============================================================
SUPPLEMENTARY MATERIAL FOR THE PAPER 
"Questioning Arbitrariness in Language: 
 a Data-Driven Study of Conventional Iconicity"
Ekaterina Abramova & Raquel Fernandez
NAACL-HLT 2016
============================================================

** Supplemental1.csv **

This file contains results from the validation procedure for all phonesthemes. 
We refer to words that share a given phonestheme as a phonesthemic cluster.

Description of the variables:

prefix: 16 prefixes that were tested
clusterSize: the number of words in the phonesthemic cluster
lemmaCount: the number of lemmas in the phonesthemic cluster
lemmaRation: the lemma ratio calculated as clusterSize/lemmaCount
avgSim: average semantic similarity of the phonesthemic cluster
numSuccesses: the number of pairwise similarity t-test successes
avgEffect: average effect size of pairwise similarity t-tests
significance: statistical significance of pairwise similarity t-tests


** Supplemental2.pdf **

This file contains:
- semantic glosses for all 16 phonesthemes that were tested
- gold labels for 6 validated phonesthemes derived from semantic glosses 
  and Roget's thesaurus
- top 25 labels for 6 validated phonesthemes generated by 3 methods


** Supplemental3.pdf ***

This file contains a screenshot of the instructions to the participants in the 
human evaluation of the centroid induced labels run on CrowdFlower.