Larisa Avram, born in Bucharest, Romania, in 1959. Ph.D. from the University of Bucharest. Associate Professor at the Faculty of Foreign Languages at the University of Bucharest.
Mellon Fellow (1 September 2002 - 31 January 2003)
During my five-month stay at NIAS, I began work on a book on the acquisition of the Romanian language. I focussed on the emergence and development of Romanian determiners and pronominal clitics with a view to answering the following questions:
- Which developmental patterns characterise the acquisition of Romanian determiners and object clitics
- Are these patterns comparable to those identified for other languages?
- Can we trace early clitic omission and indefinite determiner omission to the same cause?
- How can we account for the identified patterns within current theoretical models?
Analysis revealed similarities to the patterns identified for Italian articles and clitics (Caselli et al. 1993), Swedish articles (Bohnacker 1997, Santelmann 1998) and for German DP (Penner & Weissenborn 1996, Eisenbeiss 2000). The following developmental route in Romanian was identified: a) definite determiners seem to be present from the onset of language acquisition; b) both indefinite determiners and pronominal clitics are completely absent during the early stages; c) indefinite determiners and clitics emerge at the same time; d) there is a time-lapse between the emergence of indefinite determiners and pronominal clitics and their adult usage (both clitics and indefinite determiners continue to be randomly omitted or, when used, do not always have the correct morphological form).
The developmental pattern of pronominal clitics and determiners that we found could be explained by the delayed implementation of morphological knowledge, in particular agreement morphology. The child 'selects' the agreement feature (or all the features comprised by agreement) at a very early stage but the valuation of this feature is delayed. The result is early under-specification of Agreement. The claim is that the under-specification of Agreement derives from restricted derivational space in the early stage of language acquisition. To value the agreement features requires a multiple Spell-Out derivation (Uriagereka 2002), i.e. it requires a large derivational space that is not available at the outset of acquisition. This means that valuation of the agreement feature(s) is delayed. I believe this delay is the cause of the early omission as well as early agreement errors of articles and clitics. The valuation of one feature in one domain may reflect on other domains. This seems to be the case with agreement and clitics. The valuation of the agreement feature(s) will percolate into the clitic system, triggering the emergence of these pronominal elements. Once the child has acquired the complete set of agreement features in the nominal domain, s/he will start using determiners in an adult-like fashion and Accusative clitics emerge. Initially the child acquires agreement in the nominal domain, which explains the higher percentage of correct determiners at an early stage.
This work is the basis for a chapter in a forthcoming book entitled: 'Feature selection and feature valuation in acquisition'.