I won't go into the details right now.

Here's an example:

Code: Select all

Itaamu ijau       a   ikjamii kuura
Itamu  be.sitting LOC river bank
Itamu is sitting by the river = Itamu is by the river

Locative statements using posture verbs are negated with miwa:

Code: Select all

itaamu miwa ijau       a   ikjamii kuura
Itamu  NEG  be.sitting LOC river   bank
Itamu is not (sitting) by the river.

Existential statements

An existential statement is, more or less, a locative statement with an indefinite subject; it can also be possible to omit the actual locative phrase. There is a restriction though: existential statements almost always use one of the core posture verbs, and under no circumstances can use a transitive verb. (Locative statements are not restricted in this way.)

Because the subject is indefinite, it cannot go before the verb. Normally to make up for that something else will step up, usually the locative phrase itself. One complication: the locative preposition a cannot occur initially, so it gets dropped.

Code: Select all

ikjamii kuura pijatu     ijaisa
river   bank  be.hanging bat
There is a bat (/ are bats) by the river.

Existential statements are negated with miwa; the subject often takes the negative indefinite determiner mai:

Code: Select all

ikjamii kuura miwa pijatu     mai ijaisa
river   bank  NEG  be.hanging no  bat
There are no bats by the river

Presentative statements

Existential statements can be used to introduce a new discourse topic, and when they are, we can consider them presentative statements. Presentative statements are a bit more flexible than existentials in the verbs they can use; wamai "to come" is especially common.

Note that the ban on indefinite subjects in Akiatu means that presentative sentences are very common.

Code: Select all

ikjamii kuura wamai hja    itunaki
river   bank  come  arrive someone
Someone came to the river.

Presentative statements often make use of secondary predicates:

Code: Select all

ikjamii kuura wamai itunaki mwi tawaru
river   bank  come  someone SS  sing
Someone was approaching the river, singing.

(Note that the indefinite itunaki "someone" is still counted the subject in this sentence, licensing the same-subject subordinator mwi.)

Possessive statements

Possessive statements can follow the pattern of either existential or locative statements, normally using either ijau "to be sitting" or pijatu "to be hanging." This is straitforward following the locative pattern:

Code: Select all

apaatu ijau       a   itaamu
spear  be.sitting LOC Itamu
The spear is on Itamu (= Itamu has it)

Because a cannot occur initially, for the existential pattern you need some other way to turn a possessor into a sort of location. Use the deictic elements su, kisa, and wati in one of two ways:

Code: Select all

itaamu=wati ijau       apaatu
Itamu  DIST be.sitting spear
Itamu has a spear

itaamu watiiwi  ijau       apaatu
Itamu  environs be.sitting spear
Itamu has a spear

Other statements

Of course many statements will be built around a regular verb. There's going to be some trickery with direct objects, but for now I'll just put the object before the verb:

Code: Select all

itaamu jisaaka makjai
Itamu  fish    to.spear
Itamu is fishing

The sentences discussed above were all negated with miwa; that was because they were all stative. Non-stative sentences are negated with hwi / hwai:

Code: Select all

itaamu hwi jisaaka makjai
Itamu  NEG fish    to.spear
Itamu is not fishing

Speech-act particles

Edit: this was too rushed, and anyway I have to think about it more.

Sentences of all sorts can occur with a speech-act particle, most often in initial position:

[code]
kja itaamu jisaaka makjai
PRT Itamu fish to.spear
Itamu is fishing.
[/code]

kja marks assertions, lending some emphasis (it is also used in finite declarative subordinate clauses).

awa can be used to soften a statement:

[code]
awa itaamu jisaaka makjai
PRT Itamu fish to.spear
I guess Itamu is fishing.
[/code]

There are a few other options here, but I won't pursue them for now.

Questions

Polar questions use the sentence-initial particle hasai:

Code: Select all

hasai itaamu jisaaka makjai
Q     Itamu  fish    to.spear
Is Itamu fishing?

The question words for content questions are most often formed with cu + reduced noun:

• cunaki, who?
• cumija, what?
• cuwakwai, where? (< awajakwai "place, camp, settlement")
• cumaita, why? for what end? (< mainutawa "goal, purpose, target")
• ...and so on

These get put at the front of the sentence, and are followed by the particle sai:

Code: Select all

cunaaki sai jisaaka makjai
who     Q   fish    to.spear
Who is fishing?

Code: Select all

cuwakwai sai itaamu jisaaka makjai
where    Q   Itamu  fish    to.spear
Where is Itamu fishing?

Prepositions and such cannot be stranded, so pied piping can be necessary:

Code: Select all

sati cunaki sai itaamu jisaaka makjai
COM  who    Q   Itamu  fish    to.spear
Who is Itamu fishing with?

More details eventually, of course.

[Frislander]

Looks decent, fairly Austronesian in feel, but word of advice; please don't post so much stuff all at once it makes it hard to comment on individual things.

[akam chinjir]

Yeah, good advice, I was getting carried away.

[burke]

This has been a fun read so far! I really like how the prepositions and definiteness force interesting variety in the sentences, but you present it in an easy to follow manner.

About the question word "hasai": it feels like one of those things that could develop from an overly contracted phrase meaning something like "is it that...". Not sure if that was the goal, but a nice turn none the less

[akam chinjir]

Hey, thanks!

About the question word "hasai": it feels like one of those things that could develop from an overly contracted phrase meaning something like "is it that...". Not sure if that was the goal, but a nice turn none the less

To be honest, I just picked a word form I liked, and didn't think too much about diachronics. There's one detail, though, in support of analysing hasai as ha + sai, namely that sai by itself occurs with content question words. So maybe you've got:

ha sai itaamu jisaaka makjai?
Is Itamu fishing? ( = Is it that Itamu is fishing)

And:

cunaaki sai jisaaka makjai
Who is fishing? ( = Who is it that is fishing)

...with both questions being clefts, with sai being the complementiser or relativiser or whatever.

[dewrad]

I honestly have nothing to say here aside from the fact that I really like what you're doing. I'm really digging on this language.

[mèþru]

Same

[akam chinjir]

Resultatives

Thanks! I'm glad you're enjoying it.

Here's another instalment, about resultative constructions. It's fairly directly inspired by Mandarin, though it starts looking a bit Englishy at a couple of points.

So it's common in Akiatu to have two verbs in sequence, the first one describing an event or action or something, and the second one specifying a result. Here's an example:

Code: Select all

witaamwi hakjaaru ajja
tree     burn     be.gone
The tree burned down

One reason why this is so common is that the resultative complement gives the clause a perfective meaning. Without the resultative aja, the above sentence would be:

Code: Select all

witaamwi hakjaaru
tree     burn
The tree is burning

All finite clauses must be either perfective or imperfective, and the use or nonuse of a resultative is the main way to mark the distinction between them. (But more on that another time.)

Now, the example above shows a very common kind of case: an intransitive resultative that says what happens to the subject of an intransitive main verb. Here is another example:

Code: Select all

itaamu suwaasu rawwu
Itamu  sleep   feel.good
Itamu slept well

rawu a bit suggests the image of someone who is satisfied after eating, and is very common.

(Reminder: I cite words giving the underlying phonology, but when using them in context---and assuming I'm paying attention---I'll lengthen the stressed vowel if necessary. That's why I cite rawu but use rawwu, for example.)

Another example with rawu:

Code: Select all

itaamu ikiihwa rawwu
Itamu  leave   feel.good
Itamu left and was pleased

It's also possible for an intransitive resultative to add an argument of its own when added to an intransitive main verb; the resulting meaning is often causative:

Code: Select all

itaamu hjaaci tawaaru rawwu
Itamu  Hjaci  laugh    feel.good
Itamu laughed, pleasing Hjaci.

In English you can get away with saying Itamu laughed Hjaci happy and the meaning would be pretty close. The English rule (as far as I can tell) is that a resultative must describe an object, so it will license an object even with intransitive verbs.

Akiatu is different. As we saw above, resultatives can comfortably describe subjects. The pair of an intransitive main verb and an intransitive resultative is simply ambitransitive: taawaru rawwu is transitive in the sentence just quoted, but could also be intransitive:

Code: Select all

itaamu tawaaru rawwu
Itamu  laugh   feel.good
Itamu laughed herself happy

(Individual verbs are rarely ambitransitive, though.)

With a transitive main verb, an intransitive resultative will always describe the effects on the object:

Code: Select all

itaamu jisaaka piwwa ajja
Itamu  fish    eat   away
Itamu ate (up) (the) fish.

The intransitive resultative won't describe the subject of a transitive main verb even if that would make a lot more semantic sense:

Code: Select all

itaamu jisaaka piwwa rawwu
Itamu  fish    eat   feel.good
Itamu ate fish, and the fish was satisfied

Probably what you want to say is that Itamu was satisfied, but this is not a way to say that.

It does not help to drop the object, since the verb will still be interpreted as a transitive verb:

Code: Select all

itaamu ∅ piwwa rawwu
Itamu  3 eat   feel.good
Itamu ate it, and it was satisfied

What you need here is a construction I'll call an antipassive, which makes use of the reflexive pronoun kitikwa (< reduced form of tijakwara face):

Code: Select all

itaamu kitikwa piwwa rawwu
Itamu  REFL:AP eat   feel.good
Itamu ate herself full

You'll see that once again English allows a similar construction, though again there's a difference: you use the reflexive in English because an English resultative can only describe the object, so you need the reflexive to get the right object; and you can use the reflexive in the same way with plainly intransitive verbs, as in Itamu ran herself ragged. In the English cases the reflexive is a genuine argument of the verb, and (when this makes semantic sense) it can be replaced by an arbitrary object: Itamu ran the horse ragged. (Bad example, there are no horses in this conworld.)

The Akiatu reflexive here produces a genuinely intransitive verb, as the following example demonstrates:

Code: Select all

itaamu kitikwa piwwa rawwu     a   jisaaka
Itamu  REFL:AP eat   feel.good LOC fish
Itamu ate herself full on fish

Even with the reflexive, an explicit object can be present, though it must occur with the locative preposition a.

A resultative complement can also be transitive. With an intransitive main verb, the two verbs will share a subject:

Code: Select all

itaamu kipajja tawaaru wupaisu
Itamu  Kipaja  laugh   scare
Itamu laughed, scaring Kipaja.

When both verbs are transitive, they can share both subject and object:

Code: Select all

itaamu pisaaka makjai pwaiku
Itamu  fish    spear  hit
Itamu speared the fish.

Here, Itamu both speared and struck the fish (and the striking was the result of the spearing).

A ditransitive result is also possible, making use of an indirect object:

Code: Select all

itaamu apaatu ajamijja pwaiku i   witaamwi
Itamu  spear  throw    hit    DAT tree
Itamu throw the spear and hit the tree.

Which resultative to use on a particular occasion is mostly down to semantics. However, some verbs are especially common as resultatives (some of them in fact cannot be used as main verbs). Here are some of the most important ones:

• aja: leave; away; out (resultative only)
• hja: arrive (< main verb ihjatu)
• jaku: be fixed in place
• jakwa: be dead
• kahawa: to go (used as a detransitiviser)
• kuu: fail (resultative only)
• nai: (resultative only, mostly bleached of semantic content but implies agency)
• pwaiku: hit, strike
• rawu: feel good, be satisfied or pleased
• saka: succeed, finish, complete (resultative only)
• tau: come together, meet (resultative only)
• tima: be available, ready for use
• tisa: finish (only with path verbs like pisuru go across or over)

saka succeed and kuu fail are special in that they are typically followed by another resultative:

Code: Select all

itaamu ki apaatu kausita  tiima
Itamu  KI spear  look.for be.ready
Itamu found her spear

Code: Select all

itaamu ki apaatu kausita  saka    tiima
Itamu  KI spear  look.for succeed be.ready
Itamu was able to find her spear

Edit, better translation: Itamu managed to find her spear.

Code: Select all

itaamu hwi ki apaatu kausita  kuu  tiima
Itamu  NEG KI spear  look.for fail be.ready
Itamu could not find her spear

Edit, better translation: Itamu didn't manage to find her spear.

If you know some Mandarin, you'll probably see the similarity to 找得到 and 找不到 (zhao de dao and zhao bu dao)---except that the extra resultative is optional here, and kuu fail must be licensed by a clausal negator such as hwi (which is semantically a bit funny, because it's not the clause as a whole that's getting negated).

kahawa to go has a somewhat shifted meaning when used as a resultative. Here's an example:

Code: Select all

itaamu makjai   kahawa a   jisaaka
Itamu  to.spear go     LOC fish
Itamu thrust her spear at the fish

makjai is normally a transitive verb, and jisaaka makjai to spear (the) fish implies success: the fish actually gets hit by the spear. With kahawa you can still mention the fish as the target of the action, but without the implication that the spearing was successful.

Finally, there are some lexicalised combinations of a main verb with a resultative. For example, on its own wañi is to think, but with aja it becomes to say:

Code: Select all

itaamu waañi     mwi jisaaka makjai
Itamu  think/say SS  fish    to.spear
Itamu thinks she will go fishing.

Code: Select all

itammu waañi     ajja     mwi jisaaka makjai
Itamu  think/say out(PFV) SS  fish    to.spear
Itamu said she will/would go fishing.

And the fact that nakauti to ask regularly takes jaku to settle in place as its resultative complement is probably not predictable.

[akam chinjir]

More about resultatives perfectives

I was going to post more about resultative complements, to set up a discussion of aspect; but then I decided I should rename them as perfective complements (I'll explain why in a moment); and a certain amount of stuff about aspect got mixed in. Here's what I came up with.

Let's start with this example:

Code: Select all

itaamu suwaasu rawwu
Itamu  asleep  feel.good(PFV)
Itamu slept (and was refreshed).

Here the complement indicates a culminating state: Itamu slept, and as a result of that, and after that, felt good. rawu here is a genuinely resultative complement.

But I want these complements to be the main way of getting a perfective meaning, and it won't always make sense to mention a result. Suppose I want to say that Itamu slept all night, using a perfective. (Maybe I want to go on and say that she then woke up and went swimming.) I can't use a resultative complement like rawu here because giving both the result and the duration (all night) doesn't really make sense.

So I'll have to allow some complements that yield a perfective meaning without mentioning a result. Hence the change in terminology: these are now officially all called perfective complements.

Here's how it can work with suwasu:

Code: Select all

itaamu suwaasu jaaku        kamiwaapi ahiwwa
Itamu  asleep  settled(PFV) night     one:whole
Itamu slept all night.

Syntactically speaking, jaku is just like rawu. The difference (which at this point I'm basically just stipulating) is that rawu does and jaku does not specify a result.

There's a third option with stative verbs like suwasu: use the verb's own reduced form as the complement, yielding an inchoative meaning.

Code: Select all

itaamu suwaasu waasu
Itamu  asleep  REDUP(PFV)   ← I'll gloss this as a sort of reduplication
Itamu fell asleep.

Here, the complement does in a way describe a result, but the main verb doesn't describe a process or activity leading up to that result: the wasu isn't really a result of the suwasu.

So there are at least three options for these perfective complements:

• suwaasu waasu: focus on the inception of the state
• suwaasu jaaku: focus on the whole duration of the state
• suwaasu rawwu: focus on the result of the state

The first of these is only available for stative verbs, but process verbs will often have both of the other two options. Here are examples with ajamii weep:

Code: Select all

itaamu ajamii wukau
Itamu  weep   tired
Itamu wept herself tired.

(Aside: I'll actually be coming back to that nifty English use of a reflexive.)

Code: Select all

itaamu ajamii ajja
Itamu  weep   out/away
Itamu wept.

The first of these two examples is a resultative: Itamu was tired (or used up, or worn out) as a result of weeping. In the second, aja is not really a resultative; it does reflect the fact that weeping is somehow centrifugal, but mostly it gives the phrase a perfective meaning. (Etymologically, ajamii actually looks like aja + mii: outwater?)

I mentioned last time the distinction between wañi think and wañi aja say. That also reflects the centrigal or outward-directed sense of aja, without really generating a resultative meaning.

(Aside: if you want a perfective with the meaning think, thought you can use wañi jaku or wañi nai, though in neither case would the result be unambiguously about thought rather than speech.)

aja can also have a narrowly resultative significance, as we have seen:

Code: Select all

witaamwi hakjaaru ajja
tree     burn     away
The tree burned down

There's an interesting point here, actually. Consider these examples:

Code: Select all

itaamu ikiihwa ajja
Itamu  leave   away
Itamu left.

Code: Select all

itaamu ikiihwa rawwu
Itamu  leave   feel.good
Itamu left.

One difference here is that rawu adds the sense that Itamu welcomed her departure. But also: the first example with aja leaves the narrative focus on the place that Itamu left, whereas the second example has the focus follow Itamu.

I'll raise one last issue, the connection between these complements and what's sometimes called lexical aspect. I showed one such connection above, with stative verbs. There are a couple of other things worth mentioning, though.

First, some verbs describe events whose durations are usually insignificant. These include both semelfactives (like sneeze or knock) and accomplishments (like arrive or find). You might think it strange for verbs with meanings like these to require a complement in order to be perfective. That is, you might expect a verb meaning sneeze to be perfective on its own, given the sort of event it describes.

What sets these verbs apart in Akiatu is that (at least in finite clauses) they cannot occur on their own: they must be explicitly marked for aspect, whether perfective or imperfective, and perfectives require a perfective complement. For example, you can't simply sneeze, you have to sneeze out:

Code: Select all

hjaaci haticii ajja
Hjaci  sneeze  out
Hjaci sneezed.

With accomplishments it's actually often better to use a perfective complement with the required meaning:

Code: Select all

kipajja wamai hja
Kipaja  come  arrive(PFV)
Kipaja arrived.

Finally, some perfective complements are only possible with telic verbs (or telic phrases, this looks like it might be complicated). saka manage to, finish and kuu fail to, not manage to are two; another is siwa, which occurs with path verbs and indicates that the path has been completed:

Code: Select all

kipajja iwaari ma  amawaitu  siwwa    a   tamwipaaku
Kipaja  float  SUB go.around COMP:PFV LOC canoe
Kipaja swam around the canoe.

Three additional notes:

1. When glossing perfective complements, I've started giving both a hint of their meaning (e.g., rawu is feel.good) and an explicit (though parenthesised) mention of their grammatical role (PFV for perfective). I hope that's not confusing.
2. I've also adopted the convention of translating perfectives with the past tense and imperfectives with the present tense. I hope that's not convincing, given that Akiatu doesn't have a grammaticalised way to indicate tense.
3. I changed my gloss of suwasu from sleep to (be) asleep because it's supposed to be stative.

[akam chinjir]

Direct objects

So the idea is that the direct object usually precedes the verb, but there are various obvious signs that it's moved there from a position following the verb. (That's where you'd expect it to be given other word-order facts about the language.)

Talking about this will require some attention to information structure and prosody, and I don't feel especially on top of those topics, so some of the details below are quite tentative.

There are four positions that the direct object can occupy in an Akiatu clause:

• Right up the front, as a topic. (I won't talk further about topicalisation for now.)
• Before the verb complex, potentially separated from the verb by certain adverbs (including the negators miwa and hwai). I'll call this focus position. When a direct object occurs here, it's being presented as new information; strictly speaking the focus might be on the predicate (understood here to include the object) or on the sentence as a whole, or focus might be unmarked, but it won't be on some other constituent.
• Immediately before the verb, in what I'll call incorporated position. When the object is here, nothing can come between it and the verb complex. I don't know enough to be sure that a noun in this position is strictly incorporated, but only a bare noun is possible and it looks like head movement to me. A noun here is being backgrounded. This can be because the specific identity of its referent doesn't matter, or because it is already sufficietly established as a discourse topic. In the latter case the object will be definite, semantically speaking; in fact even proper nouns can go in this position, and object pronouns normally do. (However, given the promiscuity with which Akiatu drops arguments, definite but unfocused objects often go completely unmentioned.)
• Immediately following the verb. The object can occur here only if some other constituent (an indirect object, maybe, or a locative expression) raises to occupy the focus position. Also, if a bare noun occurs in incorporated position, it can strand other elements of the DP here.

Now, I've given a bunch of simple examples more or less like this:

Code: Select all

itaamu jisaaka makjai
Itamu  fish    to.spear
Itamu is  fishing

How can you decide whether a bare preverbal object, like jisaka fish in this sentence, is in focus position or in incorporated position?

The key here is prosody: if the object is in focus position, then it will be more prominent, prosodically speaking, than the verb; but in incorporated position it will be less prominent. For now I won't get into the question of what exactly this means, but it'll involve at least a higher pitch on the syllable with main stress.

In fact, though, an utterly bare noun before the verb is most likely in incorporated position, because an object in focus position will usually have at least one or both of the following elements:

• If it is indefinite, it will probably be followed by itu one, some. (In this use itu does not imply anything about number, see below.)
  

  Code: Select all

  itaamu jisaaka itu       piwwa ajja
  Itamu  fish    one:INDEF eat   away
  Itamu ate a fish (or: some fish)
  

• If it is focused, it will probably be followed by su or wati. (These are the proximal and distal deictic elements that also show up in demonstratives.)
  

  Code: Select all

  itaamu wipakijja wati kiwaita waita
  Itamu  butterfly FOC  see     REDUP(PFV)
  Itamu caught sight of the butterfly
  

  (Normally su will show up with animate objects and wati with inanimate ones, but the animate/inanimate distinction might not always get drawn the way you expect. Changed my mind about this.)
  
  You can get both an indefinite itu and a focus marker:

  Code: Select all

  itaamu wipakijja itu       wati kiwaita waita
  Itamu  butterfly one:INDEF FOC  see     REDUP(PFV)
  Itamu caught sight of a butterfly
  

Okay, so what about a definite direct object that is not focused? In fact an object like that does not really belong in focus position in the first place: it should be topicalised, if appropriate, or otherwise either put into the incorporated position or simply dropped.

I noted above that putting a bare noun in incorporated position can strand other parts of the DP after the verb. Here is an example:

Code: Select all

itaamu jisaaka makjai mawa      haaku wati
Itamu  fish    spear  find(PFV) five  WATI
Itamu caught five fish

The haku five cannot go with the head noun jisaka because only bare nouns are possible in incorporated position, so it remains after the verb, obligatorily followed by wati. (I don't know how to gloss wati in this usage.)

wati can occur even if there are no other (overt) elements of the DP left after the verb:

Code: Select all

itaamu jisaaka makjai wati
Itamu  fish    spear  WATI
Itamu is fishing

Because of the wati here, you know that the object, jisaka, is in incorporated position; that's it's only semantic significance.

(I say something to relate the apparent noun incorporation here with Mithun's well-known typology but that would likely require rather a lot of work and this is already long. Maybe it's worth mentioning that I looked for inspiration especially to languages like Maori that have something like noun incorporation in an analytic context, as well as to languages like Turkish or Persian with differential object marking.)

A note on itu

itu one is used for counting and as a constituent in higher numbers such as haku itu six. But when it occurs directly with a noun it indicates indefiniteness rather than number---it's use is consistent with plural and mass interpretations.

itu can follow the noun (after any adjectives), in the usual place for numbers, or precede it, in the determiner slot. There is a subtle difference: postnominal itu can only be used with specific indefinites, whereas preverbal, determiner itu cannot be used when the identity of the head noun's referent is known to the speaker. (So janaki itu is some people, maybe I know who but you don't whereas itu janaki is some people, neither of us know who.)

This itu can also be used with reduced forms to yield indefinite pronouns: naki itu or itu naki someone.

To actually indicate that you are dealing with one of something, use the adjective ahiwa one, alone, solitary, unique, whole, which can co-occur with itu:

Code: Select all

itaamu jisaaka ahiwwa    iitu      piwwa
Itamu  fish    one:whole one:INDEF eat
Itamu was eating a whole fish

[akam chinjir]

Passives

(A change of custom: I've decided it's probably more confusing than helpful to represent orthographically the lengthening of short vowels that get main stress, so I'm going to try not doing that.)

In this post I'll talk about a few ways to construct passives in Akiatu.

Now, Akiatu allows constituents to topicalise quite freely, and also allows arguments to be dropped. So if you have something like this:

Code: Select all

hjaci itamu ikjahuu
Hjaci Itamu admire
Hjaci admires Itamu

Topicalisation and argument-dropping could yield this:

Code: Select all

Itamu wai ikjahuu
Itamu, (he) admires her.

Given that, you might wonder why you need a passive voice in the first place.

The answer is twofold. First, a dropped argument is still semantically present, and will be interpreted as either definite or generic. (So piwa eat without an explicit object is eat it not eat something---in Akiatu it's an implicit argument, not a valency alternation.) So if you don't want an agent at all, simply not mentioning it won't be enough; you probably want a passive.

Second, topicalising the object is a good way to shift discourse topics, but it's a lousy way to keep talking about the same topic. For that---for a series of sentences about the same person or thing---it's more appropriate to put the continuing topic in subject position.

The most productive (and grammaticalised) way to form passives is by using one of the verbs anatu meet and kiwaita see. The main difference between them is that anatu does and kiwaita does not imply that the subject of the passive is significantly changed by the reported event or situation. (You could maybe think of this as marking a distinction between patients and themes.)

Here are two simple examples:

Code: Select all

jisaka anatu      piwa aja   (← the eating affects the fish, so anatu)
fish   meet(PASS) eat  away
The fish was eaten

Code: Select all

itamu kiwaita ikjahuu          (← maybe Itamu doesn't even know about
Itamu see     admire              the admiration, so kiwaita)
Itamu is admired

The semantic subject is not mentioned in either of these, but you can't really get away from the implications that something did the eating and someone did the admiring. That sort of implication can be canceled in anatu passives with some verbs:

Code: Select all

witamwi siwi  anatu kwaipitu wuhau
tree    small meet  break    ruined(PFV)
The branch broke

kwaipitu break is normally transitive, but with anatu and no explicit subject you get a kind of middle voice, with no inherent implication that something did the breaking.

To make the agent explicit in a passive construction, use the ablative preposition hu with anatu and the locative preposition a with kiwaita:

Code: Select all

jisaka anatu piwa aja  hu  itamu
fish   meet  eat  away ABL Itamu
The fish was eaten by Itamu

Code: Select all

witamwi siwi  anatu kwaipitu kakjai       hu  itamu
tree    small meet  break    useless(PFV) ABL Itamu
The branch broke

Code: Select all

itamu kiwaita ikjahuu a   hjaci
Itamu see     admire  LOC Hjaci
Itamu is admired by Hjaci

Unlike in the V+V constructions I discussed in previous posts, anatu and kiwaita can be separated from the main verb by certain adverbs, including for example ikijiku often:

Code: Select all

itamu anatu ikijiku     kjaitiwa kaiku     (hu   kipaja)
Itamu meet  often(IPFV) poke     down(PFV)  ABL  Kipaja
Itamu is often insulted (by Kipaja)

Passives can be formed on the direct object (theme) of ditransitive verbs like hwati give. Here's an active sentence:

Code: Select all

kipaja apatu=wati hwati tima       i   itamu
Kipaja spear=FOC  give  ready(PFV) DAT Itamu
Kipaja gave the spear to Itamu

And here is the passive:

Code: Select all

apatu kawaita   hwati tima       i   itamu (a   kipaja)
spear see(PASS) give  ready(PFV) DAT Itamu  LOC Kipaja
The spear was given to Itamu (by Kipaja).

(The recipient can also become the subject of a passive via a process called argument raising that I'll talk about in another post.)

There is another way to form passives using certain stative verbs---superficially it's similar, but the structure is actually quite different. Typically (though not always) these are verbs that can take a clausal complement:

Code: Select all

jiku    na (naki) ki jisaka piwa aja
be.good DS (one)  KI fish   eat  away(PFV)
It is good to eat fish, it is good for one to eat fish

The verbs in question can take a VP complement, and will give that VP a passive interpretation:

Code: Select all

jisaka jiku    piwa
fish   be.good eat
Fish are good to eat (= fish are nourishing, auspicious...)

Code: Select all

jisaka wakiru  piwa
fish   be.easy eat
Fish are easy to eat (= fish are tasty)

This construction allows postverbal arguments:

Code: Select all

jisaka wakiru hwati i   hjaci
fish   easy   give  DAT hjaci
Fish are easy to give to Hjaci

(Cf. wakiru na ki jisaka hwati i hjaci, it is easy to give fish to Hjaci.)

The semantic subject can be given in this construction using the locative preposition a:

Code: Select all

jisaka wakiru hwati i   hjaci a   itamu
fish   easy   give  DAT Hjaci LOC Itamu
Fish are easy for Itamu to give to Hjaci

Such constructions are possible with aku too, too much, too far and ai enough, so:

Code: Select all

aku taricuma piwa
too be.big   eat
It is too big to eat

Code: Select all

ai     taricuma piwa
enough be.big   eat
It is big enough to eat

Many of these constructions contrast with active counterparts that use the same-subject complementiser mwi:

Code: Select all

Itamu wakiru  mwi jisaka piwa
Itamu be.easy SS  fish   eat
Itamu is easy to eat fish (= Itamu likes to eat fish)

Here is another series of examples, showing a fuller paradigm (but ignoring some nuances for now), using itaja be possible, be able to:

Code: Select all

itaja       kja  itamu jisaka piwa aja
be.possible COMP Itamu fish   eat  away(PFV)
It is possible that Itamu ate fish

Code: Select all

itaja       na ki jisaka piwa
be.possible DS KI fish   eat
It is possible to eat fish

Code: Select all

itaja       na itamu ki jisaka piwa
be.possible DS Itamu KI fish   ea
It is possible for Itamu to eat fish

Code: Select all

Itamu itaja       mwi jisaka piwa
Itamu be.possible SS  fish   eat
Itamu can eat fish

Code: Select all

jisaka itaja       piwa
fish   be.possible eat
Fish are edible

(This paradigm is a larger topic, and I'll come back to it eventually. I also haven't properly introduced the complementisers kja, na, and mwi.)

[akam chinjir]

Argument raising and impersonal passives

This post is about an operation that I'll call argument raising, and an impersonal sort of passivisation that it can feed.

I'll start with an example that'll look like dative shift. Here's a ditransitive sentence with neutral word order:

Code: Select all

kipaja  apatu hwati mawa      i   itamu
Kipaja  spear give  find(PFV) DAT Itamu
Kipaja gave the spear to Itamu

The theme is encoded as a direct object, before the verb, and the recipient is an indirect object, flagged with the dative preposition i.

Corresponding to that sentence we can also have this:

Code: Select all

kipaaja itamu=su  hwati mawa      apatu
Kipaja  Itamu=FOC give  find(PFV) spear
Kipaja gave Itamu a spear

Four things have changed:

• The recipient noun phrase now appears before the verb; to be precise, it is in focus position.
• The recipient takes =su, indicating that it is focused. Unlike with direct objects in this position, focus is obligatory (though it need not be marked with =su).
• The dative preposition i has been dropped.
• The theme noun phrase now occurs after the verb.

All in all, this looks a lot like dative shift in English (as the translations are meant to imply).

One thing: I'm assuming that when (as usual) the direct object ends up before the verb, it raises to that position from a spot after the verb; but when some other argument gets raised to focus position, that somehow blocks movement of the object, which then stays in place after the verb.

Anyway the Akiatu construction is much more general than dative shift. Here's an example with a destination rather than a recipient, first with the destination in its unmarked place after the verb:

Code: Select all

hjaci kahawa hja         i   mikuwi hatau
Hjaci go     arrive(PFV) DAT waters great
Hjaci went to the ocean

Here is how it looks with the destination raised to focus position:

Code: Select all

hjaci mikuwi hatau=su  kahawa hja
Hjaci waters great=FOC go     arrive(PFV)
To the ocean Hjaci went.

Since kahawa go is intransitive, there's no object to strand in a postverbal position, but otherwise this example works exactly like the preceding one. (I've tried to indicate the shift in focus by fronting the destination phrase in my translation, and I'll play similar translation tricks below, but this is not meant to be more than suggestive.)

You can do the same thing with a location. In unmarked position:

Code: Select all

itamu iwari iwaru parawara a   ikjamii kihamii
Itamu float REDUP wander   LOC river   in.water
Itamu is swimming in the river

And with the locative raised to focus position:

Code: Select all

itamu ikjamii kihamii =su  iwari iwaru parawara
Itamu river   in.water=FOC float REDUP wander
In the river Itamu is swimming

Like the dative preposition i, the locative preposition a gets dropped when the locative phrase is raised. In fact it's a general rule that neither preposition can occur before the verb.

But other prepositions do not observe this rule:

Code: Select all

itamu kañi  wañi  aja      niwa wanawapi ahiwa
Itamu pride say   out(PFV) PER  daytime  one:whole
Itamu spoke about pride all day

Code: Select all

itamu niwa wanawapi ahiwa    =su  wañi aja      kañi
Itamu PER  daytime  one:whole=FOC say  out(PFV) pride
All day Itamu spoke about pride

sati (comitative, with) raises a complication. Here's how it goes:

Code: Select all

itamu jakwanaiwi wañi sati kipaja
Itamu ancestors  say  COM  Kipaja
Itamu is talking about the ancestors with Kipaja

Code: Select all

itamu sati kipaja=su  wañi jakwanaiwi
Itamu COM  Kipaja=FOC say  ancestors
With Kipaja Itamu is talking about the ancestors

Here's the complication: itamu sati kipaja is perfectly fine as a conjunction: Itamu and Kipaja. And in fact you can also have this:

Code: Select all

itamu sati kipaja=su  jakwanaiwi wañi
Itamu and  Kipaja=FOC ancestors  say
Itamu and Kipaja are talking about the ancestors

This and the previous example are distinguished by the position of the object, raised to incorporation position in this example but left after the verb in the previous one. However, similar pairs with sati are possible both with implicit objects and with intransitive verbs, so there's not always an object present to disambiguate. What then?

The two sentences also differ prosodically. The main difference is that in the second example focus is on the whole conjunction itamu sati kipaja; itamu is the head of that conjunction, so that's the word that ends up most prominent, prosodically speaking. By contrast, in the first example focus is just on kipaja, so that's the most prominent word. (I'm not yet sure of the details, but in this context most prominent is going to translate into a higher pitch somehow, possibly a high tone on the syllable with primary stress.)

Now, not every postverbal constituent can get raised to focus position. In particular, you couldn't do that with subjects or objects that have been demoted to oblique positions by valency-changing operations. But other than that, the operation of argument raising is (as you can see) pretty general.

There's another thing: argument raising can feed passivisation: an argument that has moved to focus position by argument raising can then become the subject in a passive construction. Here is how that works:

Code: Select all

itamu kiwaita   hwati mawa      apatu (a   kipaja)
Itamu see(PASS) give  find(PFV) spear  LOC Kipaja
Itamu was given the spear (by Kipaja)

English (some varieties of English anyway) allows dative shift to feed passivisation, so this example might seem fairly familiar. Two things to note, though: raised arguments can be passivised only using kiwaita (not anatu), and the object left behind after the verb---here apatu spear---remains there in this construction.

The operation can result in an impersonal sort of passive. Here is one, starting with neutral word order:

Code: Select all

tamwipaku iwari a   ikjamii kihamii
canoe     float LOC river   in.water
The canoe is floating in the river

Here it is with the locative raised to focus position:

Code: Select all

tamwipaku ikjamii kihamii =su  iwari
canoe     river   in.water=FOC float
In the river the canoe is floating

And here is the passivised version:

Code: Select all

ikjamii kihamii  kiwaita   iwari (a   tamwipaku)
river   in.water see(PASS) float  LOC canoe
In the river is floating (a canoe)

(My translation makes the canoe indefinite in the third version; it's no longer subject, so it no longer has to be definite.)

This construction is even possible with 0-transitive weather verbs:

Code: Select all

ipajatuu niwa kamiwapi
blow     PER  night
(The wind) was blowing through the night

Code: Select all

niwa kamiwapi=su  ipajatuu
PER  night   =FOC blow
Through the night (the wind) was blowing.

Code: Select all

niwa kamiwapi kiwaita   ipajatuu
PER  night    see(PASS) blow
Through the night was blowing (the wind).

(I've included parenthetical "the wind" not because the Akiatu harbours an implicit argument but just to make the meaning clearer.)

I'll end with another example, Akiatu's first song (a bit of it anyway). It's supposed to sound sort of mournful, and to be sung at night, around a big fire.

Code: Select all

niwa kamiwapi                Through the night
kiwaita ipajatuu             was (the wind) blowing
kiwaita sawaru               were (the waves) lapping
kiwaita timarikau            were (the people) chanting
kiwaita hakjaru a itamu      was Itamu burning

[akam chinjir]

Subject control; "mwi"

This post is about mwi, a complementiser used with nonfinite subordinate clauses in same-subject constructions (I gloss it with SS for same subject). Put another way, it's about subject control.

To fix ideas, I'll start with an English sentence: Hjaci decided to go to the ocean. Superficially, "Hjaci" looks like the subject just of the matrix verb, "decided." But semantically, it actually has to be the subject of both verbs: if you understand the sentence, you know not only that Hjaci is (supposedly) making a decision, but also that the (supposed) decision is that Hjaci go to the ocean. Hjaci is being represented as both the decider and the potential goer.

Here is one reasonable (though quite shallow) way to represent what's going on:

Code: Select all

Hjaci₁ decided ____₁ to go to the ocean

The gap marks the place where you might expect to find the subject of "to go"; the subscript indicates that that subject is, or at least is coreferential with, the "Hjaci" that is the subject of the matrix clause.

Here is how this looks in Akiatu:

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

The structure is not exactly the same as in the English, but I expect it's not too mysterious. One thing: mwi doesn't occupy the same position in the structure of the sentence as does the English to; it's in CP rather than IP, which puts it before rather than after the gap where you might expect to find a subject. Another thing: in the embedded clause, wamau come, go has no perfective complement, so you might expect it to have an imperfective sense. In fact I'm inclined to think the perfective/imperfective distinction will get neutralised in nonfinite clauses.

Here are two more examples, I hope it's reasonably clear how they work:

Code: Select all

itamu₁ kamaisu mwi ____₁ pahupahu papa
Itamu  begin   SS        be.angry REDUP(INC)
Itamu was beginning to get angry

Code: Select all

itamu₁ akimau mwi ____₁ Amwakitu=su  ikjahuu
Itamu  stand  SS        Amwakitu=FOC honour
Itamu₁ has the authority ____₁ to sacrifice to Amwakitu

Some Akiatu verbs will allow subject control even though their obvious English translations don't, verbs of speech for example:

Code: Select all

hjaci₁ wañi aja      mwi ____₁ wamau a   mikuwi hatau ka
Hjaci  say  out(PFV) SS        go    LOC waters great TRANS
Hjaci₁ said that she₁ would go to the ocean

Code: Select all

hjaci₁ hwati hakwai kwai       mwi ____₁ wamau a   mikuwi hatau ka
Hjaci  CAUS  know   REDUP(PFV) SS        go    LOC waters great TRANS
Hjaci₁ said that she₁ went to the ocean

(Cross-linguistically it's pretty normally for such verbs to allow subject control, this is different from English but not weird.)

The following example raises a slightly subtle issue:

Code: Select all

itamu₁ itakitai mwi ____₁ rawu
Itamu  seem     SS        feel.good
Itamu₁ seems ____₁ to be satisfied

Plenty of text books will tell you that the English sentence here is a case of raising rather than of control. The difference is that with control, the syntactic subject of the main verb is a semantic subject of both the main verb and the verb in the subordinate clause; but with raising, it is the semantic subject only of the second verb. That is, supposedly "Itamu seems to be satisfied" does not say that Itamu does any seeming, it says only that there is seeming, and what is seemed (as it were) is that Itamu is satisfied. On this account, "Itamu seems to be satisfied" has the same meaning (the same truth conditional meaning anyway) as "It seems that Itamu is satisfied."

(Contrast an earlier example: "Hjaci decided to go to the ocean." This says both that Hjaci did some deciding and that the deciding in question was that Hjaci go to the ocean; it does not mean the same as "It was decided that Hjaci go to the ocean," or any other construction in which "Hjaci" is the subject only of the lower verb.)

Long story short: pure raising doesn't happen in Akiatu. The Akiatu sentence above does say that in some sense Itamu did some seeming (or anyway some itakitai-ing). Maybe Itamu is standing up near the elders, but off to the side a bit, by herself; it's been a good feast, and Itamu has just finished speaking; you can tell by looking at her that she feels her words were powerful, and have had an effect; she stands proudly. She seems---she gives the appearance of being, she looks---satisfied.

Notice that in English you can say "Hjaci seems to not be here," and that's not too strange (and "Hjaci seems to be away" is perfectly fine). That wouldn't work in Akiatu: *hjaci itakitai mwi miwa suwi is no good at all, because there's no way to construe Itamu's apparent absence as a characteristic specifically of her. She doesn't look absent, or give the appearance of being absent.

For what it's worth, here's a correct way to say that Itamu seems to not be here:

Code: Select all

itakitai kja  hjaci miwa suwi
seem     COMP hjaci NEG  here
It seems that Hjaci is not here

There's no raising: "itamu" stays in the subordinate clause, the subject of the main verb is, if anything, a covert expletive pronoun.

There's some interesting room to play with idioms here. For example:

Code: Select all

itamu₁ wakiru  mwi ____₁ iwari iwari parawara
Itamu  be.easy SS        float REDUP wander
Itamu₁ is easy ____₁ to swim

It's not inevitable that this should be a meaningful sentence (the English isn't, at least for me). To decide that it would be meaningful, I asked myself in what sense of wakiru easy Itamu could be wakiru in relation to Itamu swimming. This couldn't mean that swimming is easy for Itamu, because that makes swimming rather than Itamu wakiru easy. What I decided is that the sentence would mean that Itamu likes swimming.

[Xwtek]

Your question particle is too long, I think. It tends to be contracted to something like hai. But overall, it looks interesting. Not on phonology, but on grammar.

[akam chinjir]

Thanks for the comment! It's definitely part of my plan for this language to keep the phonology simple so I can concentrate on syntax and such. (I can get bogged down in phonology and morphophonology, which I seem to love but not be very good at.)

In general I want longish words so that they can collapse in different ways in different languages. hasai → sai is definitely plausible, especially given the language's habit of dropping unstressed syllables, but for now I want the full form. (hasai actually contrasts with sai, mwi sai, and na sai, and sometimes I write it as ha sai, two morphemes.)

For what it's worth, the WALS article on question-particle position mentions three bisyllabic question particles, French est-ce que, Niuean nakai, and Hunde mbéni.