Re: Analyzing Ayeri's syntactic and voice alignment (long)
|From:||David J. Peterson <dedalvs@...>|
|Date:||Wednesday, March 19, 2008, 10:10|
[Note: This is a long reply.]
I have started to rewrite the grammar of my conlang from scratch some
time ago, this time sticking to the table of contents that Describing
Morphosyntax suggests. So far it has been going nice, but now that I
have come to the point of describing the language's syntactic
alignment I have run into problems. I want to be sure that what I
have thought about makes sense.
Before I read on, let me just suggest, if you haven't taken a look
at it, Matt Pearson's talk from LCC1--a link is provided below:
If you don't want to watch the video, or don't have the time,
much of his talk can be reconstructed from his handout, which
you can download here:
Essentially, Matt had Tokana, and then decided, after several
years in linguistics, to go back and rework his grammar. He
kept the cases he came up with, but stripped their names and
functions, and reimagined it. What he did was quite remarkable.
Before I comment, let me just say that this is the most important
If someone could help me make sense out of this mess (well, at least
it's a mess in my head)
Emphasizing the parenthetical comment, I'd say you're done!
As I said (and Ray quoted), just because the conlang trigger
system doesn't exist in the real world doesn't mean it's a bad--or
even necessarily unrealistic--system. It simply means that it's
unattested. If you're system makes sense to you, then I don't
see what the problem is.
Regarding the confusion of the description, part of the problem
one always runs into is that there are three different systems of
description one has to deal with:
Grammatical Roles: Subject, Object, etc.
Semantic Roles: Agent, Experiencer, Stimulus, Theme, Patient...
Language-Specific: Nominative Case, Accusative Case, etc.
In English, for example, in the pronouns, we have object,
possessive, and "regular", for lack of a better term. It's nice
when you have a sentence like this:
(1) I hit him.
The "I" is a subject, an agent, and in the "nominative" case, and
the "him" is a direct object, a patient, and in the objective case.
Very simple! In a sentence like (2), though, things get murky:
(2) I am seen by him.
Now "I" is in the "nominative", is a stimulus, and is the subject;
"him" is the experiencer, in the "objective", and is an oblique (or
the object of a preposition). When you're coming up with a
description, and you have the same form (this isn't a special "I"
or a special "him"; they're the same as in (1)) with different
roles (either grammatical or semantic), you have a problem.
Fortunately, it's not a problem for the language, it's a problem
for the one describing it. That's what I'd suggest is going on
with Ayeri: there's nothing wrong with the language, just the
words one uses to describe it.
So, a couple of questions. First, (1) and (2) display an accusative
pattern, and are fairly straightforward. That's fine. Georgian
looks like a very simple accusative language if you only look at
the non-past screeves. When it comes to description, here's
where questions arise (below: Carsten's example):
(3a) Ang maliya malinoas.
ang mali-(i)ya-Ø malino-as
AGTFOC sing-3s:m.FOC song-PAT
'He sings a song'
My first question: Is there such a thing as "Maliyang malinoas"?
That is, just "He sings a song" without any focus? If so, that
has certain implications; if not, it has others.
Here's the second question (Carsten's example):
(3c) Maliyo malinoas (yari).
mali-yo malino-as (yari)
sing-3s:n song-PAT (3s:m.INS)
'A song is sung (by him).'
This one is, indeed, a bit strange. If this is a passive, it's totally
unmarked. It's kind of like:
(4a) He hits me.
(4b) Him hits (by me).
Where (4b) is the passive of (4a).
So, I guess, the way the passive is marked is in a shift in agreement:
The verb agrees with the subject, but the subject is marked with
its *semantic* case. So even though it's passive, the only change
is: (1) agreement, and (2) the case marking of the agent. What it
seems we have, then, is a system that marks semantic roles, as
opposed to grammatical relations (though some semantic roles
have different forms, which has grammatical implications). The
semantic roles, however, are language-dependent (as shown in
example 2--not all languages will call the subject of "He stumbled"
With the last examples...
(4a) Ang ilya migorayas yeyam.
ang il-(i)ya-Ø migoray-as yeyam
AGTFOC give-3s:m-FOC 3s:f.BEN
'He gives her a flower.'
(4b) Le ilyāng migoray yeyam.
le il-yāng migoray-Ø yeyam
PATFOC give-3s:m.AGT flower-FOC 3s:f.BEN
'A flower he gives to her.'
(4c) Yam ilyāng migorayas ye.
yam il-yāng migoray-as ye-Ø
BENFOC give flower-PAT 3s:f-FOC
'To her he gives a flower.'
...we see again the semantic role marking pattern. Aside from
the passive construction, this is a system that marks semantic
roles only, and has language-specific definitions for semantic
roles. The focus part of it is a separate issue that need not be
tangled up with the description of the case system.
Before moving on...
I have been told that the promoting of indirect objects (in this
case, of 'yeyam' to 'yam ... ye') is called 'applicative voice', but
again, (3c) is not strictly a passive or an applicative sentence then
-- I think.
Not exactly. When an indirect object is promoted to the subject,
and the subject is demoted, that's a passive. We have it in English:
(5a) I gave her a flower.
(5b) A flower was given her (by me).
(5c) She was given a flower (by me).
The applicative is when a non-direct object is promoted up one
step. This is a kind of faux-applicative in English:
(6a) I gave a flower to her.
(6b) I gave her a flower.
I really wouldn't call this an actual applicative, but it's a useful
example: "her" was an indirect object (preceded by a preposition),
but was promoted to direct object (no preposition; place right
after the verb). In a language with a true applicative, it works
like this (this is an example from Kamakawi):
(7a) Ka hava ei i nawa ti kopu.
/past eat I obj. fish ins. hand/
"I ate a fish by hand (with my hands)."
(7b) Ka havaka ei i kopu (ti nawa).
/past eat-app. I obj. hand ins. fish/
"I ate with my hands (a fish)."
In this case, the old direct object is demoted, and becomes
non-essential (this doesn't always happen with applicatives).
From here, you can passivize it so you get something like,
"Hands were eaten with" (man, that barely made sense in
But here's the thing: when you're describing a language, you
kind of need to stick with the way the language works. For
example, it just doesn't feel right to call the alternation in (6)
an applicative process in English, simply because English doesn't
really have an applicative. It's certainly not a part of the language
the way an applicative is a part of Swahili--and it's certainly
not as widespread, or as productive (e.g., that's well and good
for "to", but what about for "about": I told a man about you > *I
told you a man?). Based on the information you've given us
here, I'd say it wouldn't make sense to talk about an applicative
in Ayeri. Passive, sure, but not applicative (passives can target
lots of different things: object, indirect objects, possessors,
instruments--even subjects, in some languages).
So if you look at the Ayeri case system in terms of Ayeri-specific
semantic roles, I'd say it's pretty explicable. Now we come to the
The first question to answer is whether or not it's obligatory
when you have more than one argument. If it is, that is very
much like the conlang trigger system. If it isn't, then it seems
like what you have is an optional focus like many other languages.
The only difference here is that instead of fronting *the word*,
you front *the case*. Here's a similar system in Kamakawi:
(8a) Ka hava ei i nawa ti kopu. (See above.)
(8b) I nawa ka hava ei ti kopu. "I ate *a fish* with my hands."
(8c) Ti kopu ka hava ei i nawa. "I ate a fish *with my hands*."
(8d) Ei ka hava i nawa ti kopu. "*I* ate a fish with my hands."
[Note: I'm going to pretend that (8d) is grammatical for this
example, but for some reason, that doesn't look right... I don't
think you can do that with the subject.]
If you treat Kamakawi as a case language with prepositional
case markers, then the Ayeri-like versions of (8b) and (8c)
would look like this:
(8b') I ka hava ei nawa ti kopu. "I ate *a fish* with my hands."
(8c') Ti ka hava ei kopu i nawa. "I ate a fish *with my hands*."
The thing that makes the system in Ayeri surprising is that the
case markers are affixes, not phonologically independent words.
In the wrong hands, a Chomskyan syntactician might describe
this as some sort of movement operation (say the spec of F needs
to be filled by something, but the noun doesn't have the right
features that need to be checked, so only the affix can move).
So, when it comes to description, it seems like there are three
things to describe:
(1) The case marking system: based on the semantics of Ayeri.
All nouns are obligatorily marked for their semantic case.
(2) The passive: the agent changes its case from Agent Case 1
to Agent Case 2, and the verb agrees with the PAT (or the
BEN--whatever is the new "subject", though I wouldn't use
the word "subject" to describe it).
(3) The focus system: Either optionally or obligatorily, when
more than one argument is present, one argument is focused
by moving its case marker to the front.
But to your original question, about describing the syntactic
alignment, these are the sentences we need to answer the
(9a) He(i) beats him(j). (No focus.)
(9b) He(i) beats him(j). (i focus.)
(9c) He(i) beats him(j). (j focus.)
Here are your examples from (1) (renumbered 10):
(10b) i. Haruyās.
'He is beaten.'
(10b) ii. Haruyāng.
These are all intransitive. The only transitive sentences you've
shown us have only one pronominal argument. The pronominal
argument in Ayeri is privileged because it attaches right to the
verb. So, what happens when you have two pronominal arguments?
Does the agent attach? Does the patient? Do they both? What's
the order? Does the "focus" change it? The answers to these
questions should shed some light on the alignment question.
Remember that a language can have different alignments, and
that one of the places *where* it can differ is with types of NP's
(e.g., pronoun, regular, proper, possessed, definite). Consider
(11a) I walk.
(11b) I see her.
(11c) She sees me.
(12a) The dog walks.
(12b) The dog chases the cat.
(12c) The cat chases the dog.
Nothing. Of course, you can figure out it's an accusative system
through other means, but let's just confine it to the morphology.
The pronouns participate in a system that's separate from the
rest of the nouns. The same thing may happen in Ayeri. Further,
the *agreement* system may differ from the case marking
system. An example from Sathir:
(13a) kadworos t_hogo.
"The man sleeps."
(13b) kanduna t_hogo.
"I eat the man."
(13c) kaduna ame teTogo.
/eat-3sg.pres. I-abs. man-erg./
"The man eats me."
Here, the verb agrees with the subject (sole argument of intransitive,
and the agent of a transitive), but the case marking is ergative.
So, for example, the case marking of Ayeri may be basically
accusative, but the agreement may be ergative in places. You
have to look at the tests cases (separate the pronouns from the
regular nouns; agreement from case marking), and test them
all. Once you do that, you should be able to explain the alignment
of the system. Like most languages, though, the answer won't
be "Ayeri is X". It'll probably end up being: "The pronouns of
Ayeri are X; the agreement is Y; the case marking is Z", etc.
If this was the implication, I'd like to suggest that you don't *need*
to change system (whether you want to is a separate matter, of
course). The system isn't broken; just different. I rather like
"A male love inevivi i'ala'i oku i ue pokulu'ume o heki a."
"No eternal reward will forgive us now for wasting the dawn."