LINGUIST List 15.600

Sat Feb 14 2004

Diss: Psycholing: Somashekar: 'DEVELOPMENTAL...'

Editor for this issue: Takako Matsui <takolinguistlist.org>

Directory

1. ss71, DEVELOPMENTAL TRENDS IN THE ACQUISITION OF RELATIVE CLAUSES

Institution: Cornell University
Program: Human Development
Dissertation Status: Completed
Degree Date: 1999

Author: Shamitha Somashekar

Dissertation Title: DEVELOPMENTAL TRENDS IN THE ACQUISITION OF
RELATIVE CLAUSES: CROSS-LINGUISTIC EXPERIMENTAL STUDY OF TULU
	
Linguistic Field: Psycholinguistics, Syntax, Cognitive Science,
Language Acquisition 

Subject Language: Tulu (code: TCY)
	
Dissertation Director 1: Barbara Lust
Dissertation Director 2: James Gair
Dissertation Director 3: Claire Cardie
	
Dissertation Abstract:

Conflicting theories exist on the fundamental nature of processes
involved in first language acquisition. Some claim that children
induce specific-language-grammar (SLG) from input-data using
principles operating directly on data. Others argue that children,
guided by abstract principles and parameters provided by Universal
Grammar (UG) construct SLG using actual data only indirectly. This
dissertation examines these claims from a cross-linguistic perpective,
studying relative clause (RC) structure and acquisition in Tulu, a
left-branching, Dravidian, SOV (Subject-Object-Verb) language, in the
light of previous acquisition studies in other languages.
	
After various forms of Tulu RCs were grammatically analyzed,
production of four kinds of RCs was tested in two experiments with 60
children from Mangalore, India, (age range: 2 yrs; 5 months to 6 yrs;
6 months, mean 4 yrs; 3 months), using controlled experimental design.
We varied overtness of relative operator, and form of RC verb
(fully-inflected canonical verb, or one lacking certain
inflection-features). We also varied whether the RC was gapped or not,
was adjacent to its head or not, and was in canonical-SOV form or not.
The design allowed contrast of relatively 'data-transparent'
structures to 'grammar-transparent' structures, i.e., most accessible
to certain UG principles. Specifically, the design ay"lowed analysis
of knowledge of inflection, and capacity to integrate infle�tion
components required for particular RC forms.
	
Results reveal that (1) Less data-transparent structures were most
accessible in RC development. (2) Children provided grammatically
correct analysis of inflection-features, integrating them according to
RC structure produced. (3) Cross-linguistic analyses showed first
language acquisition of certain Tulu-RC structures remarkably in
advance of development of English and French lexically-headed
restrictive relatives tested previously.
	
These results argue that children in first language acquisition do not
merely operate on surface-data inductively, but are guided by
grammatical principles which may be universal, in their hypotheses
about particular constructions in a language, to construct SLG.
Evidence supports not "data-transparency" but "grammar-transparency",
not "data-mapping" (from surface-data to SLG) but "grammar-mapping"
(to SLG from universally available grammatical principles).
Theoretically, results provide specific evidence that children in the
"initial-state" integrate knowledge of CP with language-specific
aspects of DP and "WH" in acquisition of relativization.