Season 3 (2021.01-02)

Part 1: 10 am–12 pm, Thursday January 28, 2021 [JST: Japan Standard Time]

Youngah Do (The University of Hong Kong)

"Variation awaiting bias: substantively biased learning of vowel harmony variation"

This study examines whether child learners alter a variable phonological pattern in an artificial language towards a phonetically-natural form. We address acquisition of a variable rounding harmony pattern through the use of two artificial languages; one with dominant harmony pattern, and another with dominant non-harmony pattern. Overall, children favor harmony pattern in their production of the languages. In the language where harmony is non-dominant, children’s subsequent production entirely reverses the pattern so that harmony predominates. This differs starkly from adults. Our results compare to the regularization found in child learning of morphosyntactic variation, suggesting a role for naturalness in variable phonological learning.

Adam McCollum (Rutgers University)

"Tutrugbu and the nature of unbounded feature spreading"

The nature of unbounded spreading patterns has been the topic of significant discussion. In particular, a number of authors have argued that unbounded feature spreading is myopic, meaning that spreading some feature(s) from x onto y in a string /x…y….z/ cannot depend in any way on z or other downstream information (Wilson 2003, 2006; McCarthy 2009; Mascaro 2019). This proposed constraint on feature spreading has generated several architectural revisions to Optimality Theory, specifically Wilson’s (2003) targeted constraints as well as Harmonic Serialism (McCarthy 2009). Furthermore, this proposed myopic nature of unbounded spreading has been marshalled to support the computational claim that phonology is subregular on the Chomsky hierarchy (Heinz 2011a,b, 2018). In this talk, I describe and analyze vowel harmony in Tutrugbu, a Ghana-Togo Mountain language of southeastern Ghana. In Tutrugbu, the application of regressive ATR harmony on prefix low vowels depends on the height of the initial-syllable vowel. In other words, ATR harmony in Tutrugbu is non-myopic, which in turn suggests that myopia is a tendency rather than a categorical constraint on feature spreading. Some implications for constraint-based and formal language theoretic analysis are discussed.

Part 2: 10 am - 12 pm, Thursday, February 11, 2021 [JST]

Juliet Stanton (New York University)

"Phonetic rhythm in -ization"

The rhythmic constraints *Clash and *Lapse are commonly assumed to evaluate syllable-sized constituents: a sequence of two adjacent stressed syllables (óó) violates *Clash, while a sequence of two stressed syllables, separated by two stressless syllables (óooó), violates *Lapse (see e.g. Prince 1983, Gordon 2002 for *Clash; Green & Kenstowicz 1995, Gordon 2002 for *Lapse). In this talk I propose that *Clash and *Lapse can be evaluated gradiently: speakers calculate violations off of a phonetically realized output representation. The closer the two stressed syllables, the greater the violation of gradient *Clash; the further away the two stressed syllables, the greater the violation of gradient *Lapse. Evidence for this claim comes from patterns of secondary stress in American English -ization: in this class of form, the inner suffix (-ize) is more likely to bear stress the further away it is from the rightmost stem stress. Supporting evidence comes from corpus, forced-choice, and production studies.

Amanda Rysling (University of California, Santa Cruz)

"Toward an account of the phonetic origins of vowel harmony"

Sound change has been a topic of eternal interest for phonologists. But constructing a full account of even one sound change is difficult, because an analyst must be able to explain both speakers' and listeners' behaviors as that sound change was taking place. Speakers must have had reason(s) to produce forms differently from how they had been producing them, and listeners must have accepted and represented those forms as different from how they had represented them before the change. That is, speakers must have provided acoustic evidence that listeners did not perceptually correct away, and the analyst must have a reason why both speakers and listeners acted as they did in order to explain why that language changed. In this talk, I present work in progress on an account of how vowel harmony likely arises in the phonologies of the world's languages. Speakers' contributions to phonologizing vowel harmony have long been understood: vowel-to-vowel coarticulation by speakers provides the best possible source of acoustic evidence from which the users of a language may begin to harmonize their vowels (Ohala, 1994; i.a.). The contribution of this talk, then, is to present a novel account of the behavior of a listener who cooperates in harmonizing vowels. I argue that perceptual generalization of the kind demonstrated by Chladkova and colleagues (Chladkova, 2014; Chladkova, Boersma, & Benders, 2015) — in which listeners use their knowledge of cues to a contrast between two categories in one part of a language's vowel space in order to begin to hear a new, analogous contrast in an as-yet unutilized part of that vowel space — provides the best evidence we have for how listeners begin to represent new vowel categories, which systematically differ from other vowels in the language in the same way as other pairs of vowels do. Once such analogously different pairs are representationally available to a listener, they can hear and learn that vowels are harmonized with each other across spans of vowel-to-vowel coarticulation. Listeners can then use this knowledge to become speakers with new articulatory targets, which encode vowel harmony even more strongly than the previous mere coarticulation from which they learned. This account makes predictions that need refinement, but which are not a priori implausible: whether a language will come to phonologize vowel harmony depends on (i) whether its vowel inventory has enough unutilized space in which analogous contrasts can be projected, and (ii) whether its realization of metrical prominence permits the kind of strong coarticulation necessary for acoustically harmonic spans to occur. Some evidence from the languages of the world which do have harmony, do not have harmony, or are not clear, suggests that these predictions are on the right track.