The Department of Linguistics at Stanford University and the Stanford Humanities Center will host a one-day workshop dedicated to exploring the typological limits of probabilistic phonological grammars. The workshop is partially funded by the France-Stanford Center for Interdisciplinary Studies as part of the project The Mathematics of Language Universals.
Location: Stanford University
Workshop date: Saturday, September 22, 2018
Invited speaker: Prof. Bruce Hayes, UCLA
More information: https://sites.google.com/site/analyzingtypologicalstructure/
Organizers: Arto Anttila (Stanford) and Giorgio Magri (CNRS)
Call for Papers:
A basic question in theoretical phonology is what a theory includes and what it excludes. A good theory should be flexible enough to closely fit the data at hand, but it should also have empirical typological content and exclude unnatural patterns. In terms of empirical fit modern phonological theories are ambitious and successful. In terms of typological content their predictions are often obscure and sometimes unknown.
The typological limits of phonological theories have been studied from various perspectives, including formal language theory (Johnson 1972, Kaplan & Kay 1994, Chandlee & Heinz 2017), factorial typologies (Prince and Smolensky 1993), Property Theory (Alber, DelBusso, & Prince 2016), algebraic methods (Merchant & Riggle 2016), and T-orders (Anttila & Magri 2018). These theoretical developments have in turn produced useful software, including finite-state tools (Beesley and Karttunen 2003, Huldén 2017), OTSoft (Hayes, Tesar, & Zuraw 2017), OTHelp (Staubs, Becker, Potts, Pratt, McCarthy, & Pater 2010), OTKit (Biró 2010), PyPhon (Riggle, Bane, & Bowman 2011), OTWorkplace (Prince, Tesar, & Merchant 2012), T-Order Generator (Anttila and Andrus 2006), and OTOrder (Djalali & Jeffers 2015), among others. These tools make it possible to explore the typological predictions of large and complex models that progressively approximate the empirical complexity of natural language phonology.
A major obstacle that stands in the way of progress is that typological analysis tools usually only apply to categorical models. Over the past two decades many phonologists have turned to quantitative data and worked extensively on patterns of stochastic variation and gradient acceptability. Such analyses often invoke probabilistic grammars, such as Stochastic OT (Boersma and Hayes 2001), Noisy Harmonic Grammar (Boersma and Pater 2016), and MaxEnt (Goldwater and Johnson 2003, Hayes and Wilson 2008). This work typically has the goal of showing that the models are rich enough to avoid undergeneration, but less attention has been paid to the question of overgeneration. The key question is how to analyze the typological structure induced by probabilistic models. The question is not trivial: while the typologies predicted by categorical phonology are usually finite, probabilistic frameworks generate an infinite family of different probability distributions.
We invite abstracts (1-2 pages, pdf) for a 30-minute talk, followed by a 15-minute discussion. We welcome submissions that address questions of the following type:
– What do probabilistic typologies look like?
– How can one effectively compute probabilistic typologies?
– Do probabilistic grammars overgenerate?
– How can one tell whether probabilistic typologies contain crazy grammars?
– How do Optimality Theory, Harmonic Grammar, and MaxEnt differ typologically?
– What is the relationship between learnability and overgeneration?
– Do learnability arguments trump tight typological predictions?
Abstracts should be emailed to anttila
stanford.edu (Anttila) and magrigrggmail.com (Magri)
Abstract submission deadline: June 25, 2018, 11:59pm PST
Notification of acceptance: July 9, 2018.
Our goal is to have a relatively small number of talks and plenty of time for informal interaction.
More information: https://sites.google.com/site/analyzingtypologicalstructure/
Organizers: Arto Anttila (Stanford) and Giorgio Magri (CNRS)