1

Syntactic Universals and Usage Frequency

(MARTIN HASPELMATH, Leipzig Spring School on Linguistic Diversity, March 2008)
2. Object marking, definiteness and animacy

1. The major (monotransitive) alignment types

(1)a. S nominative b. S c. S
absolutive

A P accusative A P ergative A P

accusative alignment neutral alignment ergative

alignment

S = the single argument of an intransitive clause

A = the most agent-like argument of a transitive clause
P = the most patient-like argument of a transitive clause

two further logical possibilities, hardly attested:

(2)a. S b. S

A P A P

horizontal alignment tripartite alignment

Universal 1:
Case alignment is practically always neutral, accusative, or ergative.
Horizontal alignment and tripartite alignment are extremely rare.

Explanation in terms of efficiency:

The horizontal alignment type requires the same coding effort as the accusative
and ergative alignments, but fails to make the important distinction between A and
P. It is equally costly but less distinctive, and therefore clearly less efficient.
The tripartite alignment requires more coding effort, but the distinction between S
and A, and S and P is redundant because S+A and S+P do not occur together in the
same clause. It is more costly but not more distinctive, and therefore clearly less
efficient.

Universal 2:
In neutral alignment, the single case is always zero-coded; in accusative
alignment, the nominative case is usually zero-coded; in ergative alignment,
the absolutive case is almost always zero-coded.

Explanation in terms of efficiency:

In all these cases, the zero-coded case is by far the most frequent case. The most
frequent case is the one hearers expect, and efficient coding systems only use overt
coding for unexpected meanings.
 2

2. Differential Object Marking (DOM)

= a difference in the form of overt case marking that depends on the intrinsic
properties of the (direct) object, not on its semantic or syntactic role

(more precisely: Differential P Marking)

Universal 3: UA#217
If any P is overtly case-marked, then all Ps that are higher on the animacy
scale, the definiteness scale, or the person scale are marked at least to the
same extent.
(Silverstein 1976)

2.1. The animacy scale: human > animate > inanimate

(3) Spanish (only human)

El director busca el carro/el perro/a su hijo.
'The director is looking for the car/the dog/his son.'

(4) Russian (animate and human)

Mia uvidel dom/kot-a/brat-a.
'Misha saw the house/the cat/the brother.'

(5) Hungarian (inanimate, animate and human -- no split!)

Table 1. human animate inanimate

Vietnamese
Spanish m
Russian m m
Hungarian m m m

2.2. The definiteness scale: pronoun > proper noun > definite > specific >
nonspecific

(6) English (only pronoun)

Leyla saw hi-m/Yusuf/the boy/a boy.

(7) older German (pronouns, proper nouns)

Friedrich sah ih-n/Gertrud-en/das Kind/ein Kind.
' Friedrich saw him/Gertrud/the child/a child.'

(8) Persian (pronouns, proper nouns, and definite)

Hasan u-ra/ Ali-ra/ ketab-ra/ ketab did.
Hasan he-ACC Ali-ACC book-ACC book saw
'Hasan saw him/Ali/the book/a book.'
 3

Table 2. pronoun proper n. definite specfic nonspecif.

Vietnamese
English m
older German m m
Persian m m m
Turkish m m m m
Hungarian m m m m m

2.3. The person scale: 1st/2nd person > 3rd person

(9) Dyirbal (1st/2nd person only)

nadya inu-na/ balagara balgan
I.NOM you-ACC they.DU hit
'I hit you/them.'
Table 3. 1st/2nd 3rd
Vietnamese
Dyirbal m
Hungarian m m

2.4. Older explanations for differential marking of animate/definite objects

Caldwell (1856: 271)

"...the principle that it is more natural for rational beings to act than to be acted
upon; and hence when they do happen to be acted upon when the nouns by
which they are denoted are to be taken objectively [i.e. are used as objects] it
becomes necessary, in order to avoid misapprehension, to suffix to them the
objective case-sign."

Thompson (1912:75)
"...wenn die Sprache ein transitives Verb besitzt, in gewissen Fllen der Patiens als solcher durch
sprachliche Mittel zur Unterscheidung von Agens gekennzeichnet werden mu, weil er sonst vom
Hrer als Agens aufgefat werden wrde. Zu dieser flschlichen Auffassung ist der Hrer fter dann
disponiert, wenn das Objekt eine bestimmte Person bezeichnet. Ist andererseits die Person oder ein
Tier Agens und ein unbelebtes Ding Patiens, so ist auch ohne sprachliche Bezeichnung ein solches
Miverstndnis fast ganz ausgeschlossen."

[...if a language has a transitive verb, in certain cases the patient needs to be marked as such
by linguistic means to distinguish it from the agent, because otherwise the hearer would
interpret it as agent. The hearer is frequently inclined toward this wrong interpretation if
the object denotes a definite human being. If, on the other hand, the human being or an
animal is the agent and an inanimate thing is the patient, such a misunderstanding is almost
completely excluded even without any linguistic marking.]
(Caldwell and Thompson cited after Filimonova 2005)

Silverstein (1976:113)
"This hierarchy expresses the semantic naturalness for a lexically-specified noun
phrase to function as agent of a true transitive verb, and inversely the naturalness
of functioning as patient of such."

Comrie (1989:128)
"...the most natural kind of transitive construction is one where the A is high in
animacy and definiteness, and the P is lower in animacy and definiteness; and any
deviation from this pattern leads to a more marked construction...
the construction which is more marked in terms of information flow should also be
more marked formally"
 4

Dixon (1994:85):
"Those participants at the left-hand end of the hierarchy are most likely to be
agents..., and those at the right-hand end are most likely to be patients..."

(the term differential object marking is from BOSSONG 1985, 1991, 1998; see also
Croft 1988, 2003:166-75, Lazard 2001)

3. Aissen's (2003) explanation in terms of "iconicity constraints"

and an "economy constraint"
"The challenge then is to develop a theory of DOM [=differential object marking]
which expresses the generalization in [Universal 3], and at the same time allows
for the various ways in which DOM can be implemented in particular
languages." (p. 437)

i.e. Aissen wants to conflate explanation of universals with language-

particular description, in the manner characteristic of generative linguistics.

This is typical of Optimality Theory more generally:

McCarthy (2002:1)
"One of the most compelling features of OT, in my view, is the way that it unites
description of individual languages with explanation of language typology...
OT is inherently typological: the grammar of one language inevitably incorporates
claims about the grammars of all languages. This joining of the individual and the
universal...is probably the most important insight of the theory."

Aissen simply presupposes that the universals should be explainable by

generative linguistic theory:

"The fact that DOM is characterized in many languages by a great deal of

apparent fuzziness has perhaps reenforced the feeling that the principles
underlying DOM are not part of core grammar. However, the exclusion of DOM
from core grammar comes at a high cost, since it means that there is no account
forthcoming from formal linguistics for what appears to be an excellent
candidate for a linguistic universal."

OT's strategy in many cases: take a known typological generalization, turn it

into an OT constraint, and account for cross-linguistic variation by inserting
counteracting constraints in different positions.

McCarthy (2002:40)
"Descriptive universals rarely make good constraints, but descriptive tendencies
often do. Indeed, the success of OT in incorporating phonetic or functional
generalizations is largely a consequence of its ability to give a fully formal status
to the otherwise fuzzy notion of a cross-linguistic tendency. Tendencies, then, are
a good place to start theorizing about constraints..."

Aissen starts with the Relational scale and the Animacy scale:

"The analysis rests on a set of proposed universal prominence scales which are
part of universal grammar." (AISSEN 1999:679):
 5

(10) Relational cale: Su > Oj (Subject > Object)

(11) Animacy scale: Hum > Anim > Inan (Human > Animate >
Inanimate)

"Harmonic alignment" yields "markedness hierarchies" (increasing

markedness of associations from left to right):

(12) a. Su/Hum > Su/Anim > Su/Inan

b. Oj/Inan > Oj/Anim > Oj/Hum

Markedness hierarchies can be "implemented" as fixed/universal constraint

subhierarchies (p. 443):

(13) a. *SU/INAN >> *SU/ANIM >> *SU/HUM

b. *OJ/HUM >> *OJ/ANIM >> *OJ/INAN

This expresses the fact that inanimate subjects and human objects are
generally disfavored. But in fact they do occur, though languages pay the
price of additional marking. What's really excluded is "marked" associations
of relation and animacy that are not case-marked:

implemented as local conjunction with *C ("STAR ZERO CASE"):

ASE

(14) *OJ/HUM & *C >> * OJ/ANIM & *C >> *OJ/INAN & *C

ASE ASE ASE

(15) *OJ/PRO & *C >> * OJ/PN & *C >> *OJ/DEF & *C

ASE ASE ASE

>> * OJ/SPEC & *C >> *OJ/NSPEC & *C

ASE ASE

"The effect of local conjunction here is to link markedness of content (expressed

by the markedness subhierarchy) to markedness of expression (expressed by
*). That content and expression are linked in this way is a fundamental idea of
markedness theory (Jakobson 1939; Greenberg 1966). In the domain of
Differential Object Marking, this is expressed formally through the constraints
[shown immediately above]. Thus they are ICONICITY CONSTRAINTS: they favor
morphological marks for marked configurations." (Aissen 2003:448)

"Iconicity" must be limited by "economy", otherwise all objects would get

case. Hence, we need an economy constraint: *STRUCcase ("STAR STRUCTURE
CASE").

This constraint is inserted among the constraints of the subhierarchy, thus

yielding the different language types:

(16) a. Vietnamese
*STRUCcase >> *OJ/HUM & *CASE >> * OJ/ANIM & *CASE >> *OJ/INAN & *CASE
b. Spanish
*OJ/HUM & *CASE >> *STRUCcase >> * OJ/ANIM & *CASE >> *OJ/INAN & *CASE
c. Russian
*OJ/HUM & *CASE >> * OJ/ANIM & *CASE >> *STRUCcase >> *OJ/INAN & *CASE
d. Hungarian
*OJ/HUM & *CASE >> * OJ/ANIM & *CASE >> *OJ/INAN & *CASE >> *STRUCcase
 6

A language such as "Anti-Spanish", which only case-marks inanimate objects,

cannot be described in this system, because the constraints in the
subhierarchy cannot be reranked.

Thus, Aissen achieves explanation by constrained description.

Method:
Use the concepts of functional-typological linguistics (scales, harmonic
association, iconicity, economy),
translate them into OT (e.g. by adjusting the TYPEFACE),
and claim that progress has been made.

"OT provides a way...to reconcile the underlying impulse of generative grammar

to model syntax in a precise and rigorous fashion with a conception of DOM
which is based on prominence scales. The purpose ... is to develop an approach...
that is formal and at the same time expresses the functional-typological
understanding of DOM" (Aissen 2003:439)

But why do we need "constrained description"? Why not opt for a division of
labor? (some universals are explained functionally, others in terms of innate
constraints from the cognitive code/UG)

Different underlying impulses of generative grammar:

( use fancy abbreviations and notational conventions)
explain as many facts as possible with the generative method ("explain universals")
focus on arguments from the poverty of the stimulus ("explain acquisition")
reduce the formal apparatus of UG as much as possible ("explain UG")

If Aissen's story differential object marking is successful, it could itself be an

argument in favor of the general approach.

4. Problems with Aissen (2003)

(i) How are language-particular idiosyncrasies dealt with?

e.g. in German, DOM in noun inflection is found only in one small subclass of
masculine nouns (Haspelmath 2002:245):
MASCULINE FEMININE NEUTER
NOMINATIVE Lwe Mann Garten Frau Nase Kind Buch
ACCUSATIVE Lwe-n Mann Garten Frau Nase Kind Buch
'lion' 'man' 'garden' 'woman' 'nose' 'child' 'book'

(ii) The contrast may not be zero-overt, but short-long:

Dyirbal: NOM ACC
1sg adya ayguna
2sg inda inuna (cf. Carnie 2005)

(iii) How do we know how the scales are aligned harmonically?

Answer: "The basic principle is that prominent structural positions attract
elements which are prominent on other dimensions." (p. 476)

This principle needs to be part of UG as well, and we need a general

definition of "prominence" across the scales.
 7

(iv) What is "markedness"? Aissen treats it as a primitive concept that

everyone understands and that everyone agrees on.

"The OT account of DOM requires...constraints which characterize the

relative markedness of various associations of grammatical function with
animacy and definiteness." (p. 440)

But in fact, the term "markedness" stands for a highly diverse range of
different (often related) concepts, none of which is needed (see Haspelmath
2006). In this case, "markedness of associations of grammatical function with
animacy/definiteness" can easily be replaced by "rarity".

Oj/Inan > Oj/Anim > Oj/Hum = human objects are rarer than inanimate
objects

(v) Iconicity: "Iconicity constraints: they favor morphological marks for

marked configurations" (p. 448)
But there is no need for a concept or principle of "iconicity as markedness
matching"; all such cases can be explained by appealing to frequency and
economy (Haspelmath 2008)

(vi) Iconicity would have to be built into the OT machinery (i.e. into UG) as
well for Aissen's system to work:

"It should be acknowledged that constraint conjunction is a powerful operation

which, if unrestricted, will generate constraints that are clearly undesirable. For
example, if the subhierarchies of [4] were conjoined with *STRUCcase rather than
with *case, all the predictions made by the present analysis would be neutralized.
One possibility is to appeal to functional reasoning: although constraints formed
by conjunction of the subhierarchies with *STRUCcase might exist, grammars in
which they were active would be highly dysfunctional since marking would be
enforced most strenuously exactly where it is least needed. (Aissen 2003:447-8, n.
12)"

In other words: Aissen's system is not restrictive enough, but overgenerates

vastly. To explain why certain languages predicted by her OT account do not
exist, she needs to "appeal to functional reasoning".

This totally undermines the whole effort, because it is far simpler to "appeal
to functional reasoning" from the very beginning (thus dispensing with all the
constraints, the prominence principle, the alignment mechanism and the
subhierarchies).

5. Differential object marking as efficient coding

5.1. The fundamental insight: statistical associations in language use

The non-harmonic associations of syntactic role and animacy/definiteness

are rare in discourse. Therefore more overt coding of non-harmonic situations
is efficient. Inefficient languages are unattested or rare because they are
inefficient, not because they are not learnable.
 8

statistical data:

Thompson 1909 (for Russian): agents: 75% human

inanimates: 10% agents (cf. Filimonova 2005:78)

Zeevat & Jger 2002, Jger 2004: SAMTAL corpus of spoken Swedish
Table 4.
animate inanimate pronoun NP definite indefinite
subject 2948 203 2984 167 3098 53
object 317 2834 1512 1639 1830 1321

p < 0.01%

5.2. In what sense is DOM efficient?

Frequencies lead to expectations, e.g. animate arguments are mostly subjects,

and only rarely objects. Hence hearers expect an animate NP to be a subject.
Object marking tells us then that against their expectations, the NP is to be
understood as an object.

Inanimate arguments are mostly objects, so that hearers expect an animate

argument to be an object. Marking it as such is relatively redundant. A coding
system that exploits the redundancy is efficient.

5.3. Does DOM serve ambiguity avoidance?

Aissen (2003:437)
"An intuition which recurs in the literature on DOM is that it is those direct
objects which are most in need of being distinguished from subjects that get
overtly case-marked. This intuition is sometimes expressed as the idea that the
function of DOM is to disambiguate subject from object."

No, the threat of ambiguity is not sufficient:

(continued:) "There may be cases in which DOM is motivated precisely by the

need to disambiguate, but it is also clear that DOM is required in many instances
where the absence of case-marking could not possibly lead to ambiguity."

And of course many languages tolerate an amount of ambiguity, because the

context usually gives enough further clues.

DOM is about maximizing distinctiveness with minimal effort, or minimizing

confusion with maximal economy:

Comrie (1977:9)
"Given the general tendency in languages [for subjects to be definite/animate and
objects to be indefinite and inanimate], instances where confusion will be
particularly likely will be where one has either indefinite...and/or inanimate
subjects, or where one has definite...and/or animate direct objects."
 9

5.4. How do languages come to have efficient case-marking?

The functional factors assert themselves in language use. Language use affects
language structure through language change. Where they have a choice,
speakers will tend to prefer more efficient coding strategies, and these usage
preferences may become part of language structure.

Morphosyntactic innovations tend to eliminate inefficient patterns created

by phonological change (cf. Bossong 1985):

Latin > Old French > Middle French

NOM mur-us mur-s mur 'wall'
ACC mur-um mur mur

Introduction of more distinctive patterns may be limited by perceived

redundancy:

e.g. Spanish introduced a new direct-object marker a (by semantic extension

from the dative a) which is first used where it is most needed (with personal
pronouns), then spreads to all animate objects, but hasn't spread further yet.

Veo a ti. Veo a Juan. Veo a mi marido. *Veo a mi perro.

'I see you.' 'I see Juan.' 'I see my husband.' 'I see my dog.'

e.g. Old High German extended the accusative suffix -an from pronouns to
personal names, where it is most needed, but not further.

NOM er dese hwer Hartmuot > Hartmuot

ACC inan desan hwenan Hartmuot Hartmuot-an
'he' 'this' 'who'

Elimination of distinctive patterns may be limited by non-redundancy:

e.g. in the Old High German n-declension, animate and inanimate nouns alike had a
distinction between nominative and accusative (cf. 15). Then the nominative-accusative
distinction was lost in inanimate nouns (following the pattern of the other declsension types),
and in Modern German only animates preserve the zero-marking in the nominative
(Haspelmath 2002:245).

(17) Old High German > Modern German

NOM. SG affo knoto Affe Knoten
ACC.SG affon knoton Affen Knoten
'ape' 'knot' 'ape' 'knot'

5.5. How does the efficiency-based approach avoid the problems of Aissen
2003?

(i) Language-particular idiosyncrasies: No problem, because the explanation

is separate from the description.

(ii) Zero-overt vs. short-long: The efficiency explanation predicts short-long,

and zero-overt only as a special case of this.
 10

(iii) How do we predict harmonic associations? The explanation uses

observed text distributions as a point of departure. It would also work if we
had no explanation for the text distributions.
But it's easy to speculate about explanations: Humans are more interested in events initiated
by humans, so they talk much more about such events than about other types of events.
Agents tend to be topics and therefore definite because when we talk we adopt the point of
view of the agent, etc.

(iv) The role of markedness. "Markedness" plays no role.

(v-vi) The role of iconicity. Iconicity plays no role.

5.6. Conclusion

Aissen's story on DOM is not successful.

Hence, there is no reason not to adopt Caldwell's (1856) and Thompson's

(1909/1912) approach and expain DOM in functional, efficiency-based terms.

If desired, this old functionalist approach can be formalized in terms of

plain (functional) OT (Zeevat & Jger 2002),
bidirectional stochastic OT (Jger 2004, Morimoto & de Swart 2005)
Evolutionary Game Theory (Jger 2007)

6. Differential Subject Marking

mirror image of Universal 3:

Universal 4: UA#217
If any A is overtly case-marked (with "ergative" case), then all Ps that are
lower on the animacy scale, the definiteness scale, or the person scale are
marked at least to the same extent.
(Silverstein 1976)

Table 5. 1st/2nd 3rd proper human inanimate

Lezgian m m m m m
Dyirbal m m m m
Guugu Yimidhirr m m m
Gumbainggir m m
Lakhota m
Hungarian

Kiparsky 2008: argues extensively against Garrett's (1990) proposal that "NP-
split ergativity" (=differential use of overt ergative case only for lower
NP types) has a purely diachronic explanation

"All diachronic roads lead to the same synchronic Rome, where ergative case is
lacks a morphological mark in high-D nominals. Far from explaining this
syncretism pattern, the various changes themselves require a motivation for
the pattern as part of their explanation. The invisible hand of historical
evolution nudges morphological systems towards certain optimal states, and
 11

part of the job of morphological theory is to say what those states are...
Historical mechanisms by themselves cannot explain why languages undergo
the particular kinds of reanalyses that result in split ergativity but not other, a
priori equally imaginable kinds of reanalyses. The D-hierarchy must in some
sense be part of the design of language. (Kiparsky 2008:3.2)"

Two models of "change resulting in typological generalization":

Model 1 (Kiparsky?) Model 2 (Haspelmath)

Acquisition, Acquisition,
variation, use variation, use

Random change Functional Functionally

motivation guided change

Typological Typological
generalization generalization

"(continued:) The D-hierarchy is a linguistic universal and SHOULD be expressed

in the synchronic theory of grammar because:
(23) a. The hierarchy is inviolable.
b. There are multiple sources of split ergative case marking.
c. The hierarchy is a pathway of analogical change.
d. The hierarchy is manifested spontaneously in child language.
e. The hierarchy must be encoded in the grammar because it intersects
with other hierarchies (notably definiteness) and because it plays a role in the
distribution of other morphological categories (notably number and agreement)."

the hierarchy is NOT inviolable (various exceptions have been noted to

DOM and DSM, Filimonova 2005)
the hierarchy must be part of a ("synchronic") functional motivation (and in
this sense perhaps part of the "design of language"), but it need not be
part of the cognitive code/Universal Grammar!

References
Aissen, Judith. 1999. "Markedness and subject choice in Optimality Theory." Natural
Language and Linguistic Theory 17: 673-711.
Aissen, Judith. 2003. "Differential object marking: Iconicity vs. economy." Natural
Language and Linguistic Theory 21.3: 435-483.
Bossong, Georg. 1985. Differenzielle Objektmarkierung in den neuiranischen Sprachen.
Tbingen: Narr.
Bossong, Georg. 1991. "Differential Object Marking in Romance and Beyond." In D.
Wanner and D. Kibbee (eds.), New Analyses in Romance Linguistics: Selected
Papers from the XVIII Linguistic Symposium on Romance Languages, Urbana-
Champaign, April 79, 1988. Amsterdam: Benjamins, 143170.
Bossong, Georg. 1998. "Le marquage diffrentiel de l'objet dans les langues d'Europe".
In: Feuillet, Jack (ed.) Actance et valence dans les langues de l'Europe. Berlin:
Mouton de Gruyter, 193-258.
Caldwell, Robert. 1856. A comparative grammar of the Dravidian or South-Indian family of
languages. London: Harrison.
Carnie, Andrew. 2005. "Some Remarks on Markedness Hierarchies: A Reply to
Aissen 1999 and 2003." Coyote Working Papers in Linguistics 14. (available from
http://dingo.sbs.arizona.edu/~carnie/Pages/Papers.html)
 12

Comrie, Bernard. 1977. Subjects and direct objects in Uralic languages: a functional
explanation of case-marking systems. tudes Finno-Ougriennes 12 (1975): 517.
Comrie, Bernard. 1981/1989. Language Universals and Linguistic Typology. Oxford:
Blackwell.
Croft, William. 1988. Agreement vs. case marking in direct objects. In Barlow, M. and
Ferguson, C. A., editors, Agreement in Natural Language: approaches, theories,
descriptions, pages 15980. Stanford: CSLI.
Croft, William. 2003. Typology and universals. 2nd ed. Cambridge: Cambridge
University Press.
de Hoop, Helen & Andrej Malchukov. 2005. "Case Marking Strategies." Paper
presented at the Workshop on Recent Developments in OT Syntax & Semantics,
Harvard University, 2 July 2005.
Dixon, R.M.W. 1994. Ergativity. Cambridge: Cambridge University Press.
Filimonova, Elena. 2005. "The noun phrase hierarchy and relational marking:
problems and counterevidence." Linguistic Typology 9.1: 77-113.
Garrett, Andrew. 1990. The Origin of NP Split Ergativity. Language 66: 261-296.
Greenberg, Joseph. 1966. Language Universals with Special Reference to Feature
Hierarchies, Mouton, The Hague.
Haspelmath, Martin. 2002. Understanding morphology. London: Arnold.
Haspelmath, Martin. 2006. "Against markedness (and what to replace it with)".
Journal of Linguistics 42.1
Haspelmath, Martin. 2008. Frequency vs. iconicity in the explanation of grammatical
asymmetries. To appear in Cognitive Linguistics
Jger, Gerhard. 2004. "Learning constraint sub-hierarchies: The Bidirectional Gradual
Learning Algorithm." In R. Blutner & H. Zeevat (eds.), Pragmatics in OT.
Palgrave MacMillan, 251-287.
Jger, Gerhard. 2007. "Evolutionary Game Theory and Typology: A Case Study."
Language 83(1), 74-109.
Jakobson, Roman. 1939. Signe Zro, Melanges de Linguistique Offerts Charles
Bally sous les Auspices de la Facult des Lettres de Luniversit de Genve por
des Collegues, des Confrres, des Disciples Reconnaissants, Georg et cie, s.a.,
Genve.
Kiparsky, Paul. 2008. "Universals constrain change, change results in typological
generalizations." In: Good, Jeff (ed.) Language change and language universals.
Oxford: Oxford University Press, 23-53.
Lazard, Gilbert. 2001. "Le marquage diffrentiel de l'objet." In: Haspelmath, Martin et
al. (eds.) Language typology and language universals: an international handbook. Vol.
II. Berlin: de Gruyter, 873-85.
McCarthy, John J. 2002. A thematic guide to Optimality Theory. Cambridge: Cambridge
University Press.
Morimoto, Y. & P. de Swart. 2005. "Modelling Synchrony and Diachrony of
Differential Object Marking." Paper presented at the Workshop on Recent
Developments in OT Syntax & Semantics, Harvard University, 2 July 2005.
Silverstein, Michael. 1976. "Hierarchy of features and ergativity." In: Dixon, R.M.W.
(ed.), Grammatical categories in australian languages, 112-171. Canberra:
Australian Institute of Aboriginal Studies.
Thompson, Alexander. 1909/1912. "Beitrge zur Kasuslehre I/IV." Indogermanische
Forschungen 24:293-307/30:65-79.
Zeevat, Henk & Gerhard Jger. 2002. "A reinterpretation of syntactic alignment." In D.
de Jongh, H. Zeevat and M. Nilsenova (eds.), Proceedings of the 3rd and 4th
International Symposium on Language, Logic and Computation. Amsterdam: ILLC.
(available from: http://www.uni-bielefeld.de/lili/personen/gjaeger/)