This study examines the acquisition of head-final relative clauses (RCs) in Turkish, by monolingual children, with attention given also to the acquisition of correlative RC constructions and adverbial clauses. The results of experimental tasks (based on elicited production) with monolingual children (mean age 4;6) are used to draw generalizations on the acquisition of head-final RCs in Turkish. Our results show that children perform better with Subject-RCs than with Object-RCs and worst with Postpositional RCs for head-final RCs in Turkish, with respect to: (i) more correct RC responses, (ii) fewer RC-related ungrammatical responses; (iii) fewer avoidances of RC structure. These results support a hierarchical approach to the syntactic structure of RCs, challenging linear sequence-based approaches. We further reassess data in the CHILDES database, which had formed the basis of claims in Slobin (1986) and related work, claiming that the acquisition of head-final RCs (which are nominalized and gapped) in Turkish lags considerably behind the acquisition of Indo-European head-initial, tensed RCs. The supposedly late age in acquiring gapped, nominalized, head-final RCs is ascribed to a hypothesized difficulty in parsing, and thus acquiring, RCs (and embedded clauses in general) which have morpho-syntactic properties different from root clauses, e.g. nominalization markers on the verb instead of tense markers, different (“possessive”) agreement markers, gaps corresponding to the RC-head, and genitive subjects rather than nominative ones. A reconsideration of these data aligns neatly with our experimental results, showing that Turkish children (even around age 4;6, i.e. relatively early) exhibit competence in head-final, nominalized headed as well as non-headed RCs at an age comparable to children acquiring Indo-European head-initial and nonnominalized RCs. In addition, we show instances of nominalized adjunct (i.e. adverbial) clauses which are used correctly by the children. Yet another type of construction found in CHILDES concerns correlative RCs: Those are not nominalized (thus have nominative rather than genitive subjects, their verbs have “finite” conditional morphology and finite agreement forms rather than “possessive” agreement), have no gaps, and do have a “wh like” pronoun in the target’s position. This makes them look very much like simple clauses in Turkish, as well as like Indo-European non nominalized RCs. Based on Slobin’s assumptions, they should be acquired earlier than their nominalized counterparts. However, the database offers no evidence of this; both from the point of view of numbers of tokens and from that of age when used in conversation, correlative RCs are similar to nominalized pre-nominal gapped RCs. Such data shed additional doubt on morphology-centered claims attributing any acquisitional difficulty to the nominalization and genitive subject case of embedded clauses in Turkish.
Lunedì 27 maggio 2019, ore 14.30
Sala Lauree del Dipartimento di Psicologia, U6 (3° piano)
Tutti gli interessati sono invitati a partecipare.
Prof. Claudio Luzzatti