Calendar

In connection with the appointment of Johan Rooryck as honorary doctor at UiT Norway's Arctic University, the University Library is pleased to invite you to an open seminar.

Johan Rooryck is a linguist, particularly known for his work in promoting Open Access over many years as a journal editor and politically engaged. He is currently executive director of Plan S.

The lectures will be about open science in its entirety, and emphasize what the professional communities themselves can do to realize the ideals of openness and transparency in research and teaching.

PROGRAM

Library Director Johanne Raade Welcome

Professor Gillian Ramchand Johan Rooryck as a researcher and journal editor

Professor Tore Nesset Open Educational Resources, with My Russian journey as a case

Professor Laura Janda The need for Open Data Archives, with TROLLing as a case

First librarian Helene N. Andreassen The Open Handbook of Linguistic Data Management

University librarian Jan Erik Frantsvåg Helping Diamond Open Access grow: The DIAMAS Project

Doctor honoris causa Johan Rooryck Closing remarks

Recent work on phonological learning has questioned the traditional viewthat innate principles guide and constrain language development in children andexplain universal properties cross-linguistically. In this talk I focus on aparticular universal, the Sonority Sequencing Principle (SSP), which governspreferences among sequences of consonants syllable-initially. Experimentalevidence indicates that English, Mandarin, and Korean speakers exhibitsensitivity to the SSP even for consonant sequences that never occursyllable-initially in those languages (such as [nb] vs. [bn] in English). Thereis disagreement regarding the implications of this finding. Berent et al.(2007) argue that these results can only be explained with reference to aninnate principle; however, Daland et. al (2011) show that computational modelscapable of inferring statistical generalizations over sound classes can detectevidence for these preferences based on related patterns in the language input(and therefore no reference to innate principles is required). Building onthese studies, I argue that English is the wrong test case: it does notdifferentiate predictions of these two hypotheses. I examine learning ofsyllable structure phonotactics in Polish, a language with very differentsonority sequencing patterns from English. Polish provides a crucial test casebecause the lexical statistics contradict the SSP, at least in part. I reviewdevelopmental evidence indicating that children acquiring Polish arenonetheless sensitive to the SSP, producing larger sonority rises moreaccurately in spontaneous production (Jarosz 2017). I then present results fromtwo experiments investigating adult Polish native speakers’ phonotacticknowledge (Jarosz & Rysling 2017, in prep). The findings indicate thatPolish native speakers’ phonotactic preferences are sensitive to the SSP andthat this SSP sensitivity is not predicted by the computational models thatsucceeded for languages like English, Mandarin, and Korean. These resultssuggest a crucial role of an inherent bias or a constraint on generalizationfrom the input. However, in a recent twist, Nelson (2022) developed a newfamily of phonotactic learning models and showed that some of these models dosucceed in predicting Polish speakers' generalizations about sonoritysequencing. I discuss implications of these modeling results for future work.

Language acquisition proceeds on the basis ofincomplete, ambiguous linguistic input, and one source of this ambiguity ishidden phonological structure. Due to recent developments in computationalmodeling of phonological learning, there now exist numerous approaches forlearning of various kinds of hidden phonological structure from incomplete,unlabeled, and noisy data. These computational models make it possible toconnect the full representational richness of phonological theory with noisy,ambiguous corpus data representative of language learners' linguisticexperience to make detailed and experimentally testable predictions aboutlanguage learning and generalization. In this talk, I briefly review thesecomputational developments and then discuss two ongoing projects that utilizethese mutually-informing connections between computation, phonological theory,and experimental data to test hypotheses about the abstract representationsthat underlie phonological knowledge.