Object Asymmetry in Korean

In this paper(fn1), I argue that Korean exhibits object asymmetry by showing that there is only one argument that demonstrates the "primary object" syntactic properties of case marking, reciprocalisation and the passive, despite clauses with more than one argument bearing accusative marking such as in the morphological causative construction.(fn2)

Evidence for object asymmetry comes from case marking, reciprocalisation and the passive. One instance that exhibits object asymmetry in Korean is case marking. The primary object obligatorily has the accusative marker; no other case marker can replace it, as shown in (1).(fn3) The accusative marker, on the other hand, can mark arguments other than primary objects, as shown in (2). The accusative marking in (2) associates with the semantic notion of 'affectedness'(fn4), as Yang (1998) suggests; that is, the primary object bears the accusative marking by default, but the accusative marker does not necessarily indicate a primary object.

(1)	Bill-i	John-*eykey/ul	ttayli-ess-ta.
	Bill-N	John-*D/A	hit-PA-DEC
	'Bill hit John.'

(2)	a.	Mary-ka	maru-eyse/lul	ki-ess-ta.
		Mary-N	floor-L/A	crawl-PA-DEC
		'Mary crawled on the floor.'
	b.	Mary-ka	tol-lo/ul	tokki-lul/*lo	matul-ess-ta.
		Mary-N	stone-I/A	ax-A/*I	make-PA-DEC
		'Mary made the ax with the stone.'
	c.	John-i	Bill-eykey/ul	chaky-*eykey/ul	cwu-ess-ta.
		John-N	Bill-D/A	book-*D/A	give-PA-DEC
		'John gave the book to Bill.'

Reciprocalisation also displays object asymmetry in Korean because, while an indirect object cannot be reciprocalised, a primary object can, as shown in (3b) and (3c), even if it may bear accusative marking:

(3)	a.	John-i	kak	ai-eykey	selo-uy	chayk-ul	cwu-ess-ta.
		John-N	each	child-D	each other-G	book-A	give-PA-DEC
		'John gave each child each other's book.'
	b.*	John-i	selo-uy	ai-eykey	kak	chayk-ul	cwu-ess-ta.
		John-N	each other-G	child-D	each	book-A	give-PA-DEC
		INT: John gave each child each other's book.
	c.*	John-i	selo-uy	ai-lul	kak	chayk-ul	cwu-ess-ta.
		John-N	eachother-G	child-A	each	book-A	give-PA-DEC
		INT: John gave each child each other's book.

Another instance of object asymmetry in Korean is shown by the passive. The primary object can be the passive subject, as shown by (4a). However, the accusative marked indirect object cannot be, as shown in (4b):

(4)	a.	chayk-i	(John-eyuhaye)	Bill-eykey/*ul	cwu-e	ci-ess-ta.
		book-N	John-by	Bill-D/*A	give-COMP	become-PA-DEC
		'The book was given to Bill.'
	b.*	Bill-i	(John-eyuhaye)	chayk-i/ul	cwu-e	ci-ess-ta.
		Bill-N	John-by	book-N/A	give-COMP	become-PA-DEC

As shown by case marking, reciproclisation and the passive, Korean is an object asymmetry language because it has only one primary object in a single clause. The claim extends to the morphological causative construction, which is monoclausal.(fn5) In the morphological causative construction of a dyadic verb, the causee and the non-causee arguments may bear accusative making, as in (5b). The accusative-marked causee may associate with the semantic notion of 'affectedness', compared to the dative-marked causee.

Yet, the morphological causative construction is consistent with the claim, just as in non-causative double accusative-marked clauses. That the object of the base verb bears accusative but not any other case marking by default, as shown in (5b), suggests that it is the primary object in the clause:

(5)	a.	ai-ka	sakwa-lul	mek-ess-ta.
		Child-N	apple-A	eat-PA-DEC
		'The child ate the apple.'
	b.	Mary-ka	ai-eykey/lul	sakwa-*eykey/lul	mek-I-ess-ta.
		Mary-N	child-D/A	apple-*D/A	eat-CAUS-PA-DEC
		'Mary fed the child the apple.'

Reciprocalisation also shows us that the object of the base verb is the primary object because it can be reciprocalised, as shown in (6a):

(6)	a.	Mary-ka	kak	ai-eykey	selo-uy	sakwa-lul	mek-I-ess-ta.
		Mary-N	each	child-D	each other-G	apple-A	eat-CAUS-PA-DEC
		'Mary fed each child each other's apple.'
	b.*	Mary-ka	selo-uy	ai-eykey	kak	sakwa-lul	mek-I-ess-ta.
		Mary-N	each other-G	child-D	each	apple-A	eat-CAUS-PA-DEC
		INT: Mary fed each child each other's apple.'
	c.*	Mary-ka	selo-uy	ai-lul	kak	sakwa-lul	mek-I-ess-ta.
		Mary-N	each other-G	child-D	each	apple-A	eat-CAUS-PA-DEC
		INT: Mary fed each child each other's apple.

In the passive of a morphological causative, the "goal" argument cannot bear accusative marking, as shown in (7): the causee has the dative marker by default:

(7)	a.	sakwa-ka	(Mary-eyuhaye)	ai-eykey/*lul	mek-I-e	ci-ess-ta.
		Apple-N	Mary-by	child-D/*A	eat-CAUS-COM	become-PA-DEC
		LIT: The apple was eaten to the child (by Mary).
	b.*	ai-ka	(Mary-eyuhaye)	sakwa-ka/ul	mek-I-e	ci-ess-ta.
		child-N	Mary-by	apple-N/A	eat-CAUS-COMP	become-PA-DEC

The syntactic opaqueness of a construction with more than one argument bearing accusative marking conforming to these patterns supports the claim that Korean is an object asymmetry language. The fact that Korean has only one primary object can be captured in the theory of AOP (Asymmetrical Object Parameter) in Bresnan and Moshi (1996) and Alsina (1996), because there is only one internal argument that is semantically unrestricted, having [-r] which maps onto a primary object. The asymmetries of double accusative-marked objects can be accounted for if we assume that the semantically restricted argument associated with the semantic notion of [+affectedness] yields double accusative marking at the level of constituent structure, but its grammatical function remains invariant at the level of functional structure: that is, the accusative marker on the causee associates with the semantic notion, but does not necessarily associate with the grammatical function of direct object; case marking may vary in the given context, while case feature remains still.

References

• Alsina, A. and S. Joshi. 1991. Parameters in Causative Construction. Proceedings of the Chicago Linguistic Society 27: 1-15. Alsina, A. 1996. Passive Types and the Theory of Object Asymmetries. NLLT 14: 673-723.
• Baker, M.C. 1988. Incorporation: A Theory of Grammatical Function Changing. Chicago: The University of Chicago Press.
• Bresnan, J. and L. Moshi. 1990. Object Asymmetries in Comparative Bantu Syntax. LI 21: 145-185.
• Hong, Ki-Sun. 1991. Argument Selection and Case Marking in Korean. Doctoral Dissertation. Stanford, CA: Stanford University.
• Mohanan, T. 1988. Causatives in Malayalam. Stanford, CA: Stanford University. Department of Linguistics. MS.
• Mohanan, T. 1990. Argument Structure in Hindi. Doctoral Dissertation. Stanford, CA: Stanford University.
• Yang, Insun. 1998. Object Marker in Korean. The Twenty-Fourth LACUS Forum: 238-250.
• Zaenen. A., J. Maling, and H. Thrainsson. 1990. Case and Grammatical Functions. Syntax and Semantics 24: Modern Icelandic Syntax: 95-134.

1. The earlier version of the paper of this abstract appeared in the proceedings of the LSK International Conference held in Seoul, Korea, in August 2002, and since then has been revised based on feedback I received from the audience at that conference. I thank the audience for their comments on the word order test in the earlier version.
2. I would like to thank Arto Anttila, Vivienne Fong, K.P. Mohanan, and my supervisor Tara Mohanan for their valuable comments and suggestions on the earlier version. All the shortcomings or mistakes are of course my own.
3. Abbreviation: N: Nominative D: Dative A: Accusative I: Instrument L: Locative PA: Past DEC: Declarative CAUS: Causative morpheme COMP: Complementiser
4. For the semantic notion of 'affectedness', please refer to Yang (1998).
5. Evidence for the monoclausality of the morphological causative comes from patterns of subject honorification, reflexive binding, the domain of negative polarity items, and control.