L2 French analyses
Coughlin
September 2, 2015
Reaction Time analyes for L2 French (KU) in behavioral study
Load in files
Set working directory This folder is the parent folder for the individual task folders
setwd("C:/Users/Katie/Desktop/Research/Dissertation/STUDIES/Studies 1 and 2 Behavioral/RESULTS/All tasks raw excel files/L2 French/R Markdown Files")
list.files()## [1] "all L2 french markdown.Rmd" "all_L2_french_markdown.html"
## [3] "all_L2_french_markdown.Rmd" "major paper stim properties.csv"
## [5] "p101.csv" "p102.csv"
## [7] "p103.csv" "p104.csv"
## [9] "p105.csv" "p106.csv"
## [11] "p107.csv" "p108.csv"
## [13] "p109.csv" "p110.csv"
## [15] "p111.csv" "p112.csv"
## [17] "p113.csv" "p114.csv"
## [19] "p115.csv" "p116.csv"
## [21] "p117.csv" "p118.csv"
## [23] "p119.csv" "p120.csv"
## [25] "p121.csv" "p122.csv"
## [27] "p123.csv" "p124.csv"
## [29] "p125.csv" "p126.csv"
## [31] "p127.csv" "p128.csv"
## [33] "p129.csv" "Participant Data.csv"
## [35] "rsconnect" "wf101.csv"
## [37] "wf102.csv" "wf103.csv"
## [39] "wf104.csv" "wf105.csv"
## [41] "wf106.csv" "wf107.csv"
## [43] "wf108.csv" "wf109.csv"
## [45] "wf110.csv" "wf111.csv"
## [47] "wf112.csv" "wf113.csv"
## [49] "wf114.csv" "wf115.csv"
## [51] "wf116.csv" "wf117.csv"
## [53] "wf118.csv" "wf119.csv"
## [55] "wf120.csv" "wf121.csv"
## [57] "wf122.csv" "wf123.csv"
## [59] "wf124.csv" "wf125.csv"
## [61] "wf126.csv" "wf127.csv"
## [63] "wf128.csv" "wf129.csv"#setwd("D:/Users/c944c978/Dropbox/Dissertation Data/R Markdown Files")Compile individual files Priming files compiled to “data” datafile
file_list <- list.files(pattern="p1")
for (file in file_list){
# if the merged dataset doesn't exist, create it
if (!exists("dataset")){
dataset <- read.csv(file, header=TRUE)
}
# if the merged dataset does exist, append to it
if (exists("dataset")){
temp_dataset <-read.csv(file, header=TRUE)
dataset<-rbind(dataset, temp_dataset)
rm(temp_dataset)
}
}
data<-dataset
dataset<-NULLWord-familiarity files compiled to “familiar” datafile
file_list <- list.files(pattern="wf")
for (file in file_list){
# if the merged dataset doesn't exist, create it
if (!exists("dataset")){
dataset <- read.csv(file, header=TRUE)
}
# if the merged dataset does exist, append to it
if (exists("dataset")){
temp_dataset <-read.csv(file, header=TRUE)
dataset<-rbind(dataset, temp_dataset)
rm(temp_dataset)
}
}
familiar<-dataset
dataset<-NULL
head(familiar)## Subject Trial Familiar.Keyboard.IsCorrect Familiar.Keyboard.Responses
## 1 101 1 0 4
## 2 101 2 0 7
## 3 101 3 0 5
## 4 101 4 0 3
## 5 101 5 0 1
## 6 101 6 0 6
## Familiar.Keyboard.ResponseTime TrialTable.Condition TrialTable.Item
## 1 4355.42 Morph D013
## 2 1563.29 Orth D072
## 3 1370.40 Orth D068
## 4 1289.76 Sem D095
## 5 1059.53 Morph D019
## 6 2021.16 Morph D023
## TrialTable.Target_or_Prime TrialTable.Word
## 1 Target fonde
## 2 Target commence
## 3 Target charge
## 4 Target dore
## 5 Target valse
## 6 Target chasseIndividual properties file
ind_props<-read.csv("Participant Data.csv",header=TRUE)
ind_props## Subject Cloze LexTale Sex AoA YrsInstr PercentFrench Immersion
## 1 101 20 0.58750 Male 7 7.0 10.0 0.00
## 2 102 23 0.65625 Female 18 4.0 15.0 0.25
## 3 103 23 0.66875 Female 16 3.0 10.0 0.00
## 4 104 11 0.57500 Female 19 2.0 2.0 0.75
## 5 105 20 0.53750 Female 18 2.0 2.5 0.00
## 6 106 22 0.62500 Male 0 18.0 32.0 24.00
## 7 107 21 0.60000 Male 14 5.0 5.0 2.00
## 8 108 30 0.58125 Female 12 7.0 5.0 0.00
## 9 109 23 0.59375 Female 14 6.0 35.0 0.50
## 10 110 39 0.83750 Male 27 15.0 20.0 38.00
## 11 111 31 0.70000 Female 18 3.0 5.0 1.25
## 12 112 32 0.70625 Male 18 6.0 10.0 10.00
## 13 113 34 0.71875 Male 18 5.0 20.0 17.00
## 14 114 28 0.56875 Female 12 5.0 2.0 0.00
## 15 115 27 0.59375 Female 18 4.0 15.0 2.00
## 16 116 17 0.49375 Female 12 9.0 5.0 0.25
## 17 117 26 0.57500 Male 20 2.5 3.0 0.00
## 18 118 28 0.56875 Female 16 3.0 25.0 0.15
## 19 119 18 0.55625 Female 12 8.0 10.0 0.00
## 20 120 22 0.65000 Female 18 2.0 3.0 1.50
## 21 121 15 0.53125 Male 16 5.0 10.0 0.00
## 22 122 21 NA Female 13 7.0 5.0 0.00
## 23 123 23 0.65000 Female 12 7.0 10.0 0.00
## 24 124 21 0.55000 Female 14 6.0 20.0 0.00
## 25 125 25 0.53000 Female 14 8.0 5.0 0.00
## 26 126 34 0.68750 Female 15 6.0 30.0 12.50
## 27 127 20 0.66875 Male 13 7.0 5.0 1.50
## 28 128 19 0.60625 Female 17 5.0 10.0 4.00
## 29 129 14 0.56875 Female 13 5.0 10.0 0.00Predict cloze score for participant 122, who could not do the cloze test or lextale this chunk of code is run independent of the RT analysis; disregard it being here
#proficiency<-data[,c(2,18,19,20,21,22,23,24)]
#proficiency<-unique(proficiency)
#cloze_pred<-lm(cloze~aoa+yrsinstr+percentfrench+immersion,data=proficiency)
#coef(cloze_pred)Lexical properties file
props<-read.csv("major paper stim properties.csv",header=TRUE)
head(props)## Item Condition Word Letters Frequency Syllables tar.rel.unrel
## 1 D136 ID pulse 5 0.20 1 rel
## 2 D144 ID moule 5 0.68 1 rel
## 3 D129 ID manie 5 1.55 2 rel
## 4 D124 ID abuse 5 1.62 2 rel
## 5 D139 ID borde 5 2.03 1 rel
## 6 D117 ID nage 4 2.36 1 rel
## rel.overlap word.nonce repeat.
## 1 1 word 1
## 2 1 word 1
## 3 1 word 1
## 4 1 word 1
## 5 1 word 1
## 6 1 word 1Cleaning data
Remove participant 112: this participant did not follow instructions to go as fast as possible. 101 & 106 grew up with a French-speaking mother. 124 grew up speaking vietnamese
data<-data[data$Subject!=112 &data$Subject!=106 &data$Subject!=101 &data$Subject!=124,]Remove breaks and practice items. This organizes the messy e-prime outpt files The factor lines just completely remove the ignored levels after subsetting
data<-data[data$Running !="PracList",]
data$Running<-factor(data$Running)
data<-data[data$Procedure != "BreakProc",]
data$Procedure<-factor(data$Procedure)
data$Condition<-factor(data$Condition)
data$Related<-factor(data$Related)
data$Slide1.ACC<-factor(data$Slide1.ACC)
data$PrimeCondition<-factor(data$PrimeCondition)Remove unneeded columns
data<-data[,c(1,2,16,18,20,30,31,33,35,42,44)]Give some cols more transparent names
library(dplyr)
data<-rename(data,List=ExperimentName)
data<-rename(data,TrialOrder=Block)
data<-rename(data,Accuracy=Slide1.ACC)
data<-rename(data,RT=Slide1.RT)Accuracy analyses
Load necessary packages
library(stats)
library(lme4)
library(lmerTest)Summarize accuracy data Accuracy is currently a factor, have to change to numeric 0,1. Same with RT column
data$Accuracy<-as.numeric(as.character(data$Accuracy))
data$RT<-as.numeric(as.character(data$RT))
aggregate(Accuracy~Condition,data=data,FUN=mean)## Condition Accuracy
## 1 ID 0.8300000
## 2 Morph 0.8466667
## 3 Nonce 0.7275000
## 4 Orth 0.8000000
## 5 Sem 0.8188889Logit model to analyze accuracy data
acc1<-glm(Accuracy~ Condition*Related+TrialOrder,data=data,family=binomial)
summary(acc1)##
## Call:
## glm(formula = Accuracy ~ Condition * Related + TrialOrder, family = binomial,
## data = data)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.0042 0.5537 0.6318 0.7854 0.8414
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.5538450 0.1369131 11.349 < 2e-16 ***
## ConditionMorph 0.1687974 0.1854712 0.910 0.363
## ConditionNonce -0.7028524 0.1423934 -4.936 7.97e-07 ***
## ConditionOrth -0.1125413 0.1767540 -0.637 0.524
## ConditionSem -0.1287882 0.1763633 -0.730 0.465
## RelatedUnrelated -0.0812305 0.1775872 -0.457 0.647
## TrialOrder 0.0004752 0.0003387 1.403 0.161
## ConditionMorph:RelatedUnrelated -0.0884976 0.2567855 -0.345 0.730
## ConditionNonce:RelatedUnrelated 0.1691561 0.1943378 0.870 0.384
## ConditionOrth:RelatedUnrelated -0.1709978 0.2440230 -0.701 0.483
## ConditionSem:RelatedUnrelated 0.0960883 0.2480168 0.387 0.698
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 7665.2 on 7199 degrees of freedom
## Residual deviance: 7554.8 on 7189 degrees of freedom
## AIC: 7576.8
##
## Number of Fisher Scoring iterations: 4Further clean and organize data for RT analyses
Remove nonce items
data<-data[data$Condition!="Nonce",]
head(data)## List Subject TrialOrder Condition Item Prime PrimeCondition
## 615 List2 102 13 Sem D085 prêtons unr
## 616 List2 102 14 Orth D041 humons unr
## 617 List2 102 15 Orth D061 sonnons orth
## 620 List2 102 18 Orth D048 donnons unr
## 623 List2 102 21 Sem D107 piquons sem
## 628 List2 102 26 Sem D081 voguons unr
## Related Accuracy RT Target
## 615 Unrelated 1 977 RACONTE
## 616 Unrelated 1 1033 JUGE
## 617 Related 1 917 SONGE
## 620 Unrelated 1 870 COUPE
## 623 Related 1 1655 PERCE
## 628 Unrelated 1 801 ÉPOUSELoad lexical property file
props<-read.csv("major paper stim properties.csv",header=TRUE)
head(props)## Item Condition Word Letters Frequency Syllables tar.rel.unrel
## 1 D136 ID pulse 5 0.20 1 rel
## 2 D144 ID moule 5 0.68 1 rel
## 3 D129 ID manie 5 1.55 2 rel
## 4 D124 ID abuse 5 1.62 2 rel
## 5 D139 ID borde 5 2.03 1 rel
## 6 D117 ID nage 4 2.36 1 rel
## rel.overlap word.nonce repeat.
## 1 1 word 1
## 2 1 word 1
## 3 1 word 1
## 4 1 word 1
## 5 1 word 1
## 6 1 word 1Organizing property files in prep of integrating to data file
targets<-props[props$tar.rel.unrel=="target" & props$word.nonce=="word",]
#primes<-props[props$tar.rel.unrel!="target"& props$word.nonce=="word",]
target_length<-targets[,c(3,4)]
target_freq<-targets[,c(3,5)]
target_syllables<-targets[,c(3,6)]
#primes_length<-primes[,c(3,4)]
#primes_length<-primes_length[unique(primes_length$Word),]
#primes_freq<-primes[,c(3,5)]
#primes_freq<-primes_freq[unique(primes_freq$Word),]
#primes_syllables<-primes[,c(3,6)]
#primes_syllables<-primes_syllables[unique(primes_syllables$Word),]Integrate lexical properties to data file
library(qdap)
data$targetfreq<-lookup(data$Target,target_freq)
data$targetlength<-lookup(data$Target,target_length)
data$targetsyllables<-lookup(data$Target,target_syllables)
#data$primefreq<-lookup(data$Prime,primes_freq)
#data$primelength<-lookup(data$Prime,primes_length)
#data$primesyllables<-lookup(data$Prime,primes_syllables)Log transform RTs and target frequency
data$logRT<-log(data$RT)
data$logtargetfreq<-log(data$targetfreq)Clean data in following order
* remove 3000+ RTs
* calculate zscores
* remove 2.5+ zscores
* keep only accurate items
* re-level the Related column so that Unrelated is treated as the baselinelibrary(stats)
data<-data[data$RT<=3000,]
data$zRT<-ave(data$RT,data$Subject,FUN=scale,na.rm=T)
data<-data[data$zRT<=2.5,]
data<-data[data$Accuracy==1,]
data$Related<-as.factor(data$Related)
data$Related<-relevel(data$Related,ref="Unrelated")Adding L2 ind_props data to dataframe
cloze<-ind_props[,c(1,2)]
lextale<-ind_props[,c(1,3)]
sex<-ind_props[,c(1,4)]
aoa<-ind_props[,c(1,5)]
yrsinstr<-ind_props[,c(1,6)]
percentfrench<-ind_props[,c(1,7)]
immersion<-ind_props[,c(1,8)]
data$cloze<-lookup(data$Subject,cloze)
data$lextale<-lookup(data$Subject,lextale)
data$sex<-lookup(data$Subject,sex)
data$aoa<-lookup(data$Subject,aoa)
data$yrsinstr<-lookup(data$Subject,yrsinstr)
data$percentfrench<-lookup(data$Subject,percentfrench)
data$immersion<-lookup(data$Subject,immersion)Adding word familiarity to dataframe.
familiar_target<-familiar[familiar$TrialTable.Target_or_Prime=="Target",]
familiar_target$TrialTable.Target_or_Prime<-factor(familiar_target$TrialTable.Target_or_Prime)
familiar_target$sub_word<-paste(familiar_target$Subject,familiar_target$TrialTable.Word)
familiar_prime<-familiar[familiar$TrialTable.Target_or_Prime=="Prime",]
familiar_prime$TrialTable.Target_or_Prime<-factor(familiar_prime$TrialTable.Target_or_Prime)
familiar_prime$sub_word<-paste(familiar_prime$Subject,familiar_prime$TrialTable.Word)
target_ratings<-familiar_target[,c(10,4)]
target_ratings<-rename(target_ratings,Rating=Familiar.Keyboard.Responses)
prime_ratings<-familiar_prime[,c(10,4)]
prime_ratings<-rename(prime_ratings,Rating=Familiar.Keyboard.Responses)
ratings<-rbind(prime_ratings,target_ratings)
data$data_sub_target<-paste(data$Subject,data$Target)
data$data_sub_prime<-paste(data$Subject,data$Prime)
data$data_sub_target<-tolower(data$data_sub_target)
data$know_target<-lookup(data$data_sub_target,ratings)
data$know_prime<-lookup(data$data_sub_prime,ratings)Center cloze score for normal distribution
data$center_cloze<-scale(data$cloze, center = TRUE, scale = FALSE)Summarize data frame to be used in analyses
str(data)## 'data.frame': 2863 obs. of 29 variables:
## $ List : Factor w/ 2 levels "List1","List2": 2 2 2 2 2 2 2 2 2 2 ...
## $ Subject : int 102 102 102 102 102 102 102 102 102 102 ...
## $ TrialOrder : int 13 14 15 18 26 27 28 34 36 37 ...
## $ Condition : Factor w/ 5 levels "ID","Morph","Nonce",..: 5 4 4 4 5 1 4 2 5 2 ...
## $ Item : Factor w/ 289 levels "D001","D002",..: 85 41 61 48 81 137 68 8 88 7 ...
## $ Prime : Factor w/ 487 levels "abjurons","aboyons",..: 431 380 460 344 283 440 322 463 260 422 ...
## $ PrimeCondition : Factor w/ 7 levels "id","morph","orth",..: 6 6 3 6 6 1 3 6 6 6 ...
## $ Related : Factor w/ 2 levels "Unrelated","Related": 1 1 2 1 1 2 2 1 1 1 ...
## $ Accuracy : num 1 1 1 1 1 1 1 1 1 1 ...
## $ RT : num 977 1033 917 870 801 ...
## $ Target : Factor w/ 298 levels "ABUSE","ACCEPTE",..: 234 149 258 66 94 239 55 15 247 169 ...
## $ targetfreq : num 54.1 29.8 17 17.8 7.3 ...
## $ targetlength : num 7 4 5 5 6 6 6 5 5 4 ...
## $ targetsyllables: num 2 1 1 1 2 2 1 2 1 1 ...
## $ logRT : num 6.88 6.94 6.82 6.77 6.69 ...
## $ logtargetfreq : num 3.99 3.39 2.83 2.88 1.99 ...
## $ zRT : num 0.986 1.246 0.708 0.49 0.17 ...
## $ cloze : num 23 23 23 23 23 23 23 23 23 23 ...
## $ lextale : num 0.656 0.656 0.656 0.656 0.656 ...
## $ sex : Factor w/ 2 levels "Female","Male": 1 1 1 1 1 1 1 1 1 1 ...
## $ aoa : num 18 18 18 18 18 18 18 18 18 18 ...
## $ yrsinstr : num 4 4 4 4 4 4 4 4 4 4 ...
## $ percentfrench : num 15 15 15 15 15 15 15 15 15 15 ...
## $ immersion : num 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 ...
## $ data_sub_target: chr "102 raconte" "102 juge" "102 songe" "102 coupe" ...
## $ data_sub_prime : chr "102 prêtons" "102 humons" "102 sonnons" "102 donnons" ...
## $ know_target : num 5 2 4 5 5 6 2 6 1 2 ...
## $ know_prime : num NA NA 1 NA NA 6 1 NA NA NA ...
## $ center_cloze : num [1:2863, 1] -1.07 -1.07 -1.07 -1.07 -1.07 ...
## ..- attr(*, "scaled:center")= num 24.1head(data)## List Subject TrialOrder Condition Item Prime PrimeCondition
## 615 List2 102 13 Sem D085 prêtons unr
## 616 List2 102 14 Orth D041 humons unr
## 617 List2 102 15 Orth D061 sonnons orth
## 620 List2 102 18 Orth D048 donnons unr
## 628 List2 102 26 Sem D081 voguons unr
## 629 List2 102 27 ID D137 récite id
## Related Accuracy RT Target targetfreq targetlength
## 615 Unrelated 1 977 RACONTE 54.12 7
## 616 Unrelated 1 1033 JUGE 29.80 4
## 617 Related 1 917 SONGE 17.03 5
## 620 Unrelated 1 870 COUPE 17.77 5
## 628 Unrelated 1 801 ÉPOUSE 7.30 6
## 629 Related 1 657 RÉCITE 3.72 6
## targetsyllables logRT logtargetfreq zRT cloze lextale sex
## 615 2 6.884487 3.991204 0.9864706 23 0.65625 Female
## 616 1 6.940222 3.394508 1.2464091 23 0.65625 Female
## 617 1 6.821107 2.834976 0.7079650 23 0.65625 Female
## 620 1 6.768493 2.877512 0.4898023 23 0.65625 Female
## 628 2 6.685861 1.987874 0.1695209 23 0.65625 Female
## 629 2 6.487684 1.313724 -0.4988924 23 0.65625 Female
## aoa yrsinstr percentfrench immersion data_sub_target data_sub_prime
## 615 18 4 15 0.25 102 raconte 102 prêtons
## 616 18 4 15 0.25 102 juge 102 humons
## 617 18 4 15 0.25 102 songe 102 sonnons
## 620 18 4 15 0.25 102 coupe 102 donnons
## 628 18 4 15 0.25 102 épouse 102 voguons
## 629 18 4 15 0.25 102 récite 102 récite
## know_target know_prime center_cloze
## 615 5 NA -1.066713
## 616 2 NA -1.066713
## 617 4 1 -1.066713
## 620 5 NA -1.066713
## 628 5 NA -1.066713
## 629 6 6 -1.066713summary(data)## List Subject TrialOrder Condition Item
## List1:1606 Min. :102.0 Min. : 11.0 ID :724 D006 : 25
## List2:1257 1st Qu.:109.0 1st Qu.: 85.0 Morph:738 D008 : 25
## Median :115.0 Median :158.0 Nonce: 0 D016 : 25
## Mean :115.5 Mean :157.8 Orth :686 D023 : 25
## 3rd Qu.:122.0 3rd Qu.:231.0 Sem :715 D029 : 25
## Max. :129.0 Max. :301.0 D031 : 25
## (Other):2713
## Prime PrimeCondition Related Accuracy
## agréons : 14 id : 369 Unrelated:1415 Min. :1
## aidons : 14 morph: 374 Related :1448 1st Qu.:1
## amusons : 14 orth : 350 Median :1
## apprenons: 14 Orth : 0 Mean :1
## brillons : 14 sem : 355 3rd Qu.:1
## brumons : 14 unr :1415 Max. :1
## (Other) :2779 Unr : 0
## RT Target targetfreq targetlength
## Min. : 282.0 ACCEPTE: 25 Min. : 0.001 Min. :3.000
## 1st Qu.: 618.0 ACCUSE : 25 1st Qu.: 6.010 1st Qu.:5.000
## Median : 723.0 AIME : 25 Median : 16.820 Median :5.000
## Mean : 797.2 AMUSE : 25 Mean : 30.851 Mean :5.335
## 3rd Qu.: 884.5 BRÛLE : 25 3rd Qu.: 32.030 3rd Qu.:6.000
## Max. :2271.0 CHASSE : 25 Max. :257.570 Max. :8.000
## (Other):2713
## targetsyllables logRT logtargetfreq zRT
## Min. :1.000 Min. :5.642 Min. :-6.908 Min. :-1.8110
## 1st Qu.:1.000 1st Qu.:6.426 1st Qu.: 1.793 1st Qu.:-0.6736
## Median :1.000 Median :6.583 Median : 2.823 Median :-0.3768
## Mean :1.286 Mean :6.632 Mean : 2.509 Mean :-0.1742
## 3rd Qu.:2.000 3rd Qu.:6.785 3rd Qu.: 3.467 3rd Qu.: 0.1514
## Max. :2.000 Max. :7.728 Max. : 5.551 Max. : 2.4865
##
## cloze lextale sex aoa
## Min. :11.00 Min. :0.4938 Female:2147 Min. :12.00
## 1st Qu.:20.00 1st Qu.:0.5687 Male : 716 1st Qu.:13.00
## Median :23.00 Median :0.5938 Median :16.00
## Mean :24.07 Mean :0.6188 Mean :15.99
## 3rd Qu.:28.00 3rd Qu.:0.6687 3rd Qu.:18.00
## Max. :39.00 Max. :0.8375 Max. :27.00
## NA's :107
## yrsinstr percentfrench immersion data_sub_target
## Min. : 2.000 Min. : 2 Min. : 0.000 Length:2863
## 1st Qu.: 3.000 1st Qu.: 5 1st Qu.: 0.000 Class :character
## Median : 5.000 Median :10 Median : 0.250 Mode :character
## Mean : 5.452 Mean :11 Mean : 3.726
## 3rd Qu.: 7.000 3rd Qu.:15 3rd Qu.: 2.000
## Max. :15.000 Max. :35 Max. :38.000
##
## data_sub_prime know_target know_prime center_cloze.V1
## Length:2863 Min. :1.00 Min. :1.000 Min. :-13.066713
## Class :character 1st Qu.:4.00 1st Qu.:2.000 1st Qu.: -4.066713
## Mode :character Median :6.00 Median :4.000 Median : -1.066713
## Mean :5.42 Mean :4.234 Mean : 0.000000
## 3rd Qu.:7.00 3rd Qu.:7.000 3rd Qu.: 3.933287
## Max. :7.00 Max. :7.000 Max. : 14.933287
## NA's :1415Priming effects
Here are the priming effects if you just look at mean RTs for related vs unrelated items
morph<-data[data$Condition=="Morph",]
orth<-data[data$Condition=="Orth",]
sem<-data[data$Condition=="Sem",]
id<-data[data$Condition=="ID",]
morph_means<-aggregate(RT~Related,data=morph,FUN=mean)
orth_means<-aggregate(RT~Related,data=orth,FUN=mean)
sem_means<-aggregate(RT~Related,data=sem,FUN=mean)
id_means<-aggregate(RT~Related,data=id,FUN=mean)
morph_priming<-morph_means[2,2]-morph_means[1,2]
orth_priming<-orth_means[2,2]-orth_means[1,2]
sem_priming<-sem_means[2,2]-sem_means[1,2]
id_priming<-id_means[2,2]-id_means[1,2]
priming<-rbind(id_priming,morph_priming,orth_priming,sem_priming)
priming## [,1]
## id_priming -54.565983
## morph_priming -37.312349
## orth_priming -7.701310
## sem_priming 8.884272Linear mixed-effects models
- For Condition, Identity is treated as the baseline
For Relatedness, Unrelated is treated as the baseline
The below models use log-transformed RT as the dependent variable. Fixed-effects are added one-by-one and models are compared using ANOVAs to decide the best model fit.
Centered cloze score is used
Log-transformed target frequency is used
Random slopes are added and tested for improved model fit with ANOVA after the best fixed-effects are decided
test1<-lmer(logRT~Condition*Related*center_cloze+TrialOrder+(1|Item)+(1|Subject),data)
test2<-lmer(logRT~Condition*Related*center_cloze+logtargetfreq+TrialOrder+(1|Item)+(1|Subject),data)
test3<-lmer(logRT~Condition*Related*center_cloze+logtargetfreq+know_target+TrialOrder+(1|Item)+(1|Subject),data)
# test 3 is best (better to include target freq and target familiarity)
test4<-lmer(logRT~Condition*Related*center_cloze-Condition:Related:center_cloze+logtargetfreq+know_target+TrialOrder+(1|Item)+(1|Subject),data)
test5<-lmer(logRT~Condition*Related*center_cloze-Condition:Related:center_cloze-center_cloze:Condition+logtargetfreq+know_target+TrialOrder+(1|Item)+(1|Subject),data)
test6<-lmer(logRT~Condition*Related*center_cloze-Condition:Related:center_cloze-center_cloze:Related+logtargetfreq+know_target+TrialOrder+(1|Item)+(1|Subject),data)
test7<-lmer(logRT~Condition*Related+center_cloze+logtargetfreq+know_target+TrialOrder+(1|Item)+(1|Subject),data)
# when adding random slopes you check if AIC is lowered by at least 2 when slope is added to decide if it's better to keep it (M. Weiling, 2015 lecture)
test8<-lmer(logRT~Condition*Related+center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1|Subject),data)
test9<-lmer(logRT~Condition*Related+center_cloze+logtargetfreq+know_target+TrialOrder+(1|Item)+(1+logtargetfreq|Subject),data)
test10<-lmer(logRT~Condition*Related+center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq|Subject),data)
test11<-lmer(logRT~Condition*Related+center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq+Related|Subject),data)
# test 9 is best
summary(test9)## Linear mixed model fit by REML t-tests use Satterthwaite approximations
## to degrees of freedom [lmerMod]
## Formula: logRT ~ Condition * Related + center_cloze + logtargetfreq +
## know_target + TrialOrder + (1 | Item) + (1 + logtargetfreq |
## Subject)
## Data: data
##
## REML criterion at convergence: 60.1
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.1398 -0.6351 -0.1427 0.4975 3.8174
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## Item (Intercept) 0.0039455 0.06281
## Subject (Intercept) 0.0308450 0.17563
## logtargetfreq 0.0002829 0.01682 -0.60
## Residual 0.0532789 0.23082
## Number of obs: 2863, groups: Item, 144; Subject, 25
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 6.890e+00 4.294e-02 4.690e+01 160.445
## ConditionMorph -2.005e-02 2.293e-02 2.339e+02 -0.874
## ConditionOrth -4.082e-02 2.342e-02 2.398e+02 -1.743
## ConditionSem -1.573e-03 2.309e-02 2.350e+02 -0.068
## RelatedRelated -8.456e-02 1.731e-02 2.705e+03 -4.886
## center_cloze -8.169e-03 4.496e-03 2.310e+01 -1.817
## logtargetfreq -1.991e-02 5.393e-03 5.180e+01 -3.692
## know_target -2.394e-02 3.186e-03 1.963e+03 -7.514
## TrialOrder -2.475e-04 5.256e-05 2.776e+03 -4.709
## ConditionMorph:RelatedRelated 3.768e-02 2.436e-02 2.706e+03 1.547
## ConditionOrth:RelatedRelated 6.656e-02 2.481e-02 2.712e+03 2.682
## ConditionSem:RelatedRelated 8.776e-02 2.459e-02 2.717e+03 3.570
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## ConditionMorph 0.382765
## ConditionOrth 0.082686 .
## ConditionSem 0.945744
## RelatedRelated 1.09e-06 ***
## center_cloze 0.082185 .
## logtargetfreq 0.000537 ***
## know_target 8.64e-14 ***
## TrialOrder 2.61e-06 ***
## ConditionMorph:RelatedRelated 0.121932
## ConditionOrth:RelatedRelated 0.007356 **
## ConditionSem:RelatedRelated 0.000364 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnO CndtnS RltdRl cntr_c lgtrgt knw_tr TrlOrd
## ConditnMrph -0.269
## ConditnOrth -0.284 0.492
## ConditionSm -0.246 0.501 0.477
## RelatedRltd -0.212 0.386 0.378 0.384
## center_cloz 0.054 -0.001 -0.002 0.001 0.001
## logtargtfrq -0.395 -0.006 0.095 -0.074 0.000 0.014
## know_target -0.313 -0.002 -0.018 0.007 0.011 -0.045 -0.248
## TrialOrder -0.184 0.015 -0.008 -0.007 0.008 -0.003 0.002 -0.024
## CndtnMrp:RR 0.150 -0.542 -0.269 -0.273 -0.711 0.001 -0.001 0.000 -0.018
## CndtnOrt:RR 0.146 -0.269 -0.543 -0.266 -0.697 0.002 -0.009 0.003 -0.001
## CndtnSm:RlR 0.149 -0.272 -0.266 -0.536 -0.705 -0.003 0.002 -0.007 -0.009
## CnM:RR CnO:RR
## ConditnMrph
## ConditnOrth
## ConditionSm
## RelatedRltd
## center_cloz
## logtargtfrq
## know_target
## TrialOrder
## CndtnMrp:RR
## CndtnOrt:RR 0.495
## CndtnSm:RlR 0.502 0.490Summarizing the linear-mixed effects models
The best mode (test5) gave the following results:
- Effect of relatedness. This means that in the ID condition, the targets with related primes had faster RTs than when the targets had unrelated primes
- Effect of log-transformed target frequency. This means that as target frequency increased, RTs got faster overall
- Effect of target familiarity. This indicates that targets that are more familiar are responded to more quickly overall
- Effect of trial order. As the experiment progressed, participants got faster overall
- Interaction of Condition x Related for the Orth condition. This means that the effect of relatedness found (in the ID) condition differs from the effect of relatedness in the Orth condition
- Interaction of Condition x Related for the Sem condition. This means that the effect of relatedness found (in the ID) condition differs from the effect of relatedness in the Sem condition
Marginal (p=0.57) interaction of Related x Cloze. The effect of relatedness may change as proficiency changes
The lack of interaction of Condition x Related for Morph indicates that the effect of relatedness found in the ID condition does not differ from the effect found in the Morph condition
Follow-up models
look at effect of relatedness in each condition
ID model
id1<-lmer(logRT~Related*center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq|Subject),id)
summary(id1)## Linear mixed model fit by REML t-tests use Satterthwaite approximations
## to degrees of freedom [lmerMod]
## Formula: logRT ~ Related * center_cloze + logtargetfreq + know_target +
## TrialOrder + (1 + center_cloze | Item) + (1 + logtargetfreq |
## Subject)
## Data: id
##
## REML criterion at convergence: 109
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.9242 -0.6303 -0.1110 0.4729 3.7466
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## Item (Intercept) 2.162e-03 0.046498
## center_cloze 4.834e-06 0.002199 -1.00
## Subject (Intercept) 4.119e-02 0.202944
## logtargetfreq 3.347e-04 0.018294 -0.74
## Residual 5.522e-02 0.234979
## Number of obs: 724, groups: Item, 36; Subject, 25
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 6.867e+00 5.718e-02 5.580e+01 120.097
## RelatedRelated -8.444e-02 1.769e-02 6.632e+02 -4.773
## center_cloze -7.053e-03 5.378e-03 2.630e+01 -1.312
## logtargetfreq -2.505e-02 9.675e-03 2.900e+01 -2.589
## know_target -1.694e-02 6.122e-03 3.183e+02 -2.767
## TrialOrder -2.473e-04 1.068e-04 6.807e+02 -2.317
## RelatedRelated:center_cloze -4.011e-03 2.614e-03 6.526e+02 -1.535
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## RelatedRelated 2.23e-06 ***
## center_cloze 0.20099
## logtargetfreq 0.01488 *
## know_target 0.00598 **
## TrialOrder 0.02081 *
## RelatedRelated:center_cloze 0.12537
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) RltdRl cntr_c lgtrgt knw_tr TrlOrd
## RelatedRltd -0.181
## center_cloz 0.068 0.002
## logtargtfrq -0.393 0.008 -0.031
## know_target -0.444 0.025 -0.058 -0.306
## TrialOrder -0.306 0.019 -0.001 0.043 -0.010
## RltdRltd:c_ -0.002 0.005 -0.238 0.012 -0.003 -0.002Morph model
morph1<-lmer(logRT~Related*center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq|Subject),morph)
summary(morph1)## Linear mixed model fit by REML t-tests use Satterthwaite approximations
## to degrees of freedom [lmerMod]
## Formula: logRT ~ Related * center_cloze + logtargetfreq + know_target +
## TrialOrder + (1 + center_cloze | Item) + (1 + logtargetfreq |
## Subject)
## Data: morph
##
## REML criterion at convergence: 8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.9390 -0.6738 -0.1605 0.5430 3.7637
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## Item (Intercept) 4.066e-03 6.376e-02
## center_cloze 2.608e-09 5.107e-05 -1.00
## Subject (Intercept) 2.679e-02 1.637e-01
## logtargetfreq 1.281e-05 3.579e-03 -1.00
## Residual 4.770e-02 2.184e-01
## Number of obs: 738, groups: Item, 36; Subject, 25
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 6.922e+00 5.305e-02 7.020e+01 130.491
## RelatedRelated -4.665e-02 1.626e-02 6.817e+02 -2.869
## center_cloze -6.325e-03 5.038e-03 2.660e+01 -1.256
## logtargetfreq -3.532e-02 1.106e-02 3.730e+01 -3.195
## know_target -2.167e-02 6.039e-03 4.842e+02 -3.588
## TrialOrder -4.222e-04 9.780e-05 6.994e+02 -4.316
## RelatedRelated:center_cloze 4.144e-04 2.487e-03 6.748e+02 0.167
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## RelatedRelated 0.004239 **
## center_cloze 0.220208
## logtargetfreq 0.002843 **
## know_target 0.000368 ***
## TrialOrder 1.82e-05 ***
## RelatedRelated:center_cloze 0.867739
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) RltdRl cntr_c lgtrgt knw_tr TrlOrd
## RelatedRltd -0.160
## center_cloz 0.068 0.011
## logtargtfrq -0.389 -0.001 0.014
## know_target -0.454 0.023 -0.079 -0.277
## TrialOrder -0.268 -0.034 0.004 0.017 -0.031
## RltdRltd:c_ 0.008 -0.036 -0.250 0.015 -0.012 -0.006Orth model
orth1<-lmer(logRT~Related*center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq|Subject),orth)
summary(orth1)## Linear mixed model fit by REML t-tests use Satterthwaite approximations
## to degrees of freedom [lmerMod]
## Formula: logRT ~ Related * center_cloze + logtargetfreq + know_target +
## TrialOrder + (1 + center_cloze | Item) + (1 + logtargetfreq |
## Subject)
## Data: orth
##
## REML criterion at convergence: 158.3
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.2579 -0.6188 -0.1646 0.5386 3.6207
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## Item (Intercept) 2.722e-03 0.0521730
## center_cloze 3.273e-07 0.0005721 -1.00
## Subject (Intercept) 2.720e-02 0.1649361
## logtargetfreq 3.739e-04 0.0193358 -0.53
## Residual 5.884e-02 0.2425748
## Number of obs: 686, groups: Item, 36; Subject, 25
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 6.861e+00 5.159e-02 9.320e+01 132.982
## RelatedRelated -1.726e-02 1.872e-02 6.090e+02 -0.922
## center_cloze -4.716e-03 4.872e-03 2.820e+01 -0.968
## logtargetfreq -1.003e-02 6.668e-03 3.130e+01 -1.505
## know_target -3.415e-02 6.586e-03 4.437e+02 -5.185
## TrialOrder -1.154e-04 1.155e-04 6.401e+02 -0.999
## RelatedRelated:center_cloze -5.154e-03 2.794e-03 5.969e+02 -1.845
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## RelatedRelated 0.3571
## center_cloze 0.3413
## logtargetfreq 0.1424
## know_target 3.29e-07 ***
## TrialOrder 0.3180
## RelatedRelated:center_cloze 0.0656 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) RltdRl cntr_c lgtrgt knw_tr TrlOrd
## RelatedRltd -0.205
## center_cloz 0.100 0.000
## logtargtfrq -0.126 -0.018 0.032
## know_target -0.581 0.028 -0.128 -0.321
## TrialOrder -0.345 0.015 0.005 -0.030 -0.016
## RltdRltd:c_ -0.020 0.003 -0.293 0.001 0.056 -0.052Sem model
sem1<-lmer(logRT~Related*center_cloze+logtargetfreq+know_target+TrialOrder+(1+center_cloze|Item)+(1+logtargetfreq|Subject),sem)
summary(sem1)## Linear mixed model fit by REML t-tests use Satterthwaite approximations
## to degrees of freedom [lmerMod]
## Formula: logRT ~ Related * center_cloze + logtargetfreq + know_target +
## TrialOrder + (1 + center_cloze | Item) + (1 + logtargetfreq |
## Subject)
## Data: sem
##
## REML criterion at convergence: 64.9
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.4327 -0.6080 -0.1084 0.4881 3.5112
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## Item (Intercept) 6.826e-03 0.0826174
## center_cloze 1.170e-07 0.0003421 -1.00
## Subject (Intercept) 2.311e-02 0.1520105
## logtargetfreq 1.989e-04 0.0141018 -0.39
## Residual 5.094e-02 0.2256988
## Number of obs: 715, groups: Item, 36; Subject, 25
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 6.906e+00 6.101e-02 6.420e+01 113.185
## RelatedRelated 1.755e-03 1.716e-02 6.480e+02 0.102
## center_cloze -1.054e-02 4.696e-03 2.740e+01 -2.245
## logtargetfreq -3.717e-02 1.487e-02 3.510e+01 -2.500
## know_target -1.926e-02 6.757e-03 5.082e+02 -2.851
## TrialOrder -1.956e-04 1.045e-04 6.683e+02 -1.871
## RelatedRelated:center_cloze 2.878e-03 2.552e-03 6.474e+02 1.127
## Pr(>|t|)
## (Intercept) < 2e-16 ***
## RelatedRelated 0.91854
## center_cloze 0.03302 *
## logtargetfreq 0.01723 *
## know_target 0.00454 **
## TrialOrder 0.06175 .
## RelatedRelated:center_cloze 0.25998
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) RltdRl cntr_c lgtrgt knw_tr TrlOrd
## RelatedRltd -0.137
## center_cloz 0.069 -0.002
## logtargtfrq -0.548 0.004 0.043
## know_target -0.369 -0.003 -0.126 -0.307
## TrialOrder -0.258 -0.010 -0.002 0.018 -0.053
## RltdRltd:c_ -0.029 -0.017 -0.276 -0.018 0.046 0.048Summarizing the follow-up models
- The effect of relatedness is found in the ID and Morph models, but it is not found in either the Orth or the Sem models.
Summarizing the study (so far)
The big model (test5) and the follow-up models demonstrate that adult L2 learners show evidence of morphological priming that cannot be attributed to orthographic or semantic priming. The priming effect from morphologically related words does not differ from the effect of repetition priming.
Holy cow, it worked!
Proficiency Analyses
plot(ind_props)
plot(data$cloze,data$lextale,ylim=c(0,1),xlab="Cloze score",ylab="lextale score")
plot(data$yrsinstr,data$cloze)
plot(data$yrsinstr,data$lextale)
Comparing other studies
why didn’t Clahsen’s work? ALL of his targets were 4 words. Try limiting my dataset to only targets that are 3 or 4 words
data_short<-data[data$targetlength<=4,]
short1<-lmer(logRT~Condition*Related+center_cloze+(1|Subject)+(1|Item),data_short)No effect of relatedness, even in the ID condition. Doesn’t look like stimuli length can explain the difference in studies.