Re: not un-/anti-passive
|From:||Eldin Raigmore <eldin_raigmore@...>|
|Date:||Monday, June 23, 2008, 18:57|
On Sun, 22 Jun 2008 14:38:41 +0300, JR <fuscian@...> wrote:
>On Fri, Jun 20, 2008 at 8:24 PM, Eldin Raigmore <eldin_raigmore@...>
>>In applicative languages, one of the reasons to promote an oblique
>>argument to direct (or primary) object, is so that passivization can be
>>applied to the result, and thus the oblique argument can wind up as
>Indeed. I came across another reason yesterday while looking through my
>grammar of Yimas, a Papuan language with six applicatives (in addition to
>possessor raising to dative). According to the author, when something can be
>expressed with an applicative or without one, the former is much more
>likely, just because it's more polysynthetic. The closer you can get to a
>one word sentence, the better!
>>>I am under the impression that "argument"
>>means "required by the verb", and that "core" means "occupies a
>>grammatical relation (or morphosyntactically-licensed argument position). All
>>core participants are arguments, and all adjuncts (non-arguments) are
>>oblique (non-core), but there are oblique arguments.
>>In a language with only 2 GRs (subject and (only) object), and with no
>>clause allowed more than one of any GR, a "ditransitive" clause would have
>>to have one of its arguments be oblique. If that's the recipient, then
>>recipients are oblique arguments; they aren't objects, because "object"
>>means "any core argument other than the subject".
>You're probably right then. I'm no expert - I don't think I ever learned
>this stuff properly, and I've seen inconsistent usage that's left me a bit
Ain't that the truth.
I think we conlangers are much more interested in a theory-neutral, cross-
linguistically-consistent terminology, than the pros seem to be.
Much of my terminology comes from Larry Trask, by way of Ray Brown here on
See this thread:
< http://tech.groups.yahoo.com/group/conlang/message/148940 >
and prequelae and sequelae and links.
A lot of what I say about Grammatical Relations comes from Patrick Farrel's
book on them. See e.g.
< http://tinyurl.com/27cgj9 >
Among the possible problems are, what exactly is meant by "valency"?
Is the valency of a verb (while sitting, unused but ready for use, in the
* the number of _arguments_ it _allows_ (including oblique arguments)?
* the number of _arguments_ it _requires_ (including obligatory obliques)?
* the number of _core_ arguments it _allows_ ?
* the number of _core_ arguments it _requires_ ?
Or does "valency" refer to the verb as actually used in a given clause?
* the number of _arguments_ actually showing up (including oblique
* the number of _core_ arguments actually showing up?
In a language with only one grammatical relation (that is, Subject, but no
Objects (not direct ones, not indirect ones, not primary ones, and not
secondary ones), if another argument -- necessarily an _oblique_ argument --
gets "promoted" to Subject, the "original subject" would have to be demoted --
perhaps to a particular oblique, perhaps omitted, perhaps there would be a
choice left up to the speaker.
Does Jeffrey's voice do that?
OTOH, suppose there are two GRs in the language; and Jeffrey's voice applies
only to clauses whose "basic", pre-"voice" form happens to have only one GR
filled -- the Subject. Jeffrey wants to promote a non-explicit experiencer to
Subject, and demote the "erstwhile Subject" to the second GR. This is rather
similar to what morphological Causativization does to the agent-of-effect or
Causee of instransitive clauses -- at least, when it is valency-raising, which it
In most languages, most of the promoting-and-demoting activity has the last
or lowest GR as the source or destination of the promoted or demoted
participant. (Most of the rest of that activity has the first or highest GR --
namely the Subject -- as the source or destination; but that's saying a lot
less, since it would naturally be informative only for languages with at least
three GRs.) So, for instance, languages with 3 GRs tend to have "dative
applicatives", but not "applicatives"; to get an oblique argument to the 2nd
position (frequently "direct object"), one must apply first "dative
applicativization", and then "dative movement". And, in a language with 2 GRs,
one can't (usually) directly make an oblique argument become the Subject in
just one step; first one must applicativize, then passivize.
If a language has 3 GRs, and has morphological causatives, and one has in it a
basic monotransitive clause with the first two GRs (Subject and primary or
direct Object) filled, and one causativizes that clause, often the Agent
(whether it's the former Subject or the Ergative Object) will be demoted to the
3rd GR (indirect object). This is not necessarily what usually happens. And if
something similar is the only reason to think there is a third GR some people
don't think it's enough of one.
Morphological Causativization is one example of "promoting-and-demoting
activity" where the destination of the Causer and the source of the Causee
tends to be the 1st or highest GR -- the Subject -- at least in Accusative
>Terminology aside, I was trying to see if these experiencers were
>lexically required by the verb or not. It could also have been that they
>were required for some verbs and not others, so I asked for different
>versions of three sentences that I thought would be representative, and
>Jeffrey left the experiencers out of all three. To use your (and probably
>the most correct) terminology now, I'd say that without this operation with
>the ho- affix, they're neither core nor oblique arguments (and as Jeffrey
>now says, they're not even expressable at all); but with it, they become
>core arguments, namely Subjects. Would you agree with that?
(1) Short, directly-responsive answer: Yes.
(2) Jeffrey seems to be "promoting" the Experiencers _from_ completely
outside the basic clause. They don't explicitly show up at all in the basic
clause. Can they still be regarded as "arguments" if they are implicit?
Professionals seem to sometimes write as if "Yes" and sometimes as if "No".
(3) Another important point is; does Jeffrey's operation result in something
with one more core argument than it used to have?
(4) And another: what happens to the "old" core arguments?
>>To me, you look right. (Is that "to me" an ethical dative?)
>I'm open to correction, but I'd say no, because 'to me' fills a certain role
>there. 'Looking' a certain way has to be in the eyes of someone or other
>(whether it's expressed or not). If there were no one who it looked right
>'to', the statement would be false. I thought the ethical dative is just an
>aside, a comment, to something that stands on its own.
To me, that looks right, too.
>>If your analysis is correct, he could also borrow whatever term grammarians
>>ofTagalog use to describe the voice that has this meaning.
>I'm not aware that there is anything quite like this in Tagalog.
Does Tagalog have no way to make the Recipient the "trigger" participant?
>>In 3-GR (or 3-MAP) languages, if an oblique argument is promoted to the 3rd
>>GR ("indirect object" in many languages), this is often called "dative
>>applicative". If the resulting clause then undergoes "dative movement",
>>that originally-oblique argument might wind up in the 2nd GR (primary
>>object ordirect object).
>What languages have dative applicatives?
< http://www.sfu.ca/linguistics/people/faculty/gerdts.html >
< http://www.sfu.ca/~gerdts/papers/directional_applicatives.pdf >
< http://www.trylus.com/p20.htm >
< http://www.wm.edu/linguistics/creek/gouge/analysis/glosses.pdf >
< http://www.summerschoolaltconference.it/ALT/Abstracts/Workshop.htm >
< http://tinyurl.com/5klmw8 >
< http://www.stanford.edu/~bresnan//bresnan-typology.pdf >
>< http://tinyurl.com/5pm6a8 >
>You're speaking of something other than possessor raising, right?
IMO "possessor raising" might be reasonably regarded as a particular kind
of "dative applicative", perhaps depending on other structural features and
transformations in the language. If it's the only kind in a particular language
the grammarian is unlikely to mention that it's a kind of "dative applicative",
>>Tagalog and (some? Many? Most? All?) other "trigger" languages appear to
>>have only one GR or MAP; the subject. They don't have objects; all
>>arguments other than the subject are oblique arguments. A voice that
>>promotes an oblique argument has to promote it directly to subject.
>Tagalog & friends are a little too difficult and contentious for me to deal
>with these days. Everyone has their own theory and it's hard to argue, and
>when you do it can get ugly, sort of like cross and crown. I also don't
>remember the details so well now, and perhaps I never knew them well
>I took a course in college on the structure of Tagalog though - my prof
>there thought it had an ergative-absolutive alignment, an active and an
>antipassive voice, and the other "voices" were applicatives. No triggers in
>What do you use GR and MAP to stand for? Grammatical relation? And ...?
GR stands for Grammatical Relation. It's a theory-neutral term; what's not
theory-neutral is, whether it's a useful term or not.
MAP stands for Morphosyntactically-licensed Argument Position. It's used by
Mapping Theory. Mapping Theory attempts to rescue what's useful from
Relational Grammar and cure it of its flaws. Both proponents and many
opponents of Mapping Theory think Relational Grammar had flaws; some don't
think anybody can cure them, though, and they don't like Mapping Theory
>>I'm not sure whether Malagasay actually has indirect objects. Or objects at
>>all. Maybe it's a 2-GR language, and some clauses we'd think of as
>>"ditransitive", are actually "dechticaetiative" (or something); the (only,
>>hence primary) object is the recipient, and the theme is an oblique
>>argument. Or maybe it's a 1-GR language, and both the theme and the
>>recipient are oblique arguments, and to promote either to subject requires
>>the same marking on the verb, the one you've called "passive voice". Or
>>maybe it's a 3-GR language.
>>Have you considered all of those? What do your sources tell you that would
>>support or contradict each possibility? Or is the most-preferable analysis
>>one I omitted?
There wouldn't be any IOs. All Os woud be D.
"Give" might, or might not, have two forms; give1 and give2 (like Smi5th's
name in the show "Taxi", where "the 5 is silent").
"I give1 you" (object is recipient; theme is implicit or adjunct or oblique)
"I give2 the answer" (object is theme; recipient is implicit or adjunct or
Note that both have only one Object. There is only one kind of Object; the
terms "direct object" and "indirect object" make no sense when talking about
this kind of language.
>I can't say I've looked into these matters enough to have formed my own
>opinions. For indirect objects in Malagasy, I think I was going according to
>the analysis in this paper:
>If it were 2-GR, and IOs were really oblique,
>why would they use the same passive as the DO, instead of the
If it were a 2-GR language, we might have
"You were give1n (by me)" as passivization of "I give1 you",
"The answer was give2n (by me)" as passivization of "I give2 the answer".
Both would be passive; the same voice would be applied, but to two different
verbs. In one of those verbs (give2), the "original" object would have been
the Theme, but in the other (give1), the "original object" would have been the
If it were a 1-GR language, then all "voices" would be promoting an oblique
argument to Subject. In a 1-GR language that might be what "passivization"
means; obviously "passivization" has to mean something slightly different when
discussing languages with 2 or more GRs than when discussing languages with
Such a language might distinguish whether the promoted oblique -- the "new
subject" -- was the theme or the recipient, by marking the passivized verb
somehow; others might not. In the latter case, "the same voice would be
used" for both themes and recipients. In a 1-GR language I see no good
reason to distinguish between the terms "passive voice" and "circumstantial
voice", at least not based on our discussion so far; one would need to re-
define these terms so that there meaning, when talking about the 1-GR
language, would be clear.
>If you really want to know about Mal., though, ask Matt Pearson - he used
>to be on this list, though I haven't seen him in a while.
>I'm not familiar with this 1-GR notion. Are there any languages where such
>an analysis is generally accepted,
Short answer: "Generally" accepted? No.
Not everyone thinks there are GRs.
Aristotle did, but Panini didn't use them.
Not everyone who thinks there are GRs thinks every language has them.
Among people who think there are GRs, there are some languages about which
there is some controversy over how many the language has. Most languages
seem to have either 2 or 3 GRs, according to these people; but there are some
languages about which not enough is known yet to decide. Amongst such
languages are included a few about which a great deal is known, yet this
question is still undecided, and hence probably undecidable.
Among people who think some languages have GRs and some don't, there are
some languages about which there is disagreement over whether they have
them or not. Tagalog (and its friends) is one such; some people think it has
one and only one GR (the Subject), some think it has no GRs at all, and some
can't make up their minds yet. I think the idea that it has two or more GRs
has much less support than the idea that it has one and only one.
Languages which appear to probably have exactly one GR, as well as
languages which appear to probably have four GRs, seem to be rare.
Among people who even admit any languages at all have any GRs at all, there
is disagreement about whether or not some languages don't have them; some
people think every language has Subjects, some think a few don't.
There are people who don't think 4 is the upper limit on number of GRs. The
writers I've read who say that are close to thinking that the idea of GRs is not
very useful in any universal-grammar sense, partly because they think some
languages don't really have an upper limit.
>preferably not Austronesian? :-)
I do not know of any non-Austronesian ones.
I just tried to search, and didn't find one in the first 7 pages returned.
< http://www.chrisdb.me.uk/wiki/doku.php?id=grammatical_relations >
>>ObYourConlang: Is Khafos in the same genetic grouping as Malagasay and
>No relation, I'm afraid!