General

This file belongs to the dataset published in:
Van Boven, Cindy. 2023. Annotations of plural reduplication in NGT (corpus & elicited data). University of Amsterdam / Amsterdam University of Applied Sciences. DOI: 10.21942/uva.23260814.

Title: Phonological restrictions on nominal pluralization in Sign Language of the Netherlands: Evidence from corpus and elicited data

Author: C. van Boven

Contact: c.m.j.vanboven@uva.nl (Cindy van Boven)

Funding: This study is part of the project “Morphological reduplication in Sign Language of the Netherlands: A typological and theoretical perspective”, part of the research programme PhDs in the Humanities with project number PGW19.003, funded by the Dutch Research Council (NWO) (PhD student: C. van Boven; supervisors: dr. R. Pfau, dr. S. Hamann)

Present research

Metadata

Data collection period: January - February 2020

Data collected by: C. van Boven

Topic

This study investigates nominal reduplication to express plurality in Sign Language of the Netherlands (NGT). The aim is to offer a comprehensive description of nominal pluralization processes in the language, focusing mostly on reduplication, adressing potential restrictions on this process.

Research questions

1) Do phonological noun types in NGT differ with respect to the pluralization strategies they undergo, i.e. does NGT pluralization involve phonologically triggered allomorphy?

2) Do numerals/quantifiers ‘block’ reduplication, i.e., does NGT allow for NP-internal number agreement?

3) Does the number of syllables in the mouthing influence the number of repetitions in reduplication?

In addition to statistical analyses aiming to answer these questions, this file contains an analysis of the inter-rater agreement for the annotations of mouthings.

Methods

This study combines two methods: corpus analysis and data elicitation.

Corpus search: The starting point is a corpus search in the Corpus NGT (Crasborn et al. 2008; Crasborn and Zwitserlood 2008). The annotated part of the corpus was searched for plural nouns, which, according to the Corpus NGT Annotation Conventions, are annotated for ‘.PL’ in their gloss (Crasborn et al. 2015: 15). I therefore searched for “.pl” on the Gloss tier. I also searched for signs that appear in a plural context, but are not glossed for .PL: I searched for the plural of 12 frequent Dutch nouns on the Translation tier.

Data elicitation: For the purpose of this study, a gap-filling task was designed: Participants were presented with signed (carrier) sentences in which the plural noun was omitted and replaced by a question mark sign. Participants were asked to repeat the sentence and fill in the gap, based on a picture that shows the targeted plural noun. 21 nouns with different phonological specifications were targeted. Each noun was targeted twice: once in a sentence without a numeral/quantifier, and once preceded by a numeral/quantifier, resulting in a total of 42 carrier sentences for plural nouns. Moreover, 11 sentences eliciting singular nouns were added, in order to (i) ensure that participants did not simply reduplicate all signs, because they realized that the task targets plurals, and (ii) elicit the singular forms of the nouns that have an inherent repetition in their citation form for comparison.

References:
Crasborn, Onno & Inge Zwitserlood. 2008. The Corpus NGT: An online corpus for professionals and laymen. In Onno Crasborn, Thomas Hanke, Eleni Efthimiou, Inge Zwitserlood & Ernst Thoutenhoofd (eds.), Construction and exploitation of sign language corpora. 3rd workshop on the representation and processing of sign languages, 44–49. Paris: ELDA.
Crasborn, Onno, Inge Zwitserlood & Johan Ros. 2008. The Corpus NGT: A digital open access corpus of movies and annotations of Sign Language of the Netherlands. http://hdl.handle.net/hdl:1839/00-0000-0000-0004-DF8E-6 (accessed 26 March 2020).
Crasborn, Onno, Richard Bank, Inge Zwitserlood, Els van der Kooij, Anne Meijer & Anna Sáfár. 2015. Annotation conventions for the Corpus NGT, version 3. https://doi.org/10.13140/RG.2.1.1779.4649.

Data

297 plural nouns were extracted from the corpus and divided into four phonological noun types: 88 body-anchored nouns, 194 lateral nouns, 11 midsagittal nouns, and 4 complex movement nouns. 189 nominal plurals were elicited from five deaf native NGT signers (one male, four female, age range 25–62, mean age 38.4). The elicited nominal plurals were divided in the same categories: 97 body-anchored nouns, 30 lateral nouns, 26 midsagittal nouns, and 36 complex movement nouns.

Loading data

data.plurals.all <- read.csv("plural_reduplication _annotations.csv", header = TRUE, sep = ";")


class(data.plurals.all)

## [1] "data.frame"

head(data.plurals.all)

##   Participant Data_type    Noun Noun_type Noun_type..spec Strategy1 Strategy2 Zero Num Number_repetitions Mouthing Syllables_mouthing Corpus.NGT.file.number      Corpus.NGT.time.code
## 1        s001    corpus DING.PL         L            <NA>    simple    simple    0   0                 NA     none                  0               CNGT0098 00:01:13.520-00:01:14.720
## 2        s001    corpus KIND.PL         L            <NA>       sim       sim    0   0                  0    kind'                  1               CNGT0098 00:04:56.360-00:04:56.805
## 3        s001    corpus  SCHOOL         L            <NA>      zero      zero    1   0                  0  school'                  1               CNGT0099 00:04:09.760-00:04:10.280
## 4        s001    corpus  SCHOOL         L            <NA>      zero      zero    1   0                  0  unclear                 NA               CNGT0099 00:04:17.280-00:04:17.400
## 5        s001    corpus KIND.PL         L            <NA>       sim       sim    0   0                  0    kind'                  1               CNGT0099 00:00:42.960-00:00:43.960
## 6        s002    corpus KIND.PL         L            <NA>  sideward  sideward    0   0                  2  unclear                 NA               CNGT0098 00:02:45.160-00:02:45.600

colnames(data.plurals.all)

##  [1] "Participant"            "Data_type"              "Noun"                   "Noun_type"              "Noun_type..spec"        "Strategy1"              "Strategy2"              "Zero"                   "Num"                    "Number_repetitions"     "Mouthing"               "Syllables_mouthing"     "Corpus.NGT.file.number" "Corpus.NGT.time.code"

data.plurals.all$Participant <- as.factor(data.plurals.all$Participant)
data.plurals.all$Data_type <- as.factor(data.plurals.all$Data_type)
data.plurals.all$Noun <- as.factor(data.plurals.all$Noun)
data.plurals.all$Noun_type <- as.factor(data.plurals.all$Noun_type)
data.plurals.all$Noun_type..spec <- as.factor(data.plurals.all$Noun_type..spec)
data.plurals.all$Strategy1 <- as.factor(data.plurals.all$Strategy1)
data.plurals.all$Strategy2 <- as.factor(data.plurals.all$Strategy2)
data.plurals.all$Zero <- as.factor(data.plurals.all$Zero)
data.plurals.all$Num <- as.factor(data.plurals.all$Num)
data.plurals.all$Number_repetitions <- as.numeric(data.plurals.all$Number_repetitions)
data.plurals.all$Mouthing <- as.factor(data.plurals.all$Mouthing)
data.plurals.all$Syllables_mouthing <- as.numeric(data.plurals.all$Syllables_mouthing)
data.plurals.all$Corpus.NGT.file.number <- as.factor(data.plurals.all$Corpus.NGT.file.number)
data.plurals.all$Corpus.NGT.time.code <- as.factor(data.plurals.all$Corpus.NGT.time.code)


#Check structure in data.plurals.all
str(data.plurals.all)

## 'data.frame':    486 obs. of  14 variables:
##  $ Participant           : Factor w/ 64 levels "p02","p03","p04",..: 6 6 6 6 6 7 7 7 7 8 ...
##  $ Data_type             : Factor w/ 2 levels "corpus","elicited": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Noun                  : Factor w/ 58 levels "(KLEIN)KIND",..: 14 25 47 47 25 25 25 25 47 51 ...
##  $ Noun_type             : Factor w/ 4 levels "B","C","L","M": 3 3 3 3 3 3 3 3 3 3 ...
##  $ Noun_type..spec       : Factor w/ 5 levels "alt","circ","contact",..: NA NA NA NA NA NA NA NA NA NA ...
##  $ Strategy1             : Factor w/ 6 levels "other","sideward",..: 5 4 6 6 4 2 2 6 5 2 ...
##  $ Strategy2             : Factor w/ 5 levels "other","sideward",..: 4 3 5 5 3 2 2 5 4 2 ...
##  $ Zero                  : Factor w/ 2 levels "0","1": 1 1 2 2 1 1 1 2 1 1 ...
##  $ Num                   : Factor w/ 2 levels "0","1": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Number_repetitions    : num  NA 0 0 0 0 2 2 0 1 3 ...
##  $ Mouthing              : Factor w/ 119 levels "'meisjesss' (I think s spreads over PTround)",..: 85 47 100 110 47 110 53 47 100 38 ...
##  $ Syllables_mouthing    : num  0 1 1 NA 1 NA 3 1 1 4 ...
##  $ Corpus.NGT.file.number: Factor w/ 131 levels "CNGT0004","CNGT0007",..: 17 17 18 18 18 17 17 17 18 1 ...
##  $ Corpus.NGT.time.code  : Factor w/ 290 levels "00:00:04.640-00:00:06.720",..: 136 271 257 260 76 214 235 233 191 231 ...

summary(data.plurals.all)

##   Participant     Data_type         Noun     Noun_type  Noun_type..spec        Strategy1      Strategy2   Zero    Num     Number_repetitions      Mouthing   Syllables_mouthing Corpus.NGT.file.number                Corpus.NGT.time.code
##  p02    : 41   corpus  :297   KIND.PL :105   B:185     alt      : 10    other       :  2   other   :  2   0:338   0:330   Min.   :0.0000     none     : 68   Min.   :0.000      none    :189           none                     :189      
##  p03    : 40   elicited:189   MENS.PL : 43   C: 40     circ     : 17    sideward    :169   sideward:197   1:148   1:156   1st Qu.:0.0000     unclear  : 56   1st Qu.:1.000      CNGT0256: 10           00:02:22.400-00:02:22.720:  3      
##  p04    : 39                  DING.PL : 36   L:224     contact  : 60    sideward.sim: 28   sim     :  6                   Median :1.0000     kinderen': 38   Median :1.000      CNGT0298: 10           00:00:14.370-00:00:14.700:  2      
##  p06    : 36                  SCHOOL  : 24   M: 37     nocontact: 37    sim         :  6   simple  :133                   Mean   :0.7824     kind'    : 32   Mean   :1.484      CNGT0014:  9           00:00:33.040-00:00:33.240:  2      
##  p05    : 33                  PROBLEEM: 23             rep      :  9    simple      :133   zero    :148                   3rd Qu.:1.0000     mens'    : 26   3rd Qu.:2.000      CNGT1684:  7           00:00:34.800-00:00:35.000:  2      
##  s014   : 22                  VROUW   : 15             NA's     :353    zero        :148                                  Max.   :4.0000     mensen'  : 19   Max.   :4.000      CNGT0136:  6           00:01:42.360-00:01:42.800:  2      
##  (Other):275                  (Other) :240                                                                                NA's   :77         (Other)  :247   NA's   :56         (Other) :255           (Other)                  :286

Trimming the data: Removing nouns for which number of repetitions and/or mouthing was unclear

#First remove irrelevant column with many NA's:
data.plurals.trimmed <- subset (data.plurals.all, select = -Noun_type..spec)
#data.plurals.trimmed

#Then remove all rows with NA's:
data.plurals.withoutNA <- na.omit(data.plurals.trimmed)
#data.plurals.withoutNA

#Check structure in data.plurals.withoutNA
str(data.plurals.withoutNA)

## 'data.frame':    363 obs. of  13 variables:
##  $ Participant           : Factor w/ 64 levels "p02","p03","p04",..: 6 6 6 7 7 7 8 8 8 8 ...
##  $ Data_type             : Factor w/ 2 levels "corpus","elicited": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Noun                  : Factor w/ 58 levels "(KLEIN)KIND",..: 25 47 25 25 25 47 51 25 25 14 ...
##  $ Noun_type             : Factor w/ 4 levels "B","C","L","M": 3 3 3 3 3 3 3 3 3 3 ...
##  $ Strategy1             : Factor w/ 6 levels "other","sideward",..: 4 6 4 2 6 5 2 3 2 6 ...
##  $ Strategy2             : Factor w/ 5 levels "other","sideward",..: 3 5 3 2 5 4 2 2 2 5 ...
##  $ Zero                  : Factor w/ 2 levels "0","1": 1 2 1 1 2 1 1 1 1 2 ...
##  $ Num                   : Factor w/ 2 levels "0","1": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Number_repetitions    : num  0 0 0 2 0 1 3 1 3 0 ...
##  $ Mouthing              : Factor w/ 119 levels "'meisjesss' (I think s spreads over PTround)",..: 47 100 47 53 47 100 38 53 53 85 ...
##  $ Syllables_mouthing    : num  1 1 1 3 1 1 4 3 3 0 ...
##  $ Corpus.NGT.file.number: Factor w/ 131 levels "CNGT0004","CNGT0007",..: 17 18 18 17 17 18 1 2 4 6 ...
##  $ Corpus.NGT.time.code  : Factor w/ 290 levels "00:00:04.640-00:00:06.720",..: 271 257 76 235 233 191 231 63 102 221 ...
##  - attr(*, "na.action")= 'omit' Named int [1:123] 1 4 6 13 16 19 22 31 33 40 ...
##   ..- attr(*, "names")= chr [1:123] "1" "4" "6" "13" ...

summary(data.plurals.withoutNA)

##   Participant     Data_type         Noun     Noun_type        Strategy1      Strategy2   Zero    Num     Number_repetitions      Mouthing   Syllables_mouthing Corpus.NGT.file.number                Corpus.NGT.time.code
##  p02    : 31   corpus  :220   KIND.PL : 60   B:164     other       :  1   other   :  1   0:238   0:244   Min.   :0.0000     none     : 45   Min.   :0.000      none    :143           none                     :143      
##  p03    : 29   elicited:143   MENS.PL : 41   C: 22     sideward    :105   sideward:120   1:125   1:119   1st Qu.:0.0000     mens'    : 26   1st Qu.:1.000      CNGT0014:  7           00:00:14.370-00:00:14.700:  2      
##  p04    : 29                  PROBLEEM: 21   L:148     sideward.sim: 15   sim     :  6                   Median :1.0000     kinderen': 24   Median :1.000      CNGT1684:  7           00:00:34.800-00:00:35.000:  2      
##  p05    : 28                  SCHOOL  : 21   M: 29     sim         :  6   simple  :111                   Mean   :0.8058     kind'    : 20   Mean   :1.485      CNGT0136:  5           00:01:42.360-00:01:42.800:  2      
##  p06    : 26                  DING.PL : 15             simple      :111   zero    :125                   3rd Qu.:1.0000     mensen'  : 19   3rd Qu.:2.000      CNGT0254:  5           00:02:22.400-00:02:22.720:  2      
##  s008   : 12                  VROUW   : 14             zero        :125                                  Max.   :4.0000     school'  : 17   Max.   :4.000      CNGT0137:  4           00:02:48.120-00:02:49.000:  2      
##  (Other):208                  (Other) :191                                                                                  (Other)  :212                      (Other) :192           (Other)                  :210

Loading annotations of mouthings by both annotators to check inter-rater agreement

Mouthings.both.annotators <- read.csv("Mouthings_bothannotators.csv", header = TRUE, sep = ";")

Mouthings.both.annotators #This data set shows the type of mouthing each annotator observed: no mouthing (none); a singular Dutch word (singular); a plural Dutch word (plural); or a reduplicated Dutch word (reduplication).

##                          Source        Rater1        Rater2
## 1  CNGT0299, s017, 00:00:23.040          none      singular
## 2  CNGT1684, s069, 00:02:53.320          none          none
## 3  CNGT0099, s001, 00:04:09.760      singular      singular
## 4  CNGT0255, s014, 00:06:26.800      singular        plural
## 5  CNGT0438, s022, 00:01:59.280        plural        plural
## 6  CNGT1628, s067, 00:00:52.305        plural        plural
## 7  CNGT0333, s015, 00:00:37.600 reduplication reduplication
## 8  CNGT0065, s005, 00:00:05.720         other         other
## 9                           p06          none          none
## 10                          p06          none          none
## 11                          p05        plural        plural
## 12                          p03        plural        plural
## 13                          p02 reduplication      singular
## 14                          p06 reduplication reduplication
## 15                          p02      singular      singular
## 16                          p06      singular      singular

#Removing column with source file, which is not necessary to check inter-rater agreement
Mouthings.ratings <- subset (Mouthings.both.annotators, select = -Source)

Mouthings.ratings

##           Rater1        Rater2
## 1           none      singular
## 2           none          none
## 3       singular      singular
## 4       singular        plural
## 5         plural        plural
## 6         plural        plural
## 7  reduplication reduplication
## 8          other         other
## 9           none          none
## 10          none          none
## 11        plural        plural
## 12        plural        plural
## 13 reduplication      singular
## 14 reduplication reduplication
## 15      singular      singular
## 16      singular      singular

Results

Mixed-effects model 1: Noun type

Set contrasts

#setting contrasts for noun type

contrasts.nountype <- cbind(c(-1/4, +3/4, -1/4, -1/4), c(+1/4, +1/4, -1/4, -1/4), c(+3/4, -1/4, -1/4, -1/4)) # B, C, L, M
colnames (contrasts.nountype) <- c("-BLM+C", "-LM+BC", "-CLM+B")
contrasts (data.plurals.all$Noun_type) <- contrasts.nountype
contrasts (data.plurals.all$Noun_type)

##   -BLM+C -LM+BC -CLM+B
## B  -0.25   0.25   0.75
## C   0.75   0.25  -0.25
## L  -0.25  -0.25  -0.25
## M  -0.25  -0.25  -0.25

#setting contrasts for data type
contrasts.datatype <- cbind(c(-0.5, +0.5))
colnames (contrasts.datatype) <- c("-corpus+elicited")
contrasts (data.plurals.all$Data_type) <- contrasts.datatype
contrasts (data.plurals.all$Data_type)

##          -corpus+elicited
## corpus               -0.5
## elicited              0.5

Run model

Zero.nountype <- glmer(Zero ~ Noun_type + Data_type + (Noun_type | Participant), glmerControl(calc.derivs = FALSE, optCtrl=list(maxfun=1e6)), data=data.plurals.all, family=binomial)

## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient

summary(Zero.nountype)

## Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
##  Family: binomial  ( logit )
## Formula: Zero ~ Noun_type + Data_type + (Noun_type | Participant)
##    Data: data.plurals.all
## Control: glmerControl(calc.derivs = FALSE, optCtrl = list(maxfun = 1000000))
## 
##      AIC      BIC   logLik deviance df.resid 
##    579.2    637.8   -275.6    551.2      472 
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.2279 -0.6729 -0.3790  0.8809  3.4473 
## 
## Random effects:
##  Groups      Name            Variance Std.Dev. Corr             
##  Participant (Intercept)     0.3990   0.6317                    
##              Noun_type-BLM+C 0.7889   0.8882   -0.86            
##              Noun_type-LM+BC 0.8635   0.9293   -0.50  0.14      
##              Noun_type-CLM+B 0.3594   0.5995   -0.70  0.96 -0.10
## Number of obs: 486, groups:  Participant, 64
## 
## Fixed effects:
##                           Estimate Std. Error z value    Pr(>|z|)    
## (Intercept)                -0.8425     0.1657  -5.086 0.000000366 ***
## Noun_type-BLM+C             0.6250     0.3936   1.588       0.112    
## Noun_type-LM+BC             2.3263     0.5495   4.233 0.000023031 ***
## Data_type-corpus+elicited   0.2088     0.3407   0.613       0.540    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) N_-BLM N_-LM+
## Nn_ty-BLM+C  0.122              
## Nn_ty-LM+BC -0.465 -0.033       
## Dt_typ-crp+  0.281 -0.390 -0.330
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 1 column / coefficient

Calculate odds ratio & confidence intervals

#Odds ratio of the logit scale coefficients
Zero.nountype_coef <- round(coef(summary(Zero.nountype)), 3)
Zero.nountype_OR <- exp(Zero.nountype_coef[,1])
Zero.nountype_OR

##               (Intercept)           Noun_type-BLM+C           Noun_type-LM+BC Data_type-corpus+elicited 
##                  0.430848                  1.868246                 10.236912                  1.232445

#Confidence intervals
Zero.nountype_CI <- exp(confint(Zero.nountype, method="Wald"))
Zero.nountype_CI

##                               2.5 %     97.5 %
## .sig01                           NA         NA
## .sig02                           NA         NA
## .sig03                           NA         NA
## .sig04                           NA         NA
## .sig05                           NA         NA
## .sig06                           NA         NA
## .sig07                           NA         NA
## .sig08                           NA         NA
## .sig09                           NA         NA
## .sig10                           NA         NA
## (Intercept)               0.3112589  0.5958415
## Noun_type-BLM+C           0.8637362  4.0412373
## Noun_type-LM+BC           3.4877573 30.0657228
## Data_type-corpus+elicited 0.6319469  2.4027282

Mixed-effects model 2: Numerals/quantifiers