Resolution of structural ambiguities during real-time l2 sentence processing: Evidence from online and offline measures

The investigation of the ways second language (L2) learners process the L2 input has provided valuable insights into the understanding of human cognition and developing instructional interventions that would bring about better L2 learning. In this respect, the exploration of the mechanisms L2 learners employ to process linguistic data has allowed researchers to examine the aspects of L2 acquisition that still remain unresolved. Over the past few decades, researchers have addressed the investigation of L2 processing mechanism by referring to the way L2 learners resolve structural ambiguities real time. This dissertation also aims to draw a clear picture of the L2 sentence processing mechanism by comparing native English speakers and Turkish learners of L2 English at different proficiency levels in terms of their use of varios information sources and effects of transfer during the processing of two different structural ambiguities, namely, the processing of relative clause (RC) attachment ambiguities and prepositional phrase (PP) attachment ambiguities, through a series of offline and online experiments. For each type of ambiguity, the offline experiments tested the attachment preferences of L2 English learners and native English speakers via the use of pencil-and-paper comprehension questionnaires, whereas the online experiments investigated real-time processing of the target syntactic structures through the use of eye-tracking methodology. The offline experiment on the resolution of RC attachment ambiguities revealed that Turkish and English RC attachment preferences differed in that native English speakers preferred to attach the RC to the local noun phrase (low attachment preference), whereas the Turkish participants preferred the nonlocal NP to attach the RC (high attachment preference). Even though the same offline task fell short of detecting an attachment bias for the Turkish participants in the target language, English, the findings from a more sensitive online measure, the eye tracking task, indicated that the Turkish learners of L2 English tended to transfer the Turkish high attachment preference when processing English RC attachment ambiguities. As for the PP attachment ambiguities, the rationale was to investigate L2 learners' attachment preferences in situations when direct transfer effects from the learners' L1 are not probable as the equivalents of the PP attachment test items used in the offline task are not ambiguous in learners' L1, Turkish. This would render the opportunity to exclude the confounding factor of L1 transfer, which indeed substantially influenced L2 learners' processing decisions in RC ambiguity resolution. The results from both offline and online tasks indicated a clear low attachment bias for the native speakers and L2 learners groups, which suggests that L1 and L2 processing are both guided by structure-based processing strategies/biases in the computation of abstract relationships among syntactic constituents and that processing decisions can be made basing on the hierarchically deep structural information. The findings in general were interpreted to offer insights into several issues including how L2 grammatical processing could take place, whether and to what extent the transfer of L1 processing routines could influence real-time L2 sentence comprehension, whether it is possible for L2 learners to achieve native-like processing in the L2 and whether L2 processing mechanism is subject to a change with the increasing proficiency.