LINGUIST List 14.1768

Tue Jun 24 2003

Sum: Ergativity, Word Order, and Agreement

Editor for this issue: Karen Milligan <>


  1. Jose-Luis Mendivil, ergativity, word order and agreement

Message 1: ergativity, word order and agreement

Date: Mon, 23 Jun 2003 14:51:31 +0000
From: Jose-Luis Mendivil <>
Subject: ergativity, word order and agreement

Three weeks ago I posted a query (Linguist 14.1529) in order to obtain
supporting statistical data on these three statements:

a) OV languages show object agreement more usually than VO languages
b) Most ergative languages display OV order
c) Most ergative languages show object agreement

The following wise and kind people shared their knowledge and opinions
with me (in order of 'appearance'):

- Andres Enrique-Arias
- Leo Connolly
- Jonathan David Bobaljik
- Mario van de Visser
- Maren Pannemann
- Hsiu-chuan Liao

Not a long list, but qualitatively relevant.

A summary of the responses follows.

The main conclusion is that correlations (a) and (c) are stronger than
correlation (b).

Leo Connoly observed that all the correlations are true 'on the
whole', though there are exceptions. Both Connoly and Bobaljik
referred to the need of a clear definition of what counts as 'object
agreement' in correlation (c). Of course, I meant Dixon's O agreement
(the absolutive argument in a transitive construction), not S
agreement (the absolutive argument in an intransitive construction).

On correlation (b) Bobaljik mentioned Dixon's 1994 book 'Ergativity',
where the claim is that verb-medial order is ''seldom ... preferred''
in ergative languages.

Van der Visser remarked that the correct generalization is that
ergative languages do not display verb medial word order (I'd found
this statement in Anderson's 1976 article on 'ergative
subjects'). Mario van der Visser referred to two papers by Mahajan:
1994 and 1997. And, in the same vein, Hsiu-chuan Liao mentioned two
statements, one by Trask (1979): ''The basic word order is SOV, can be
VSO, never SVO'' and one by Mahajan (1994): ''SVO languages are never
ergative. Ergativity is found only in verb final and verb initial
languages''). Liao observes he does not believe that the claim that
''most ergative languages display OV order'' is true. He observed that
although a lot of well-known ergative languages are verb-final,
ergative Polynesian languages are verb-initial. He adds that if we
consider Philippine languages and other Philippine-type languages as
ergative, we will have a number of VO ergative languages. So he
proposes that the more accurate statement would be that ''most
ergative languages display either verb initial or verb final order,
but never SVO.'' Liao observes that this statement is similar to
Mahajan's claim, but different in that it allows OVS ergative
languages (he mentions Balinese, an Austronesian language that was
claimed to be ergative and that exhibits an OVS order).

As to correlation (c) Pannemann draw my attention to the work of
Neeleman and Weerman (1999). In Neeleman and Weerman's framework
arguments are either licensed by means of a case shell (dependent
marking) or by verbal agreement (head marking). Marked case arguments
such as ergative or accusative arguments carry a case shell and are
therefore licensed by case. However both nominative and absolutive
arguments would be a manifestation of absence of case: so these
arguments do not carry a case shell and must be licensed by verbal
agreement. This would account for the correlation in (c). But,
according to Liao, correlation (c) does not hold, at least not
clearly. Liao (as well as other respondents) has a reasonable problem
with the word 'most' I used in my original query. According to Liao,
statement (c) seems to suggest that most ergative languages exhibit
verbal agreement, but he observes that it is not an easy job to
measure out the total number of ergative languages in the world. Of
course, if we are not sure about the total number of ergative
languages, we cannot state anything like ''most ergative languages

As to (c) Liao proposes the following modification: ''In an ergative
language, if the verb agrees with only one argument per clause, it
tends to agree with the S (of an intransitive verb) and the O (of a
transitive verb) rather than the A (of a transitive clause).'' Or, in
other words: ''if an ergative language exhibit verbal agreement, the
verb tends to exhibit absolutive agreement.'', which seems to be in
the same spirit of Neeleman and Weerman's treatment. (Nevertheless,
Liao mentions one exception: Papuan language Enga, an ergative
language with S/A agreement only).

The complete list of sources I was addressed to is the following:

Dixon, R.W. (1994): Ergativity. Cambridge, Cambridge University Press.

Enrique-Arias, Andr�s (2002): ''Accounting for the Position of Verbal
Agreement Morphology with Psycholinguistic and Diachronic Explanatory
Factors.'' In Studies in Language 26.1

Mahajan, Anoop. 1994. ''The ergative parameter: Have-be alternation,
word order and split ergativity''.� NELS 24: 317-331.

Mahajan, Anoop (1997): ''Universal Grammar and the Typology of
Ergative Languages''. In Alexiadou and Hall, eds. (1997): Universal
Grammar and Typological Variation. John Benjamins, 35-57.

Neeleman, Ad and Fred Weerman (1999): Flexible Syntax. A theory of
Case and Arguments, Kluwer, Dordrecht.

Siewierska, Anna and Bakker, Dick (1994): ''The distribution of
subject and object agreement and word order type.'' In Siewierska,
Anna (ed.), Eurotype Working Papers 6: 83-126.

Trask, R. L. 1979. On the origin of ergativity. In Ergativity,
ed. by Frans Plank, 385-404.

Mario van de Visser recommend me to take a look at the ''Universals
Archive'' on the website of the University of Konstanz (but I have not
tried yet): 

He said that there can be found language universals in a very broad
sense, references, counterexamples, etcetera: ''Just typing keywords
like 'ergativity' and 'word order' should provide you with the
relevant information''.

Thanks again to all respondents and to the Linguist List for making
all this possible.

Jose-Luis Mendivil. 
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue