###########################################################################
# vowelsInMHR.R
#
# From "Notes on Probability and Statistics for Analyzing the 
# Sounds of Languages" -- a companion textbook to Peter Ladefoged's
# "Vowels and Consonants: An Introduction to the Sounds of Languages"
# 2nd edition (Blackwell, 2005).
#
# (c) 2007, Grant McGuire, Fangfang Li, Eunjong Kong, & Mary E. Beckman 
# Department of Linguistics, Ohio State University
#
# R code for evaluating the distribution of vowels in different types of 
# syllables in the wordlist derived from the Moe, Hopkins, and Rush (1982) 
# corpus.  The Moe et al. study is described in the following book, which 
# also prints the word list and associated frequencies. 
# 
#    Alden J. Moe, Carol J. Hopkins, & R. Timothy Rush (1982). The 
#    vocabulary of first-grade children. Springfield, IL: Thomas.
# 
# This book should be cited whenever you use the wordlist data file. 
#

###########################################################################
# Start by downloading the file Moe.wp from the course web site and reading 
# it into R. 
# 
# This sets the working directory to where you have put the data file.  You 
# will need to replace the path name in argument with the right path name 
# for how you have organized your computer.    
setwd('C:/children/LexiconAdult/english')
mhr=read.table("Moe.wp",header=TRUE,sep="\t")
setwd('C:/Lx286/dataAnalysisReports/reportNo5')
names(mhr)
# [1] "orth"      "pron"      "freq"      "noLetters" "noPhons"   "noSyls"

###########################################################################
# Count the vowel type frequencies in stressed syllables versus unstressed 
# syllables in the MHR ....
#
# Set up a vector of different vowel types.
vowels=c("c","U","W","a","O","e","u","o","&","E","Y","i","I","x","R")

# Set up a table for the counts, with the vowel types as row names.  
V.table=data.frame(all=rep(0,length(vowels)), stressed=0, unstressed=0)
rownames(V.table)=vowels

# Loop through the rows of the table, counting the syllables for the 
# three columns.  
for (i in rownames(V.table)) {
	# Pick out the subsets of words that include the target vowel in any 
	# syllable, in a stressed syllable, and in an unstressed syllable.
	temp=as.character(mhr$pron[grep(i,as.character(mhr$pron))])
	V1=paste(i,"[12]",sep="")
	temp1=temp[grep(V1,temp)]
	V0=paste(i,"0",sep="")
	temp0=temp[grep(V0,temp)]
	# Set N, N1, and N0 to the number of these words.
	N=length(temp)
	N1=length(temp1)
	N0=length(temp0)
	# Increment by the number of words in which the target vowel occurs 2 times.
	N=N+length(grep(paste(i,".*",i,sep=""),temp))
	N1=N1+length(grep(paste(V1,".*",V1,sep=""),temp1))
	N0=N0+length(grep(paste(V0,".*",V0,sep=""),temp0))
	# Increment by the number of words in which the target vowel occurs 3 times.
	N=N+length(grep(paste(i,".*",i,".*",i,sep=""),temp))
	N1=N1+length(grep(paste(V1,".*",V1,".*",V1,sep=""),temp1))
	N0=N0+length(grep(paste(V0,".*",V0,".*",V0,sep=""),temp0))
	# Increment by the number of words in which the target vowel occurs 4 times.
	N=N+length(grep(paste(i,".*",i,".*",i,".*",i,sep=""),temp))
	N1=N1+length(grep(paste(V1,".*",V1,".*",V1,".*",V1,sep=""),temp1))
	N0=N0+length(grep(paste(V0,".*",V0,".*",V0,".*",V0,sep=""),temp0))
	# Increment by the number of words in which the target vowel occurs 5 times.
	N=N+length(grep(paste(i,".*",i,".*",i,".*",i,".*",i,sep=""),temp))
	N1=N1+length(grep(paste(V1,".*",V1,".*",V1,".*",V1,".*",V1,sep=""),temp1))
	N0=N0+length(grep(paste(V0,".*",V0,".*",V0,".*",V0,".*",V0,sep=""),temp0))
	# Increment by the number of words in which the target vowel occurs 6 times.
	N=N+length(grep(paste(i,".*",i,".*",i,".*",i,".*",i,".*",i,sep=""),temp))
	N1=N1+length(grep(paste(V1,".*",V1,".*",V1,".*",V1,".*",V1,".*",V1,sep=""),temp1))
	N0=N0+length(grep(paste(V0,".*",V0,".*",V0,".*",V0,".*",V0,".*",V0,sep=""),temp0))
	# Set that cell of the table to be 
	V.table[i,"all"]=N
	V.table[i,"stressed"]=N1
	V.table[i,"unstressed"]=N0
}

# Do a Chi-squared test to see if the distribution of vowels is independent 
# of the syllable type.  
chisq.test(V.table[,c("stressed","unstressed")])
#         Pearson's Chi-squared test
# 
# data:  V.table[, c("stressed", "unstressed")] 
# X-squared = 4419.355, df = 14, p-value < 2.2e-16

# Also, evaluate the distribution overall against a model in which every vowel
# is equally likely to occur in any syllable in the MHR words. 
V.model=data.frame(all=V.table[,"all"],model=round(sum(V.table$all)/length(vowels)))
chisq.test(V.model)
#         Pearson's Chi-squared test
# 
# data:  V.model 
# X-squared = 2902.707, df = 14, p-value < 2.2e-16

# Make a barplot for each of the columns for both of the Chi-squared tests.
windows(height=4, width=6, pointsize=12)
par(family="serif", oma=c(1,0,0,0), mar=c(3,2.2,2,0.1), mgp=c(1.8,0.5,0), mfrow=c(1,4))
vowels2=c("oj","U","aw","a","O","e","u","o","ae","E","aj","i","I","^","r")
xlim=c(0,2200)

barplot(rev(V.model[,"model"]), horiz=T, xlim=xlim);box(); mtext("model syllables",line=0.5)
text(rep(-1,length(vowels2)),seq(0.8,18,1.2),rev(vowels2),xpd=T,adj=c(1,0.5),cex=1.5)
barplot(rev(V.table[,"all"]), horiz=T, xlim=xlim);box(); mtext("all syllables",line=0.5)
text(rep(-1,length(vowels2)),seq(0.8,18,1.2),rev(vowels2),xpd=T,adj=c(1,0.5),cex=1.5)
barplot(rev(V.table[,"stressed"]), horiz=T, xlim=xlim);box(); mtext("stressed syllables",line=0.5)
text(rep(-1,length(vowels2)),seq(0.8,18,1.2),rev(vowels2),xpd=T,adj=c(1,0.5),cex=1.5)
barplot(rev(V.table[,"unstressed"]), horiz=T, xlim=xlim);box(); mtext("unstressed syllables",line=0.5)
text(rep(-1,length(vowels2)),seq(0.8,18,1.2),rev(vowels2),xpd=T,adj=c(1,0.5),cex=1.5)
mtext("type counts in Moe, Hopkins, and Rush (1982) words", side=1, line=-1, outer=T)

savePlot("vowelsInMHRsyllables", type="jpg")