Use this URL to cite or link to this record in EThOS:
Title: Multiple object tracking supports thematic role features for language
Author: Jessop, Andrew Allen
ISNI:       0000 0004 7656 7783
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
A central feature of language is the ability to generalise beyond our previous experience to understand unfamiliar input and produce novel sentences (Hockett & Hockett, 1960). It is suggested that thematic roles support this ability by labelling the participants of an event with functions such as the agent and patient (Jackendoff, 1972). However, the nature of these roles is unclear, as language-based theories (e.g., Dowty, 1991) have often relied upon abstract conceptual features that are difficult to define or are not always available in the input. Yet, connectionist models of language have been unable to demonstrate systematic behaviour without including such role variables (Chang, 2002). This thesis aimed to address this issue by considering whether thematic roles in language are supported by the visual system, which allows viewers to track multiple objects in parallel (Pylyshyn & Storm, 1988) and identify their roles in causal events from their movement features (Scholl & Tremoulet, 2000). To test this connection, a total of nine experiments were conducted with adult participants using a modified version of the multiple object tracking paradigm. In this task, pairs of white circles engaged in causal push events. The participants were required to track the visually indistinguishable targets from these push events while they moved in random patterns for a short period, before producing a sentence to describe their interaction. Chapter 1 provides an introduction to these experiments. The function of thematic roles in reaching abstract representations of language is discussed, as well as issues with current theories regarding the nature of these roles. The visual perception of causality and the psychophysical features underpinning these impressions is presented as a potential solution, followed by a summary of the multiple object tracking literature. Chapter 2 reports the first three experiments of the thesis, which collectively provide evidence that the multiple object tracking system can support agent and patient features, even when the attentional demands of the task are manipulated. These studies found that the participants were able to retain the target identities to produce transitive sentences at test that accurately labelled the agent and patient of the interaction. Chapter 3 presents the results of a further six experiments, which used the same multiple push tracking task to examine the processing asymmetry between subject-extracted and object-extracted relative clauses (Gordon & Lowder, 2012). Here, the participants tracked visually identical targets that played the same role in two pushes or switched roles between the events. The results of these six experiments showed that visual events with consistent role-referent bindings are easier to retain and describe in language than those with no repetition or inconsistent bindings. This suggests that subject-extracted relative clauses may be easier to process since the subject referent plays the same role in both events. It also indicates that non-linguistic processing may contribute to the relative clause asymmetry, as the effect was observed when the studies also varied the sentence structures the participants used to describe the events. Finally, chapter 4 discusses the implications of these findings and whether thematic roles are supported by multiple object tracking in vision.
Supervisor: Chang, Franklin ; Bertamini, Marco Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral