Cognitive and Computational Modeling of Handwriting
General Material Designation
[Thesis]
First Statement of Responsibility
Ellenblum, Gali
Subsequent Statement of Responsibility
Nozari, Nazbanou
.PUBLICATION, DISTRIBUTION, ETC
Name of Publisher, Distributor, etc.
The Johns Hopkins University
Date of Publication, Distribution, etc.
2019
GENERAL NOTES
Text of Note
152 p.
DISSERTATION (THESIS) NOTE
Dissertation or thesis details and type of degree
Ph.D.
Body granting the degree
The Johns Hopkins University
Text preceding or following the note
2019
SUMMARY OR ABSTRACT
Text of Note
Handwritten letters can be produced in many different ways - there is nothing inherent in the shape of a letter that dictates in which order and direction the strokes ought to be produced. For example, many people write uppercase 'A' starting at the top-left, while others choose to start it at the bottom-left. But uppercase 'A' could also be written starting with the horizontal line, yet this stroke pattern is never observed. Why are some stroke patterns observed while others are not? And how is one pattern chosen over others in a given instance? The systematicity in the observed productions of letters has brought researchers in the past (e.g., Goodman & Levine, 1973; Van Sommers, 1984), to propose that rules govern the way letters are written. But the rules that have been proposed, for example 'start at the top' or 'no pen lifts,' are often violated and occasionally come into conflict. In this dissertation, we present an account that can deal with these conflicts and with rule violations, using a novel application of Optimality Theory (OT; Prince & Smolensky, 1993). At the center of OT is the notion that rules can be violated, and that conflicts between them are resolved by a ranking of the rules. In addition to OT, we also implemented Harmonic Grammar (HG; Legendre, Miyata, & Smolensky, 1990a), a sister-framework to OT that requires only a weighting of the rules, rather than a ranking. We develop the theoretical framework needed to apply OT and HG to handwriting, and show that we can use these systems to model both prescribed and participant writing, in English and in Hebrew, by right-handed and left-handed individuals. We further use the framework of violable rules to resolve previously unsettled debates, and to shed light on some of the cognitive mechanisms underlying handwriting.