Publication:
Restrictive Tier Induction

Simple item page
dc.contributor.advisorMichael Becker
dc.contributor.advisorGaja Jarosz
dc.contributor.authorKim, Seoyoung
dc.contributor.departmentUniversity of Massachusetts Amherst
dc.date2024-03-27T18:46:38.000
dc.date.accessioned2024-04-26T15:55:41Z
dc.date.available2024-04-26T15:55:41Z
dc.date.submittedSeptember
dc.date.submitted2022
dc.description.abstractThis dissertation proposes the Restrictive Tier Learner, which automatically induces only the tiers that are absolutely necessary in capturing phonological long-distance dependencies. The core of my learner is the addition of an extra evaluation step to the existing Inductive Projection Learner (Gouskova and Gallagher 2020), where the necessity and accuracy of the candidate tiers are determined. An important building block of my learner is a typological observation, namely the dichotomy between trigram-bound and unbounded patterns. The fact that this dichotomy is attested in both consonant interactions and vowel interactions allows for a unified approach to be used. Another important piece of information is that only unboundedness implies trigram-boundedness, and not vice versa. These typological observations together shed light on the critical role of trigrams in phonological learning. The premise that there is no other distance at which a restriction holds than these two lets us safely assume that searching only up to trigrams might actually be a near-exhaustive search for local interactions. On top of that, the fact that interaction beyond a trigram window, which we need tiers for, always implies interaction within a trigram window guarantees that all necessary tiers can be discovered by looking at trigram constraints. Hence, a learner can confidently search up to trigrams for local interactions and expand its search for non-local ones from the discovered trigrams. I present several case studies to test the abilities of the Restrictive Tier Learner in capturing various long-distance dependencies that are attested in natural languages. The current version of the learner maintains all the strengths of the previous learning algorithms while showing improved performance in critical cases.
dc.description.degreeDoctor of Philosophy (PhD)
dc.description.departmentLinguistics
dc.identifier.doihttps://doi.org/10.7275/30930862
dc.identifier.orcidhttps://orcid.org/0000-0001-6334-1644
dc.identifier.urihttps://hdl.handle.net/20.500.14394/18993
dc.relation.urlhttps://scholarworks.umass.edu/cgi/viewcontent.cgi?article=3746&context=dissertations_2&unstamped=1
dc.source.statuspublished
dc.subjectlong-distance dependencies
dc.subjectcomputational phonology
dc.subjectphonological tier
dc.subjectphonotactic learning
dc.subjectComputational Linguistics
dc.subjectPhonetics and Phonology
dc.titleRestrictive Tier Induction
dc.typeopenaccess
dc.typearticle
dc.typedissertation
digcom.contributor.authorisAuthorOfPublication|email:seoyoungkimk@gmail.com|institution:University of Massachusetts Amherst|Kim, Seoyoung
digcom.identifierdissertations_2/2641
digcom.identifier.contextkey30930862
digcom.identifier.submissionpathdissertations_2/2641
dspace.entity.typePublication
Files
Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Restrictive_Tier_Induction.pdf
Size:
580.81 KB
Format:
Adobe Portable Document Format
Name: Restrictive_Tier_Induction.pdf
Size: 580.81 KB
Kind: Adobe PDF
Collections
Dissertations