Khemehekis wrote: ↑19 Jun 2022 11:49
@Kiwikami: Your last update was in November of 2021. I want to see more shiny amazingness in Alál grammar!
...I really do just pop in twice a year to dump grammar on y'all and then vanish for months on end, don't I?
Whoops. In my defense, I
should be writing a dissertation. Instead here are some words. Some of the glosses are probably wrong and much may be unclear; I'll will be editing this in pieces over the next few days. Or weeks. Or eventually.
-------------
Alál: The Poo Principle
or
59 Circumstantials and an Antipassive
A disclaimer: I used to draw a line in my notes between verbs that are "morphologically (in)transitive" - i.e. agree for one/two arguments - and "syntactically (in)transitive", i.e. are actually mono/divalent. (Likely not the best way to describe this; in my defense, I'm a phoneticist.) This is because morphologically (in)transitive verbs do have restrictions on the number of
required arguments (one or two), but technically, most verbs in Alál (the exception being non-volitive intransitives) have the potential to be maximally
trivalent, able to take one agent-, patient-, and oblique-case argument each. This is independent of their person marking, which can only agree for one or two of these. Nowadays, when I use the words "transitive" and "intransitive" in reference to Alál verbs, I specifically mean verbs whose person-marking
agrees for two and one arguments, respectively. The level of valency / total number of arguments, agreed-with or otherwise, will be referenced where it's relevant.
The
actual semantic roles of those arguments depend on:
- Whether the verb agrees for one or two arguments (i.e. is "intransitive" or "transitive")
- Whether the verb is volitional or non-volitional
- Whether the verb has an incorporated axis marker
- Whether I've had coffee this morning
- Probably the number of times a butterfly in Thailand flapped its wings last Tuesday
This is about seven years worth of shenanigans I have known but not tried to write down before. Bear with me.
(or a semanticist, or a morphologist)
---
Intransitive
Intransitive verbs are indicated by a person-marker that agrees with a single argument. Most verb roots appear in only a transitive or intransitive form by default, with exceptions; these exceptions, while generally semantically related, aren't predictably derivable (e.g. intransitive
kaıtı 'he gives off light' vs. transitive
kataı 'he watches it', or intransitive
zílı 'it (a river) flows' vs. transitive
zılà 'it (ice, etc.) caused him to slip').
The person markers are
az(v), ul(v), vù, ıv (í), and
ì, for the first exclusive, first inclusive, second, third definite, and "fourth" (third indefinite) respectively. As usual,
v represents the word's root vowel; the third definite marker has an irregular
í form if the root vowel is
ı.
Our intransitive verb example, with person agreement underlined, will be:
Sıuxı. He is afraid.
This verb is
non-volitional. As mentioned before, Alál is Split-S along lexical volition; the subjects of lexically non-volitional intransitive verbs take the patientive (P) case, while those of lexically volitional ones take the agentive (A). There are methods of changing volition on verbs, but we won't talk about that here. The person agreement here thus refers specifically to that P, which we can add overtly if we like (or not, hurrah for pro-drop):
Sıuxı kaúh. The man.P is afraid.
I like to line up how the person markers map onto the actual arguments. The marker
ıu points to our subject, which is the lone P-marked argument.
The morpheme:
ıu
Points to an argument marked as:
P
With the semantic role of:
(p)atient
Thus:
ıu
P
p
I want to make this very clear:
P here does not mean the semantic patient. It means specifically
an argument that if overt must be in the patientive case. Lower-case
p references the semantic role of patient. For now, those are the same thing; this will not always be the case. Could I have called these cases ergative-absolutive to avoid confusion? Yes probably, split-ergativity aside. But I did not, for I am a fool.
---
Intransitive + Oblique
We can add another argument here. One might imagine every verb having an unmarked, floating amorphous oblique-case argument just waiting to be referenced and thus brought into existence. This may added overtly via a noun in the oblique case without making any morphological change to the verb. By default this argument will have a benefactive meaning. For reference, argument order is usually
VOPA. Or, you know, VOS.
Sıuxı muîṭamah kaúh. The man.P is afraid
for the student.O.
Again we can line up markers to arguments, but the oblique has no morphemes on the verb referencing it at all:
ıu
P O
p b(enefactive)
Why do I consider this an argument and not an adjunct? Good question. The short answer is because it interacts with voice. We'll get there.
If the verb fulfills either of these criteria…
(1) Is a verb with an incorporated axis-of-motion marker
(2) Is marked for tense (via an axis-of-time marker)
...then the oblique argument must instead be the element referenced by that marker.
For example, take the past tense marker
lı. All axis markers,
including tense, may be expressed periphrastically as prepositional phrases, where the object (in the oblique case) is some element that the direction of motion or time is relative to:
Lı·Sıuxı ḷáa. He is afraid
before the battle.
However, these can be incorporated directly into the verb with no overt oblique argument. This is a pretty typical use of axis markers, as described in my last few posts here. In that case, the element that the direction or time is relative to is some context-sensitive reference point, generally the speaker at the time and place of speaking. For
lı, we thus we get a pretty typical past-tense reading:
Sıuxlıı kaúh. He
was afraid ("is afraid before now").
But
if you re-add the oblique argument, it takes its place as the reference point again:
Sıuxlıı ḷáa. He is afraid
before the battle.O.
Now
ḷáa isn't the oblique argument of a preposition anymore; it's the oblique argument of a verb. The mapping of person-marker to argument remains the same – it points to a subject, which here is P – but now there is also an axis marker pointing to the O. I'll use
x here to refer to any semantic role that isn't an agent, patient, or benefactive, since they all function the same here and are introduced via axis.
ıu lı
P O
p x
What if you have an incorporated axis marker (e.g. tense) and still want a benefactive? No problem! The benefactive acts like all other axis markers and has a prepositional equivalent:
xa. (Sort of. We'll get to what's up with
xa later.)
Xa·Sıuxlıı muîṭamah kaúh. The man.P was afraid
for the student.O.
And yes, you
could incorporated that
xa where you would any other axis. The following are thus equivalent.
Sıuxı muîṭamah kaúh. The man.P is afraid
for the student.O.
Sıuxxaı muîṭamah kaúh. The man.P is afraid
for the student.O.
This is also useful if you want to talk about an unspecified beneficiary:
Sıuxxaı kaúh. The man.P is afraid
for someone.
Axis markers can and do stack, in a fixed order. The
rightmost marker will point to the oblique.
Sıuxlıxaı kaúh. The man.P was afraid for someone.
Sıuxlıxaı muîṭamah kaúh. The man.P was afraid
for the student.O.
*The man.P was afraid before the student.O.
Specifying the reference points for other axis markers must be done by popping them out into their prepositional forms, using the usual multi-prepositional-phrase constructions:
Sıuxı muîṭamah lı·s láa kaúh. The man.P is afraid for the student.O
before the battle.O.
Lı·rà láa Sıuxı muîṭamah kaúh. The man.P is afraid for the student.O
before the battle.O.
So, to recap: Intransitives agree for a single argument, marked as either A or P. Adding an oblique-marked argument O usually forces a benefactive, but if there's an axis marker on the verb, the oblique argument will instead be the reference point for the motion or time described by that marker. Got it? Cool. Glad one of us does.
---
Intransitive + Volition
Recall that this verb is
non-volitional. Volitional intransitives will consider the single argument as an agent rather than a patient, and thus put it in the agentive case. We may use
Ṭıuhaı 'he shouts' as our intransitive volitional example. Same person maker, but now the argument its pointing to is both the semantic agent and in the agentive case.
Ṭıuhaı kaıh. 'The man.A shouts'
ıu
A
a
Now. The oblique here behaves exactly as expected... for incorporated axis markers.
Ṭıuhlıaı ḷáa kaıh. The man.A shouts
before the battle.O.
ıu
A O
a x
However. The specific case of the
benefactive in this situation is slightly different:
?Ṭıuhaı kası kaıh. The man.A shouts
for the child.O.
This is dispreferred. Why? Well, the short and slightly inaccurate version is that when it comes to assigning case to nouns in the context of a verb's arguments, Alál tries to fill slots in the order A>P>O. If there's already a P, the next one up is an O. But if there's only an A, it doesn't matter that the word is semantically a benefactive, it wants to fill that P slot. And so instead we get what I've called the P-benefactive, where the benefactive argument is granted patientive-case marking, and moved to the phrase-final position:
Ṭıuhaı kaıh kıús. The man.A shouts
for the child.P.
ıu
A P
a b
Phrase-final position is historically rather volition-heavy and carries a kind of implication that the beneficiary requested the event to happen. (E.g. The man shouts
on the behest of the child.) If that connotation is not desired, the benefactive can still preposition-ify as normal:
Ṭıuhaı xa·s kası kaıh. The man.A shouts
for the child.O.
But you can't get around this by using the
incorporated benefactive axis marker
xa. Or, at the very least, it sounds quite odd:
???Ṭıuhxaaı kası kaıh. The man.A shouts
for the child.O.
Why? Well... so... just table that for now. We'll get there, I promise.
---
Transitive
As with "intransitives", these
can have as many as three arguments; the verb, however, must agree in person with exactly two of them. These are a bit simpler than intransitives, at least in the active voice. There is a big chart of all the transitive person markers, which are merged; a single marker references both subject and object. For now, only worry about the marker
a/ı/u - third person subject, third person singular object. (Yes, only object plurality is marked here, we'll get into that
later.)
Our example transitive verb will be
Kataı 'He watches it'. Transitive verbs don't care one whit about volition. For your average transitive, the subject if overt shows up in the agentive case and the object in the patientive, and their semantic roles match. Thus, the structure is this:
Kataı ḷaúr kaıh. The man watches the battle.
a
A P
a p
Note that I arrange the A and P agreement this way (A-P) in the structure despite the order of their overt arguments being P-A (that is, there's verb-object-subject word order). This is because, as so many things are, of voice. We'll get there.
Just as with non-volitive intransitives, we can add that floating oblique for a default benefactive:
Kataı mlîaaru ḷaúr kaıh. The man.A watches the battle.P
for his general.O.
a
A P O
a p b
Or we can specify the referent of the rightmost axis marker, including tense:
Katlıaı hmaḳuùala ḷaúr kaıh. The man.A watches the battle.P
before the false victory.O
a lı
A P O
a p x
Great. So that's done.
---
Transitive + Inverse
Where things get... let's charitably say "interesting"... is in Alál's three distinctly-marked voices: active, inverse, and causative. The active is unmarked and is the default. The inverse is marked via a circumfix, one of
í-ıka,
í-(a)k,
vì-(a)t, or
vì-ıta, depending on the verb's lexical volition. Don't worry about that for now.
Put simply: The inverse will swap
the semantic roles of the first two arguments with "first two" determined by their positioning relative to the marking on the verb. (Why yes, this did first arise all those years ago in the Navajo Language Academy when I learned about the
yi-bi alternation, how did you guess?) That is, this structure, the standard transitive:
A P
a p
Becomes this:
A P
p a
Case marking is not swapped. The agentive-marked argument is now the semantic patient:
Kataı ḷaúr kaıh. The man.A watches the battle.P. [Active voice, volitive verb, VPA word order]
Íkatık kauh ḷaí. The battle.A is watched by the man.P. [Inverse voice, volitive verb, VPA word order]
a
A P
p a
So why have this voice? It's not a passive, it doesn't reduce valency. It does serve several purposes in discourse; first, it ascribes a bit more agency to the thing being watched, which due to the VOS word order is now in that, as mentioned above, historically volitional phrase-final position. Second, while there are a couple of relative-clause-like constructions, the most common by far uses an attributive verb.
The noun modified by such a verb must be acting as its subject. Thus, we can produce the following, using the active and inverse voices:
Hıuṭaḳa kaús katáaı. The man
who is watching him laughs.
Hıuṭaḳa kaús íkatáık. The man
who is being watched (by him) laughs.
That is, there is no direct equivalent to 'The man whom he is watching laughs'.
What is particularly important here, however, is this isn't an A-P swap; this is a
first two arguments swap. This means the inverse functions just fine in the intransitive - if you add an oblique argument onto that.
---
Intransitive + Inverse
Recall that intransitive verbs can take an oblique-marked argument, the default benefactive, without any over marking on the verb. If you apply the inverse voice to an intransitive verb, it yoinks that implied oblique into existence and swaps it with the only overtly marked argument. Thus, let's take a look at the inverse of old intransitive friend:
Sıuxı muîṭamah kaúh. The man.P is afraid for the student.O.
a
P O
p b
Uìsıuxıt kaha muîṭamàh. The student.P is feared for by the man.O.
a
P O
b p
Voila! A
circumstantial voice! This is what all this has been leading to!
✧・゚: * jazz hands * :・゚✧
What type of circumstantial, exactly, will depend entirely on whether there is an axis marker, and if so, which. Technically, since there are 57 spatio-temporal axis markers and two oddballs, there are... arguably that many circumstantials. But that just sounds silly.
For non-volitional intransitives, when there is no axis marker and the argument is the default benefactive, we have a patientive-case benefactive and an oblique-case patient/experiencer/performer of the action as seen above. When there
is an axis marker, the reference point for the axis becomes the subject (patientive case if involitive, agentive case if volitive), and the former subject becomes oblique:
Hıuṭlıḳa ḷáa kaıh. 'The man.A laughs before the battle.O'
ıu lı
A O
a x
Íhıuṭlıḳak kaha ḷaí. 'The battle.A is laughed before by the man.O'
ıu lı
A O
x a
This, in combination with attributive verbs, allows such lovely constructions as:
ḷáa íhıuṭúlıḳak the battle before which there was laughter
And with the highly-productive nominalization of verbs, we can get quite creative:
íhıuṭlıraḳak that which was predicated by laughter
íhıuṭùḳ·ruuḷak the machine which stops laughter (through the use of which one stops laughing)
uìhıuṭut·saḳıta the place in which one laughs uncontrollably
...and so on
Now, remember, this inversion applies strictly to
the first two arguments; we can get interesting results in verbs with multiple incorporated axes, which… alright, look, in my notes I call it the Poo Principle, I am a grown adult, that's just how it is. Have some motion verbs to illustrate. (Don't worry about the vowels, they're doing their own thing):
Ḷîḳa. He falls.
Ḷîḳtıà. He falls
backwards (relative to his starting point).
Ḷîḳtıà lakı. He falls
back to/behind the tree.O.
Ǐḷıḳtìàt lıúk. The
tree.P is fallen
behind.
Now add in an incorporated benefactive. Remember, only the
rightmost O can ever actually be specified by an overt oblique-case argument:
Ḷîḳtıxaaı. He falls backwards, for something.
í tì xa
P O O
p o b
The inverse swaps the first two, effectively popping the semantic patient out of existence.
The second O, here a benefactive, is left alone and stays in the oblique case.
Ǐḷıḳtìxaàk. It is fallen behind, for something.
í tì xa
P O O
o p b
This means, even if you were to have two overt arguments, the semantic patient would never appear; this is the Poo Principle:
Ǐḷıḳtìxaàk ḳamàa lıúk. The tree.P is fallen behind, for science.O.
í tì xa
P O O
o p b
And again we can look at attributive and nominalized forms of this for fun:
muîṭamah uìsıuxúlıxaıt the student who was feared for
muîṭamah uìsıuxúlıxaıt kaha the student who was feared for, by the man
An oblique noun providing a new reference point for incorporated axes works just fine in attributive verbs – not so much in nominalized ones. Nominalized verbs can take no arguments.
uìsıuxmaıt the feared-for one
*
uìsıuxmaıt kaha the feared-for (by the man) one
---
Circumstantial Transitives
The next question, then, is whether it is possible to promote an oblique argument to subject position in a
transitive verb. That'd be useful, wouldn't it, given how important this oblique-promotion is for those relative clause constructions. But the inverse clearly cannot do this alone, since the existence of both an A and a P argument necessitates that those be swapped. Swapping A and O directly is a plain and simple impossibility, according to the Poo Principle.
Alright. So.
You may recall that volitional intransitives (i.e. those whose single argument is A) really don't like having a default benefactive in the oblique. In fact, they aren't fans of anything involving
xa, except when explicitly used as a preposition. Remember:
Ṭıuhaı xa·s kası kaıh. The man.A shouts for the child.O.
...but...
???Ṭıuhxaaı kası kaıh. The man.A shouts for the child.O.
(The inverse voice works just fine on the P-benefactive, for the record, wonky word order and all:)
Ṭıuhaı kaıh kıús. The man.A shouts for the child.P.
ıu
A P
a b
Íṭıuhık kıs kaúh. The child.A is shouted for by the man.P.
ıu
A P
b a
The reason that an intransitive verb whose only argument must be an A-marked semantic agent doesn't like
xa is because I lied.
Xa doesn't actually point to the benefactive. It points to the
next semantic role in sequence.
Earlier on, in explanation for why the P-benefactive happens, I said this: "The short and slightly inaccurate version is that when it comes to assigning case to nouns in the context of a verb's arguments, Alál tries to fill slots in the order A>P>O. If there's already a P, the next one up is an O. But if there's only an A, it doesn't matter that the word is semantically a benefactive, it wants to fill that P slot."
It's time for the longer and more accurate version: Alál
does like the order A>P>O when it comes to case marking, and that explanation
was true - historically. But entirely separately from case, when it comes to semantic roles, it likes the order
causer>agent>patient>oblique (well, benefactive>other.oblique). I call this
capo. The "default oblique", or really anything introduced by
xa, is in fact just
the next not-yet-marked role in that sequence. The P-benefactive is actually a separate, historical benefactive construction based on that phrase-final volition-heavy position, which is used for volitive intransitive verbs because the usual default or
xa method of introducting a benefactive there...
would actually introduce a semantic patient.
Non-volitive intransitive, marked argument is
p, default oblique adds next in
capo order:
o (b).
Sıuxı muîṭamah kaúh. The man.P is afraid
for the student.O.
ıu
P O
p b
Volitive intransitive, marked argument is
a, default oblique adds next in
capo order:
p.
*
Hıuṭlıḳa muîṭamah kaıh. *The man.A laughs
the student.O.
ıu
A O
a p
This doesn't make sense for morphologically intransitive volitive verbs. They've only got one argument, it's not a patient, that
p has no business being there! There are other ways of going about reflexives and reciprocals if that's what you're aiming for! Thus, this does not happen. If you want a benefactive in a volitive intransitive verb, you have to make use of that old historical oddity - that the "reason" or "purpose" or "beneficiary" or "instigating force" of an event can appear in patientive case in the phrase-final position.
Recall, though, that most verb stems are either intransitive or transitive, and it is very rare to have one appear as both. (Even when they do, they don't normally share volition, aspect, etc.) So, and hear me out: What if you took a transitive verb, but gave it intransitive person-marking, popping out the P argument and reintroducing it as that
xa-marked
p oblique?
Well that sounds like a very strange way of going about an...
...
Antipassive Voice? Apparently?
In simple terms, the antipassive is formed by incorporation of the axis marker
xa into a transitive verb stem conjugated as if intransitive. This demotes the object to an oblique. It occurs for exactly two reasons: as a means of indicating general (transitive) actions an entity takes regardless of target (I yearn. For what? I don't know, I just yearn.) and in conjunction with inversion to allow for transitive circumstantial constructions.
Step one: Take a transitive verb, regardless of volition.
Step two: Conjugate it as if it was (a volitional) intransitive (subject = agent).
Step three: Reintroduce the removed patient using
xa, since
p is next in
capo order now.
Step four: Add whatever axes you like.
Step five: Invert. If you want.
There is one other small note which is that
when this construction is used, incorporated
xa becomes
xà. This isn't so much a means of dealing with ambiguity so much as a quirk that comes from interactions with the conclusive aspect
à, which shows up a lot in these kinds of verbs, but it's a useful marker to tell at a glance when this antipassive-ish thing is happening.
To illustrate, we'll do this with
Kataı 'he watches it'. While
KAT is one of the verb roots that does also have a normal intransitive form, that form is explicitly non-volitive, so there's no ambiguity.
a
A P
a p
The intransitive third person here would be
*Kaıtaı. This on its own is ungrammatical.
aı
A
a
Kaıtxàaı. He watches [things]. Adding
xa re-adds the thing(s) that is/are watched, floating amorphously about. If we wanted, we could add an overt oblique argument specify what it is. This is does have some widespread use, as a means of avoiding specifying anything - including definiteness and plurality, both of which the person-agreement otherwise indicates - about the object of an action. "It eats some [unspecified] thing" is
Rıẓàx, but I would translate the title of the second Welcome to Night Vale novel,
It Devours, as
Rîẓxaax.
aı xà
A O
a p
If we were to invert this now, we'd get
Íkaıtxàık. It/he is watched [by unspecified things]. This is grammatically technically fine, if rare, and does the same thing as above - allowing one to specify that a thing has something done to it while eliminating all information including definiteness and number from the agent. Often used to describe actions that happened spontaneously or on their own, or in cases where the (agentive-case-marked) semantic patient had some influence on their fate.
Íkxıùtàk - he was killed, vs.
Íkxıutxàak - he got himself killed.
aı xà
A O
p a
But now, let's add another axis marker. Recall that these can stack, and in a fixed order (ZCPXYRExz, for the record - that's
xa there second-to-last). Now with the addition of
lı we already have an
o, but
xa still adds
p because that slot still isn't filled. The rightmost points to the overt oblique argument, if there is one - but the
leftmost is what will swap with
a when inverted.
Kaıtlıxàaı. He watch
ed [things].
aı lı xà
A O O
a o p
And... invert.
Íkaıtlıxàık. It/he has something watching something else before it.
aı lı xà
A O O
o a p
Let's add some overt arguments to clear that up:
Katlıaı mamu zuaz kaıh. The man.A watches the sea.P before the storm.O (started).
Íkaıtlıxàık záz mıum. The storm.A had the sea.O being watched before it (started).
Sutràax makàr ḳîuúl. He(.A) buried his horse.P near the grave.O.
Ísutràak makàr ḳîuıul. His horse.A was buried near the grave.O.
Ísıutràxàak ḳîualu mkíra. The grave.A has his horse.O buried near it.
So, doesn't this mean there's no way to reintroduce the agent? Once it swapped with that first oblique, the transitive verb's original subject was lost to the middle O of the POO, never to return. Ah, but you see, semantic agents, like secret agents, tend to appear where you least expect them. And rather like corn, they have a remarkable tendency to come through poo relatively unscathed.
(I am a grown adult, I swear it.)
After all, the agent is a pretty volition-heavy position. And we have a perfectly well-historically-established construction to deal with that.
So, behold, the P-agent:
Íkaıtlıxàık záz mıum kaúh. The storm.A had the sea.O being watched before it (started) by the man.P.
Sıutràxàax makàr ḳîuúl htumàl. The grave.A has his horse.O buried under it by the mortician.P.
Welcome to the Möbius Frankenpassive, where the agent if overt is in the patientive case, the patient in the oblique, and the oblique in the agentive. I love and hate it in equal measure, and it's probably not how natural language works. Uncertain. See comment from Bones above. No matter - it's fun and kinda neat.
Have some nice phrases using that relative clause construction:
Suthàax lakı amìzzaí. - My friend.A buried him under the tree.
Ísuthàak lakı amìzzaí. - My friend.A was buried by him under the tree.
lakı ísıutûhaxàak amǐzzaa - the tree under which my friend.O was buried
lakı ísıutûhaxàak amìzzaúr - the tree under which my friend.P buried (something)
And some nominalized verbs:
íríẓurxà·haıàk - the fog in which things are devoured
ílaíkáraxà·ruàk - the contraption from which people are hanged
ìrıumutxà·saàt - the place where things are (naturally, over time) destroyed
íhmaılahxà·maàk - the person who thinks (in whose eyes) things are conquered
---
Wait a second, said no one. If
xa is being used to mean
p, we can introduce a benefactive using a prepositional phrase, but how do we
promote a benefactive to subject position in a transitive verb?
I don't think you can. I don't really want to allow stacking multiple instances of
xa. Unless I figure out a solution, transitive benefactives are just a little defective, and don't have a circumstantial form. Or. Well. There's this thing with
za introducing a kind of benefactive outside of causative constructions, but that...
Sut(xa)àx akìzlaṭa ahùḳa amìzzaí. - My
friend buried the
bird for my
sibling.
Sutzaàx akìzlaṭa ahùḳa amìzzaí. - My friend buried the bird, as instructed by my sibling.
Ísut(xa)àk akìzlaṭa amìzzaúr híka. - The bird was buried for my sibling by my friend.
Ísıutzaxàak hàk akìzlaıṭ amìzzaúr. - My sibling had the bird buried for them by my friend.
ıu za xà
A C O
b a p
akìzlaṭa ísıutúzaxàak - my sibling for whom something was buried
ísıutúzaxà·maàk - the person for whom something was buried
Oh, that works. That works nicely. I like that. We'll unpack that later. When we deal with the...
---
Causative
Nah, I'm going to go make coffee.
See you all in either two days or six months, there is no in-between!