Markedness and the acquisition of preposition pied-piping and stranding: A longitudinal perspective.

Abstract

This thesis represents an exploratory study of longitudinal performance data of second language acquisition, augmented by sentences from a grammaticality judgement task. The focus is on preposition pied-piping and stranding constructions as documented in a generative framework. The purpose of the study was to investigate these data within in the context of markedness theory associated with Universal Grammar in order to determine whether or not the postulated developmental sequence was reflected in the data. Preposition pied-piping and stranding have been investigated recently within this framework in cross-sectional studies and have provided researchers with inconclusive results. For the purposes of this thesis it was hypothesized that longitudinal data would provide clarification of the development and use of these constructions in the acquisition of English as a second language by adult learners. It was further assumed that the longitudinal study would permit additional insight into the acquisition of the constructions concerned that would lead to the formulation of hypotheses for future research. In addition to the longitudinal data from the learners, data from native speaker control groups were collected in order to validate the authenticity of the constructions the learners provided and to ensure that preposition stranding is used as anticipated in informal interaction. The results provide support for the developmental sequence postulated by markedness theory. They show that the learners go through a brief stage during which they accept and produce preposition pied-piping, but explicitly reject stranding. The analysis of the data shows that the five learners use the constructions under investigation less frequently than might be expected. This seems to suggest that transfer from the native languages involved, which have the unmarked but not the marked construction, is not an important factor although it cannot be ruled out. Some learners produced stranded versions but rejected them on the grammaticality task. This combined with the low frequency and the limited syntactic and lexical range of the utterances suggests that the learners produced at least some of the early stranded versions as unanalyzed chunks. Evidence of avoidance of the pied-piping and stranding constructions was also found. The native speaker control groups, who were given pied-piped and stranded versions of each of the learner utterances in order to decide which one they would likely have chosen, strongly favoured stranded constructions over pied-piped ones in most instances. In conclusion, the data provide support for the developmental sequence postulated in a theory of markedness. They also suggest a number of potential hypotheses for future research.