Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: THEORY: Alignment of ditransitive with monotransitive case roles.

From:tomhchappell <tomhchappell@...>
Date:Friday, August 5, 2005, 20:02
Hello, conlangers.
I have been re-reading the July 2005 draft of
"Argument Marking in Ditransitive Alignment Types" by
Martin Haspelmath of
the Max Planck Institute for Evolutionary Anthropology.
(This can be found online at

In this paper Haspelmath considers 100 example languages.
These languages come in
four of the six types of transitive-to-intransitive alignment, and
four of the six types of ditransitive-to-monotransitive alignment,
that Sierwieska spoke of as I quoted her below.
He includes examples of
accusative, ergative, tripartite, and neutral alignment
of transitive to intransitive, and he includes examples of
indirective, secundative, tripartite, and neutral alignment
of ditransitive to monotransitive.
He does not include any examples of Active/Stative ("split-S"),
nor of Hierarchical (inverse/direct?),
transitive-to-intransitive alignment.
He does not include any examples of split-P, nor Hierarchical,
ditransitive-to-monotransitive alignment.

So, the possibility exists for Haspelmath's paper to include example
natlangs in 16 of the 36 combinations-of-types I asked about below.
In fact, his sample does include examples of 13 of these combinations.

In this paper Haspelmath is interested in
the marking of the arguments.
He distinguishes between what he calls "flagging" --
which others might call "dependent-marking",
and yet others "declension",
where the nominal itself may have some morpheme attached to it
to mark it as being in one of the cases --
and what he calls "indexing" --
which others might call "head-marking",
and yet others "agreement" or "concord",
where the verb is marked to agree with the nominal differently,
depending on which case the nominal is in.

Some languages "flag" according to one pattern,
and "index" according to a different pattern.
So, it's possible for a language to have
two different patterns of transitive-to-intransitive alignment;
one for flagging, the other for indexing.
It's also possible for a language to have
two different patterns of ditransitive-to-monotransitive alignment;
one for flagging, the other for indexing.
As a result, a single given language might fit into just one,
or perhaps two, or, at a maximum, four,
of the type-combinations I am seeking.

Here are the type-combinations
Haspelmath claims to have found in his 100-language sample,
together with some of the languages in each combination of types.

Indirective Accusative:
Urubu-Kaapor, Apurina I, Purepecha Tarascan.

Indirective Ergative:
Trumai, Kipea

Indirective Tripartite:
Hixkaryana, Semelai

Indirective Neutral:
Warao, Hixkaryana

Secundative Accusative:
Wari, Awa Pit

Secundative Ergative:
Ika, Greenlandic West

Secundative Tripartite:
Sahaptin, Maricopa, Greenlandic West, Mangarayi

Secundative Neutral:
Wari, Tagalog II

Tripartite Accusative:
Awa Pit, Imonda

Neutral Accusative:
Urubu-Kaapor, Barasano

Neutral Ergative:
Trumai, Shipibo-Konibo, Lakota

Neutral Tripartite:
Sahaptin, Maricopa, Mangarayi, Ainu

Neutral Neutral:
Warao, Araona, Ojibwa Ottawa


Note that Sierwieska also considered languages
to fall into more than one type.
For instance,
if a language was accusative for pronouns but ergative for nouns,
or accusative for present-tense but ergative for past-tense,
she would have counted it both
as an accusative language and as an ergative language.
I am sure -- in fact I think she intends to say so in "Person" --
that it is only thus that she was able
to "fill in" all 36 combination-types.



Kabardian uses the Ergative Case for Donors, Recipients, and Agents.
If you remember way back when I started this thread,
I wondered if  natlangs had variable alignment of the A, D, and R
as they do of the P, R, and T and of the S, A, and P.
Kabardian is an Indirective Ergative language.
Knowing it is Indirective lets us know that
the P and the T receive the same case as each other,
and the R receives a case different from
the case assigned to the P and the T.
(Because it is an Ergative language,
the P and the T will receive the Absolutive case, along with the S.)

However, nothing in the phrase "Indirective Ergative" tells us that
the A, the D, and the R will all receive the same (Ergative) case.

The fact that, in Kabardian, it does, makes me wonder anew
whether or not it might actually be a worthwhile typological question
about natlangs, how they align the cases of the A, D, and R roles.


Thanks for any comments.

Tom H.C. in MI

--- In, tomhchappell <tomhchappell@Y...>
> wrote: > [snip] > Siwierska says six main kinds > of Transitive-to-Intransitive alignment > are attested by natlangs; accusative, ergative, active (split-S), > hierarchical, tripartite, > and neutral (all three roles -- S, A, & P -- marked the same). > Crediting Dryer, Croft, and Haspelmath, she says there are also six > main kinds of Ditransitive-to-Monotransitive alignment attested by > natlangs; indirective, secundative, split-P, hierarchical, > tripartite, and neutral. > She says: > "All the possible combinations of the major monotransitive and > ditransitive alignment types are attested." > However she does not give an example of each, ... > [snip] > Does anyone know of a natlang example of each combination of the > alignment types she speaks of? > That would be 36 combinations-of-types ... > [snip]


tomhchappell <tomhchappell@...>