ergative to accusative alignment

Fri Jan 10 18:16:11 UTC 2014

The idea that alignment is a property of constructions rather than 
languages is especially important to remember when looking at the 
creation of innovative main clause constructions, as the alignment of 
the new constructions need not match the alignment of the pre-existing 
constructions.

A good example of innovative constructions with both ergative-absolutive 
and nominative-accusative patterns -- which I know much better myself 
(and have described in more detail in Gildea 2012) -- is the Cariban 
family: the etymologically oldest main clause alignment system is has 
hierarchical indexation on the verb, but alongside modern reflexes of 
this main clause type, there are innovative main clauses in mutliple 
Cariban languages that have ergative-absolutive alignment patterns (from 
three to five distinct etymological source constructions), as well 
several more innovative main clauses that have nominative-accusative 
alignment patterns (again from three to four distinct etymological 
source constructions).  Most languages have at least two different main 
clause constructions, each belonging to a different alignment category, 
and several have three or more.  Given that the original main clause 
construction in Proto-Cariban did not have ergative alignment (at least 
not as most linguists define ergative pattern), these examples do not 
show a change from ergative to accusative alignment, but rather the 
creation of new constructions with accusative alignment *regardless* of 
the alignment of the old main clause constructions.

So we should be asking two completely different questions, both relevant 
to the original question.

First, can languages with ergative-absolutive main clauses innovate a 
new kind of main clause that is nominative-accusative?  The answer to 
this is clearly yes, as many of the posts here have already indicated. 
To the examples already given, I think we should add the example of 
someMayan languages, in which the older type of main clause construction 
has ergative-absolutive personal indexation on the verb, but 
nominalizations have nominative indexation (they are possessed by A and 
S); reanalysis of a biclausal construction with a main verb (> 
auxiliary) and nominalization (> new main verb) leads to split 
alignment, with the old system ergative-absolutive and the new system 
nominative-accusative. For what it's worth, the creation of innovative 
ergative-absolutive main clauses from older biclausal constructions is 
also well-attested, so there is no particular directionality to this 
mechanism.

Second, do individual constructions change internally such that 
ergative-absolutive alignment properties can become 
nominative-accusative alignment properties? Here again, the answer is 
yes.  In the passive-to-ergative reanalysis in Cariban (Gildea 1997), 
the innovative ergative main clause has absolutive verbal indexation, 
ergative case-marking, and absolutive control of coreference with 
subjects of corrdinate clauses, reflexive possessors, etc.  In some 
languages, these new ergative main clauses have changed such that 
control of coreference is now with the A/S. This is still early in the 
process of shift, though, so no morphological changes are attested. In 
Indic, the older verb agrees with the absolutive for animacy, but this 
has been replaced by nominative agreement for person and number in, 
e.g., Nepali. This leaves a situation in which the only ergative pattern 
remaining in the construction is the ergative case-marker.  At this 
point, we still call it an "ergative construction" (and some would also 
call Nepali and "ergative Language"), but this does raise the question 
of whether that notion should be more graded, as the construction (and 
the language) clearly has fewer ergative patterns today than its 
ancestors did in the past. If one believes the historical scenarios 
posited in Estival & Myhill's 1988 list of language constructions that 
have shifted from ergative-absolutive to nominative-accusative, then 
there are multiple examples of loss of even this last hold-out, the 
ergative case-marker, removing all ergative-absolutive patterns from a 
construction that once has several.

And in another side note, this mechanism may be more directional than 
the reanalyses I discussed first.  I am aware of very few cases of a 
nominative-accusative construction gaining ergative properties 
incrementally in the other direction, i.e., innovating an ergative 
case-marker in an existing construction, then innovating absolutive 
verbal indexation, then finally shifting control of coreference from the 
nominative to the absolutive argument. All of these are logically 
possible, but I only know of a handful of cases where an ergative 
case-marker was added to a construction that did not have one (always in 
a minor construction, and always when the dominant construction in main 
clauses already had that ergative marker), and I don't know of any 
examples where a pre-existing construction added absolutive indexation, 
nor shifted from nominative to absolutive control of coreference. I 
discussed this apparent asymmetry in directionality in a proceedings 
paper bout 10 years back (Gildea 2004), and as of Queixalós & Gildea 
2010, no new examples had come along, but I still have not followed up 
satisfactorily to really test the empirical basis of that assertion.  I 
would welcome examples, if anyone has suggestions.

Best,

Spike

Gildea, Spike. 2004.Are there universal cognitive motivations for 
ergativity? /L'ergativité en Amazonie/, v. 2, ed. by F. Queixalós, 1-37. 
Brasília: CNRS, IRD and the Laboratório de Línguas Indígenas, UnB. 
Accessible online at: 
http://celia.cnrs.fr/FichExt/Documents%20de%20travail/Ergativite/Introductions_ergativite.htm

Gildea, Spike.2012. Linguistic Studies in the Cariban Family/./ 
/Handbook of South American Languages/, ed. by Lyle Campbell & Veronica 
Grondona, 441-494.Berlin: Mouton de Gruyter.

Queixalós, Francesc & Spike Gildea.2010. Manifestations of Ergativity in 
Amazonia. /Ergativity in Amazonia/, ed. by Spike Gildea & Francesc 
Queixalós, 1-25./Typological Studies in Language/, v. 89. Amsterdam: 
John Benjamins.

On 1/9/14, 10:58 AM, "Daniel Hieber  -- 
============================================================ Ljuba 
Veselinova, Associate Professor Dept of Linguistics, Stockholm 
University, S-10691 Stockholm, Sweden Phone: +46-8-16-2332 Fax: +46-8-15 
5389 URL  : http://www2.ling.su.se/staff/ljuba/  "We learn by going 
where we want to go."                                           Julia 
Cameron ============================================================" 
wrote:
> Richard, that’s a great point. Bickel et al. (2013) provide an 
> excellent illustration of this issue, which they call ‘Siewierska’s 
> Problem’ in memory of Anna and her seminal (2003) article on alignment 
> in ditransitive constructions. There she points out that verbal person 
> marking can show different patterns of alignment depending on whether 
> one examines the trigger potential, morphological form, position, or 
> conditioning factors of the person forms. Bickel et al. then show that 
> discrepancies among these different criteria are in fact extremely 
> common crosslinguistically. So the descriptive linguist needs to be 
> very specific about the details of alignment, and make sure they're 
> comparing like with like when comparing synchronic or diachronic data.
>
> References
> Bickel, Balthasar, Giorgio Iemmolo, Taras Zakharko & Alena 
> Witzlack-Makarevich. 2013. Patterns of alignment in verb agreement. In 
> Dik Bakker & Martin Haspelmath (eds.), /Languages Across Boundaries: 
> Studies in Memory of Anna Siewierska./ 15-36. De Gruyter.
>
> Siewierska, Anna. 2013. Person agreement and the determination of 
> alignment. In Dunstan Brown, Greville G. Corbett & Carole Tiberius 
> (eds.), /Agreement: A Typological Perspective/. 339-370. Wiley-Blackwell.
>
>
> Daniel W. Hieber
> Graduate Student in Linguistics
> University of California, Santa Barbara
> www.danielhieber.com
>
> Omnis habet sua dona dies. ~ Martial
>
> *From:* Richard Griscom <mailto:rgriscom at UOREGON.EDU>
> *Sent:* ‎Thursday‎, ‎January‎ ‎9‎, ‎2014 ‎8‎:‎16‎ ‎AM
> *To:* LINGTYP at LISTSERV.LINGUISTLIST.ORG 
> <mailto:LINGTYP at LISTSERV.LINGUISTLIST.ORG>
>
> This is a very interesting topic, but I would also add a word of 
> caution against making assumptions regarding entire language systems 
> conforming to a single alignment pattern. In my opinion, alignment is 
> best viewed as construction-specific rather than language-specific in 
> order to avoid inaccurate generalizations across the distribution. 
> This, of course, doesn't preclude an analysis of a shift in the 
> alignment patterns of one or more constructions in a given language.
>
> Best,
> Richard
>
>
>
> On Thu, Jan 9, 2014 at 12:32 AM, Florian Siegl <florian.siegl at gmx.net 
> <mailto:florian.siegl at gmx.net>> wrote:
>
>     A related phenomena though only partly answering the initial
>     posting is attested on Kamtchatka. Chukchi and Koryak show
>     ergative alignment, but not Itelmen. The Itelmen absolute case
>     marks S as well as A and P. Whether the Itelmen transitive verbal
>     agreement markers still follow erg-abs alignment is not settled.
>     Itelmen is not ergative but apparently not very
>     nominative-accusative either. See the relevant sections in Georg,
>     Stefan & Volodin, Alexander P. 1999. Die itelmenische Sprache -
>     Grammatik und Texte. Tunguso-Sibirca 5. Wiesbaden: Harrassowitz.
>
>     Best wishes,
>
>     Florian Siegl
>
>
>
>
>     On 9.01.2014 9:03, Don Killian wrote:
>
>         Dear Raheleh,
>
>         Depending on what you might be including by ergative, there's
>         an interesting article by Dimmendaal 2012
>         (http://www.uni-koeln.de/phil-fak/afrikanistik/mitarbeit/dimmendaal/Marked%20nominative%20in%20Eastern%20Sudanic%20130907%20DR%20comments.pdf)
>         in which he postulates the origins of Marked Nominative
>         (depending on who you ask, some consider this a subcategory of
>         accusative alignment) in Eastern Sudanic languages.
>
>         Gaahmg is particularly interesting as far as diachronic
>         developments go, as it also has both passive and antipassive
>         constructions and in fact can allow for both types of markers
>         simultaneously.  If you're curious, email me and I can send
>         you an article by Tim Stirtz.
>
>         Best,
>
>         Don
>
>
>         On 01/08/2014 02:24 PM, Raheleh Izadi Far wrote:
>
>             Dear all,
>
>             Does anybody know about languages which have changed from
>             ergative
>             alignment to accusative alignment? or does anybody know
>             about the
>             mechanisms involved in such a change? what are the studies
>             concerning
>             this issue? and if there are any, are they accessible online?
>
>             Thank you very much in advance
>
>             kind regards,
>             Raheleh Izadifar
>
>
>
>