#Setting up R packages
Installing packages needed for analysis
lmerTest uses Satterthwaite approximations for pvalues in MEMs https://www.r-bloggers.com/2014/02/three-ways-to-get-parameter-specific-p-values-from-lmer/)
###Preliminary operations before the analysis
##Load data
#need to exclude speakers with NO Overt subjects; see prodrop categorical speakers.xlsx
prodropdata <- subset(prodropdata, Speaker !="FXF11B")
prodropdata <- subset(prodropdata, Speaker !="FXF21A")
prodropdata <- subset(prodropdata, Speaker !="FXF29C")
prodropdata <- subset(prodropdata, Speaker !="FXF56B")
prodropdata <- subset(prodropdata, Speaker !="FXF56C")
prodropdata <- subset(prodropdata, Speaker !="FXF62B")
prodropdata <- subset(prodropdata, Speaker !="FXF72B")
prodropdata <- subset(prodropdata, Speaker !="FXM11B")
prodropdata <- subset(prodropdata, Speaker !="FXM23B")
prodropdata <- subset(prodropdata, Speaker !="FXM25A")
prodropdata <- subset(prodropdata, Speaker !="FXM27B")
prodropdata <- subset(prodropdata, Speaker !="FXM28A")
prodropdata <- subset(prodropdata, Speaker !="FXM34A")
prodropdata <- subset(prodropdata, Speaker !="FXM47A")
prodropdata <- subset(prodropdata, Speaker !="FXM47C")Ordering factor vectors to determine the default factor. We need to do it for the bigger models.
Subsetting data
Changing and checking factor levels in new data sets
#Summaries of each subset
## Prodrop Speaker Person Number Grammatical_gender
## Null :1338 C2F21C : 102 first :2064 sg :2541 / : 0
## Overt:2171 C1F50A : 100 second: 369 pl : 927 F : 0
## C1F50B : 100 third :1035 other: 0 generic: 0
## C1F54B : 100 NA's : 41 NA's : 41 M : 0
## C1F58A : 100 neuter : 0
## C1F74A : 100 (Other): 0
## (Other):2907 NA's :3509
## Tense Switch_disc Clause_type Preverbal_content_binary
## present :2204 different:1464 main :3134 none: 0
## nonpresent:1083 Other : 0 conjoined : 375 yes : 0
## other : 222 same :2045 subordinate: 0 NA's:3509
## non-past : 0
## past : 0
##
##
## Preverbal_content Generation Sex Age Transcription
## NA's:3509 Homeland: 708 M:1497 Min. :16 Length:3509
## Gen1 :1400 F:2012 1st Qu.:21 Class :character
## Gen2 :1401 Median :43 Mode :character
## Mean :42
## 3rd Qu.:58
## Max. :87
##
## Language Token.number EO_lang EO_culture
## Cantonese:3509 Min. : 1 Min. :-3 Min. :-1
## Faetar : 0 1st Qu.: 878 1st Qu.:-2 1st Qu.:-1
## Italian : 0 Median :1755 Median :-1 Median : 0
## Korean : 0 Mean :1755 Mean : 0 Mean : 0
## Polish : 0 3rd Qu.:2632 3rd Qu.: 1 3rd Qu.: 0
## Ukrainian: 0 Max. :3509 Max. : 3 Max. : 2
## NA's :1408 NA's :1408## Prodrop Speaker Person Number Grammatical_gender
## Null :1393 FXM77A :349 first : 261 sg :783 M :561
## Overt: 217 FXF40A :150 second: 91 pl :197 F :233
## FXF56A :111 third :1258 other: 0 Other : 41
## F1F71A : 93 NA's :630 / : 0
## F1M92A : 91 generic: 0
## F2M32A : 88 (Other): 0
## (Other):728 NA's :775
## Tense Switch_disc Clause_type Preverbal_content_binary
## non-past :1136 different:527 conjoined : 0 none:1292
## nonpresent: 474 Other : 0 main : 0 yes : 318
## other : 0 same :284 subordinate: 0
## past : 0 NA's :799 NA's :1610
## present : 0
##
##
## Preverbal_content Generation Sex Age Transcription
## Length:1610 Homeland:799 M:869 Min. :26 Length:1610
## Class :character Gen1 :578 F:741 1st Qu.:51 Class :character
## Mode :character Gen2 :233 Median :71 Mode :character
## Mean :63
## 3rd Qu.:77
## Max. :92
##
## Language Token.number EO_lang EO_culture
## Cantonese: 0 Min. : 1 Min. : NA Min. : NA
## Faetar :1610 1st Qu.: 302 1st Qu.: NA 1st Qu.: NA
## Italian : 0 Median : 552 Median : NA Median : NA
## Korean : 0 Mean : 687 Mean :NaN Mean :NaN
## Polish : 0 3rd Qu.:1217 3rd Qu.: NA 3rd Qu.: NA
## Ukrainian: 0 Max. :1648 Max. : NA Max. : NA
## NA's :1610 NA's :1610## Prodrop Speaker Person Number Grammatical_gender
## Null :1377 I1M61B : 124 first :1196 sg :1172 M : 495
## Overt: 416 I1M62A : 104 second: 94 pl : 621 F : 8
## I1M75A : 101 third : 503 other: 0 / : 0
## I2F45A : 100 generic: 0
## I2F44A : 99 neuter : 0
## I2M42A : 99 (Other): 0
## (Other):1166 NA's :1290
## Tense Switch_disc Clause_type Preverbal_content_binary
## present :955 different:874 main :1469 none:1424
## nonpresent:813 Other : 0 conjoined : 324 yes : 369
## non-past : 0 same :919 subordinate: 0
## other : 18
## past : 0
## NA's : 7
##
## Preverbal_content Generation Sex Age Transcription
## 2 : 106 Homeland:748 M:1109 Min. :14 Length:1793
## Neg : 260 Gen1 :375 F: 684 1st Qu.:35 Class :character
## none:1427 Gen2 :670 Median :47 Mode :character
## Mean :47
## 3rd Qu.:61
## Max. :75
## NA's :277
## Language Token.number EO_lang EO_culture
## Cantonese: 0 Min. : 1 Min. :-2 Min. :-1
## Faetar : 0 1st Qu.: 449 1st Qu.:-1 1st Qu.: 0
## Italian :1793 Median : 897 Median : 0 Median : 0
## Korean : 0 Mean : 897 Mean : 0 Mean : 0
## Polish : 0 3rd Qu.:1345 3rd Qu.: 1 3rd Qu.: 0
## Ukrainian: 0 Max. :1793 Max. : 2 Max. : 2
## NA's :748 NA's :748## Prodrop Speaker Person Number Grammatical_gender
## Null :780 K0F15A :217 first :753 sg :724 / : 0
## Overt:211 K2M36A :111 second: 81 pl :110 F : 0
## K2F18A : 88 third :142 other:142 generic: 0
## K1F56A : 84 NA's : 15 NA's : 15 M : 0
## K2M39A : 81 neuter : 0
## K2M16A : 71 (Other): 0
## (Other):339 NA's :991
## Tense Switch_disc Clause_type Preverbal_content_binary
## non-past : 0 different:394 main :482 none:258
## nonpresent: 0 Other : 0 conjoined :509 yes :733
## other : 0 same :577 subordinate: 0
## past : 0 NA's : 20
## present : 0
## NA's :991
##
## Preverbal_content Generation Sex Age Transcription
## NA's:991 Homeland:376 M:439 Min. :15 Length:991
## Gen1 :169 F:552 1st Qu.:15 Class :character
## Gen2 :446 Median :28 Mode :character
## Mean :31
## 3rd Qu.:40
## Max. :65
##
## Language Token.number EO_lang EO_culture
## Cantonese: 0 Min. : 1 Min. :-2 Min. :-1
## Faetar : 0 1st Qu.:248 1st Qu.:-2 1st Qu.: 0
## Italian : 0 Median :496 Median :-2 Median : 1
## Korean :991 Mean :496 Mean :-1 Mean : 0
## Polish : 0 3rd Qu.:744 3rd Qu.: 0 3rd Qu.: 1
## Ukrainian: 0 Max. :991 Max. : 1 Max. : 1
## NA's :674 NA's :674## Prodrop Speaker Person Number Grammatical_gender
## Null :762 PXF44A :108 first :676 sg :757 M :334
## Overt:225 PXF16A :101 second: 68 pl :230 F :152
## P1M44A : 87 third :243 other: 0 / : 0
## P1M55A : 76 generic: 0
## P1M88A : 61 neuter : 0
## P1M36A : 57 (Other): 0
## (Other):497 NA's :501
## Tense Switch_disc Clause_type Preverbal_content_binary
## present :464 Other : 12 main :167 none:880
## nonpresent:523 different:414 conjoined :820 yes :107
## non-past : 0 same :555 subordinate: 0
## other : 0 NA's : 6
## past : 0
##
##
## Preverbal_content Generation Sex Age Transcription
## : 1 Homeland:209 M:613 Min. :16 Length:987
## Neg :107 Gen1 :392 F:374 1st Qu.:22 Class :character
## none:879 Gen2 :386 Median :44 Mode :character
## Mean :39
## 3rd Qu.:55
## Max. :88
##
## Language Token.number EO_lang EO_culture
## Cantonese: 0 Min. : 2 Min. :-1 Min. :-2
## Faetar : 0 1st Qu.:248 1st Qu.: 0 1st Qu.: 0
## Italian : 0 Median :495 Median : 0 Median : 1
## Korean : 0 Mean :495 Mean : 0 Mean : 1
## Polish :987 3rd Qu.:742 3rd Qu.: 0 3rd Qu.: 2
## Ukrainian: 0 Max. :988 Max. : 1 Max. : 4
## NA's :320 NA's :320## Prodrop Speaker Person Number Grammatical_gender
## Null : 61 U1F57A : 25 first :174 sg :214 neuter : 8
## Overt:239 U1F85A : 25 second: 31 pl : 83 / : 1
## U1F90A : 25 third : 92 other: 0 F :104
## U1M46A : 25 NA's : 3 NA's : 3 generic: 23
## U2F13A : 25 M : 82
## U2F44A : 25 Other : 0
## (Other):150 plural : 82
## Tense Switch_disc Clause_type Preverbal_content_binary
## past :131 different:150 main :171 none: 0
## non-past :169 Other : 0 conjoined : 63 yes : 0
## nonpresent: 0 same :148 subordinate: 47 NA's:300
## other : 0 NA's : 2 NA's : 19
## present : 0
##
##
## Preverbal_content Generation Sex Age Transcription
## NA's:300 Homeland:100 M:125 Min. :13 Length:300
## Gen1 :100 F:175 1st Qu.:40 Class :character
## Gen2 :100 Median :51 Mode :character
## Mean :52
## 3rd Qu.:67
## Max. :90
##
## Language Token.number EO_lang EO_culture
## Cantonese: 0 Min. : 1 Min. :-3 Min. :-1
## Faetar : 0 1st Qu.: 76 1st Qu.:-1 1st Qu.:-1
## Italian : 0 Median :150 Median : 0 Median :-1
## Korean : 0 Mean :184 Mean : 0 Mean :-1
## Polish : 0 3rd Qu.:325 3rd Qu.: 1 3rd Qu.: 0
## Ukrainian:300 Max. :400 Max. : 1 Max. : 0
## NA's :175 NA's :175###REML=TRUE gives REML criterion at convergence; REML=FALSE (or not listed) gives AIC, LL, etc. ##Main effects for Cantonese
main_effect_cantonese <- glmer (Prodrop ~ Person +
Number +
Tense +
Switch_disc +
Clause_type +
Sex +
Generation +
(1|Speaker),
data = cantonesedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effect_cantonese, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Generation + (1 | Speaker)
## Data: cantonesedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 4015 4089 -1996 3991 3456
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -6.048 -0.830 0.414 0.697 2.969
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.483 0.695
## Number of obs: 3468, groups: Speaker, 36
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.8193 0.3319 2.47 0.0136 *
## Personsecond 1.1507 0.1664 6.92 4.7e-12 ***
## Personthird 0.3636 0.0934 3.89 1.0e-04 ***
## Numberpl -0.6129 0.0985 -6.22 5.0e-10 ***
## Tensenonpresent -0.3956 0.0933 -4.24 2.2e-05 ***
## Tenseother -0.1317 0.1676 -0.79 0.4318
## Switch_discsame -1.0383 0.0836 -12.42 < 2e-16 ***
## Clause_typeconjoined 0.3922 0.1379 2.84 0.0044 **
## SexF 0.4268 0.2531 1.69 0.0918 .
## GenerationGen1 0.0905 0.3330 0.27 0.7859
## GenerationGen2 0.3700 0.3306 1.12 0.2631
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Main effects in Faetar
main_effect_faetar <- glmer (Prodrop ~ Person +
Number +
Tense +
Grammatical_gender +
Switch_disc +
Preverbal_content_binary +
Sex +
Generation
+ (1|Speaker),
data = faetardata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(main_effect_faetar, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Tense + Grammatical_gender + Switch_disc +
## Preverbal_content_binary + Sex + Generation + (1 | Speaker)
## Data: faetardata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 151 192 -64 129 319
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.965 -0.247 -0.095 -0.054 4.221
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 2 1.41
## Number of obs: 330, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.368 2.243 -0.61 0.542
## Personthird -2.603 1.669 -1.56 0.119
## Numberpl -1.238 0.928 -1.33 0.182
## Tensenonpresent 1.117 1.116 1.00 0.317
## Grammatical_genderF 0.878 0.540 1.63 0.104
## Grammatical_genderOther -17.162 387.036 -0.04 0.965
## Switch_discsame -0.141 0.587 -0.24 0.810
## Preverbal_content_binaryyes -1.505 0.841 -1.79 0.074 .
## SexF -1.865 1.250 -1.49 0.136
## GenerationGen2 1.020 1.133 0.90 0.368
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?#checking for KOs across Faetar dataset
table(faetardata$Person,faetardata$Prodrop)##
## Null Overt
## first 205 56
## second 61 30
## third 1127 131table(faetardata$Number,faetardata$Prodrop)##
## Null Overt
## sg 709 74
## pl 177 20
## other 0 0table(faetardata$Tense,faetardata$Prodrop)##
## Null Overt
## non-past 980 156
## nonpresent 413 61
## other 0 0
## past 0 0
## present 0 0table(faetardata$Switch_disc,faetardata$Prodrop)##
## Null Overt
## different 445 82
## Other 0 0
## same 246 38table(faetardata$Grammatical_gender,faetardata$Prodrop)##
## Null Overt
## M 515 46
## F 208 25
## Other 38 3
## / 0 0
## generic 0 0
## neuter 0 0
## plural 0 0table(faetardata$Clause_type,faetardata$Prodrop)##
## Null Overt
## conjoined 0 0
## main 0 0
## subordinate 0 0table(faetardata$Preverbal_content_binary,faetardata$Prodrop)##
## Null Overt
## none 1116 176
## yes 277 41table(faetardata$Sex,faetardata$Prodrop)##
## Null Overt
## M 749 120
## F 644 97table(faetardata$Generation,faetardata$Prodrop)##
## Null Overt
## Homeland 702 97
## Gen1 500 78
## Gen2 191 42##Main effects in Italian
main_effect_italian <- glmer (Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = italiandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effect_italian, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Generation + (1 | Speaker)
## Data: italiandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1760 1832 -867 1734 1773
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.350 -0.549 -0.375 -0.179 5.597
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.38 0.617
## Number of obs: 1786, groups: Speaker, 27
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.4832 0.2373 -2.04 0.042 *
## Personsecond -0.5528 0.2772 -1.99 0.046 *
## Personthird -0.0934 0.1615 -0.58 0.563
## Numberpl -1.1521 0.1566 -7.36 1.9e-13 ***
## Tensenonpresent 0.0423 0.1287 0.33 0.742
## Tenseother -0.4571 0.6025 -0.76 0.448
## Switch_discsame -1.0604 0.1341 -7.90 2.7e-15 ***
## Clause_typeconjoined -0.2561 0.1653 -1.55 0.121
## Preverbal_content_binaryyes -0.1562 0.1506 -1.04 0.300
## SexF 0.3857 0.2858 1.35 0.177
## GenerationGen1 -0.1014 0.3885 -0.26 0.794
## GenerationGen2 -0.0453 0.3217 -0.14 0.888
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Main effects in Korean
main_effect_korean <- glmer (Prodrop ~ Person +
Number +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = koreandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficientsummary(main_effect_korean, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + Generation + (1 | Speaker)
## Data: koreandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 873 926 -426 851 945
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.185 -0.533 -0.365 -0.145 11.130
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.276 0.526
## Number of obs: 956, groups: Speaker, 16
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.9643 0.4224 -4.65 3.3e-06 ***
## Personsecond -1.5681 0.4206 -3.73 0.00019 ***
## Personthird -2.8035 0.5242 -5.35 8.9e-08 ***
## Numberpl -0.6576 0.2846 -2.31 0.02083 *
## Switch_discsame -0.4307 0.1774 -2.43 0.01520 *
## Clause_typeconjoined 0.5457 0.1785 3.06 0.00224 **
## Preverbal_content_binaryyes 0.5942 0.2221 2.68 0.00747 **
## SexF -0.0429 0.3435 -0.13 0.90050
## GenerationGen1 0.1538 0.4720 0.33 0.74445
## GenerationGen2 0.5267 0.3863 1.36 0.17273
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient##Main effects in Polish
main_effect_polish <- glmer (Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation
+ (1|Speaker),
data = polishdata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effect_polish, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + Generation +
## (1 | Speaker)
## Data: polishdata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 500 554 -237 474 473
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.244 -0.552 -0.364 0.000 5.545
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.668 0.817
## Number of obs: 486, groups: Speaker, 15
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -17.3415 66.3797 -0.26 0.7939
## Personsecond 0.4210 0.5220 0.81 0.4200
## Personthird 0.8582 0.2997 2.86 0.0042 **
## Numberpl -0.5235 0.3133 -1.67 0.0947 .
## Grammatical_genderM -0.4975 0.3659 -1.36 0.1739
## Switch_discdifferent 16.1188 66.3769 0.24 0.8081
## Switch_discsame 15.0691 66.3770 0.23 0.8204
## Clause_typeconjoined 0.0872 0.3855 0.23 0.8211
## Preverbal_content_binaryyes 0.2186 0.4890 0.45 0.6549
## SexF 0.1113 0.6707 0.17 0.8682
## GenerationGen1 0.4578 0.9107 0.50 0.6152
## GenerationGen2 0.9110 0.8527 1.07 0.2853
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Main effects in Ukrainian
main_effect_ukrainian <- glmer (Prodrop ~ Person +
Number +
Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = ukrainiandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(main_effect_ukrainian, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Grammatical_gender + Tense + Switch_disc +
## Clause_type + Sex + Generation + (1 | Speaker)
## Data: ukrainiandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 274 328 -122 244 262
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.182 0.147 0.340 0.493 1.463
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.318 0.564
## Number of obs: 277, groups: Speaker, 12
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.7107 0.7523 3.60 0.00031 ***
## Personsecond -0.0387 0.8569 -0.05 0.96401
## Personthird -0.2529 0.4223 -0.60 0.54929
## Numberpl 17.0314 173.0878 0.10 0.92162
## Grammatical_genderM 0.7818 0.6251 1.25 0.21099
## Grammatical_genderOther -0.6275 0.8938 -0.70 0.48262
## Grammatical_genderplural -17.7378 173.0879 -0.10 0.91838
## Tensenon-past -0.5931 0.3730 -1.59 0.11186
## Switch_discsame -0.9342 0.3570 -2.62 0.00888 **
## Clause_typeconjoined -0.4893 0.3872 -1.26 0.20631
## Clause_typesubordinate 0.7581 0.5672 1.34 0.18140
## SexF 0.6947 0.5764 1.21 0.22812
## GenerationGen1 -1.0055 0.6195 -1.62 0.10454
## GenerationGen2 -0.6658 0.6264 -1.06 0.28780
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?##Models for interactions. All speakers for each language
##Interactions for Cantonese
interactions_cantonese <- glmer (Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
#Grammatical_gender*Sex +
#Grammatical_gender*Generation +
Tense*Sex +
Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
#Preverbal_content_binary*Sex +
#Preverbal_content_binary*Generation +
(1|Speaker),
data = cantonesedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_cantonese, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Generation + Person * Sex + Person * Generation + Number *
## Sex + Number * Generation + Tense * Sex + Tense * Generation +
## Switch_disc * Sex + Switch_disc * Generation + Clause_type *
## Sex + Clause_type * Generation + (1 | Speaker)
## Data: cantonesedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 3985 4188 -1959 3919 3435
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -7.720 -0.792 0.397 0.675 2.846
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.495 0.703
## Number of obs: 3468, groups: Speaker, 36
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.04658 0.38892 2.69 0.00712 **
## Personsecond 1.73551 0.37783 4.59 4.4e-06 ***
## Personthird 0.49543 0.27789 1.78 0.07462 .
## Numberpl -0.39704 0.28826 -1.38 0.16840
## Tensenonpresent -1.11271 0.25459 -4.37 1.2e-05 ***
## Tenseother -2.57464 1.21885 -2.11 0.03466 *
## Switch_discsame -0.98790 0.23416 -4.22 2.5e-05 ***
## Clause_typeconjoined -0.70779 0.37942 -1.87 0.06212 .
## SexF 0.40601 0.30204 1.34 0.17887
## GenerationGen1 -0.00626 0.40145 -0.02 0.98756
## GenerationGen2 -0.03930 0.39615 -0.10 0.92097
## Personsecond:SexF -0.59300 0.33317 -1.78 0.07509 .
## Personthird:SexF 0.28479 0.20004 1.42 0.15453
## Personsecond:GenerationGen1 -0.96881 0.41949 -2.31 0.02092 *
## Personthird:GenerationGen1 -0.48495 0.27318 -1.78 0.07586 .
## Personsecond:GenerationGen2 0.05511 0.43598 0.13 0.89940
## Personthird:GenerationGen2 -0.10850 0.28630 -0.38 0.70471
## Numberpl:SexF -0.44760 0.21046 -2.13 0.03344 *
## Numberpl:GenerationGen1 -0.11538 0.28565 -0.40 0.68626
## Numberpl:GenerationGen2 0.11478 0.29659 0.39 0.69874
## Tensenonpresent:SexF 0.62398 0.19606 3.18 0.00146 **
## Tenseother:SexF 0.73852 0.36012 2.05 0.04029 *
## Tensenonpresent:GenerationGen1 0.52222 0.26204 1.99 0.04627 *
## Tenseother:GenerationGen1 2.12639 1.22141 1.74 0.08170 .
## Tensenonpresent:GenerationGen2 0.42889 0.26380 1.63 0.10399
## Tenseother:GenerationGen2 2.17967 1.20071 1.82 0.06947 .
## Switch_discsame:SexF -0.17798 0.17417 -1.02 0.30683
## Switch_discsame:GenerationGen1 -0.04628 0.23800 -0.19 0.84581
## Switch_discsame:GenerationGen2 0.17756 0.24199 0.73 0.46310
## Clause_typeconjoined:SexF 0.02594 0.32322 0.08 0.93603
## Clause_typeconjoined:GenerationGen1 1.87156 0.38985 4.80 1.6e-06 ***
## Clause_typeconjoined:GenerationGen2 1.33963 0.36211 3.70 0.00022 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Interactions for Faetar
interactions_faetar <- glmer (Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
#Clause_type +
Preverbal_content_binary +
#EO_lang +
#EO_culture +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
Grammatical_gender*Sex +
Grammatical_gender*Generation +
# Tense*Sex +
# Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
#Clause_type*Sex +
#Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = faetardata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Hessian is numerically singular: parameters are not uniquely determinedsummary(interactions_faetar, correl=F) ## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Preverbal_content_binary + Sex + Generation + Person * Sex +
## Person * Generation + Number * Sex + Number * Generation +
## Grammatical_gender * Sex + Grammatical_gender * Generation +
## Switch_disc * Sex + Switch_disc * Generation + Preverbal_content_binary *
## Sex + Preverbal_content_binary * Generation + (1 | Speaker)
## Data: faetardata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 166 246 -62 124 309
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.742 -0.251 -0.097 -0.076 4.551
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 1.62 1.27
## Number of obs: 330, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value
## (Intercept) -1.84e+01 2.10e+04 0.00
## Personthird 1.60e+01 2.10e+04 0.00
## Numberpl -1.73e+00 1.15e+00 -1.50
## Grammatical_genderF 1.06e-01 7.31e-01 0.14
## Grammatical_genderOther -1.89e+01 7.91e+03 0.00
## Switch_discsame -2.38e-01 8.04e-01 -0.30
## Preverbal_content_binaryyes -1.90e+00 1.24e+00 -1.54
## SexF 6.60e+01 4.00e+04 0.00
## GenerationGen2 6.85e-02 1.32e+00 0.05
## Personthird:SexF -6.85e+01 4.00e+04 0.00
## Numberpl:SexF -1.40e+01 6.84e+03 0.00
## Numberpl:GenerationGen2 -1.54e+01 9.62e+03 0.00
## Grammatical_genderF:SexF 3.46e-01 1.45e+00 0.24
## Grammatical_genderOther:SexF 3.73e+00 1.49e+04 0.00
## Grammatical_genderF:GenerationGen2 1.34e+00 1.17e+00 1.14
## Grammatical_genderOther:GenerationGen2 2.11e+00 1.54e+04 0.00
## Switch_discsame:SexF 4.86e-01 1.67e+00 0.29
## Switch_discsame:GenerationGen2 -6.87e-02 1.25e+00 -0.05
## Preverbal_content_binaryyes:SexF 1.82e+00 1.94e+00 0.94
## Preverbal_content_binaryyes:GenerationGen2 4.19e-02 1.87e+00 0.02
## Pr(>|z|)
## (Intercept) 1.00
## Personthird 1.00
## Numberpl 0.13
## Grammatical_genderF 0.88
## Grammatical_genderOther 1.00
## Switch_discsame 0.77
## Preverbal_content_binaryyes 0.12
## SexF 1.00
## GenerationGen2 0.96
## Personthird:SexF 1.00
## Numberpl:SexF 1.00
## Numberpl:GenerationGen2 1.00
## Grammatical_genderF:SexF 0.81
## Grammatical_genderOther:SexF 1.00
## Grammatical_genderF:GenerationGen2 0.25
## Grammatical_genderOther:GenerationGen2 1.00
## Switch_discsame:SexF 0.77
## Switch_discsame:GenerationGen2 0.96
## Preverbal_content_binaryyes:SexF 0.35
## Preverbal_content_binaryyes:GenerationGen2 0.98
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
## optimizer (bobyqa) convergence code: 0 (OK)
## unable to evaluate scaled gradient
## Hessian is numerically singular: parameters are not uniquely determined##Interactions for Italian
interactions_italian <- glmer (Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
# Grammatical_gender*Sex +
# Grammatical_gender*Generation +
Tense*Sex +
Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = italiandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficientssummary(interactions_italian, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Generation + Person * Sex +
## Person * Generation + Number * Sex + Number * Generation +
## Tense * Sex + Tense * Generation + Switch_disc * Sex + Switch_disc *
## Generation + Clause_type * Sex + Clause_type * Generation +
## Preverbal_content_binary * Sex + Preverbal_content_binary *
## Generation + (1 | Speaker)
## Data: italiandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1744 1936 -837 1674 1751
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.683 -0.544 -0.349 -0.144 6.560
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.456 0.675
## Number of obs: 1786, groups: Speaker, 27
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.2912 0.3016 -0.97 0.33429
## Personsecond -0.9390 0.4392 -2.14 0.03251
## Personthird -1.1254 0.3076 -3.66 0.00025
## Numberpl -1.2213 0.2828 -4.32 1.6e-05
## Tensenonpresent 0.2283 0.2328 0.98 0.32686
## Tenseother -1.4297 1.1167 -1.28 0.20047
## Switch_discsame -1.0498 0.2538 -4.14 3.5e-05
## Clause_typeconjoined -0.4218 0.3047 -1.38 0.16627
## Preverbal_content_binaryyes 0.1112 0.2859 0.39 0.69730
## SexF -0.0157 0.3816 -0.04 0.96712
## GenerationGen1 -0.1844 0.5140 -0.36 0.71971
## GenerationGen2 0.0574 0.4210 0.14 0.89155
## Personsecond:SexF -0.1314 0.6233 -0.21 0.83304
## Personthird:SexF -0.0570 0.3839 -0.15 0.88196
## Personsecond:GenerationGen1 1.9882 0.9054 2.20 0.02810
## Personthird:GenerationGen1 1.6509 0.5430 3.04 0.00236
## Personsecond:GenerationGen2 0.9639 0.8124 1.19 0.23540
## Personthird:GenerationGen2 1.8708 0.4034 4.64 3.5e-06
## Numberpl:SexF 0.3122 0.3607 0.87 0.38686
## Numberpl:GenerationGen1 -0.5218 0.4783 -1.09 0.27530
## Numberpl:GenerationGen2 -0.0864 0.3820 -0.23 0.82098
## Tensenonpresent:SexF 0.8904 0.2971 3.00 0.00273
## Tenseother:SexF 1.7823 1.3600 1.31 0.19000
## Tensenonpresent:GenerationGen1 -0.2951 0.3702 -0.80 0.42534
## Tensenonpresent:GenerationGen2 -1.2484 0.3194 -3.91 9.3e-05
## Switch_discsame:SexF -0.0485 0.2941 -0.16 0.86908
## Switch_discsame:GenerationGen1 -0.5117 0.3983 -1.28 0.19889
## Switch_discsame:GenerationGen2 0.0900 0.3204 0.28 0.77875
## Clause_typeconjoined:SexF 0.0679 0.3802 0.18 0.85820
## Clause_typeconjoined:GenerationGen1 0.6164 0.4638 1.33 0.18382
## Clause_typeconjoined:GenerationGen2 -0.0319 0.4016 -0.08 0.93670
## Preverbal_content_binaryyes:SexF 0.0720 0.3391 0.21 0.83185
## Preverbal_content_binaryyes:GenerationGen1 -0.2208 0.4267 -0.52 0.60491
## Preverbal_content_binaryyes:GenerationGen2 -0.7209 0.3612 -2.00 0.04595
##
## (Intercept)
## Personsecond *
## Personthird ***
## Numberpl ***
## Tensenonpresent
## Tenseother
## Switch_discsame ***
## Clause_typeconjoined
## Preverbal_content_binaryyes
## SexF
## GenerationGen1
## GenerationGen2
## Personsecond:SexF
## Personthird:SexF
## Personsecond:GenerationGen1 *
## Personthird:GenerationGen1 **
## Personsecond:GenerationGen2
## Personthird:GenerationGen2 ***
## Numberpl:SexF
## Numberpl:GenerationGen1
## Numberpl:GenerationGen2
## Tensenonpresent:SexF **
## Tenseother:SexF
## Tensenonpresent:GenerationGen1
## Tensenonpresent:GenerationGen2 ***
## Switch_discsame:SexF
## Switch_discsame:GenerationGen1
## Switch_discsame:GenerationGen2
## Clause_typeconjoined:SexF
## Clause_typeconjoined:GenerationGen1
## Clause_typeconjoined:GenerationGen2
## Preverbal_content_binaryyes:SexF
## Preverbal_content_binaryyes:GenerationGen1
## Preverbal_content_binaryyes:GenerationGen2 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients##Interactions for Korean
interactions_korean <- glmer (Prodrop ~ Person +
Number +
#Grammatical_gender +
#Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
#Grammatical_gender*Sex +
#Grammatical_gender*Generation +
#Tense*Sex +
#Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = koreandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 4 columns / coefficientssummary(interactions_korean, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + Generation + Person * Sex + Person * Generation + Number *
## Sex + Number * Generation + Switch_disc * Sex + Switch_disc *
## Generation + Clause_type * Sex + Clause_type * Generation +
## Preverbal_content_binary * Sex + Preverbal_content_binary *
## Generation + (1 | Speaker)
## Data: koreandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 886 1027 -414 828 927
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.465 -0.517 -0.340 -0.105 12.569
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.419 0.648
## Number of obs: 956, groups: Speaker, 16
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.89091 0.73854 -3.91 9.1e-05
## Personsecond -0.33493 0.85826 -0.39 0.696
## Personthird -3.10797 1.64114 -1.89 0.058
## Numberpl 0.72356 0.67947 1.06 0.287
## Switch_discsame 0.03466 0.46075 0.08 0.940
## Clause_typeconjoined 1.06463 0.45771 2.33 0.020
## Preverbal_content_binaryyes 0.58822 0.57230 1.03 0.304
## SexF 0.44351 0.64210 0.69 0.490
## GenerationGen1 1.85195 0.82620 2.24 0.025
## GenerationGen2 1.03339 0.73899 1.40 0.162
## Personsecond:SexF -0.93463 0.97566 -0.96 0.338
## Personthird:SexF 1.18690 1.47608 0.80 0.421
## Personsecond:GenerationGen1 -0.93894 1.26709 -0.74 0.459
## Personthird:GenerationGen1 0.36584 1.50105 0.24 0.807
## Personsecond:GenerationGen2 -1.74823 1.10561 -1.58 0.114
## Personthird:GenerationGen2 -1.24730 1.50913 -0.83 0.409
## Numberpl:SexF -1.07179 0.63327 -1.69 0.091
## Numberpl:GenerationGen1 -1.02773 0.84282 -1.22 0.223
## Numberpl:GenerationGen2 -1.21342 0.72655 -1.67 0.095
## Switch_discsame:SexF -0.12002 0.41519 -0.29 0.773
## Switch_discsame:GenerationGen1 -0.60910 0.56263 -1.08 0.279
## Switch_discsame:GenerationGen2 -0.66231 0.44372 -1.49 0.136
## Clause_typeconjoined:SexF -0.65421 0.41668 -1.57 0.116
## Clause_typeconjoined:GenerationGen1 0.00526 0.57641 0.01 0.993
## Clause_typeconjoined:GenerationGen2 -0.23878 0.44030 -0.54 0.588
## Preverbal_content_binaryyes:SexF 0.19018 0.51458 0.37 0.712
## Preverbal_content_binaryyes:GenerationGen1 -1.40485 0.66352 -2.12 0.034
## Preverbal_content_binaryyes:GenerationGen2 0.39840 0.57408 0.69 0.488
##
## (Intercept) ***
## Personsecond
## Personthird .
## Numberpl
## Switch_discsame
## Clause_typeconjoined *
## Preverbal_content_binaryyes
## SexF
## GenerationGen1 *
## GenerationGen2
## Personsecond:SexF
## Personthird:SexF
## Personsecond:GenerationGen1
## Personthird:GenerationGen1
## Personsecond:GenerationGen2
## Personthird:GenerationGen2
## Numberpl:SexF .
## Numberpl:GenerationGen1
## Numberpl:GenerationGen2 .
## Switch_discsame:SexF
## Switch_discsame:GenerationGen1
## Switch_discsame:GenerationGen2
## Clause_typeconjoined:SexF
## Clause_typeconjoined:GenerationGen1
## Clause_typeconjoined:GenerationGen2
## Preverbal_content_binaryyes:SexF
## Preverbal_content_binaryyes:GenerationGen1 *
## Preverbal_content_binaryyes:GenerationGen2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 4 columns / coefficients##Interactions for Polish
interactions_polish <- glmer (Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
Grammatical_gender*Sex +
Grammatical_gender*Generation +
# Tense*Sex +
# Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = polishdata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(interactions_polish, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + Generation +
## Person * Sex + Person * Generation + Number * Sex + Number *
## Generation + Grammatical_gender * Sex + Grammatical_gender *
## Generation + Switch_disc * Sex + Switch_disc * Generation +
## Clause_type * Sex + Clause_type * Generation + Preverbal_content_binary *
## Sex + Preverbal_content_binary * Generation + (1 | Speaker)
## Data: polishdata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 510 653 -221 442 452
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.719 -0.515 -0.339 0.000 4.183
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.627 0.792
## Number of obs: 486, groups: Speaker, 15
##
## Fixed effects:
## Estimate Std. Error z value
## (Intercept) -1.60e+01 5.70e+01 -0.28
## Personsecond -1.68e+00 1.28e+00 -1.31
## Personthird 1.95e+00 1.24e+00 1.57
## Numberpl -7.00e-01 1.28e+00 -0.55
## Grammatical_genderM -1.72e+00 1.62e+00 -1.06
## Switch_discdifferent 1.58e+01 5.70e+01 0.28
## Switch_discsame 1.43e+01 5.70e+01 0.25
## Clause_typeconjoined -7.26e-01 1.36e+00 -0.53
## Preverbal_content_binaryyes 1.40e-01 1.81e+00 0.08
## SexF -1.27e+00 1.70e+00 -0.74
## GenerationGen1 -1.72e+01 1.30e+02 -0.13
## GenerationGen2 -2.91e-01 8.04e+01 0.00
## Personsecond:SexF 2.32e+00 1.63e+00 1.42
## Personthird:SexF 2.40e-01 9.95e-01 0.24
## Personsecond:GenerationGen1 3.72e+00 1.64e+00 2.27
## Personthird:GenerationGen1 -1.35e+00 1.17e+00 -1.16
## Personthird:GenerationGen2 -2.01e+00 1.19e+00 -1.69
## Numberpl:SexF -2.39e-01 1.02e+00 -0.23
## Numberpl:GenerationGen1 6.79e-01 1.27e+00 0.54
## Numberpl:GenerationGen2 3.06e-01 1.19e+00 0.26
## Grammatical_genderM:SexF 7.96e-01 1.42e+00 0.56
## Grammatical_genderM:GenerationGen1 9.36e-01 1.53e+00 0.61
## Grammatical_genderM:GenerationGen2 5.74e-01 1.23e+00 0.47
## Switch_discdifferent:SexF 4.80e-01 6.59e-01 0.73
## Switch_discdifferent:GenerationGen1 -3.67e-01 9.16e-01 -0.40
## Switch_discdifferent:GenerationGen2 1.56e+00 8.04e+01 0.02
## Switch_discsame:GenerationGen2 2.49e+00 8.04e+01 0.03
## Clause_typeconjoined:SexF 4.79e-01 1.09e+00 0.44
## Clause_typeconjoined:GenerationGen1 1.80e+01 1.30e+02 0.14
## Clause_typeconjoined:GenerationGen2 -2.35e-01 1.16e+00 -0.20
## Preverbal_content_binaryyes:SexF -4.73e-04 1.28e+00 0.00
## Preverbal_content_binaryyes:GenerationGen1 -2.28e-01 1.95e+00 -0.12
## Preverbal_content_binaryyes:GenerationGen2 2.60e-01 1.53e+00 0.17
## Pr(>|z|)
## (Intercept) 0.779
## Personsecond 0.190
## Personthird 0.117
## Numberpl 0.585
## Grammatical_genderM 0.290
## Switch_discdifferent 0.782
## Switch_discsame 0.802
## Clause_typeconjoined 0.593
## Preverbal_content_binaryyes 0.938
## SexF 0.456
## GenerationGen1 0.895
## GenerationGen2 0.997
## Personsecond:SexF 0.155
## Personthird:SexF 0.809
## Personsecond:GenerationGen1 0.023 *
## Personthird:GenerationGen1 0.246
## Personthird:GenerationGen2 0.090 .
## Numberpl:SexF 0.814
## Numberpl:GenerationGen1 0.592
## Numberpl:GenerationGen2 0.797
## Grammatical_genderM:SexF 0.574
## Grammatical_genderM:GenerationGen1 0.540
## Grammatical_genderM:GenerationGen2 0.641
## Switch_discdifferent:SexF 0.467
## Switch_discdifferent:GenerationGen1 0.689
## Switch_discdifferent:GenerationGen2 0.985
## Switch_discsame:GenerationGen2 0.975
## Clause_typeconjoined:SexF 0.660
## Clause_typeconjoined:GenerationGen1 0.890
## Clause_typeconjoined:GenerationGen2 0.840
## Preverbal_content_binaryyes:SexF 1.000
## Preverbal_content_binaryyes:GenerationGen1 0.907
## Preverbal_content_binaryyes:GenerationGen2 0.866
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?##Interactions for Ukrainian
interactions_ukrainian <- glmer (Prodrop ~ Person +
Number +
Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
Grammatical_gender*Sex +
Grammatical_gender*Generation +
Tense*Sex +
Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
# Preverbal_content_binary*Sex +
# Preverbal_content_binary*Generation +
(1|Speaker),
data = ukrainiandata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients## boundary (singular) fit: see help('isSingular')summary(interactions_ukrainian, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Grammatical_gender + Tense + Switch_disc +
## Clause_type + Sex + Generation + Person * Sex + Person *
## Generation + Number * Sex + Number * Generation + Grammatical_gender *
## Sex + Grammatical_gender * Generation + Tense * Sex + Tense *
## Generation + Switch_disc * Sex + Switch_disc * Generation +
## Clause_type * Sex + Clause_type * Generation + (1 | Speaker)
## Data: ukrainiandata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 292 448 -103 206 234
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -6.428 0.000 0.252 0.494 2.266
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 277, groups: Speaker, 12
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.26e+01 1.08e+04 0.00 1.00
## Personsecond -1.90e+01 1.98e+04 0.00 1.00
## Personthird 4.38e-01 8.74e-01 0.50 0.62
## Numberpl 2.30e+01 2.92e+04 0.00 1.00
## Grammatical_genderM -2.00e+01 1.08e+04 0.00 1.00
## Grammatical_genderOther -1.73e+00 2.26e+04 0.00 1.00
## Grammatical_genderplural -4.36e+01 3.12e+04 0.00 1.00
## Tensenon-past -1.42e+00 1.03e+00 -1.38 0.17
## Switch_discsame -1.15e+00 7.74e-01 -1.49 0.14
## Clause_typeconjoined 1.11e+00 1.04e+00 1.07 0.28
## Clause_typesubordinate 6.96e-01 1.01e+00 0.69 0.49
## SexF -1.91e+01 1.08e+04 0.00 1.00
## GenerationGen1 -1.87e+00 1.43e+00 -1.31 0.19
## GenerationGen2 -2.61e-01 1.54e+00 -0.17 0.87
## Personsecond:SexF 1.72e+01 1.98e+04 0.00 1.00
## Personthird:SexF -1.19e+00 1.03e+00 -1.16 0.25
## Personsecond:GenerationGen1 1.45e+00 3.27e+00 0.44 0.66
## Personthird:GenerationGen1 1.74e-01 1.37e+00 0.13 0.90
## Personsecond:GenerationGen2 3.28e-02 2.00e+00 0.02 0.99
## Personthird:GenerationGen2 -1.68e+00 1.14e+00 -1.47 0.14
## Numberpl:SexF -3.97e+00 4.13e+04 0.00 1.00
## Numberpl:GenerationGen1 -2.87e-01 3.55e+04 0.00 1.00
## Numberpl:GenerationGen2 -5.50e+00 4.13e+04 0.00 1.00
## Grammatical_genderM:SexF 3.75e+01 1.27e+04 0.00 1.00
## Grammatical_genderOther:SexF 2.34e+00 2.26e+04 0.00 1.00
## Grammatical_genderplural:SexF 2.47e+01 4.27e+04 0.00 1.00
## Grammatical_genderM:GenerationGen1 2.09e+01 5.25e+03 0.00 1.00
## Grammatical_genderOther:GenerationGen1 4.65e-01 2.26e+00 0.21 0.84
## Grammatical_genderplural:GenerationGen1 2.29e-01 3.55e+04 0.00 1.00
## Grammatical_genderM:GenerationGen2 6.24e-01 1.40e+00 0.45 0.66
## Grammatical_genderplural:GenerationGen2 4.91e+00 4.13e+04 0.00 1.00
## Tensenon-past:SexF 5.39e-01 1.05e+00 0.51 0.61
## Tensenon-past:GenerationGen1 1.11e+00 1.24e+00 0.89 0.37
## Tensenon-past:GenerationGen2 3.17e-01 1.22e+00 0.26 0.79
## Switch_discsame:SexF 1.22e+00 9.96e-01 1.23 0.22
## Switch_discsame:GenerationGen1 -1.42e+00 1.15e+00 -1.23 0.22
## Switch_discsame:GenerationGen2 -5.07e-01 1.04e+00 -0.49 0.63
## Clause_typeconjoined:SexF -2.28e+00 9.79e-01 -2.33 0.02 *
## Clause_typesubordinate:SexF -4.30e-01 1.59e+00 -0.27 0.79
## Clause_typeconjoined:GenerationGen1 -9.10e-01 1.34e+00 -0.68 0.50
## Clause_typesubordinate:GenerationGen1 7.82e-01 1.55e+00 0.50 0.61
## Clause_typeconjoined:GenerationGen2 -2.51e-01 1.17e+00 -0.22 0.83
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')#Subsetting the data for Homeland
cantonesehomelanddata <- subset(prodropdata, Language =="Cantonese" & Generation =="Homeland")
faetarhomelanddata <- subset(prodropdata, Language =="Faetar" & Generation =="Homeland")
italianhomelanddata <- subset(prodropdata, Language =="Italian" & Generation =="Homeland")
koreanhomelanddata <- subset(prodropdata, Language =="Korean" & Generation =="Homeland")
polishhomelanddata <- subset(prodropdata, Language =="Polish"& Generation =="Homeland")
ukrainianhomelanddata <- subset(prodropdata, Language =="Ukrainian"& Generation =="Homeland")#Models for homeland speakers
#Main effects for homeland speakers
##Cantonese
main_effects_cant_homeland <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = cantonesehomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_cant_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Age + (1 | Speaker)
## Data: cantonesehomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 774 824 -376 752 679
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -6.891 -0.734 0.355 0.620 2.629
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.754 0.868
## Number of obs: 690, groups: Speaker, 8
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.1382 0.9323 2.29 0.0218 *
## Personsecond 1.4735 0.3377 4.36 1.3e-05 ***
## Personthird 0.7443 0.2282 3.26 0.0011 **
## Numberpl -0.7616 0.2382 -3.20 0.0014 **
## Tensenonpresent -0.6677 0.2116 -3.16 0.0016 **
## Tenseother -1.8962 1.1736 -1.62 0.1061
## Switch_discsame -1.1133 0.1957 -5.69 1.3e-08 ***
## Clause_typeconjoined -0.6642 0.2618 -2.54 0.0112 *
## SexF -0.2784 0.7504 -0.37 0.7107
## Age -0.0161 0.0163 -0.98 0.3252
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Faetar
main_effects_faet_homeland <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
#Clause_type +
#Switch_disc +
Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = faetarhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## boundary (singular) fit: see help('isSingular')summary(main_effects_faet_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Preverbal_content_binary +
## Sex + Age + (1 | Speaker)
## Data: faetarhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 399 439 -191 381 602
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.535 -0.347 -0.333 -0.214 5.654
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 611, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -5.5590 1.5395 -3.61 0.00031 ***
## Personsecond 4.5232 1.1686 3.87 0.00011 ***
## Personthird -0.1185 0.4222 -0.28 0.77903
## Numberpl -0.0419 0.3520 -0.12 0.90533
## Tensenonpresent 0.4997 0.3954 1.26 0.20628
## Preverbal_content_binaryyes 0.7856 0.3265 2.41 0.01614 *
## SexF 0.8454 0.6710 1.26 0.20772
## Age 0.0463 0.0195 2.37 0.01775 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Italian
main_effects_ital_homeland <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = italianhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_ital_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Age + (1 | Speaker)
## Data: italianhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 718 773 -347 694 736
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.221 -0.535 -0.345 -0.169 6.587
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.384 0.62
## Number of obs: 748, groups: Speaker, 16
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.66395 0.51571 -1.29 0.1979
## Personsecond -0.97060 0.34430 -2.82 0.0048 **
## Personthird -1.13186 0.28158 -4.02 5.8e-05 ***
## Numberpl -1.08471 0.25509 -4.25 2.1e-05 ***
## Tensenonpresent 0.50410 0.21432 2.35 0.0187 *
## Tenseother -0.38563 0.61580 -0.63 0.5312
## Switch_discsame -1.03513 0.23863 -4.34 1.4e-05 ***
## Clause_typeconjoined -0.36879 0.28091 -1.31 0.1892
## Preverbal_content_binaryyes 0.12379 0.24533 0.50 0.6138
## SexF 0.22071 0.40182 0.55 0.5828
## Age 0.00644 0.01063 0.61 0.5445
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Korean
main_effects_kor_homeland <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
#Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = koreanhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient## boundary (singular) fit: see help('isSingular')summary(main_effects_kor_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + Age + (1 | Speaker)
## Data: koreanhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 336 375 -158 316 361
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.763 -0.561 -0.370 -0.191 6.071
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 371, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.718770 0.653316 -4.16 3.2e-05 ***
## Personsecond -0.891798 0.566142 -1.58 0.1152
## Personthird -1.943190 0.749663 -2.59 0.0095 **
## Numberpl 0.073692 0.513145 0.14 0.8858
## Switch_discsame -0.011914 0.299463 -0.04 0.9683
## Clause_typemain -0.518120 0.292333 -1.77 0.0763 .
## Preverbal_content_binaryyes 0.737078 0.394957 1.87 0.0620 .
## SexF 1.353460 0.424660 3.19 0.0014 **
## Age 0.000226 0.009068 0.02 0.9801
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Polish
main_effects_polish_homeland <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
#Sex +
Age +
(1|Speaker),
data = polishhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## boundary (singular) fit: see help('isSingular')summary(main_effects_polish_homeland, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Age + (1 | Speaker)
## Data: polishhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 116 146 -47 94 108
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.880 -0.497 -0.278 -0.074 3.995
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 119, groups: Speaker, 2
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -15.8871 1694.1170 -0.01 0.9925
## Personsecond 0.5835 1.0087 0.58 0.5630
## Personthird 2.1131 0.7411 2.85 0.0044 **
## Numberpl -0.9698 0.7868 -1.23 0.2177
## Grammatical_genderM -0.7406 0.8351 -0.89 0.3752
## Switch_discdifferent 14.0647 1694.1171 0.01 0.9934
## Switch_discsame 12.1452 1694.1172 0.01 0.9943
## Clause_typeconjoined -0.2915 0.8155 -0.36 0.7207
## Preverbal_content_binaryyes 0.1730 1.2708 0.14 0.8917
## Age 0.0287 0.0264 1.09 0.2761
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Ukrainian
main_effects_ukrainian_homeland <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = ukrainianhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient## boundary (singular) fit: see help('isSingular')summary(main_effects_ukrainian_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Grammatical_gender + Tense + Switch_disc +
## Clause_type + Sex + Age + (1 | Speaker)
## Data: ukrainianhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 92 128 -32 64 85
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -6.618 0.104 0.222 0.331 1.257
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 99, groups: Speaker, 4
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.02e+01 3.74e+02 0.05 0.957
## Personsecond -1.83e+01 3.74e+02 -0.05 0.961
## Personthird 5.21e-01 8.02e-01 0.65 0.516
## Numberpl 2.04e+01 6.11e+02 0.03 0.973
## Grammatical_genderF -1.76e+01 3.74e+02 -0.05 0.962
## Grammatical_genderM -1.74e+01 3.74e+02 -0.05 0.963
## Grammatical_genderplural -3.86e+01 6.30e+02 -0.06 0.951
## Tensenon-past -1.05e+00 9.12e-01 -1.15 0.250
## Switch_discsame -7.99e-01 7.29e-01 -1.09 0.274
## Clause_typemain -5.15e-01 1.03e+00 -0.50 0.617
## Clause_typesubordinate 1.90e-01 1.32e+00 0.14 0.886
## SexF 2.01e+00 1.14e+00 1.77 0.077 .
## Age -2.16e-04 1.62e-02 -0.01 0.989
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Interactions for homeland speakers only
##Cantonese
interactions_cant_homeland <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
Age +
Person*Sex +
Person*Age +
Number*Sex +
Number*Age +
#Grammatical_gender*Sex +
#Grammatical_gender*Age +
Tense*Sex +
Tense*Age +
Switch_disc*Sex +
Switch_disc*Age +
Clause_type*Sex +
Clause_type*Age +
#Preverbal_content_binary*Sex +
#Preverbal_content_binary*Age +
(1|Speaker),
data = cantonesehomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient## Warning in optwrap(optimizer, devfun, start, rho$lower, control = control, :
## convergence code 1 from bobyqa: bobyqa -- maximum number of function
## evaluations exceeded## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge with max|grad| = 0.0235932 (tol = 0.002, component 1)summary(interactions_cant_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Age + Person * Sex + Person * Age + Number * Sex +
## Number * Age + Tense * Sex + Tense * Age + Switch_disc *
## Sex + Switch_disc * Age + Clause_type * Sex + Clause_type *
## Age + (1 | Speaker)
## Data: cantonesehomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 770 879 -361 722 666
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -7.590 -0.702 0.326 0.579 3.735
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.767 0.876
## Number of obs: 690, groups: Speaker, 8
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.53355 1.06777 1.44 0.15094
## Personsecond 1.74847 0.93958 1.86 0.06276 .
## Personthird -0.17279 0.74434 -0.23 0.81643
## Numberpl 1.00625 0.70711 1.42 0.15472
## Tensenonpresent -1.08962 0.65225 -1.67 0.09481 .
## Tenseother -0.05319 3.91927 -0.01 0.98917
## Switch_discsame 0.12485 0.59421 0.21 0.83358
## Clause_typeconjoined -2.97005 0.81782 -3.63 0.00028 ***
## SexF 0.33263 0.86401 0.38 0.70025
## Age -0.00958 0.01905 -0.50 0.61501
## Personsecond:SexF -1.17891 0.68863 -1.71 0.08690 .
## Personthird:SexF 0.10651 0.62571 0.17 0.86483
## Personsecond:Age 0.00907 0.01691 0.54 0.59179
## Personthird:Age 0.02372 0.01202 1.97 0.04843 *
## Numberpl:SexF -0.90719 0.62995 -1.44 0.14984
## Numberpl:Age -0.02971 0.01235 -2.41 0.01609 *
## Tensenonpresent:SexF 0.23874 0.52487 0.45 0.64921
## Tensenonpresent:Age 0.00290 0.01114 0.26 0.79426
## Tenseother:Age -0.07714 0.14802 -0.52 0.60227
## Switch_discsame:SexF -0.92195 0.47644 -1.94 0.05298 .
## Switch_discsame:Age -0.01779 0.01058 -1.68 0.09261 .
## Clause_typeconjoined:SexF 2.15933 0.74395 2.90 0.00370 **
## Clause_typeconjoined:Age 0.01753 0.01558 1.13 0.26053
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
## optimizer (bobyqa) convergence code: 1 (bobyqa -- maximum number of function evaluations exceeded)
## Model failed to converge with max|grad| = 0.0235932 (tol = 0.002, component 1)##Faetar
interactions_faet_homeland <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
# Tense +
# Switch_disc +
#Clause_type +
Preverbal_content_binary +
Sex +
Age +
Person*Sex +
Person*Age +
Number*Sex +
Number*Age +
# Grammatical_gender*Sex +
# Grammatical_gender*Age +
# Tense*Sex +
# Tense*Age +
# Switch_disc*Sex +
# Switch_disc*Age +
#Clause_type*Sex +
#Clause_type*Age +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Age +
(1|Speaker),
data = faetarhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## boundary (singular) fit: see help('isSingular')summary(interactions_faet_homeland, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Preverbal_content_binary + Sex +
## Age + Person * Sex + Person * Age + Number * Sex + Number *
## Age + Preverbal_content_binary * Sex + Preverbal_content_binary *
## Age + (1 | Speaker)
## Data: faetarhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 403 474 -185 371 595
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.010 -0.355 -0.351 -0.203 4.922
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 611, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.21e+00 5.24e+00 0.23 0.82
## Personsecond 1.43e+02 7.56e+04 0.00 1.00
## Personthird -7.37e+00 5.50e+00 -1.34 0.18
## Numberpl -7.60e+01 1.68e+04 0.00 1.00
## Preverbal_content_binaryyes 1.92e+00 5.62e+00 0.34 0.73
## SexF -7.77e-01 2.87e+00 -0.27 0.79
## Age -4.77e-02 6.91e-02 -0.69 0.49
## Personsecond:SexF -3.29e+01 3.99e+04 0.00 1.00
## Personthird:SexF 1.87e+00 2.94e+00 0.64 0.53
## Personsecond:Age -1.80e+00 9.82e+02 0.00 1.00
## Personthird:Age 1.01e-01 7.25e-02 1.39 0.16
## Numberpl:SexF 1.97e+01 4.59e+03 0.00 1.00
## Numberpl:Age 9.93e-01 2.18e+02 0.00 1.00
## Preverbal_content_binaryyes:SexF -5.37e-01 2.59e+00 -0.21 0.84
## Preverbal_content_binaryyes:Age -1.39e-02 7.35e-02 -0.19 0.85
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Italian
interactions_ital_homeland <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Age +
Person*Sex +
Person*Age +
Number*Sex +
Number*Age +
# Grammatical_gender*Sex +
# Grammatical_gender*Age +
Tense*Sex +
Tense*Age +
Switch_disc*Sex +
Switch_disc*Age +
Clause_type*Sex +
Clause_type*Age +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Age +
(1|Speaker),
data = italianhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in optwrap(optimizer, devfun, start, rho$lower, control = control, :
## convergence code 1 from bobyqa: bobyqa -- maximum number of function
## evaluations exceeded## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge with max|grad| = 0.0102275 (tol = 0.002, component 1)## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(interactions_ital_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Age + Person * Sex + Person *
## Age + Number * Sex + Number * Age + Tense * Sex + Tense *
## Age + Switch_disc * Sex + Switch_disc * Age + Clause_type *
## Sex + Clause_type * Age + Preverbal_content_binary * Sex +
## Preverbal_content_binary * Age + (1 | Speaker)
## Data: italianhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 724 853 -334 668 720
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.799 -0.523 -0.330 -0.123 5.360
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.353 0.594
## Number of obs: 748, groups: Speaker, 16
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.673054 0.683936 -2.45 0.014 *
## Personsecond 1.540557 1.177995 1.31 0.191
## Personthird -0.439591 0.704819 -0.62 0.533
## Numberpl -0.303432 0.642874 -0.47 0.637
## Tensenonpresent 0.277553 0.571659 0.49 0.627
## Tenseother 6.496195 5.735635 1.13 0.257
## Switch_discsame -0.312280 0.589242 -0.53 0.596
## Clause_typeconjoined 0.851596 0.700352 1.22 0.224
## Preverbal_content_binaryyes 0.533462 0.723919 0.74 0.461
## SexF 0.209640 0.545790 0.38 0.701
## Age 0.035260 0.014775 2.39 0.017 *
## Personsecond:SexF -1.025767 0.817724 -1.25 0.210
## Personthird:SexF -0.648793 0.650343 -1.00 0.318
## Personsecond:Age -0.058308 0.026321 -2.22 0.027 *
## Personthird:Age -0.016204 0.015015 -1.08 0.281
## Numberpl:SexF 0.224786 0.562311 0.40 0.689
## Numberpl:Age -0.025705 0.014923 -1.72 0.085 .
## Tensenonpresent:SexF 1.036277 0.492795 2.10 0.035 *
## Tenseother:SexF 4.051173 3.253541 1.25 0.213
## Tensenonpresent:Age -0.000772 0.012466 -0.06 0.951
## Tenseother:Age -0.430394 0.369777 -1.16 0.244
## Switch_discsame:SexF -0.105612 0.490603 -0.22 0.830
## Switch_discsame:Age -0.018876 0.013039 -1.45 0.148
## Clause_typeconjoined:SexF -1.350123 0.723852 -1.87 0.062 .
## Clause_typeconjoined:Age -0.023495 0.015525 -1.51 0.130
## Preverbal_content_binaryyes:SexF -0.057012 0.550820 -0.10 0.918
## Preverbal_content_binaryyes:Age -0.010597 0.015352 -0.69 0.490
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 1 (bobyqa -- maximum number of function evaluations exceeded)
## Model failed to converge with max|grad| = 0.0102275 (tol = 0.002, component 1)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?##Korean
interactions_kor_homeland <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
#Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Age +
(1|Speaker),
data = koreanhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient## boundary (singular) fit: see help('isSingular')summary(interactions_kor_homeland, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + Age + (1 | Speaker)
## Data: koreanhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 336 375 -158 316 361
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.763 -0.561 -0.370 -0.191 6.071
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 371, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.718770 0.653316 -4.16 3.2e-05 ***
## Personsecond -0.891798 0.566142 -1.58 0.1152
## Personthird -1.943190 0.749663 -2.59 0.0095 **
## Numberpl 0.073692 0.513145 0.14 0.8858
## Switch_discsame -0.011914 0.299463 -0.04 0.9683
## Clause_typemain -0.518120 0.292333 -1.77 0.0763 .
## Preverbal_content_binaryyes 0.737078 0.394957 1.87 0.0620 .
## SexF 1.353460 0.424660 3.19 0.0014 **
## Age 0.000226 0.009068 0.02 0.9801
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Polish
interactions_polish_homeland <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
#Sex +
#Age +
#Person*Sex +
Person*Age +
#Number*Sex +
Number*Age +
#Grammatical_gender*Sex +
Grammatical_gender*Age +
#Tense*Sex +
# Tense*Age +
#Switch_disc*Sex +
Switch_disc*Age +
#Clause_type*Sex +
Clause_type*Age +
#Preverbal_content_binary*Sex +
Preverbal_content_binary*Age +
(1|Speaker),
data = polishhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients## boundary (singular) fit: see help('isSingular')summary(interactions_polish_homeland, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Person * Age + Number *
## Age + Grammatical_gender * Age + Switch_disc * Age + Clause_type *
## Age + Preverbal_content_binary * Age + (1 | Speaker)
## Data: polishhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 123 170 -44 89 102
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.120 -0.478 -0.249 0.000 4.844
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 119, groups: Speaker, 2
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.87e+01 7.60e+03 0.00 1.00
## Personsecond 5.37e-01 1.08e+00 0.50 0.62
## Personthird 2.72e+01 4.60e+03 0.01 1.00
## Numberpl 1.48e+00 2.52e+00 0.59 0.56
## Grammatical_genderM -2.50e+01 4.60e+03 -0.01 1.00
## Switch_discdifferent 1.50e+01 7.60e+03 0.00 1.00
## Switch_discsame 1.53e+01 7.60e+03 0.00 1.00
## Clause_typeconjoined -5.38e-02 2.26e+00 -0.02 0.98
## Preverbal_content_binaryyes -2.45e+01 9.06e+03 0.00 1.00
## Age 1.24e-02 7.51e-02 0.17 0.87
## Personthird:Age -5.70e-01 1.05e+02 -0.01 1.00
## Numberpl:Age -5.85e-02 6.68e-02 -0.88 0.38
## Grammatical_genderM:Age 5.45e-01 1.05e+02 0.01 1.00
## Switch_discdifferent:Age 6.58e-02 4.48e-02 1.47 0.14
## Clause_typeconjoined:Age -6.66e-03 6.30e-02 -0.11 0.92
## Preverbal_content_binaryyes:Age 5.75e-01 2.06e+02 0.00 1.00
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Ukrainian
interactions_ukrainian_homeland <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
Age +
Person*Sex +
Person*Age +
Number*Sex +
Number*Age +
Grammatical_gender*Sex +
Grammatical_gender*Age +
Tense*Sex +
Tense*Age +
Switch_disc*Sex +
Switch_disc*Age +
Clause_type*Sex +
Clause_type*Age +
#Preverbal_content_binary*Sex +
#Preverbal_content_binary*Age +
(1|Speaker),
data = ukrainianhomelanddata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 8 columns / coefficients## boundary (singular) fit: see help('isSingular')summary(interactions_ukrainian_homeland, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Grammatical_gender + Tense + Switch_disc +
## Clause_type + Sex + Age + Person * Sex + Person * Age + Number *
## Sex + Number * Age + Grammatical_gender * Sex + Grammatical_gender *
## Age + Tense * Sex + Tense * Age + Switch_disc * Sex + Switch_disc *
## Age + Clause_type * Sex + Clause_type * Age + (1 | Speaker)
## Data: ukrainianhomelanddata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 95 170 -19 37 70
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.6344 0.0000 0.0000 0.0001 1.4169
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 99, groups: Speaker, 4
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 6.16e+01 3.97e+04 0 1
## Personsecond -6.16e+01 3.98e+04 0 1
## Personthird 6.70e+00 6.24e+03 0 1
## Numberpl 4.76e+01 5.20e+04 0 1
## Grammatical_genderF 1.74e+02 4.96e+04 0 1
## Grammatical_genderM -2.01e+01 2.84e+04 0 1
## Grammatical_genderplural -6.04e+01 5.80e+04 0 1
## Tensenon-past -2.49e+01 1.40e+04 0 1
## Switch_discsame -2.87e+01 2.14e+04 0 1
## Clause_typemain -2.08e+00 9.66e+03 0 1
## Clause_typesubordinate -6.32e+00 3.04e+04 0 1
## SexF 3.35e+01 3.06e+04 0 1
## Age -6.55e-03 6.71e+02 0 1
## Personsecond:SexF -1.46e+01 7.54e+04 0 1
## Personthird:SexF -2.81e+00 1.56e+04 0 1
## Personthird:Age -3.31e-01 3.47e+02 0 1
## Numberpl:SexF -1.10e+01 2.30e+04 0 1
## Numberpl:Age -3.22e-01 2.30e+02 0 1
## Grammatical_genderF:SexF -9.48e+01 2.73e+04 0 1
## Grammatical_genderF:Age -2.30e+00 5.05e+02 0 1
## Tensenon-past:SexF -9.33e+00 6.18e+03 0 1
## Tensenon-past:Age 3.04e-01 1.64e+02 0 1
## Switch_discsame:SexF -1.06e+01 9.69e+03 0 1
## Switch_discsame:Age 3.54e-01 2.52e+02 0 1
## Clause_typemain:SexF 2.81e+00 2.31e+04 0 1
## Clause_typesubordinate:SexF 2.13e+01 2.80e+04 0 1
## Clause_typemain:Age 1.55e-02 5.37e+02 0 1
## Clause_typesubordinate:Age 7.30e-02 6.35e+02 0 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 8 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Models for heritage speakers only
##Subsetting heritage data for each language
cantoneseheritagedata <- subset (prodropdata, Language =="Cantonese" & (Generation =="Gen1" | Generation == "Gen2"))
faetarheritagedata <- subset(prodropdata, Language =="Faetar" & (Generation =="Gen1" | Generation == "Gen2"))
italianheritagedata <- subset(prodropdata, Language =="Italian" & (Generation =="Gen1" | Generation == "Gen2"))
koreanheritagedata <- subset(prodropdata, Language =="Korean" & (Generation =="Gen1" | Generation == "Gen2"))
polishheritagedata <- subset(prodropdata, Language == "Polish" & (Generation =="Gen1" | Generation == "Gen2"))
ukrainianheritagedata <- subset(prodropdata, Language == "Polish" & (Generation =="Gen1" | Generation == "Gen2"))##Levels for heritage data
cantoneseheritagedata <- cantoneseheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'present', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There was 1 warning in `mutate()`.
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and Dfaetarheritagedata <- faetarheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'present', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There was 1 warning in `mutate()`.
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and Ditalianheritagedata <- italianheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'present', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There was 1 warning in `mutate()`.
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and Dkoreanheritagedata <- koreanheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'non-past', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Clause_type = fct_relevel(Switch_disc,'main', 'conjoined'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There were 2 warnings in `mutate()`.
## The first warning was:
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and D
## ℹ Run `dplyr::last_dplyr_warnings()` to see the 1 remaining warning.polishheritagedata <- polishheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'present', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There was 1 warning in `mutate()`.
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and Dukrainianheritagedata <- ukrainianheritagedata %>% mutate(Generation = fct_relevel(Generation, 'Gen1', 'Gen2'),
Prodrop = fct_relevel(Prodrop, 'Null', 'Overt'),
Person = fct_relevel(Person, 'first','second','third'),
Number = fct_relevel(Number, 'sg', 'pl'),
Tense = fct_relevel(Tense, 'present', 'nonpresent'),
Grammatical_gender = fct_relevel(Grammatical_gender, 'M', 'F'),
Switch_disc = fct_relevel(Switch_disc, 'S', 'D'),
Preverbal_content_binary = fct_relevel(Preverbal_content_binary, 'none', 'yes'),
Sex = fct_relevel(Sex, 'M', 'F'))## Warning: There was 1 warning in `mutate()`.
## ℹ In argument: `Switch_disc = fct_relevel(Switch_disc, "S", "D")`.
## Caused by warning:
## ! 2 unknown levels in `f`: S and D##Main effects for heritage speakers
##Cantonese
main_effects_cantonese_heritage <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = cantoneseheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_cantonese_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Generation + (1 | Speaker)
## Data: cantoneseheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 3223 3288 -1600 3201 2767
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -5.161 -0.838 0.403 0.699 2.815
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.367 0.606
## Number of obs: 2778, groups: Speaker, 28
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.6534 0.2866 5.77 8.0e-09 ***
## Personsecond 0.9898 0.1917 5.16 2.4e-07 ***
## Personthird 0.3155 0.1037 3.04 0.0023 **
## Numberpl -0.5989 0.1087 -5.51 3.6e-08 ***
## Tensenonpresent -0.3262 0.1051 -3.10 0.0019 **
## Tenseother -0.0309 0.1709 -0.18 0.8565
## Switch_discsame -1.0225 0.0934 -10.95 < 2e-16 ***
## Clause_typemain -0.8812 0.1769 -4.98 6.3e-07 ***
## SexF 0.5847 0.2467 2.37 0.0178 *
## GenerationGen2 0.2635 0.2471 1.07 0.2862
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1main_effects_cantonese_heritage_EO <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
(1|Speaker),
data = cantoneseheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_cantonese_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + EO_lang + EO_culture + (1 | Speaker)
## Data: cantoneseheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 2487 2555 -1232 2463 2076
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.542 -0.826 0.387 0.731 2.974
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.348 0.589
## Number of obs: 2088, groups: Speaker, 21
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.8002 0.3132 5.75 9.0e-09 ***
## Personsecond 1.0648 0.2206 4.83 1.4e-06 ***
## Personthird 0.3041 0.1172 2.59 0.00948 **
## Numberpl -0.5925 0.1221 -4.85 1.2e-06 ***
## Tensenonpresent -0.4425 0.1213 -3.65 0.00027 ***
## Tenseother -0.0643 0.1979 -0.32 0.74534
## Switch_discsame -1.0277 0.1050 -9.79 < 2e-16 ***
## Clause_typemain -1.0556 0.2289 -4.61 4.0e-06 ***
## SexF 0.6630 0.2827 2.35 0.01899 *
## EO_lang 0.0524 0.0857 0.61 0.54105
## EO_culture -0.1243 0.1792 -0.69 0.48770
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Faetar
main_effects_faetar_heritage <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
#Clause_type +
Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = faetarheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(main_effects_faetar_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Preverbal_content_binary + Sex + Generation + (1 | Speaker)
## Data: faetarheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 150 188 -65 130 320
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.932 -0.256 -0.093 -0.053 4.221
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 2.06 1.43
## Number of obs: 330, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.1747 1.9255 -0.09 0.928
## Personthird -2.7347 1.6695 -1.64 0.101
## Numberpl -1.2231 0.9258 -1.32 0.186
## Grammatical_genderF 0.8639 0.5404 1.60 0.110
## Grammatical_genderOther -17.1181 362.0406 -0.05 0.962
## Switch_discsame -0.0555 0.5761 -0.10 0.923
## Preverbal_content_binaryyes -1.5478 0.8368 -1.85 0.064 .
## SexF -1.8801 1.2613 -1.49 0.136
## GenerationGen2 0.8235 1.1289 0.73 0.466
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?#No EO for FAETAR##Italian
main_effects_italian_heritage <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = italianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_italian_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Generation + (1 | Speaker)
## Data: italianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1024 1078 -501 1002 1027
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.622 -0.552 -0.355 -0.158 6.370
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.415 0.644
## Number of obs: 1038, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.797 0.425 -1.87 0.0611 .
## Personsecond 0.174 0.508 0.34 0.7324
## Personthird 0.607 0.209 2.91 0.0037 **
## Numberpl -1.270 0.206 -6.17 6.7e-10 ***
## Tensenonpresent -0.278 0.168 -1.65 0.0983 .
## Switch_discsame -1.143 0.170 -6.72 1.8e-11 ***
## Clause_typemain 0.256 0.208 1.23 0.2183
## Preverbal_content_binaryyes -0.341 0.196 -1.74 0.0821 .
## SexF 0.775 0.455 1.70 0.0887 .
## GenerationGen2 -0.179 0.472 -0.38 0.7050
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1main_effects_italian_heritage_EO <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
(1|Speaker),
data = italianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_italian_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + EO_lang + EO_culture + (1 | Speaker)
## Data: italianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1020 1080 -498 996 1026
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.659 -0.543 -0.355 -0.155 6.503
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.236 0.486
## Number of obs: 1038, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.904 0.314 -2.88 0.0040 **
## Personsecond 0.195 0.508 0.38 0.7008
## Personthird 0.611 0.209 2.93 0.0034 **
## Numberpl -1.266 0.206 -6.15 7.5e-10 ***
## Tensenonpresent -0.272 0.168 -1.62 0.1049
## Switch_discsame -1.140 0.170 -6.70 2.1e-11 ***
## Clause_typemain 0.262 0.207 1.26 0.2064
## Preverbal_content_binaryyes -0.337 0.196 -1.72 0.0850 .
## SexF 0.543 0.345 1.57 0.1153
## EO_lang -0.370 0.142 -2.61 0.0089 **
## EO_culture 0.032 0.181 0.18 0.8595
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Korean
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + Generation + (1 | Speaker)
## Data: koreanheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 541 580 -262 523 576
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.133 -0.561 -0.370 -0.109 11.774
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.144 0.379
## Number of obs: 585, groups: Speaker, 10
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.126 0.436 -2.58 0.0098 **
## Personsecond -1.960 0.633 -3.10 0.0019 **
## Personthird -3.187 0.730 -4.36 1.3e-05 ***
## Numberpl -0.744 0.341 -2.18 0.0290 *
## Switch_discsame -0.581 0.225 -2.59 0.0096 **
## Preverbal_content_binaryyes 0.622 0.269 2.31 0.0209 *
## SexF -0.703 0.344 -2.04 0.0411 *
## GenerationGen2 0.418 0.379 1.10 0.2701
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients## boundary (singular) fit: see help('isSingular')## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula:
## Prodrop ~ Person + Number + Switch_disc + Clause_type + Preverbal_content_binary +
## Sex + EO_lang + EO_culture + (1 | Speaker)
## Data: koreanheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 264 301 -122 244 303
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -0.992 -0.434 -0.327 -0.205 8.200
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0 0
## Number of obs: 313, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.349 0.628 -2.15 0.032 *
## Personsecond -1.725 1.061 -1.63 0.104
## Personthird -1.462 0.764 -1.91 0.056 .
## Numberpl -0.634 0.489 -1.30 0.195
## Switch_discsame -0.759 0.343 -2.21 0.027 *
## Preverbal_content_binaryyes 0.421 0.430 0.98 0.328
## SexF -1.021 0.427 -2.39 0.017 *
## EO_lang -0.220 0.217 -1.02 0.310
## EO_culture 0.597 0.252 2.37 0.018 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Polish
main_effects_polish_heritage <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
(1|Speaker),
data = polishheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(main_effects_polish_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + Generation +
## (1 | Speaker)
## Data: polishheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 391 438 -184 367 355
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.185 -0.539 -0.370 0.572 4.377
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.819 0.905
## Number of obs: 367, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.2525 0.5009 -2.50 0.0124 *
## Personsecond 0.4455 0.6682 0.67 0.5049
## Personthird 0.2798 0.3878 0.72 0.4705
## Numberpl -0.1055 0.3722 -0.28 0.7769
## Grammatical_genderF 0.6105 0.4664 1.31 0.1906
## Switch_discOther -15.6235 170.6667 -0.09 0.9271
## Switch_discsame -0.8413 0.2779 -3.03 0.0025 **
## Clause_typemain -0.0284 0.4494 -0.06 0.9497
## Preverbal_content_binaryyes 0.1651 0.5343 0.31 0.7573
## SexF 0.0619 0.7433 0.08 0.9337
## GenerationGen2 0.3664 0.6018 0.61 0.5426
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1main_effects_polish_heritage_EO <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
(1|Speaker),
data = polishheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(main_effects_polish_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + EO_lang +
## EO_culture + (1 | Speaker)
## Data: polishheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 351 400 -163 325 291
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.211 -0.563 -0.382 0.806 3.012
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.478 0.692
## Number of obs: 304, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.167 0.429 -2.72 0.0065 **
## Personsecond 0.640 0.701 0.91 0.3611
## Personthird 0.295 0.434 0.68 0.4966
## Numberpl -0.278 0.404 -0.69 0.4909
## Grammatical_genderF 0.442 0.522 0.85 0.3969
## Switch_discOther -15.552 512.000 -0.03 0.9758
## Switch_discsame -0.666 0.291 -2.29 0.0220 *
## Clause_typemain 0.151 0.469 0.32 0.7477
## Preverbal_content_binaryyes 0.109 0.544 0.20 0.8414
## SexF 0.632 0.877 0.72 0.4710
## EO_lang 0.355 0.403 0.88 0.3779
## EO_culture 0.210 0.183 1.15 0.2502
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?##Ukrainian
main_effects_ukrainian_heritage <- glmer(Prodrop ~
Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
# # Preverbal_content_binary +
Sex +
Generation
+ (1|Speaker),
data = ukrainianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(main_effects_ukrainian_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Sex + Generation + (1 | Speaker)
## Data: ukrainianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 389 432 -184 367 356
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.187 -0.543 -0.372 0.574 4.338
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.814 0.902
## Number of obs: 367, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.2479 0.4998 -2.50 0.0125 *
## Personsecond 0.4547 0.6672 0.68 0.4956
## Personthird 0.2763 0.3872 0.71 0.4755
## Numberpl -0.0991 0.3710 -0.27 0.7893
## Grammatical_genderF 0.6029 0.4653 1.30 0.1951
## Switch_discOther -15.5665 228.9734 -0.07 0.9458
## Switch_discsame -0.8336 0.2766 -3.01 0.0026 **
## Clause_typemain -0.0185 0.4488 -0.04 0.9671
## SexF 0.0691 0.7411 0.09 0.9257
## GenerationGen2 0.3682 0.6004 0.61 0.5397
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?main_effects_ukrainian_heritage_EO <- glmer(Prodrop ~
Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
# # Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
+ (1|Speaker),
data = ukrainianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge: degenerate Hessian with 1 negative eigenvaluessummary(main_effects_ukrainian_heritage_EO, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Sex + EO_lang + EO_culture + +(1 | Speaker)
## Data: ukrainianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 349 394 -163 325 292
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.211 -0.565 -0.384 0.803 3.002
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.471 0.687
## Number of obs: 304, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.163 0.414 -2.81 0.005 **
## Personsecond 0.648 0.691 0.94 0.349
## Personthird 0.293 0.431 0.68 0.498
## Numberpl -0.274 0.401 -0.68 0.495
## Grammatical_genderF 0.436 0.517 0.84 0.399
## Switch_discOther -15.469 1665.136 -0.01 0.993
## Switch_discsame -0.660 0.288 -2.29 0.022 *
## Clause_typemain 0.160 0.464 0.34 0.730
## SexF 0.642 0.857 0.75 0.454
## EO_lang 0.355 0.391 0.91 0.364
## EO_culture 0.211 0.180 1.17 0.242
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## optimizer (bobyqa) convergence code: 0 (OK)
## unable to evaluate scaled gradient
## Model failed to converge: degenerate Hessian with 1 negative eigenvalues##Interactions for heritage speakers
##Cantonese
interactions_cant_heritage <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
#Grammatical_gender*Sex +
#Grammatical_gender*Generation +
Tense*Sex +
Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
#Preverbal_content_binary*Sex +
#Preverbal_content_binary*Generation +
(1|Speaker),
data = cantoneseheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_cant_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + Generation + Person * Sex + Person * Generation + Number *
## Sex + Number * Generation + Tense * Sex + Tense * Generation +
## Switch_disc * Sex + Switch_disc * Generation + Clause_type *
## Sex + Clause_type * Generation + (1 | Speaker)
## Data: cantoneseheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 3216 3364 -1583 3166 2753
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -7.215 -0.802 0.406 0.685 2.908
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.396 0.629
## Number of obs: 2778, groups: Speaker, 28
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.46975 0.42545 5.81 6.4e-09 ***
## Personsecond 0.67767 0.32225 2.10 0.03547 *
## Personthird -0.00686 0.17223 -0.04 0.96825
## Numberpl -0.54448 0.17610 -3.09 0.00199 **
## Tensenonpresent -0.62772 0.17385 -3.61 0.00031 ***
## Tenseother -0.44983 0.29874 -1.51 0.13213
## Switch_discsame -1.08507 0.15893 -6.83 8.7e-12 ***
## Clause_typemain -1.43945 0.34054 -4.23 2.4e-05 ***
## SexF -0.06448 0.45978 -0.14 0.88848
## GenerationGen2 -0.54021 0.45731 -1.18 0.23749
## Personsecond:SexF -0.39683 0.38741 -1.02 0.30569
## Personthird:SexF 0.31204 0.21352 1.46 0.14391
## Personsecond:GenerationGen2 1.00477 0.38897 2.58 0.00979 **
## Personthird:GenerationGen2 0.37629 0.22076 1.70 0.08829 .
## Numberpl:SexF -0.36064 0.22549 -1.60 0.10975
## Numberpl:GenerationGen2 0.21014 0.23354 0.90 0.36823
## Tensenonpresent:SexF 0.71480 0.21489 3.33 0.00088 ***
## Tenseother:SexF 0.72419 0.36070 2.01 0.04467 *
## Tensenonpresent:GenerationGen2 -0.09955 0.21586 -0.46 0.64467
## Tenseother:GenerationGen2 0.05932 0.36217 0.16 0.86989
## Switch_discsame:SexF -0.06739 0.18889 -0.36 0.72128
## Switch_discsame:GenerationGen2 0.21768 0.18960 1.15 0.25093
## Clause_typemain:SexF 0.47885 0.36290 1.32 0.18700
## Clause_typemain:GenerationGen2 0.50343 0.35831 1.41 0.16002
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1interactions_cant_heritage_EO <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
#Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
Person*Sex +
Person*EO_lang +
Person*EO_culture +
Number*Sex +
Number*EO_lang +
Number*EO_culture +
#Grammatical_gender*Sex +
#Grammatical_gender*Generation +
Tense*Sex +
Tense*EO_lang +
Tense*EO_culture +
Switch_disc*Sex +
Switch_disc*EO_lang +
Switch_disc*EO_culture +
Clause_type*Sex +
Clause_type*EO_lang +
Clause_type*EO_culture +
#Preverbal_content_binary*Sex +
#Preverbal_content_binary*Generation +
(1|Speaker),
data = cantoneseheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_cant_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Sex + EO_lang + EO_culture + Person * Sex + Person * EO_lang +
## Person * EO_culture + Number * Sex + Number * EO_lang + Number *
## EO_culture + Tense * Sex + Tense * EO_lang + Tense * EO_culture +
## Switch_disc * Sex + Switch_disc * EO_lang + Switch_disc *
## EO_culture + Clause_type * Sex + Clause_type * EO_lang +
## Clause_type * EO_culture + (1 | Speaker)
## Data: cantoneseheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 2460 2646 -1197 2394 2055
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -5.213 -0.775 0.394 0.716 3.432
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.318 0.564
## Number of obs: 2088, groups: Speaker, 21
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.81213 0.40385 4.49 7.2e-06 ***
## Personsecond 1.51846 0.32325 4.70 2.6e-06 ***
## Personthird 0.44099 0.18195 2.42 0.01537 *
## Numberpl -0.40780 0.19488 -2.09 0.03639 *
## Tensenonpresent -0.80337 0.16969 -4.73 2.2e-06 ***
## Tenseother -0.64313 0.28564 -2.25 0.02435 *
## Switch_discsame -1.05514 0.15468 -6.82 9.0e-12 ***
## Clause_typemain -1.01329 0.34712 -2.92 0.00351 **
## SexF 0.51921 0.57108 0.91 0.36326
## EO_lang 0.04488 0.19504 0.23 0.81799
## EO_culture -0.48589 0.38168 -1.27 0.20301
## Personsecond:SexF -0.69637 0.47762 -1.46 0.14483
## Personthird:SexF 0.07494 0.25001 0.30 0.76436
## Personsecond:EO_lang 0.38928 0.11746 3.31 0.00092 ***
## Personthird:EO_lang -0.01504 0.08639 -0.17 0.86179
## Personsecond:EO_culture 0.28154 0.34507 0.82 0.41457
## Personthird:EO_culture 0.72273 0.17075 4.23 2.3e-05 ***
## Numberpl:SexF -0.49956 0.25947 -1.93 0.05419 .
## Numberpl:EO_lang 0.28630 0.08899 3.22 0.00129 **
## Numberpl:EO_culture -0.52445 0.18177 -2.89 0.00391 **
## Tensenonpresent:SexF 0.80100 0.26127 3.07 0.00217 **
## Tenseother:SexF 0.92963 0.45932 2.02 0.04298 *
## Tensenonpresent:EO_lang -0.04420 0.08479 -0.52 0.60214
## Tenseother:EO_lang 0.16082 0.13827 1.16 0.24480
## Tensenonpresent:EO_culture -0.16471 0.16798 -0.98 0.32682
## Tenseother:EO_culture -0.48307 0.29526 -1.64 0.10182
## Switch_discsame:SexF 0.04761 0.21973 0.22 0.82848
## Switch_discsame:EO_lang 0.06930 0.06849 1.01 0.31166
## Switch_discsame:EO_culture 0.00346 0.14163 0.02 0.98052
## Clause_typemain:SexF -0.03114 0.48205 -0.06 0.94849
## Clause_typemain:EO_lang -0.13976 0.17313 -0.81 0.41951
## Clause_typemain:EO_culture 0.41383 0.32962 1.26 0.20930
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Faetar
interactions_faet_heritage <- glmer(Prodrop ~ Person +
# Number +
# Grammatical_gender +
# Tense +
Switch_disc +
#Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
# Number*Sex +
# Number*Generation +
# Grammatical_gender*Sex +
# Grammatical_gender*Generation +
# Tense*Sex +
# Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
#Clause_type*Sex +
#Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = faetarheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_faet_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Switch_disc + Preverbal_content_binary + Sex +
## Generation + Person * Sex + Person * Generation + Switch_disc *
## Sex + Switch_disc * Generation + Preverbal_content_binary *
## Sex + Preverbal_content_binary * Generation + (1 | Speaker)
## Data: faetarheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 653 728 -311 621 795
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.353 -0.434 -0.310 -0.216 4.929
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.354 0.595
## Number of obs: 811, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.122 0.446 -2.51 0.012
## Personsecond -0.630 0.785 -0.80 0.422
## Personthird -0.669 0.387 -1.73 0.084
## Switch_discsame -1.190 0.518 -2.30 0.022
## Preverbal_content_binaryyes -0.597 0.426 -1.40 0.161
## SexF -0.188 0.606 -0.31 0.757
## GenerationGen2 -0.590 0.765 -0.77 0.441
## Personsecond:SexF 0.562 0.823 0.68 0.494
## Personthird:SexF -0.411 0.519 -0.79 0.428
## Personsecond:GenerationGen2 1.965 0.890 2.21 0.027
## Personthird:GenerationGen2 0.792 0.645 1.23 0.219
## Switch_discsame:SexF 1.260 0.566 2.23 0.026
## Switch_discsame:GenerationGen2 0.372 0.580 0.64 0.521
## Preverbal_content_binaryyes:SexF 0.753 0.541 1.39 0.164
## Preverbal_content_binaryyes:GenerationGen2 -0.552 0.656 -0.84 0.400
##
## (Intercept) *
## Personsecond
## Personthird .
## Switch_discsame *
## Preverbal_content_binaryyes
## SexF
## GenerationGen2
## Personsecond:SexF
## Personthird:SexF
## Personsecond:GenerationGen2 *
## Personthird:GenerationGen2
## Switch_discsame:SexF *
## Switch_discsame:GenerationGen2
## Preverbal_content_binaryyes:SexF
## Preverbal_content_binaryyes:GenerationGen2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#No Faetar EO##Italian
interactions_ital_heritage <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
# Grammatical_gender*Sex +
# Grammatical_gender*Generation +
Tense*Sex +
Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = italianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_ital_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + Generation + Person * Sex +
## Person * Generation + Number * Sex + Number * Generation +
## Tense * Sex + Tense * Generation + Switch_disc * Sex + Switch_disc *
## Generation + Clause_type * Sex + Clause_type * Generation +
## Preverbal_content_binary * Sex + Preverbal_content_binary *
## Generation + (1 | Speaker)
## Data: italianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1033 1157 -492 983 1013
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.672 -0.545 -0.345 -0.121 7.178
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.467 0.683
## Number of obs: 1038, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.4051 0.5481 -0.74 0.460
## Personsecond 1.1453 0.9423 1.22 0.224
## Personthird 0.5044 0.4612 1.09 0.274
## Numberpl -1.7387 0.4087 -4.25 2.1e-05
## Tensenonpresent -0.0500 0.3061 -0.16 0.870
## Switch_discsame -1.5682 0.3282 -4.78 1.8e-06
## Clause_typemain -0.1304 0.3683 -0.35 0.723
## Preverbal_content_binaryyes -0.1309 0.3424 -0.38 0.702
## SexF 0.9704 0.7031 1.38 0.168
## GenerationGen2 -0.8383 0.7380 -1.14 0.256
## Personsecond:SexF -0.5805 1.5325 -0.38 0.705
## Personthird:SexF 0.1566 0.5178 0.30 0.762
## Personsecond:GenerationGen2 -0.7394 1.4507 -0.51 0.610
## Personthird:GenerationGen2 0.0875 0.6148 0.14 0.887
## Numberpl:SexF 0.1449 0.4966 0.29 0.770
## Numberpl:GenerationGen2 0.5496 0.5524 0.99 0.320
## Tensenonpresent:SexF 0.8003 0.3967 2.02 0.044
## Tensenonpresent:GenerationGen2 -0.8893 0.4233 -2.10 0.036
## Switch_discsame:SexF -0.1994 0.3882 -0.51 0.608
## Switch_discsame:GenerationGen2 0.7088 0.4266 1.66 0.097
## Clause_typemain:SexF -0.5993 0.4859 -1.23 0.217
## Clause_typemain:GenerationGen2 0.9047 0.5084 1.78 0.075
## Preverbal_content_binaryyes:SexF 0.0131 0.4693 0.03 0.978
## Preverbal_content_binaryyes:GenerationGen2 -0.4610 0.4909 -0.94 0.348
##
## (Intercept)
## Personsecond
## Personthird
## Numberpl ***
## Tensenonpresent
## Switch_discsame ***
## Clause_typemain
## Preverbal_content_binaryyes
## SexF
## GenerationGen2
## Personsecond:SexF
## Personthird:SexF
## Personsecond:GenerationGen2
## Personthird:GenerationGen2
## Numberpl:SexF
## Numberpl:GenerationGen2
## Tensenonpresent:SexF *
## Tensenonpresent:GenerationGen2 *
## Switch_discsame:SexF
## Switch_discsame:GenerationGen2 .
## Clause_typemain:SexF
## Clause_typemain:GenerationGen2 .
## Preverbal_content_binaryyes:SexF
## Preverbal_content_binaryyes:GenerationGen2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1interactions_ital_heritage_EO <- glmer(Prodrop ~ Person +
Number +
# Grammatical_gender +
Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
Person*Sex +
Person*EO_lang +
Person*EO_culture +
Number*Sex +
Number*EO_lang +
Number*EO_culture +
# Grammatical_gender*Sex +
# Grammatical_gender*EO_lang +
Tense*Sex +
Tense*EO_lang +
Tense*EO_culture +
Switch_disc*Sex +
Switch_disc*EO_lang +
Switch_disc*EO_culture +
Clause_type*Sex +
Clause_type*EO_lang +
Clause_type*EO_culture +
Preverbal_content_binary*Sex +
Preverbal_content_binary*EO_lang +
Preverbal_content_binary*EO_culture +
(1|Speaker),
data = italianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))
summary(interactions_ital_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Tense + Switch_disc + Clause_type +
## Preverbal_content_binary + Sex + EO_lang + EO_culture + Person *
## Sex + Person * EO_lang + Person * EO_culture + Number * Sex +
## Number * EO_lang + Number * EO_culture + Tense * Sex + Tense *
## EO_lang + Tense * EO_culture + Switch_disc * Sex + Switch_disc *
## EO_lang + Switch_disc * EO_culture + Clause_type * Sex +
## Clause_type * EO_lang + Clause_type * EO_culture + Preverbal_content_binary *
## Sex + Preverbal_content_binary * EO_lang + Preverbal_content_binary *
## EO_culture + (1 | Speaker)
## Data: italianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 1039 1202 -486 973 1005
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.641 -0.557 -0.339 -0.079 11.926
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.314 0.56
## Number of obs: 1038, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.9935 0.4400 -2.26 0.024 *
## Personsecond 0.7547 1.3009 0.58 0.562
## Personthird 0.5842 0.3520 1.66 0.097 .
## Numberpl -1.6162 0.3549 -4.55 5.3e-06 ***
## Tensenonpresent -0.6146 0.2702 -2.27 0.023 *
## Switch_discsame -1.1196 0.2558 -4.38 1.2e-05 ***
## Clause_typemain 0.4941 0.3478 1.42 0.156
## Preverbal_content_binaryyes -0.2411 0.2906 -0.83 0.407
## SexF 0.5457 0.6007 0.91 0.364
## EO_lang -0.5334 0.2683 -1.99 0.047 *
## EO_culture -0.3876 0.3098 -1.25 0.211
## Personsecond:SexF -0.8790 1.3806 -0.64 0.524
## Personthird:SexF 0.0922 0.4517 0.20 0.838
## Personsecond:EO_lang -0.5141 0.6300 -0.82 0.414
## Personthird:EO_lang -0.0886 0.1878 -0.47 0.637
## Personsecond:EO_culture -0.1004 0.9901 -0.10 0.919
## Personthird:EO_culture -0.0250 0.2109 -0.12 0.906
## Numberpl:SexF 0.3648 0.4413 0.83 0.408
## Numberpl:EO_lang -0.2151 0.1990 -1.08 0.280
## Numberpl:EO_culture -0.0613 0.2267 -0.27 0.787
## Tensenonpresent:SexF 0.3267 0.3531 0.93 0.355
## Tensenonpresent:EO_lang -0.3174 0.1665 -1.91 0.057 .
## Tensenonpresent:EO_culture -0.2762 0.1829 -1.51 0.131
## Switch_discsame:SexF 0.1089 0.3563 0.31 0.760
## Switch_discsame:EO_lang 0.2158 0.1513 1.43 0.154
## Switch_discsame:EO_culture 0.3874 0.1769 2.19 0.029 *
## Clause_typemain:SexF -0.2082 0.4459 -0.47 0.641
## Clause_typemain:EO_lang 0.2861 0.2131 1.34 0.180
## Clause_typemain:EO_culture 0.3606 0.2215 1.63 0.103
## Preverbal_content_binaryyes:SexF -0.2489 0.4029 -0.62 0.537
## Preverbal_content_binaryyes:EO_lang 0.0574 0.1775 0.32 0.746
## Preverbal_content_binaryyes:EO_culture 0.0433 0.2101 0.21 0.837
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##Korean
interactions_kor_heritage <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
#Tense +
Switch_disc +
#Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
#Grammatical_gender*Sex +
#Grammatical_gender*Generation +
#Tense*Sex +
#Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
# Clause_type*Sex +
# Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = koreanheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficientssummary(interactions_kor_heritage, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Switch_disc + Preverbal_content_binary +
## Sex + Generation + Person * Sex + Person * Generation + Number *
## Sex + Number * Generation + Switch_disc * Sex + Switch_disc *
## Generation + Preverbal_content_binary * Sex + Preverbal_content_binary *
## Generation + (1 | Speaker)
## Data: koreanheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 547 630 -254 509 566
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.163 -0.533 -0.363 -0.086 11.609
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.184 0.429
## Number of obs: 585, groups: Speaker, 10
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.5078 0.6489 -0.78 0.4339
## Personsecond -1.5206 1.3635 -1.12 0.2648
## Personthird -2.3097 1.3381 -1.73 0.0843
## Numberpl -0.1357 0.6832 -0.20 0.8425
## Switch_discsame -0.2256 0.5276 -0.43 0.6690
## Preverbal_content_binaryyes -0.5514 0.6204 -0.89 0.3742
## SexF -0.2681 0.6292 -0.43 0.6701
## GenerationGen2 -0.7180 0.6745 -1.06 0.2871
## Personsecond:SexF 0.0964 1.4379 0.07 0.9465
## Personthird:SexF 1.1120 1.4942 0.74 0.4567
## Personsecond:GenerationGen2 -0.6981 1.4311 -0.49 0.6257
## Personthird:GenerationGen2 -1.8746 1.4974 -1.25 0.2106
## Numberpl:SexF -1.2014 0.7817 -1.54 0.1243
## Numberpl:GenerationGen2 -0.1976 0.7668 -0.26 0.7967
## Switch_discsame:SexF -0.4447 0.4684 -0.95 0.3424
## Switch_discsame:GenerationGen2 -0.2189 0.5329 -0.41 0.6812
## Preverbal_content_binaryyes:SexF -0.1646 0.5737 -0.29 0.7742
## Preverbal_content_binaryyes:GenerationGen2 1.7081 0.6241 2.74 0.0062
##
## (Intercept)
## Personsecond
## Personthird .
## Numberpl
## Switch_discsame
## Preverbal_content_binaryyes
## SexF
## GenerationGen2
## Personsecond:SexF
## Personthird:SexF
## Personsecond:GenerationGen2
## Personthird:GenerationGen2
## Numberpl:SexF
## Numberpl:GenerationGen2
## Switch_discsame:SexF
## Switch_discsame:GenerationGen2
## Preverbal_content_binaryyes:SexF
## Preverbal_content_binaryyes:GenerationGen2 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients interactions_kor_heritage_EO <- glmer(Prodrop ~ Person +
Number +
#Grammatical_gender +
#Tense +
Switch_disc +
# Clause_type +
Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
Person*Sex +
Person*EO_lang +
Person*EO_culture +
Number*Sex +
Number*EO_lang +
Number*EO_culture +
#Grammatical_gender*Sex +
#Grammatical_gender*EO_lang +
#Grammatical_gender*EO_culture +
#Tense*Sex +
#Tense*EO_lang +
#Tense*EO_culture +
Switch_disc*Sex +
Switch_disc*EO_lang +
Switch_disc*EO_culture +
# Clause_type*Sex +
# Clause_type*EO_lang +
# Clause_type*EO_culture +
Preverbal_content_binary*Sex +
Preverbal_content_binary*EO_lang +
Preverbal_content_binary*EO_culture +
(1|Speaker),
data = koreanheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 5 columns / coefficients## boundary (singular) fit: see help('isSingular') summary(interactions_kor_heritage_EO, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Switch_disc + Preverbal_content_binary +
## Sex + EO_lang + EO_culture + Person * Sex + Person * EO_lang +
## Person * EO_culture + Number * Sex + Number * EO_lang + Number *
## EO_culture + Switch_disc * Sex + Switch_disc * EO_lang +
## Switch_disc * EO_culture + Preverbal_content_binary * Sex +
## Preverbal_content_binary * EO_lang + Preverbal_content_binary *
## EO_culture + (1 | Speaker)
## Data: koreanheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 262 352 -107 214 289
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.171 -0.441 -0.245 0.000 4.141
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 1.32e-38 1.15e-19
## Number of obs: 313, groups: Speaker, 6
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -3.01e+00 1.45e+00 -2.07 0.038 *
## Personsecond -1.96e+02 3.42e+05 0.00 1.000
## Personthird -1.18e+02 1.48e+05 0.00 0.999
## Numberpl 1.00e+00 1.56e+00 0.64 0.521
## Switch_discsame -6.37e-01 1.06e+00 -0.60 0.548
## Preverbal_content_binaryyes 1.67e+00 1.45e+00 1.15 0.248
## SexF 1.16e+00 1.31e+00 0.88 0.378
## EO_lang -6.64e-01 6.44e-01 -1.03 0.303
## EO_culture 3.25e-01 7.30e-01 0.44 0.656
## Personsecond:SexF 5.58e+00 5.60e+04 0.00 1.000
## Personthird:SexF 1.60e+02 2.24e+05 0.00 0.999
## Personsecond:EO_lang -9.46e+01 1.64e+05 0.00 1.000
## Personthird:EO_lang -7.10e+00 5.46e+04 0.00 1.000
## Personthird:EO_culture -9.06e+01 4.15e+04 0.00 0.998
## Numberpl:SexF -1.12e-01 1.59e+00 -0.07 0.944
## Numberpl:EO_lang 8.41e-01 7.07e-01 1.19 0.234
## Numberpl:EO_culture -1.13e+00 7.08e-01 -1.59 0.111
## Switch_discsame:SexF -5.31e-01 9.35e-01 -0.57 0.571
## Switch_discsame:EO_lang -2.57e-01 4.91e-01 -0.52 0.601
## Switch_discsame:EO_culture -6.13e-01 5.33e-01 -1.15 0.250
## Preverbal_content_binaryyes:SexF -2.25e+00 1.31e+00 -1.73 0.084 .
## Preverbal_content_binaryyes:EO_lang 4.10e-01 6.52e-01 0.63 0.530
## Preverbal_content_binaryyes:EO_culture 1.06e+00 6.98e-01 1.53 0.127
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 5 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## boundary (singular) fit: see help('isSingular')##Polish
interactions_polish_heritage <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
Grammatical_gender*Sex +
Grammatical_gender*Generation +
# Tense*Sex +
# Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
Preverbal_content_binary*Sex +
Preverbal_content_binary*Generation +
(1|Speaker),
data = polishheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge: degenerate Hessian with 1 negative eigenvaluessummary(interactions_polish_heritage, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + Generation +
## Person * Sex + Person * Generation + Number * Sex + Number *
## Generation + Grammatical_gender * Sex + Grammatical_gender *
## Generation + Switch_disc * Sex + Switch_disc * Generation +
## Clause_type * Sex + Clause_type * Generation + Preverbal_content_binary *
## Sex + Preverbal_content_binary * Generation + (1 | Speaker)
## Data: polishheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 395 492 -172 345 342
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.723 -0.516 -0.330 0.380 4.332
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.798 0.894
## Number of obs: 367, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value
## (Intercept) -1.2381 0.5503 -2.25
## Personsecond 2.0050 1.0786 1.86
## Personthird 0.5889 0.5878 1.00
## Numberpl -0.0162 0.5337 -0.03
## Grammatical_genderF 0.7982 0.8393 0.95
## Switch_discOther -18.7809 7481.1326 0.00
## Switch_discsame -1.1033 0.4421 -2.50
## Clause_typemain -18.3207 2390.3204 -0.01
## Preverbal_content_binaryyes -0.0720 1.0019 -0.07
## SexF 0.5275 1.1746 0.45
## GenerationGen2 0.1928 0.7565 0.25
## Personthird:SexF 0.2508 1.0075 0.25
## Personsecond:GenerationGen2 -3.7609 1.6385 -2.30
## Personthird:GenerationGen2 -0.6481 1.0286 -0.63
## Numberpl:SexF -0.2415 1.0234 -0.24
## Numberpl:GenerationGen2 -0.3705 0.8397 -0.44
## Grammatical_genderF:SexF -0.7965 1.4306 -0.56
## Grammatical_genderF:GenerationGen2 0.3213 1.2207 0.26
## Switch_discsame:SexF -0.4777 0.6628 -0.72
## Switch_discsame:GenerationGen2 0.5565 0.6090 0.91
## Clause_typemain:SexF -0.4725 1.0879 -0.43
## Clause_typemain:GenerationGen2 19.2773 2390.3205 0.01
## Preverbal_content_binaryyes:SexF -0.0649 1.2779 -0.05
## Preverbal_content_binaryyes:GenerationGen2 0.5408 1.3559 0.40
## Pr(>|z|)
## (Intercept) 0.024 *
## Personsecond 0.063 .
## Personthird 0.316
## Numberpl 0.976
## Grammatical_genderF 0.342
## Switch_discOther 0.998
## Switch_discsame 0.013 *
## Clause_typemain 0.994
## Preverbal_content_binaryyes 0.943
## SexF 0.653
## GenerationGen2 0.799
## Personthird:SexF 0.803
## Personsecond:GenerationGen2 0.022 *
## Personthird:GenerationGen2 0.529
## Numberpl:SexF 0.813
## Numberpl:GenerationGen2 0.659
## Grammatical_genderF:SexF 0.578
## Grammatical_genderF:GenerationGen2 0.792
## Switch_discsame:SexF 0.471
## Switch_discsame:GenerationGen2 0.361
## Clause_typemain:SexF 0.664
## Clause_typemain:GenerationGen2 0.994
## Preverbal_content_binaryyes:SexF 0.959
## Preverbal_content_binaryyes:GenerationGen2 0.690
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## unable to evaluate scaled gradient
## Model failed to converge: degenerate Hessian with 1 negative eigenvaluesinteractions_polish_heritage_EO <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
Person*Sex +
Person*EO_lang +
Person*EO_culture +
Number*Sex +
Number*EO_lang +
Number*EO_culture +
Grammatical_gender*Sex +
Grammatical_gender*EO_lang +
Grammatical_gender*EO_culture +
# Tense*Sex +
# Tense*EO_lang +
# Tense*EO_culture +
Switch_disc*Sex +
Switch_disc*EO_lang +
Switch_disc*EO_culture +
Clause_type*Sex +
Clause_type*EO_lang +
Clause_type*EO_culture +
Preverbal_content_binary*Sex +
Preverbal_content_binary*EO_lang +
Preverbal_content_binary*EO_culture +
(1|Speaker),
data = polishheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients## Warning in optwrap(optimizer, devfun, start, rho$lower, control = control, :
## convergence code 1 from bobyqa: bobyqa -- maximum number of function
## evaluations exceeded## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?summary(interactions_polish_heritage_EO, correl=F)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Preverbal_content_binary + Sex + EO_lang +
## EO_culture + Person * Sex + Person * EO_lang + Person * EO_culture +
## Number * Sex + Number * EO_lang + Number * EO_culture + Grammatical_gender *
## Sex + Grammatical_gender * EO_lang + Grammatical_gender *
## EO_culture + Switch_disc * Sex + Switch_disc * EO_lang +
## Switch_disc * EO_culture + Clause_type * Sex + Clause_type *
## EO_lang + Clause_type * EO_culture + Preverbal_content_binary *
## Sex + Preverbal_content_binary * EO_lang + Preverbal_content_binary *
## EO_culture + (1 | Speaker)
## Data: polishheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 356 482 -144 288 270
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.936 -0.599 -0.338 0.465 5.530
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.158 0.397
## Number of obs: 304, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.8218 0.4295 -1.91 0.056 .
## Personsecond -0.0399 2.2114 -0.02 0.986
## Personthird -3.1616 3.5958 -0.88 0.379
## Numberpl -0.5226 1.0798 -0.48 0.628
## Grammatical_genderF 2.3731 1.9059 1.25 0.213
## Switch_discOther -16.4679 219.4300 -0.08 0.940
## Switch_discsame -1.2285 0.5324 -2.31 0.021 *
## Clause_typemain -1.5757 1.8762 -0.84 0.401
## Preverbal_content_binaryyes -1.3787 1.7983 -0.77 0.443
## SexF 1.2823 1.4245 0.90 0.368
## EO_lang 0.3600 0.4456 0.81 0.419
## EO_culture 0.0807 0.2050 0.39 0.694
## Personthird:SexF 6.1397 5.6903 1.08 0.281
## Personsecond:EO_lang -3.5638 1.9035 -1.87 0.061 .
## Personthird:EO_lang -4.8673 3.8072 -1.28 0.201
## Personsecond:EO_culture 0.3968 0.6231 0.64 0.524
## Personthird:EO_culture 0.7030 0.7160 0.98 0.326
## Numberpl:SexF -0.4064 2.0255 -0.20 0.841
## Numberpl:EO_lang -0.5605 1.3520 -0.41 0.678
## Numberpl:EO_culture -0.1265 0.3021 -0.42 0.675
## Grammatical_genderF:SexF -4.6250 2.9967 -1.54 0.123
## Grammatical_genderF:EO_lang 2.5669 2.1943 1.17 0.242
## Grammatical_genderF:EO_culture -0.5449 0.3995 -1.36 0.173
## Switch_discsame:SexF 0.7645 0.9149 0.84 0.403
## Switch_discsame:EO_lang 0.5594 0.5095 1.10 0.272
## Switch_discOther:EO_culture 0.0863 242.5886 0.00 1.000
## Switch_discsame:EO_culture 0.2578 0.2296 1.12 0.262
## Clause_typemain:SexF 0.1398 2.0586 0.07 0.946
## Clause_typemain:EO_lang 5.7611 3.3852 1.70 0.089 .
## Clause_typemain:EO_culture 0.4569 0.8251 0.55 0.580
## Preverbal_content_binaryyes:SexF 2.9442 2.5753 1.14 0.253
## Preverbal_content_binaryyes:EO_lang -2.9048 2.3163 -1.25 0.210
## Preverbal_content_binaryyes:EO_culture 0.3348 0.5247 0.64 0.523
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients
## optimizer (bobyqa) convergence code: 1 (bobyqa -- maximum number of function evaluations exceeded)
## Model is nearly unidentifiable: large eigenvalue ratio
## - Rescale variables?##Ukrainian
interactions_ukrainian_heritage <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
Generation +
Person*Sex +
Person*Generation +
Number*Sex +
Number*Generation +
Grammatical_gender*Sex +
Grammatical_gender*Generation +
# Tense*Sex +
# Tense*Generation +
Switch_disc*Sex +
Switch_disc*Generation +
Clause_type*Sex +
Clause_type*Generation +
# Preverbal_content_binary*Sex +
# Preverbal_content_binary*Generation +
(1|Speaker),
data = ukrainianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge: degenerate Hessian with 1 negative eigenvaluessummary(interactions_ukrainian_heritage, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Sex + Generation + Person * Sex + Person *
## Generation + Number * Sex + Number * Generation + Grammatical_gender *
## Sex + Grammatical_gender * Generation + Switch_disc * Sex +
## Switch_disc * Generation + Clause_type * Sex + Clause_type *
## Generation + (1 | Speaker)
## Data: ukrainianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 389 475 -173 345 345
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.712 -0.518 -0.333 0.380 4.336
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.772 0.879
## Number of obs: 367, groups: Speaker, 13
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.2393 0.5428 -2.28 0.022 *
## Personsecond 2.0065 1.0710 1.87 0.061 .
## Personthird 0.5980 0.5872 1.02 0.309
## Numberpl -0.0209 0.5336 -0.04 0.969
## Grammatical_genderF 0.7967 0.8346 0.95 0.340
## Switch_discOther -18.6053 7530.7171 0.00 0.998
## Switch_discsame -1.1080 0.4416 -2.51 0.012 *
## Clause_typemain -18.2974 2358.5419 -0.01 0.994
## SexF 0.5343 1.1642 0.46 0.646
## GenerationGen2 0.2150 0.7463 0.29 0.773
## Personthird:SexF 0.2227 1.0011 0.22 0.824
## Personsecond:GenerationGen2 -3.7526 1.6315 -2.30 0.021 *
## Personthird:GenerationGen2 -0.6578 1.0254 -0.64 0.521
## Numberpl:SexF -0.2458 1.0197 -0.24 0.809
## Numberpl:GenerationGen2 -0.3528 0.8355 -0.42 0.673
## Grammatical_genderF:SexF -0.8050 1.4239 -0.57 0.572
## Grammatical_genderF:GenerationGen2 0.3244 1.2147 0.27 0.789
## Switch_discsame:SexF -0.4446 0.6555 -0.68 0.498
## Switch_discsame:GenerationGen2 0.5835 0.6056 0.96 0.335
## Clause_typemain:SexF -0.5130 1.0871 -0.47 0.637
## Clause_typemain:GenerationGen2 19.2742 2358.5420 0.01 0.993
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## unable to evaluate scaled gradient
## Model failed to converge: degenerate Hessian with 1 negative eigenvaluesinteractions_ukrainian_heritage_EO <- glmer(Prodrop ~ Person +
Number +
Grammatical_gender +
# Tense +
Switch_disc +
Clause_type +
# Preverbal_content_binary +
Sex +
EO_lang +
EO_culture +
Person*Sex +
Person*EO_lang +
Person*EO_culture +
Number*Sex +
Number*EO_lang +
Number*EO_culture +
Grammatical_gender*Sex +
Grammatical_gender*EO_lang +
Grammatical_gender*EO_culture +
# Tense*Sex +
# Tense*EO_lang +
# Tense*EO_culture +
Switch_disc*Sex +
Switch_disc*EO_lang +
Switch_disc*EO_culture +
Clause_type*Sex +
Clause_type*EO_lang +
Clause_type*EO_culture +
# Preverbal_content_binary*Sex +
# Preverbal_content_binary*EO_lang +
# Preverbal_content_binary*EO_culture +
(1|Speaker),
data = ukrainianheritagedata, family = "binomial", control = glmerControl(optCtrl = list(maxfun = 2e4), optimizer = "bobyqa"))## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## unable to evaluate scaled gradient## Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
## Model failed to converge: degenerate Hessian with 2 negative eigenvaluessummary(interactions_ukrainian_heritage_EO, correl=F)## Warning in vcov.merMod(object, use.hessian = use.hessian): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Warning in vcov.merMod(object, correlation = correlation, sigm = sig): variance-covariance matrix computed from finite-difference Hessian is
## not positive definite or contains NA values: falling back to var-cov estimated from RX## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: Prodrop ~ Person + Number + Grammatical_gender + Switch_disc +
## Clause_type + Sex + EO_lang + EO_culture + Person * Sex +
## Person * EO_lang + Person * EO_culture + Number * Sex + Number *
## EO_lang + Number * EO_culture + Grammatical_gender * Sex +
## Grammatical_gender * EO_lang + Grammatical_gender * EO_culture +
## Switch_disc * Sex + Switch_disc * EO_lang + Switch_disc *
## EO_culture + Clause_type * Sex + Clause_type * EO_lang +
## Clause_type * EO_culture + (1 | Speaker)
## Data: ukrainianheritagedata
## Control: glmerControl(optCtrl = list(maxfun = 20000), optimizer = "bobyqa")
##
## AIC BIC logLik deviance df.resid
## 350 462 -145 290 274
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.935 -0.601 -0.334 0.512 4.765
##
## Random effects:
## Groups Name Variance Std.Dev.
## Speaker (Intercept) 0.295 0.543
## Number of obs: 304, groups: Speaker, 11
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.932 0.435 -2.14 0.032 *
## Personsecond -0.159 2.116 -0.08 0.940
## Personthird -2.916 3.436 -0.85 0.396
## Numberpl -0.378 1.087 -0.35 0.728
## Grammatical_genderF 2.404 2.065 1.16 0.244
## Switch_discOther -16.029 3206.037 0.00 0.996
## Switch_discsame -1.152 0.518 -2.23 0.026 *
## Clause_typemain -2.238 1.773 -1.26 0.207
## SexF 1.524 1.443 1.06 0.291
## EO_lang 0.238 0.458 0.52 0.603
## EO_culture 0.127 0.217 0.59 0.558
## Personthird:SexF 5.673 5.442 1.04 0.297
## Personsecond:EO_lang -3.529 1.818 -1.94 0.052 .
## Personthird:EO_lang -4.490 3.591 -1.25 0.211
## Personsecond:EO_culture 0.454 0.605 0.75 0.453
## Personthird:EO_culture 0.662 0.697 0.95 0.342
## Numberpl:SexF -0.632 2.031 -0.31 0.756
## Numberpl:EO_lang -0.324 1.333 -0.24 0.808
## Numberpl:EO_culture -0.141 0.300 -0.47 0.637
## Grammatical_genderF:SexF -4.737 3.244 -1.46 0.144
## Grammatical_genderF:EO_lang 2.660 2.384 1.12 0.265
## Grammatical_genderF:EO_culture -0.584 0.415 -1.41 0.160
## Switch_discsame:SexF 0.767 0.900 0.85 0.394
## Switch_discsame:EO_lang 0.525 0.501 1.05 0.294
## Switch_discOther:EO_culture 0.302 2780.475 0.00 1.000
## Switch_discsame:EO_culture 0.233 0.224 1.04 0.299
## Clause_typemain:SexF 1.416 1.672 0.85 0.397
## Clause_typemain:EO_lang 3.692 1.755 2.10 0.035 *
## Clause_typemain:EO_culture 0.586 0.703 0.83 0.405
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 3 columns / coefficients
## optimizer (bobyqa) convergence code: 0 (OK)
## unable to evaluate scaled gradient
## Model failed to converge: degenerate Hessian with 2 negative eigenvalues##Models for all languages and all speakers
##Main effects for all languages
##Interactions for all languages