Kie Zuraw

UCLA Linguistics

Quantitatve models of phonological variation

Seoul National University, Department of Linguistics block seminar

September 2012

Mondays, Tuesdays, Thursdays, and Fridays 16:00-18:50 in 14-2126 now in 2-320

Excet for labs, such as Thursday Sept. 6, when we meet in 102-202

Almost everything here is a PDF file, so you need Adobe Acrobat Reader (or similar software) to view and print.

• Course information
  
• Lecture handouts
  
• Readings and reading questions
  
• Links
  

Course information

Kie's student hours: Tuesdays and Thursdays 11:00-13:00, in 3-302

Syllabus

Lecture handouts

1. Introduction to variation: lexical vs. free variation, lexical selection, multi-site variation
   Data files from class:
   • Spreadsheet for sCVC words
2. Quantifying variation in rule theories: variable rules, linear regression, logistic regression
   Data files from class:
   • Spreadsheet to play with coefficients for Labov (th)-index graph
   • Link to English verb data from Lieberman et al. Nature paper. You can see more supplementary materials and the full paper at the paper's webpage
3. In-class lab; model selection
   Data files from class:
   • The R script that I was typing in class. Change the extension from .txt to .R (UCLA won't let me post an .R file)
   Data files for in-class lab:
   • Data file from Hayes and White 2013. For more about the paper, see the Hayes and White 2013 webpage
   • Hungarian dataset
4. Model comparison, cont'd; Probability distributions over OT rankings
   Data files from class:
   • Simple OTSoft input file. You can modify it to make your own input files.
   • OTSoft input file with Anttila's Finnish data, based on Boersma's Praat input file (scroll down to "Anttila_data.txt")
5. Probability distributions over OT rankings II: Stochastic OT and the Gradual Learning Algorithm
   Data files from class:
   • Excel file for playing with Stochastic OT and GLA. Use the tabs to see the two "worksheets" in the file
   • OTSoft input file for Dutch syllables
   • OTSoft input file for mini-Navajo
   • Escudero and Boersma slides that we looked at (phonetic category learning)
6. Convergence and the GLA; lab
   Data files from class:
   • Excel file for setting up 10-fold cross-validation
7. noisy Harmonic Serialism; lab
   Data files from class:
   • Excel file for noisy HG demo
8. Maximum Entropy constraint models; lab
   Data files from class:
   • R script for multinomial logistic regression
     Change the extension from .txt to .R (UCLA won't let me post an .R file)
9. MaxEnt loose ends; harmonically bounded candidates; grammar architecture
   Data files from class:
   • Excel file for predicting candidate probabilities

Readings and reading questions

• Coetzee and Pater 2011
• Cedergren and Sankoff 1974
• Anttila 1997
  Reading question for Anttila
• Boersma and Hayes 2001
• Coetzee 2009
• Goldwater and Johnson 2003
• Martin 2007
• Guy 1991
  You might not be able to download this for free, depending on where you connect from. It will be made available through other means.

Links

• R (free statistics software)
  • Main webage for R. You can download it from here.
  • D.H. Kim's R tutorial from Illinois State University
  • R resources for Harmonic Grammar from UMass
• Machine learning
  • Andrew Ng's online course
    Free but you have to create an account. The latest offering started August 20, but if you have free time you could still catch up.
• Statistics tutorials for linguists (a small selection)
  • Drager and Hay: mixed-effects models for beginners
  • Levy and Bicknell: mixed-effects models
  • Harald Baayen's book, Analyzing Linguistic Data
  • Forian Jaeger: materials from various workshops and classes
• Software (other than R)
  • OTSoft by Bruce Hayes
  • Praat, by Paul Boersma and David Weenink
  • OT-Help by Robert Staubs, Michael Becker, Christopher Potts, Patrick Pratt, John J. McCarthy, and Joe Pater