Dissertation text analysis

TW: The corpus is based on legislations against sexual violence

Preparation

library(readxl)
library(quanteda)

## Package version: 3.3.1
## Unicode version: 14.0
## ICU version: 71.1

## Parallel computing: 8 of 8 threads used.

## See https://quanteda.io for tutorials and examples.

library(quanteda.textmodels)
library(quanteda.textplots)
library(quanteda.textstats)
setwd("~/Desktop/Rstudio")
d <- read_excel('test.xlsx')
head(d)

Creating the corpus

corp <- corpus(d, text_field = 'text')  ## create the corpus
corp

## Corpus consisting of 21 documents and 2 docvars.
## text1 :
## "Any person who by using violence or threat of violence force..."
## 
## text2 :
## "Sentenced for committing the crime of rape, with imprisonmen..."
## 
## text3 :
## "a man who has intercourse with a woman outside of marriage, ..."
## 
## text4 :
## "a man who have intercourse with a woman outside of marriage,..."
## 
## text5 :
## "a man who have intercourse with women, with the consent of t..."
## 
## text6 :
## "a man who has sexual intercourse with a woman, with the woma..."
## 
## [ reached max_ndoc ... 15 more documents ]

##Preprocessing

dtm = dfm(corp, stem = T, remove = stopwords('en'), remove_punct = T)

## Warning: 'dfm.corpus()' is deprecated. Use 'tokens()' first.

## Warning: '...' should not be used for tokens() arguments; use 'tokens()' first.

## Warning: 'remove' is deprecated; use dfm_remove() instead

## Warning: 'stem' is deprecated; use dfm_wordstem() instead

dtm

## Document-feature matrix of: 21 documents, 89 features (89.41% sparse) and 2 docvars.
##        features
## docs    person use violenc threat forc woman sexual intercours marriag shall
##   text1      1   1       2      1    1     1      1          1       1     1
##   text2      0   0       0      0    0     0      0          0       0     0
##   text3      0   0       0      0    0     2      0          1       1     0
##   text4      0   0       0      0    0     2      0          1       1     0
##   text5      0   0       0      1    0     1      0          1       0     0
##   text6      0   0       0      0    0     3      1          1       0     0
## [ reached max_ndoc ... 15 more documents, reached max_nfeat ... 79 more features ]

Analysis

Using a wordcloud to show the most frequent words used in the corpus

textplot_wordcloud(dtm,)

textstat_frequency(dtm, n = 10)                                  

Comparing Corpora

• Analyse how different terms are used and how frequently they are used in each versions of the bill (specifically focusing on the definition of rape)
• The image below shows the terms in the final document that has been used more often (shown in blue) and less often (shown in grey) compared to other documents
• Comparing all three documents, we can see that the second version is more likely to use gendered terms such as ‘woman’, ‘man’, ‘marriage’ and ‘husband’. Whereas compared to the final version, the term ‘person’ is more likely to be used so is ‘disabled’ which may illustrate the more comprehensive nature of the latter. This is supported by the negative chi2 of the terms ‘woman’, ‘man’, and ‘marriage’.

is_2023 <- docvars(dtm)$year == '2023' 
final_dtm <- dtm[is_2023,]
ts <- textstat_keyness(dtm, is_2023)
textplot_keyness(ts)

is_2015 <- docvars(dtm)$year == '2015' 
semifinal_dtm <- dtm[is_2015,]
ts2 <- textstat_keyness(dtm, is_2015)
textplot_keyness(ts2)

is_1915 <- docvars(dtm)$year == '1915' 
initial_dtm <- dtm[is_1915,]
ts3 <- textstat_keyness(dtm, is_1915)
textplot_keyness(ts3)

Keyword in-context

• we use kwic() to find how a term is used in a sentence by identifying the words around it
• kwic = key words in context
• Using KWIC for ‘woman’ shows that the term is often used for the ‘victim’ context, i.e. at the receiving end of assaults, whereas the ‘man’ is on the perpetrator side. This becomes clearer when using KWIC on ‘intercourse’ as ‘man’ is always on the right hand side of the term, and ‘woman’ on the left hand side. Whereby when exploring ‘person’ we can see that the term is used to illustrate both the victim and perpetrator status. Again showing a more gender neutral stance.

woman <- kwic(tokens(corp), 'woman', window = 5)
head(woman, 10)    ## only view first 10 results

man <- kwic(tokens(corp), 'man', window = 5)
head(man, 10)    ## only view first 10 results

intercourse <- kwic(tokens(corp), 'intercourse', window = 5)
head(intercourse, 10)    ## only view first 10 results

person <- kwic(tokens(corp), 'person', window = 5)
head(person, 10)    ## only view first 10 results