Supplemental materials for Bruce Hayes (2017), 

"Varieties of Noisy Harmonic Grammar"

Proceedings of the 2016 Annual Meeting in Phonology
University of Southern California

==================================================================

These Supplementary Materials consist of a nested folder structure 
containing files reporting the analysis of four schematic phonological 
patterns, using various versions of Noisy Harmonic Grammar as well 
as maxent grammar.

------------------------------------------------------------------

1. Cases examined

See main text for details.

A. Simple harmonic bounding
B. Local optionality:  /p/ Voicing as applied to /apapapapapapa/
C. Scalar vs. constant constraints
D. Consonant loss in an underlying CVCC syllable, from Jesney (2007)

------------------------------------------------------------------

2. Frameworks tested

A. Maxent grammars
B. Noisy harmonic grammars, exploring several binary choices, viz.:

   i. Noise is added to (constraints/tableau cells).
  ii. Noise is added (before/after) multiplication of violations by 
      weights
 iii. Cells without violations (receive/do not receive) noise
  iv. Perturbed weights (are/are not) allowed to go negative

------------------------------------------------------------------

3. Software employed

A. For maxent, the Maxent Grammar Tool, available at:

   www.linguistics.ucla.edu/people/hayes/MaxentGrammarTool/

B. For Noisy Harmonic Grammar, OTSoft version 2.5, available at:

   www.linguistics.ucla.edu/people/hayes/otsoft/

------------------------------------------------------------------

4. Files included

A. Simulation input file, in OTSoft format (also works for Maxent 
   Grammar Tool)
B. Folder containing various files generated by OTSoft.  You can 
   consult these for initial weights, other settings used in the
   simulations.
C. A spreadsheet providing results and interpretation of each 
   simulation.  In this spreadsheet:

    i. Pasted-in program output is shown in blue.
   ii. My commentaries, trying to explain why the output pattern 
       is derived, are in red. W refers to weights, N to noise 
       values.

