Use this URL to cite or link to this record in EThOS:
Title: Bridging between on-line linguistic adaptation and long-term language learning
Author: Janciauskas, M.
ISNI:       0000 0004 7656 3600
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Linguistic adaptation is a set of phenomena where language representations change in adults in response to linguistic input. Some of these adaptation effects are persistent and have been argued to be due to learning mechanisms that are similar to those used in language acquisition (Dell & Chang, 2013). Theories of language acquisition often assume that the creation of language-specific categories and structures involves bigger changes in linguistic representations than those required to support linguistic adaptation and it's not clear how their proposed mechanisms extend to explaining learning effects in adults. This thesis will examine the extent to which common mechanisms could explain both language acquisition and linguistic adaptation. In the first part of the thesis, we use a non-linguistic artificial grammar learning task to teach participants a simplified language over the course of an experiment. During the experiment, we examine the extent to which sentence-grain linguistic adaptation effects are related to the experiment-grain acquisition of the structural rules of the language. By linking these processes we provide behavioral evidence that short-term effects that occur at the level of sentence processing and longer-term effects that are characterized by structure acquisition are supported by a common mechanism. The second part of the thesis attempts to link the learning processes that take place in the artificial grammar learning tasks to the real-world language learning that takes place over the course of years and that result in long-term structure knowledge To do so, we use a computational model that is able to simulate the sentence/experiment grain changes to model how learners acquire a second language (L2). Since the data from L2 learners captures the changes that arise from years of real world learning, the model provides a link between sentence/experiment grain learning and year-grain learning. The final chapter links these two sections by examining how some of the behaviours observed in L2 speakers could be modeled within our artificial grammar learning paradigm. Chapter 1 provides an overview of the complexities that child language acquisition studies must explain and introduces the linguistic adaptation phenomena in adults highlighting the need to incorporate these processes into the theories of language learning. It will then describe the existing language acquisition theories focusing on whether they could explain linguistic adaptation effects like verb bias and abstract structural priming. The chapter will then proceed by describing a prediction-based statistical learning mechanism that is based on the assumptions of the connectionist model of language production, the Dual Path model (Chang, 2002). Referred to as Linguistic Adaptation Mechanism of Language Learning (LAMOLL), it will detail a potential unified account of language structure acquisition and linguistic adaptation in adults. To support this account, Chapter 2 will present 4 studies that use a nonlinguistic artificial grammar-learning (AGL) task to investigate if grammar acquisition and linguistic adaptations effects like structural priming could be elicited in a single grammar learning task. The studies will also test the prediction-based nature of this learning mechanism. Chapter 3 will shift attention to demonstrating that the same learning mechanism is active in children and adults. Second language studies show that people who start learning an L2 at a later age do not generally learn it to the same degree as native language learners do. This is often used to support the claim that language learning relies on language-specific learning mechanisms that are not active in adults. Such claims challenge the assumptions that the same mechanism that is responsible for language acquisition could support linguistic adaptation effects in adults. In this chapter we will explore the lack of the effect of years of language input on adult grammar knowledge reported in these studies employing a variety of methods ranging from corpus analysis and data reanalysis to modeling the results using the Dual Path model that motivated the use of AGL task to study language structure acquisition and adaptation effects in Chapter 2. Chapter 4 will report two experiments that will employ the AGL task developed in Chapter 2 to explore behaviorally the L2 effects observed in Chapter 3. This will test how the new AGL task generalized to new situations and its ability to explain linguistic effect as a method to study domain-general learning effects in language. Finally, Chapter 5 will summarize the findings from all studies discussing their implications and will provide some directions for the future studies.
Supervisor: Chang, Franklin ; Rowland, Caroline ; Lieven, Elena Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral