load("~/Documents/Classes/FAAV_Data/basic_context.Rdata")
uw <- load.vowels(c("uw","Tuw"), path = path)
## Loading required package: plyr
uw <- subset(uw, Context == "Internal" &
!PreSegTrans %in% c("W","Y","L","R") &
!FolSegTrans %in% c("L","R"))
uw$VClass <- "uw"
uw$VClass[uw$PreSegTrans %in% c("CH", "D",
"JH", "SH",
"T", "TH",
"Z", "ZH",
"N")] <- "Tuw"
head(uw)
## File Name Age Sex Eth Edu Street Year F1 F2
## 1 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 420.9 1231
## 2 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 423.9 1272
## 16 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 453.3 1052
## 18 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 438.6 1421
## 19 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 408.0 1876
## 20 PH00-1-1-JStevens.txt John S 21 m i/r 14 Buchanan 2000 414.0 1869
## F3 VClass Manner Place Voice PreSeg FolSeq
## 1 2470 uw fricative labiodental voiced nasal labial complex_coda
## 2 2255 uw fricative labiodental voiced nasal labial complex_coda
## 16 2570 uw fricative labiodental voiced nasal labial one_fol_syll
## 18 2097 uw fricative labiodental voiced nasal labial complex_coda
## 19 2396 Tuw stop apical voiceless palatal <NA>
## 20 2414 Tuw stop apical voiceless palatal one_fol_syll
## Stress_Dur Info Word Time Context PreSegTrans
## 1 1.59 MOVED /5/ 38.713 MOVED 38.71 Internal M
## 2 1.90 MOVED /5/ 47.713 MOVED 47.71 Internal M
## 16 1.10 MOVING /5/ 666.871 MOVING 666.87 Internal M
## 18 1.60 MOVED /5/ 1141.623 MOVED 1141.62 Internal M
## 19 1.50 SHOOT /6/ 1233.14 SHOOT 1233.14 Internal SH
## 20 1.60 SHOOTING /5/ 1237.203 SHOOTING 1237.20 Internal SH
## FolSegTrans Fole2SegTrans Trans FolTrans Stress Dur_msec
## 1 V D M UW1 V D sp 1 59
## 2 V D M UW1 V D F ER0 M 1 90
## 16 V AH0 M UW1 V AH0 N sp 1 100
## 18 V D M UW1 V D IH1 N 1 60
## 19 T DH SH UW1 T DH AH0 1 50
## 20 T IH0 SH UW1 T IH0 NG AH0 T 1 60
## F1.n F2.n DOB DOB0 Decade NormFile
## 1 -1.591 -0.8842 1979 91 9.1 PH00-1-1-JStevens-norm.txt
## 2 -1.554 -0.7567 1979 91 9.1 PH00-1-1-JStevens-norm.txt
## 16 -1.186 -1.4471 1979 91 9.1 PH00-1-1-JStevens-norm.txt
## 18 -1.370 -0.2884 1979 91 9.1 PH00-1-1-JStevens-norm.txt
## 19 -1.753 1.1424 1979 91 9.1 PH00-1-1-JStevens-norm.txt
## 20 -1.677 1.1191 1979 91 9.1 PH00-1-1-JStevens-norm.txt
Using the dplyr package for this.
library(dplyr)
##
## Attaching package: 'dplyr'
##
## The following objects are masked from 'package:plyr':
##
## arrange, desc, failwith, id, mutate, summarise
##
## The following objects are masked from 'package:stats':
##
## filter, lag
##
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
The %.% operator strings together dplyr functions.
Here I'm creating the valyes necessary to calculate Cohen's d.
uw_means <- uw %.%
group_by(File, DOB, Age, Sex, Eth, VClass) %.%
summarise(mu = mean(F2.n), sd = sd(F2.n), n = length(F2.n)) %.%
filter(n > 3, length(mu)==2)
uw_means
## Source: local data frame [335 x 9]
## Groups: File, DOB, Age, Sex, Eth
##
## File DOB Age Sex Eth VClass mu sd n
## 1 PH90-2-7-JDoe.txt 1951 39 m r/i Tuw -0.38507 0.3032 4
## 2 PH85-3-9-Paul.txt 1946 39 m NA Tuw 0.18702 0.5148 14
## 3 PH82-1-1-Rosy.txt 1926 56 f NA Tuw 0.71130 0.1954 4
## 4 PH81-3-6-JLewis.txt 1913 68 m i/w/g Tuw 0.04321 0.2641 5
## 5 PH81-0-5-AJones.txt 1951 30 m a Tuw -0.78569 0.1558 5
## 6 PH80-2-7-MRose.txt 1953 27 f r Tuw 0.79013 0.6738 6
## 7 PH78-4-3-MrGarbies.txt 1904 74 m w Tuw 0.61306 0.3744 6
## 8 PH73-0-5-CCappo.txt 1932 41 f r Tuw 0.45999 0.4401 7
## 9 PH77-4-1-TOHara.txt 1957 20 m r Tuw -0.04156 0.1920 4
## 10 PH96-3-2-Kay.txt 1941 55 f p Tuw -0.09579 1.0551 13
## .. ... ... ... ... ... ... ... ... ..
library(ggplot2)
ggplot(uw_means, aes(DOB, mu, color = VClass)) + geom_point() + stat_smooth()
## geom_smooth: method="auto" and size of largest group is <1000, so using loess. Use 'method = x' to change the smoothing method.
Here's the Cohen's d calculations.
uw_cohen <- uw_means %.%
filter(length(mu)==2) %.% # Only use data from speakers who have both vowels
arrange(VClass) %.% # Keep the subtraction order consistant
summarise(delta = diff(mu), # difference in means
pooled_sd = sqrt(sum(((n-1)*sd))/(sum(n)-2)), # pooled sd
d = delta/pooled_sd) %.% # difference in means/pooled sd
filter(is.finite(d)) # Filter out any NAs
uw_cohen
## Source: local data frame [114 x 8]
## Groups: File, DOB, Age, Sex
##
## File DOB Age Sex Eth delta pooled_sd
## 1 PH06-2-3-Amanda.txt 1965 41 f NA 0.48532 1.1086
## 2 PH00-1-1-JStevens.txt 1979 21 m i/r -0.53685 0.8446
## 3 PH76-3-2-AB-MrHellerman.txt 1932 44 m j -0.04796 0.6809
## 4 PH00-1-2-NJulian.txt 1948 52 f o/i -0.05465 0.8227
## 5 PH00-1-3-AB-GSalvi.txt 1982 18 m NA -1.93697 0.7067
## 6 PH04-3-2-AB-ECrane.txt 1979 25 m w/j -1.02102 0.6414
## 7 PH00-1-5-BagODonuts.txt 1930 70 m NA -1.15538 0.5888
## 8 PH82-1-3-AB-JOMalley.txt 1913 69 m r -0.06118 1.1773
## 9 PH82-3-5-RFlores.txt 1970 12 f NA 0.35060 0.6667
## 10 PH02-2-8-ABC-GJones.txt 1956 46 m a -1.25060 0.6664
## .. ... ... ... ... ... ... ...
## Variables not shown: d (dbl)
ggplot(uw_cohen, aes(DOB, d)) + geom_point() + stat_smooth()
## geom_smooth: method="auto" and size of largest group is <1000, so using loess. Use 'method = x' to change the smoothing method.
