SIGMORPHON 2019 Election Statements

The SIGMORPHON constitution requires that the members of the executive board be elected by the SIGMORPHON community every 2 years. Each candidate was asked to provide a short bio, as well as a statement describing why their election would benefit SIGMORPHON in the upcoming two years. These statements are provided below.




Jason Eisner

Jason Eisner is Professor of Computer Science at Johns Hopkins University. He is a Fellow of the Association for Computational Linguistics (and an action editor of its TACL journal). At Johns Hopkins, he is also affiliated with the Center for Language and Speech Processing, the Machine Learning Group, the Cognitive Science Department, and the national Center of Excellence in Human Language Technology. His goal is to develop the probabilistic modeling, inference, and learning techniques needed for a unified model of all kinds of linguistic structure. His 125+ papers have presented various algorithms for parsing, machine translation, and weighted finite-state machines; formalizations, algorithms, theorems, and empirical results in computational phonology; and unsupervised or semi-supervised learning methods for syntax, morphology, and word-sense disambiguation. He is also the lead designer of Dyna, a new declarative programming language that provides an infrastructure for AI research. He has received two school-wide awards for excellence in teaching.


Garrett Nicolai

Garrett Nicolai is a post-doctoral fellow at Johns Hopkins University. He obtained his PhD in Computing Science from the University of Alberta in 2017. His dissertation concerned the development of methods to transition between stems, lemmata and inflected forms, and he continues to be interested in inflectional phenomena. Recently, his work has concentrated on sub-word level transduction in low-resource settings, and information discovery with minimal and distant supervision. He has been the secretary of SIGMORPHON since 2017, was co-chair of the 2018 SIGMORPHON workshop, and is co-chairing the 2019 workshop.


Amir Zeldes

Amir Zeldes is a computational linguist specializing in work on and with corpora, including corpus linguistics studies, building freely available corpora, and creating open source annotation interfaces and NLP tools that make corpus creation easier. Following his thesis work on productive argument filling in morphology and syntax, he has taught Computational Linguistics at Georgetown University since 2014, where he leads the Corpling@GU lab. He has worked on a variety of topics including discourse relations, analyzing compounds, NLP and lexicography for low resource languages, coreference and morphological segmentation in Afro-Asiatic languages. His latest book, Multilayer Corpus Studies, explores the use of multiple concurrent corpus analyses to create multifactorial theoretical models and practical tools.


Though I am a newcomer to SIGMORPHON, I have been a member of the ACL since 2009, and I am interested in strengthening and contributing to SIGMORPHON as an at-large member of the executive committee. The current moment in Computational Linguistics is a very interesting one, with rapid progress in some areas and the availability of more resources than ever for a variety of languages. However this progress has not benefitted work on all languages equally. If elected, I would promote activities that bolster inclusiveness and cross-linguistic initiatives, such as the release of open source datasets in low resource languages and infrastructure to support them, organization of multilingual shared tasks such as the recent task on Universal Morphological Reinflection, and the explicit consideration of linguistic diversity as a positive evaluation criterion in reviewing. I believe that standardization efforts and open sharing and pooling of resources are crucial for progress in multi- and cross-linguistic NLP, and that SIGMORPHON should contribute to this development by involving researchers with shared interests but different domains of expertise in the creation of guidelines, and bringing language documentation and computational work closer together to promote the ACL’s role in supporting and protecting cultural heritage.

Elliot Moreton

Elliot Moreton is a phonologist and phonetician in the Department of Linguistics at UNC-Chapel Hill. His primary research focus is the comparative study of inductive bias in human learners when faced with structurally isomorphic problems across different domains such as phonology, morphology, and vision, by means of lab experiments and computational and mathematical modelling. He received a PhD in Linguistics in 2002 from the University of Massachusetts-Amherst. Past and current research affiliations include Brain and Cognitive Science at MIT, Cognitive Science at Johns Hopkins, and NTT Communication Science Research Labs.


Computational phonology has been expanding rapidly in recent years among theoretical linguists, owing to renewed interest in formal-language theory and to advances in the synthesis of machine-learning ideas with phonological theory. This creates opportunities for dialogue and collaboration across disciplines, and as a committee member, I would work to make SIGMORPHON a venue for them. Shared tasks, for instance, are not yet a fixture in theoretical phonology, but SIGMORPHON could make them so.

Gaja Jarosz

Gaja Jarosz is a computational and theoretical phonologist whose research falls at the intersection of phonological theory, statistical machine learning, and first language acquisition. A primary focus of her research is developing formal models of phonological learning grounded in probability and statistical learning theory and applying them to study the cognitive representations and biases underlying first language acquisition of phonology. One major thread of her work focuses on applying and adapting statistical NLP methods to model learning of the hidden representations (e.g. prosodic structure, word/morpheme boundaries, rules/constraints, underlying representations, exceptionality) posited in modern phonological theory from data that is as representative of children’s linguistic exposure as possible. A complementary thread in her work is using these and other computational models to examine learning biases in first language acquisition (experimentally and through analysis of corpora). Dr. Jarosz received her PhD in Cognitive Science from the Johns Hopkins University in 2006. She was assistant (and later associate) professor of Linguistics at Yale University (2007-2015). Since 2015, Dr. Jarosz has been an associate professor in the Department of Linguistics at the University of Massachusetts Amherst. In 2007, she co-founded the regional computational phonology conference NECPhon, now in its twelfth year, and in 2017 Jarosz co-founded the Society for Computation in Linguistics.


Integration of statistical NLP, linguistic theory, and psycholinguistics is central to my research program, and I have been working to build connections across these disciplines locally, at my home institutions (both Yale and UMass), through integrative courses and interdisciplinary research groups and student mentoring for many years. More recently, I co-founded and co-organized the first two annual meetings of the Society for Computation in Linguistics, whose central aim is to build a bridge from linguistics to NLP/ML. SCiL is co-located with the LSA, hosts invited speakers and panelists from the ACL community, and publishes paper-reviewed proceedings with the ACL Anthology. It has been very successful in attracting submissions and participants from both communities, and we expect it to continue growing next year. As a member of the SIGMORPHON executive committee, I would likewise support efforts to increase scientific exchange between the SIGMORPHON and linguistics communities.

Géraldine Walther

Géraldine Walther is a postdoctoral researcher at the university of Zurich with a PhD from the Université Paris Diderot in theoretical and computational linguistics (2013). She has been part of the organising committee for the CoNLL-SIGMORPHON shared tasks of 2017 and 2018. Géraldine Walther’s research is concerned with the dimensions and degree of variability of linguistic systems. She is investigating questions in morphology, syntax, and the lexicon from both computational and cognitive points of view. She has developed PARSLI, a formalised model of inflectional morphology with emphasis on the lexicon, and co-designed a computational implementation of the model, AlexinaPARSLI (with Benoît Sagot, Inria Paris). Her recent research has been pursuing avenues based on insights from learning theory, that take an explicitly discriminative perspective on linguistic form and function. Géraldine Walther’s work is largely based on the development and exploitation of electronic lexical resources and corpora and quantitative corpus studies. She has developed electronic lexical resources for close to 10 different languages, including previously underdocumented ones such as Tuatschin (Rhaeto-Romance) and Khaling (Kiranti), and she is currently developing large documentation and language acquisition corpora for Tuatschin. Some of those resources are based on her own original fieldwork and data collection.


My research lies at the interface between computational linguistics, cognition, and linguistic typology. Within SIGMORPHON, I could contribute to bridging initiatives across computational morphology and linguistic research, notably involving the rapidly emerging community of quantitative morphology (Blevins, Milin, Kostić, Filipović-Đurđević, Ackerman, Ramscar, Malouf…), in which I am strongly involved. I would seek to further and advertise the development of joint enterprises, combining cognitively grounded linguistic questioning and the development and use of of state-of-the-art computational methodologies. I would seek to promote discussions and collaborations involving initiatives related to computational morphology for small-resource languages and the language documentation community, of which I am also part: such languages yield a range of interesting and complex patterns that are bound to challenge the development of novel methodologies within computational morphology.

Kyle Gorman

Kyle Gorman is assistant professor of linguistics at the Graduate Center at the City University of New York and director of the master’s program in computational linguistics. He also works as a software engineer at Google, and previously was an assistant professor at the Center for Spoken Language Understanding at the Oregon Health & Science University in Portland, OR. He holds a PhD from the University of Pennsylvania, where he was advised by Charles Yang. His research interests includes computational approaches to phonology and morphology, and finite-state techniques in particular. He is the creator of the Pynini grammar compilation library.


I strive to promote three components of the SIGMORPHON enterprise. The first, which is already underway thanks to efforts by current members of the executive commitee, is to help develop high-quality standardized data sets and tasks, and promote their widespread use as benchmarks for sequence modeling. Secondly, I hope to bridge the now-substantial gap between morphophonological analysis as performed by linguists, and the computational modeling of morphology as in the SIG’s shared tasks. This is an issue I address in both in-progress and forthcoming work using error analysis. Finally, I hope to use these shared tasks and linguistically-informed error analysis so that this SIG’s meetings and tasks can attract linguists interested in the nature of words from other perspectives.

Mans Hulden

Mans Hulden is an assistant professor of Linguistics at the University of Colorado, Boulder. His research interests focus on modeling and learning of phonological and morphological structure. He has developed a number of finite-state tools, among them foma, which has been used for producing morphological analyzers for several dozen languages. He is secretary of SIGFSM, the ACL special interest group on finite state methods, and serves on its international advisory committee. He was a co-organizer of the SIGMORPHON and CoNLL-SIGMORPHON shared tasks in 2016, 2017, and 2018.


The scope of SIGMORPHON is wide and covers computational approaches of two to three traditional core subfields of linguistics. As Executive Committee member, one of my main goals would be to attempt to conserve the presence of SIGMORPHON-dedicated areas at relevant meetings of the *ACL/EMNLP and to maintain and encourage a healthy and inclusive balance between phonetics, phonology, and morphology in our own meetings and workshops.

As a relatively small SIG, one of the things we can do to maintain and promote visibility is through strategic co-operation between SIGMORPHON and other SIGs. Earlier efforts such as organizing joint events with, for example, SIGFSM, or joint shared tasks with CoNLL (2017 and 2018) are all examples of ways to maintain an active and healthy SIG, and efforts that I strongly support.

One of the recent positive developments in the field of computational linguistics in general is an increased emphasis on reproducible research. This is also happening in SIGMORPHON-related subfields, and is a good thing since it encourages researchers to release relevant code and data. Efforts to catalogue such resources and make them visible and easily accessible through SIGMORPHON would be a high-priority item for the SIG, and I would actively support such developments.

Ryan Cotterell

Ryan is a lecturer of computer science at the University of Cambridge. He specializes in natural language processing, computational linguistics and machine learning, focusing on deep learning and statistical approaches to phonology, morphology, linguistic typology and low-resource languages. He will receive his Ph.D. in Spring 2019 from the computer science department of the Johns Hopkins University, where he was affiliated with the Center for Language and Speech Processing; he was co-advised there by Jason Eisner and David Yarowsky. He has received best paper awards at ACL 2017 and EACL 2017 and two honorable mentions for best paper at EMNLP 2015 and NAACL 2016. Previously, he was a visiting Ph.D. student at the Center for Information and Language Processing at LMU Munich supported by a Fulbright Fellowship and a DAAD Research Grant under the supervision of Hinrich Schütze. His PhD was supported by an NDSEG graduate fellowship, the Fredrick Jelinek Fellowship, and a Facebook Fellowship.


I propose to push for two things if reelected.

  1. Given the growing interest in character-level NLP, we should host a workshop every year. Papers that we could be publishing are taken away by SCLeM and other [] workshops. We have a long tradition of high-quality papers that do sub-word work "before it was cool", to quote Sharon Goldwater.
  2. We should organize a shared task every year to push for more open-source phonology and morphology datasets. I am happy to take the lead in organizing these, if I remain on the committee.

Sandra Kübler

Sandra Kübler is a Professor of Computational Linguistics in the Department of Linguistics at Indiana University. She received her PhD in 2003 from the University of Tübingen, Germany. Her main research interest is in parsing morphologically rich languages and the interaction of POS and morphological information with parsing. She has also worked on coreference resolution and sentiment analysis. She is on the editorial board of Natural Language Engineering, was an area chair for IJCNLP in 2013, and has co-organized a range of workshops co-located with ACL and EACL, including the last two SIGMORPHON workshops.


SIGMORPHON is a SIG that brings together researchers from a wide range of areas. I will work on creating an inclusive environment for fostering collaborations between researchers in different areas as well as across different geographic regions. I believe that there is a vast potential for interdisciplinary work combining linguistically focused research in phonetics, phonology, and morphology with computational methods and insights from Computational Linguistics.