a-Uttes: Syntax

[Ares Land]

The a-Uttes language

The a-Uttes language is spoken in the science-fictional setting I talked about a few months earlier.

I've tried to come up with something a little alien and focus a little more on syntax.
I'd be very happy for feedback; especially if you think the alignment I propose below makes any sense, or if you think it's attested somewhere and described in a better way...

Intro

It is the commonest language in Uttes (formerly, Yttes. Using <y> was annoying), a disk-shaped orbital megastructure, with a population of about 1 trillion ða-Uttes.

The ða-Uttes are human, though genetically a little different. Their ancestry include several species in the genus Homo, including sapiens and a few other close relatives.

There is a very wide variety of languages spoken in Uttes. Many of these date back to the original colonization of the Uttes system; others were introduced as new civilizations were contacted. Languages introduced in Uttes tend to conform, over centuries, to a kind of generic, pan-Uttes phonology and syntax.

The a-Uttes language itself is the commonest one, and serves as an official language. Its ancestry and family relationship are muddled. It ultimately dates back to a language spoken in the early days of spaceflight, but it has freely borrowed lexicon and constructions ever since. It exists in a state of dynamic equilibrium with the other languages of Uttes (not unlike that which posited for Australian languages), with the added complication that earlier varieties of the languages also contribute. Lexicon and grammar constructions are freely reborrowed from all stages of the languages spoken in the last two millenia.
(That's what having translation apps, universal literacy, video and recording for two millenia does to you.)

Brief notes on phonology

This section mostly aims at providing a pronunciation guide rather than an in-depth analysis.
The phoneme inventory of a-Uttes is follows:


Consonants

	Labials	Alveolar	Post-alveolar	Palatals	Velars	Labio-velar	Glotta
Plosives	p	t			k	kw /kʷ/
Sibilant		s	š /ʃ/
Non-sibilant	v /ʋ/	ð /ð͇/		y /j/		w /w/	h /ɦ/
Nasals	m	n
Lateral		l

Vowels

	Front	Central	Back
High	i	u [ɵ ~ ʊ]	o [o ~ u]
Mid	e
Low		a

Diphtongs
au, eo, ai

The consonant inventory is fairly restricted. A few difficulties: h is always voiced (more accurately, perhaps, breathy voiced?). ð is a bit difficult. Start with the English dental [ð] and move the tip of the tongue to the alveolar ridge. It may be tapped.

As for the vowels: u is approximately [ʊ] but slightly more fronted. For o, the allophones [o ~ u] are more or less in free variation in open syllables.

Stress falls regularly on the penultimate.
Consonant length is contrastive between vowels, and also word-finally or before a consonant:
ðennsu [ðen:.su] 'animal'

Vowels are lenghtened in stressed, open syllables: yitis [ji:tis] 'vine'

Case markers, animacy, number and case

Noun phrases are followed by a case marker, encoding information about animacy, case, and number.

Animate	Singular	Plural
Transitive	wo	weo
Accusative	in	ine
Unmarked	∅	e
Inanimate	Singular	Plural
Transitive	to	twe
Accusative	ten	twen
Unmarked	∅	we

Animacy
Animacy is semantic. Basically, human beings, animals and anything having a will of its own are animate. Anything else is inanimate.
Living things are generally animate (including plants and microorganism), but the exceptions are sometimes tricky.
Dangerous things: fire, vaccum, radiation, weapons are animate.

Inanimate markers can be used with human beings. This is either endearing or as deeply insulting.
Animate markers are used with inanimates when their behavior is erratic, sudden or unexpected.

The ða-Uttes don't anthropomorphise computers or AI much: if they work as expected, they take inanimate markers.

Number
The plural is used for a definite, but unspecified number. Generic terms, or plural entities considered as a set are singular.
ða 'a person, people in general, human beings'
ða e 'several people, some people'
If there's a number or a quantifier, the singular is used instead:

mastamm we 'several robots' > manna mastamm 'many robots'
ða e 'several people' > haa ða 'two people'

Morphosyntactic alignment and case usage

OK, now we can get to the fun part.

The transitive case
The transitive case is used for the agent and patient of a transitive verb:

ða-Uttes wo pono a-senši to unne.

ða-Uttes wo pono a-senši twe unne

person-Uttes TRANS meat of-culture TRANS eat

'The ða-Uttes eat cultured meat.'

An animate NP in the transitive case is always the agent; an inanimate NP in the transitive case is always the patient. In effect, the transitive case tells us that NPs behave in their default role.

Unmarked NPs
Animate patient, animate agent.
What happens, though, if an animate acts on another animate? In that case, the animate patient is unmarked:

Ðeenos wo ðennsu unne.

ðeenos wo ðennsu unne

cat TRANS animal eat

'Cats eat animals.'

Inanimate agent
What if an inanimate acts on an animate, or for that matter on another inanimate?
The inanimate agent can be left unmarked:

Mastamm ðeenos ennisos

mastamm ðeenos ennisos

robot cat chase.away

Or introduced with a postposition:
Unla wu ðanne to šimon

unla wu ðanne šimon

diamond by glass TRANS work

'Diamond cuts glass'

The patient is unmarked (if animate) or marked with the transitive (if inanimate)

Instrumentals

The above construction can be turned into an instrumental, simply by adding an animate NP in the transitive case:
diamond by glass TRANS work --> 1s TRANS diamond by glass TRANS work
Unla wu ðanne to šimon --> Ni wo unla wu ðanne šimon.
'I cut glass with diamond.'

robot cat chase.away --> young TRANS robot cat chase.away
Mastamm ðeenos ennisos. --> Pwenn wo mastamm ðeenos ennisos.
'The teenager had the cat chased away by the robot.'

Inanimates are never conceived of as agents -- in Mastamm ðeenos ennisos., the robot isn't conceived of as acting of its own: it is only an instrument for an (unstated) animate agent.
Getting back at our notes on animacy, this explains why, if the robot chases the cat of its own, either as an unexpected consequence of its programming or because it's malfunctioning, a transitive marker is used:

Mastamm wo ðeenos ennisos.

mastamm wo ðeenos ennisos

robot TRANS.ANIM cat chase.away

'The malfunctioning robot chased the cat away.' ~ 'The damn robot chased the cat away.'

Verbs of perception


With verbs of perception, the observer is unmarked. An observed object is in the transitive case; an observed animate is unmarked as well.

Ni nen leða.

ni nen leða

I you hear

'I hear you.'

Ðeenos mastamm to šini

ðeenos mastamm to šini

cat robot TRANS see

'The cat sees the robot.'

You can introduce a transitive animate argument. This has a causative meaning:

Ni wo nen mastamm to šini

I TRANS you robot TRANS see

'I show you the robot.'

With two animates, the relationship is symmetrical: Ni nen leða. 'I hear you' implies 'you hear me'.

The accusative

The accusative case is used for patients when the patient changes as a result of this action. This usually translates as a perfective:

Ni wo ðanne ten šimon.

ni wo ðanne ten šimon

1s TRANS glass ACC cut

'I cut, I have cut the class'

Mastamm ðeenos in ennisos

mastamm ðeenos in ennisos

robot cat ACC chase.away

'The robot chased the cat away.'

Ðeenos wo ðennsu in unne.

ðeenos wo ðennsu in unne

cat TRANS animal ACC eat

'The cat ate the animal.'

As the last example shows, the accusative case tends to imply a specific object. The accusative is never used for habitual or general meanings; or, in other words, it also marks telicity:

Ilðen wo nahða to un.

Ilðen wo nahða to un

Ilðen TRANS ship TRANS build

'Ilðen is building a ship / Ilðen builds ships'

Ilðen wo nahða ten un.
'Ilðen built the ship.'

Dative constructions

The indirect object, or beneficiary is unmarked; the direct object is either in the transitive or accusative case:

Ilðen wo ni e nahða to un.

Ilðen wo nahða to un

Ilðen TRANS 1p PL ship TRANS build

'Ilðen is building us a ship'

Ilðen wo ni e nahða to un.

Ilðen wo nahða to un

Ilðen TRANS 1p PL ship ACC build

'Ilðen is building us a ship'

Ilðen wo ni ðennsu in ve

Ilðen wo ni ðennsu in ve

Ilðen TRANS me animal ACC give

'Ilðen gave me a cat.'

[bradrn]

Ooh, excellent! I’ve been waiting for more information on Yttes Uttes now, sorry. But let that not stop me from giving my usual critique:

There is a very wide variety of languages spoken in Uttes. Many of these date back to the original colonization of the Uttes system; others were introduced as new civilizations were contacted. Languages introduced in Uttes tend to conform, over centuries, to a kind of generic, pan-Uttes phonology and syntax.

Mind giving some more information about this Sprachbund?

[a-Uttes] exists in a state of dynamic equilibrium with the other languages of Uttes (not unlike that which posited for Australian languages)

I’ve read a bit about Australian languages, and I don’t recall seeing anything about this; where did you see it?

Lexicon and grammar constructions are freely reborrowed from all stages of the languages spoken in the last two millenia.
(That's what having translation apps, universal literacy, video and recording for two millenia does to you.)

I particularly like this detail! This, everyone, is what it looks like when someone has properly thought about the relationship between language and culture.

Morphosyntactic alignment and case usage

Hmm, this seems reasonable enough to me, except that it’s completely backwards.

Let me elaborate. W.r.t. alignment, an NP is functionally unmarked if it is in the grammatical relation appropriate for its semantics. The unmarked situation for animate nouns is as agents; the unmarked situation for inanimate nouns is as patients. (i.e. animates act on and affect inanimates.) Which agrees with what you say (my emphasis):

An animate NP in the transitive case is always the agent; an inanimate NP in the transitive case is always the patient. In effect, the transitive case tells us that NPs behave in their default role.

But wait — now you’ve got NPs which are formally marked in the functionally unmarked role, and formally unmarked in the functionally marked role. This is certainly known, but usually it’s rare and not nearly as pervasive. (The example which comes to mind is Arrernte, in which 1s is ergative and everything else is accusative, but even there, ‘wrong’ marking is restricted to just one pronoun.) I suggest switching this around: mark inanimate agents and animate patients with the transitive case, and keep the ‘default role’ unmarked.

Another issue: I don’t like the analysis where wo~weo and to~twe are both unified into a single ‘transitive’ case. Looking at their distribution: wo is used for animate nouns in A position, and to is used for inanimate nouns in O position. That is, animate nouns are unmarked in S and O and take wo when A, and inanimate nouns are unmarked in S and A and take to when O. I suggest that this is simply an animacy-based split — animate nouns take ergative alignment with ergative marker wo, and inanimate nouns take accusative alignment with accusative marker to. (Which, as I already said, is the wrong way around, though let’s ignore that for now.) I really see no reason to analyse this as one case-marker.

Lest I sound too harsh, one thing I really do like about this alignment is the special ‘affected accusative’ marker, which sounds exactly like the sort of thing a natlang would do. (In fact, I’m pretty sure it is indeed attested somewhere, though where I couldn’t say for sure.)

[Ares Land]

Ooh, excellent! I’ve been waiting for more information on Yttes Uttes now, sorry. But let that not stop me from giving my usual critique:

Thanks!

Mind giving some more information about this Sprachbund?

Sure. As soon as I figure out what exactly I meant by that
More seriously, though, you'd probably expect a small-ish consonant inventory, case markers and a similarly ass-backwards alignment system.

[a-Uttes] exists in a state of dynamic equilibrium with the other languages of Uttes (not unlike that which posited for Australian languages)

I’ve read a bit about Australian languages, and I don’t recall seeing anything about this; where did you see it?[/quote]
I meant punctuated equilibrium, sorry! as postulated by Dixon. Oversimplifying things, languages will not fit the traditional tree model.

I particularly like this detail! This, everyone, is what it looks like when someone has properly thought about the relationship between language and culture.

Heh, thanks.

Hmm, this seems reasonable enough to me, except that it’s completely backwards.

(...)
This is certainly known, but usually it’s rare and not nearly as pervasive.

Yes, the language is supposed to feel sort of alien, and I'll try to break a few universals if I can get away with it.

I justify this diachronically: they derive from auxiliaries or co-verbs. Synchronically, TAM markers still attach to these. (The verb itself doesn't do much except sit there. But more on that later.)

Another issue: I don’t like the analysis where wo~weo and to~twe are both unified into a single ‘transitive’ case. Looking at their distribution: wo is used for animate nouns in A position, and to is used for inanimate nouns in O position. That is, animate nouns are unmarked in S and O and take wo when A, and inanimate nouns are unmarked in S and A and take to when O. I suggest that this is simply an animacy-based split — animate nouns take ergative alignment with ergative marker wo, and inanimate nouns take accusative alignment with accusative marker to. (Which, as I already said, is the wrong way around, though let’s ignore that for now.) I really see no reason to analyse this as one case-marker.

Tell you what, calling that case transitive was completely wrong, because I'm pretty sure they are used with intransitive verbs. *slams forehead*

The distribution is:
Animate: S => wo, A => wo, O=unmarked
Inanimate: S => to, A => unmarked, O=to

(Not in all cases, though. That depends on the semantics of the verb.)
But more on that later. I need to figure it out more precisely.

Lest I sound too harsh, one thing I really do like about this alignment is the special ‘affected accusative’ marker, which sounds exactly like the sort of thing a natlang would do. (In fact, I’m pretty sure it is indeed attested somewhere, though where I couldn’t say for sure.)

No problem at all: it's exactly the kind of feedback I'm after!

The accusative marker is inspired both by the Finnish partitive/accusative distinction and the Mandarin bǎ construction.

[Vardelm]

Morphosyntactic alignment and case usage

I would swear I've seen another conlang posted here in the past year or so that had a vaguely similar alignment, at least in that what case was used was all over the place based on whether the agent & patient were animate or not. Maybe that was another of yours, or perhaps Bradrn?

Either way, the alignment seems like a weird duck, but at the same time it feels pretty logical. I couldn't necessarily predict which case was going to be used as I read it, but when you explained it it made sense. The tidbit about adding causativity was a particularly nice tough. I give it 4.5 thumbs-up out of 5!

[bradrn]

Hmm, this seems reasonable enough to me, except that it’s completely backwards.

(...)
This is certainly known, but usually it’s rare and not nearly as pervasive.

Yes, the language is supposed to feel sort of alien, and I'll try to break a few universals if I can get away with it.

Sure, but I’m feeling pretty confident that even non-human languages wouldn’t break this particular universal (as long as they still use S/A/O), since it’s based on more general principles of markedness rather than being just some arbitrary fact.

I justify this diachronically: they derive from auxiliaries or co-verbs. Synchronically, TAM markers still attach to these. (The verb itself doesn't do much except sit there. But more on that later.)

How do the diachronics work?

Another issue: I don’t like the analysis where wo~weo and to~twe are both unified into a single ‘transitive’ case. Looking at their distribution: wo is used for animate nouns in A position, and to is used for inanimate nouns in O position. That is, animate nouns are unmarked in S and O and take wo when A, and inanimate nouns are unmarked in S and A and take to when O. I suggest that this is simply an animacy-based split — animate nouns take ergative alignment with ergative marker wo, and inanimate nouns take accusative alignment with accusative marker to. (Which, as I already said, is the wrong way around, though let’s ignore that for now.) I really see no reason to analyse this as one case-marker.

Tell you what, calling that case transitive was completely wrong, because I'm pretty sure they are used with intransitive verbs. *slams forehead*

The distribution is:
Animate: S => wo, A => wo, O=unmarked
Inanimate: S => to, A => unmarked, O=to

(Not in all cases, though. That depends on the semantics of the verb.)
But more on that later. I need to figure it out more precisely.

Ah, now this makes more sense! In this case, animate nouns take accusative alignment and inanimate nouns take ergative alignment, which is the right way round.

(Well, strictly speaking, it’s marked-nominative and marked-absolutive, which generally pattern opposite to ‘normal’ accusative and ergative alignments, so this possibly is still wrong, but I know little enough about marked-S that I’m willing to buy this pattern as something possible.)

Morphosyntactic alignment and case usage

I would swear I've seen another conlang posted here in the past year or so that had a vaguely similar alignment, at least in that what case was used was all over the place based on whether the agent & patient were animate or not. Maybe that was another of yours, or perhaps Bradrn?

Laqar, perhaps?

[Vardelm]

Ah, now this makes more sense! In this case, animate nouns take accusative alignment and inanimate nouns take ergative alignment, which is the right way round.

(Well, strictly speaking, it’s marked-nominative and marked-absolutive, which generally pattern opposite to ‘normal’ accusative and ergative alignments, so this possibly is still wrong, but I know little enough about marked-S that I’m willing to buy this pattern as something possible.)

/agree

Split-ergativity regarding animate vs inanimate, with both marked accusative & ergative. Groovy.

Laqar, perhaps?

Yeah, I think that was it. Similar-ish deal, but using volition instead of animation.

[Ares Land]

How do the diachronics work?

A very good question; and it was a great opportunity to actually write the stuff down!

Let's go back to pre-a-Uttes (or an approximation thereof: I haven't worked out the language beyond what was strictly necessary.)
The folloowing was a common construction in PAU:

[gloss="human do eat-NMLZ']taa hā umnit-a[/quote]
In effect, 'humas do the eating' instead of 'humans eats.'

umnit-a here is really the object of hā, 'do'. Compare 'do the dishes', 'do the cleaning' in English...

This construction came to replace most of the synthetic verb forms. The problem is, how do you handle transitives?

One strategy was to treat it as an indirect object, with a postposition:

(1) [gloss="human do meat out.of eat-NMLZ']taa hā pānuʔa tā umnit-a[/gloss]
'Humans eat meat'; tā being here a partitive marker.

Another is to use a coverb:
(2)

taa hā pānuʔa ma umnit-a.

human do meat do.ABS eat-NMLZ

And yet another was to use a topic-comment construction:
(3)

pānuʔa taa hā umnit-a

meat human do eat-NMLZ

Finally, one last trick was to use a possessive:
(4)

taa hā umnit-a-a pānuʔa

human do eat-NMLZ-POSS meat

(1) works very well for inanimates. But it was very marked semantically when dealing with animates, for instance:
[gloss="human do meat out.of eat-NMLZ']taa hā tanʔatus tā umnit-a[/gloss]
'Humans are eating out of animals.'
(2) works with both animates and inanimate (except you want in, not ma with animates) but has implications in terms of telicity, aspect and definiteness.

So that left (3) & (4) as a default strategy.
Let's look at (4) now. The possessive suffix was often reinterpreted as a genitive initial clitic, with a requirement that the genitive always follows its head noun.
So the construction was something like:

taa hā umnit a-tanʔatus

human do eat-NMLZ GEN-animal

'Humans eat animals'

PAU often fronted constituent at this stage, with sentences often following the order Topic - Focus - S V
Except:
(**)

nipantis a-tanʔatus taa hā umnit

planet GEN-animal human do eat-NMLZ

'On planets, it is animals that human eat' is ambiguous; it's ambiguous whether in nipantis a-tanʔatus umnit, the head of a-tanʔatus 'animal's' is nipantis, 'planet' or umnit 'eat'. Patients ended up losing the prefix, on the model of sentences like (4).

As for inanimate subjects...
Typically, inanimate are (and were) unlikely to be agents.
Typical constructions involving inanimates were:

(6)

tamānaʔata timaʔ-ʔūn

glass cut-ABS

'Glass is cut'.
or
(7)

tamānaʔata tā timaʔ-ʔūn

glass out.of cut-ABS

'Someglass is cut'.

In a sentence such as 'diamond cuts glass', diamond is S in English; but not in a-Uttes or PAU. It's either an instrument, topic or focus:

(7)

tamānaʔata timaʔ-ʔūn ʔūnʔulaʔa hū

glass cut-ABS diamond by

or
(8)

ʔūnʔulaʔa tamānaʔata timaʔ-ʔūn

diamond / glass cut-ABS

Absolutive agreement markers such as ʔūn were either lost or interpreted as part of the verb.

Sure, but I’m feeling pretty confident that even non-human languages wouldn’t break this particular universal (as long as they still use S/A/O), since it’s based on more general principles of markedness rather than being just some arbitrary fact.

In English (for instance), most if not all utterances share the same pattern as actions. A-Uttes mostly doesn't.
S/A/O is restricted to cases where something physical and voluntary occurs.
I know, that's very unclear... I'm still trying to figure out the implications!

The default state, so to speak, of a noun isn't that of an argument, but that of an adjunct.

[bradrn]

This construction came to replace most of the synthetic verb forms. The problem is, how do you handle transitives?

One strategy was to treat it as an indirect object, with a postposition:

(1) [gloss="human do meat out.of eat-NMLZ']taa hā pānuʔa tā umnit-a[/gloss]
'Humans eat meat'; tā being here a partitive marker.

Another is to use a coverb:
(2)

taa hā pānuʔa ma umnit-a.

human do meat do.ABS eat-NMLZ

And yet another was to use a topic-comment construction:
(3)

pānuʔa taa hā umnit-a

meat human do eat-NMLZ

Finally, one last trick was to use a possessive:
(4)

taa hā umnit-a-a pānuʔa

human do eat-NMLZ-POSS meat

I can’t say I’m following this whole post especially well, but as it happens this is one area I know a bit about. (I have a fondness for languages with closed verb classes.) From what I can see, most languages simply have a special transitive construction with an extra slot for the adjunct, e.g. Kalam (Pawley 2011:47):

Meg

tooth

tug

by.hand

jupin

remove:PF:1s

I pulled out a tooth.

This is somewhat close to your (1), but without the adposition; the verbal adjunct tug occurs with both a subject and an object. (And we know it’s a verbal adjunct and not some sort of object because Kalam has only one ditransitive verb, and ju- isn’t it.) In Kalam, adjuncts are a separate word class, but it’s entirely possible to merge them with nominals; e.g. verb adjuncts in Komnzo are a subclass of the property nouns (Döhler 2018:87):

Efothf

sun:ERG

bad

ground

twof

heat

wäfiyokwr

2|3sg:sbj>3sg.fem:obj:npst:ipfv/make

The sun makes the ground hot.

From my limited understanding, it seems that human languages are more likely to prefer your (1) when they have a large open verb class, and are more likely to prefer a special construction when they have a small or closed verb class and make heavy use of such adjuncts.

[Ares Land]

I can’t say I’m following this whole post especially well

Hmm. Probably a good sign much of the case analysis needs to be reworked and redone.

From my limited understanding, it seems that human languages are more likely to prefer your (1) when they have a large open verb class, and are more likely to prefer a special construction when they have a small or closed verb class and make heavy use of such adjuncts.

In any case, I think of that particular development in the ancestor language as having nothing exceptional about it. Your posts makes me think I'd probably should go take a look at languages with closed verb classes.

[bradrn]

I can’t say I’m following this whole post especially well

Hmm. Probably a good sign much of the case analysis needs to be reworked and redone.

No, it’s more likely a sign that I have several hours of unwatched lectures I need to catch up on, and the associated worrying is making it difficult for me to concentrate too well on anything right now…

(I hate online learning.)

From my limited understanding, it seems that human languages are more likely to prefer your (1) when they have a large open verb class, and are more likely to prefer a special construction when they have a small or closed verb class and make heavy use of such adjuncts.

In any case, I think of that particular development in the ancestor language as having nothing exceptional about it. Your posts makes me think I'd probably should go take a look at languages with closed verb classes.

Yes, do — they’re very interesting! Some good ones to look at: the Papuan languages Kalam and Komnzo, the Australian languages Jingulu and Ngan’gityemerri (depending on how you define ‘verb’), and the west African languages Eʋegbe, Fongbe and Kinauri Kanuri (aargh, confused them again). Pawley (2006) is also a good introduction to closed verb classes. You may also be interested in Wutung, a language having only 32 verbs which interestingly constitute a clearly open class.

[Civil War Bugle]

It is the commonest language in Uttes (formerly, Yttes. Using <y> was annoying),

Nooo, I rather liked the spelling with a y!

I have no profound commentary right now, but I have been following updates on these guys whenever I notice a new thread about them, so keep it up.

[Rounin Ryuuji]

Just given this a read-through; it's very interesting, but I, too, have no very long feedback to offer.

[Ares Land]

Thanks everyone for the interest!

I'm reworking this quite a bit -- in many ways, because of your feedback, so thanks!


In The Case for Case, Fillmore suggests the following semantic roles are near-human universal: (the list itself is from Fillmore, and I did read The Case for Case, but you needn't really bother. Zompist's treatment in the SCL is far more readable and gets right at the good bits.)

Agentive(A), typically animate perceived instigator of theaction identified by the verb.
Instrumental (I), inanimate force or object causallyinvolved in the action or state identified by the verb.
Dative(D),animate being affected by the state or action identified by the verb
Factitive(F), object or being resulting from the action orstate identified by the verb, or understood as a part of the meaning of the verb
Locative(L), location or spatial orientationof the state or action identified by the verb
Objective(O), semantically most neutral, [the object affected]

That is, any NP in a sentence in any given human language can be assigned one of these roles (there are a few others, and others have contributed to the list, but whatever) which then map to case according to morphosyntactic alignment, voice, and so on.

a-Uttes doesn't quite follow the same semantic mapping. In the languages of Uttes, NPs tend to fit a subtly different set of roles:

• State -- an NP being assigned a property.
• Process -- an NP involved in a specific process
• Result -- an NP assigned a property as a result of a process.
• Frame of reference: -- the frame of reference according to which the statement is valid. (This is often, but not always similar to a topic)
• Force -- the inanimate direct cause of the state, result or process.
• Origin, position, direction, manner -- exactly what it says on the tin!

Now, let's run these through a list of examples. (Credits where credits due, many examples are borrowed from zompist's treatment)

Planet earth (State) is blue.
Jupiter (State) is the largest planet in the solar system (Frame)
An elephant's (Frame) ears (State) are very big
Ilden (Frame) sees the planet. (State) Here a paraphrase helps: 'the planet is visible to Ilden'
I (Frame) like Ike (State) 'To me, Ike is likeable'.
Ilden(Frame) likes soup (State)

Ilden(Process) eats
Ilden (Process) eats his soup(Process)
Ilden (Process) finished his soup (Result)
Ilden (Process) builds a sattelite (Process)
That sattelite (Result) was made by Ilden (Process)
The satellite (State) orbits the Earth (Location)
On Earth (Frame) the sun (state) moves in the sky (Location) from east (Origin) to West(Direction)

The general(State) is a rebel
The general(Process) rebels

Ilden (Frame) sees Earth (State)
Ilden (Process) looks at Earth (State)
Ilden (Process) shows Earth (Process) to the general (Direction)
Ilden gives (Process) his soup (Process) to the dog (Direction)

Bill (State) died
Ilden (Process) is baking a cake (Process)
The cake (Process) is baking
Ilden (Process) is baking
The password (Force) opens the file (Process)
Ilden opens the file (Process) with the password (Force)


Anything that is capable of consciousness is animate. (This is close to the metaphyiscal concept of the subject, but that's incredibly confusing.)
Semantically, the languages of Uttes assume that a Process animate is necessarily a participant -- in our terms, an agent or co-agent.

So, for instead:
Ilden (Process) fights Alis (Process) implies that Alis fights back.

If she doesn't, the semantic roles assigned are as follows:

Ilden (Process) fights Alis (Direction)

Though, if the process is completed, the Result role can be assigned:

Ilden (Result) fought Alis (Result)

With animates, the Result roles is used for past experience as well:

Ilden (State) is on Earth.
Ilden (Result) has been to Earth

Ilden (Process) persuades the general (Direction) to rebel ==> The general isn't rebelling yet.
Ilden (Result) persuaded the general (State/Process) to rebel ==> The general is rebelling.
Ilden (Result) persuaded the general (Result) to rebel ==> the general has rebelled.

The Force role is often assigned to inanimate:

Gravity (Force) pushes Ilden down

But not necessarily so:

The drunk da-Uttes (Force) scared the Earth people(Result)
I (Force) pushed Ilden down

You guessed it, this happens with involuntary actions. We can carry it a little further:

Ilden (Result) persuaded the General (Force) to rebel (the general isn't aware he's rebelling)
Ilden (Process) fights Alis (Force) (Alis fights him, though reluctantly.

What these last examples means is that, given the same English sentence, a translation in da-Uttes would use different cases or voices, depending on how willing or successful the participants are.

Of course describing this seems a little awkward with English examples. Next time, hopefully, we'll run through the same examples in da-Uttes.

[bradrn]

I like it! Especially how it’s just subtly different enough from human languages. Some thoughts:

• Process seems a bit vague — ‘an NP involved in a specific process’ could cover just about anything. But from what I can see, it looks like Process subsumes what we Earthlings call ‘agents’ and ‘undergoers’, i.e. things which effect an action and things which are affected by it. Is that correct?
• I like how Process/Result maps to telicity, which in human languages is typically a verbal category. However, this isn’t unattested: Finnish does express this as a nominal case distinction. Also, you may want to look at Kayardild, which also uses case for some verbal categories, though not telicity. (Actually, you should just look at Kayardild in general; that language has some weird case systems. Yes, plural. Plus four levels of case stacking.)
• I fail to see the difference between your Force and Fillmore’s Instrumental. Is there any?
• I’m not overly fond of the names — to me names like ‘Process’ indicate something verbal rather than nominal. I suggest: Possessor, Participant, Assigned, Frame, Instrumental, Origin, Position, Direction, Manner. Not that the name is particularly important, though.

[Ares Land]

I like it! Especially how it’s just subtly different enough from human languages. Some thoughts:

• Process seems a bit vague — ‘an NP involved in a specific process’ could cover just about anything. But from what I can see, it looks like Process subsumes what we Earthlings call ‘agents’ and ‘undergoers’, i.e. things which effect an action and things which are affected by it. Is that correct?

Thanks!
Yep -- that's about it. (Generally 'Process' implies something physical going on. But I think examples will make that clearer.)

[*] I like how Process/Result maps to telicity, which in human languages is typically a verbal category. However, this isn’t unattested: Finnish does express this as a nominal case distinction. Also, you may want to look at Kayardild, which also uses case for some verbal categories, though not telicity. (Actually, you should just look at Kayardild in general; that language has some weird case systems. Yes, plural. Plus four levels of case stacking.)[/list]

Ah, thanks: I'd never heard of Kayardild and it looks extremely interesting and in some places very close to what I'd like to do.

[*] I fail to see the difference between your Force and Fillmore’s Instrumental. Is there any?

Nope. In fact I believe I saw 'force' used for instrumentals somewhere. (Force fits better the 'mechanics' metaphor there, though)

[/quote]

[*] I’m not overly fond of the names — to me names like ‘Process’ indicate something verbal rather than nominal. I suggest: Possessor, Participant, Assigned, Frame, Instrumental, Origin, Position, Direction, Manner. Not that the name is particularly important, though.

Yeah, I agree -- the names suck, and the suggestions are gratefully accepted.

[Vardelm]

[*] I’m not overly fond of the names — to me names like ‘Process’ indicate something verbal rather than nominal. I suggest: Possessor, Participant, Assigned, Frame, Instrumental, Origin, Position, Direction, Manner. Not that the name is particularly important, though.

Yeah, I agree -- the names suck, and the suggestions are gratefully accepted.

To me, those names seem to be broken out more than the categories you have. For instance, is "Origin, position, direction, manner" just 1 single category? If you have 6 total, I would suggest a slight change to something like this:

• State -- Object
• Process -- Participant
• Result -- Result
• Frame of reference: -- Frame
• Force -- Force
• Origin, position, direction, manner -- Path

"Object" is the only one that I'm hesitant about, but I don't know another term for "something that is in a state". I think the definition of grammatical objects leaves enough room that it would work relatively well here. You could maybe switch to something like "instance", "item", "specimen", or other synonyms, but I doubt that's necessary.

[Ares Land]

For instance, is "Origin, position, direction, manner" just 1 single category?

Ah, sorry, that was unclear. There are four categories, not just one: I just grouped together the ones that work pretty much as you'd expect.
I like your names! Using 'Object' makes sense, but the potential for confusion is I think pretty high.

[Vardelm]

Ah, sorry, that was unclear. There are four categories, not just one: I just grouped together the ones that work pretty much as you'd expect.
I like your names!

Ah, OK. In that case, yeah, I'd just use those exact same words for those 4.

Using 'Object' makes sense, but the potential for confusion is I think pretty high.

Possible, yes, but as always if you just describe what that title means - as you pretty much always have to do with case - then you're covered. If you can find a better word for it, I'll be interested to see what that is. It's a tough one.

[Ares Land]

Morphology

I'll begin this by dropping some tables on you, to serve as future reference.
There are three major lexical categories in a-Uttes: content words, anaphora and postpositions.

The main syntactic slots are argument ("noun-like") and predicate ("verb-like"). Predicates can occupy either indifferently.

What we'll cover here is argument morphology: the endings a word can take when occupying the argument slot.

a-Uttes words inflect for case and specificity. More on specifity later, but for the moment, let's just say it's not unlike definiteness.

Inflection will feel familiar; it's mostly suffixing, with fused case/specificity endings, and nouns belong in a number of declension classes.

Animate nouns belong to one of four declension classes (two of which have subclasses), as follows:

anim_declension.JPG (82.12 KiB) Viewed 7000 times

Which declension a word belongs to is entirely predictable from the generic object:

• If it ends in -os, it belongs to the 4th declension. If the ending is preceded by a consonant cluster, it follows the -os, -aus subtype.
• Words in -ak belong to the 3rd declension.
• Words in -in belong to the 2nd declension.
• Words in -a belong to the 1st declension (-a, -au subtype)
• Everything else belongs to the 1st declension.

Inanimates belong to one of three classes:

inan_declension.JPG (70.75 KiB) Viewed 7000 times

The rules are likewise pretty simple:

• If it ends in -a > first declension (in -a)
• If it ends in -o > first declension (in -o)
• In -i > second declension.
• Everything else belongs to the third declension. Words ending in a consonant cluster or long consonant form the participant case -o.

Animacy is mostly semantic. Sentient or living beings, and by extension places are animate; anything else is inanimate.

The good news is, there are no irregular words. Any word can and will fit one of the declension patterns given above.

The bad news is, this doesn't quite extend to borrowings or reborrowings . Loanwords may use the case/specificity endings of the source language. This is generally in free variation(*) with the expected, native a-Uttes form.

For instance:

• mastamm, 'robot' > mastammo 'the robot' in free variation(*) with mastammah 'the robot.'
• poahn 'junior, apprentice' > poahne or poahnit

This is not unlike using scenarii instead of scenarios, but this is rare and not a little pedantic in English, whereas a-Uttes uses it fairly commonly.

Borrowings from unfamiliar languages, or language that don't fit in the da-Uttes sprachbund will take the old a-Uttes ending instead:
Amerikan 'American, Americans' > Amerikanit 'the American.'
Again, the regular, expected a-Uttes form is never incorrect: Amerikane

(*) not quite free variation: using 'borrowed' case endings implies a meaning closer to that in the source language: mastammah is more properly 'the automaton'. Likewise using the 'native' a-Uttes endings implies familiarity: Amerikane is an American you know personally.

The borrowed case endings are mostly unpredictable to the outsider, and I won't bore you with yet another set of tables, but generally the average da-Uttes knows enough Old a-Uttes to get the endings mostly right.


Anaphora

Anaphora are distinguished from regular content words by being a relatively closed class, and by particular declension patterns.
I'll just give the personal pronouns here; we'll cover the other anaphora as they come up.

pronouns.JPG (55.29 KiB) Viewed 7000 times

[bradrn]

The main syntactic slots are argument ("noun-like") and predicate ("verb-like"). Predicates can occupy either indifferently.

…somehow I’m not particularly surprised by this. Let me guess, it’s predicate-initial and uses relational nouns as well, right?

a-Uttes words inflect for case and specificity. More on specifity later, but for the moment, let's just say it's not unlike definiteness.

You’re welcome to borrow my explanation of specificity in the latest Hlʉ̂ post if you want. (Though I’m not sure I got it 100% correct, so corrections would also be welcome.)