Statistics on Grundtvig's use of bible references

Nikolaj Frederik Severin Grundtvig is one of the most influential persons in Danish history. He was a polymath and a very prolific man, a pastor, author, poet, philosopher, historian, teacher and politician, as wikipedia describes him.

This project includes a statical analysis of his use of bible references, as found in https://tekster.kb.dk/. The data comes from the Center for Grundtvigforskning which is the publisher of Grundtvigs Værker referred to as GV.

I use the counts of individual bible references to compare all the years of Grundtvig's professional life. The philologists at the Grundtig centre has identified 11499 references to the bible in the GV. They refer to 4637 different locations in the bible, i.e., he uses each location 2.5 times on the average. In reality there are large number of bible locations that just appear once in a reference, whereas there are a number of favourates which is cited 50-60 in the corpus.

What you find below is an outline of my analysis of of his bible references from a technical point of view, with references to scripts and data. The paper.

Prerequisites

Most scripts require saxon be available. Do this by

SAXON_JAR="/usr/share/maven-repo/net/sf/saxon/Saxon-HE/9.9.1.5/Saxon-HE-9.9.1.5.jar"
SAXON="java -jar $SAXON_JAR "

or source parameters.sh. This might have to be modified to your installation, obviously.

Find references from texts using find and xslt which requires that you have access to the TEI source files. I cannot provide you with that, since they are not ownded by me or my employer.

Any way, my data extraction is done with a shell one-liner and the xslt script

find build/text-retriever/gv/  \
	-name 'txt.xml' -exec xsltproc \
	--stringparam file {}  ../bible-references/explore-references.xsl {} \;  > \
	../bible-references/table-all.xml

The result is stored in table-all.xml, which is really a huge html table.

Aggregating references by year

I then wrote a shell script, aggregate.sh which runs an xsl transform counting the number of references per year given that table-all.xml. It also makes that for given bible references, which can be given as a parameter.

#!/bin/bash

source "parameters.sh"

$SAXON table-all.xml aggregate-per-year.xsl > aggregated-references-per-year.text

$SAXON table-all.xml aggregate-per-year-for-given-reference.xsl > selected_ref_1.text
$SAXON ref='1 Kor 13,12'   table-all.xml aggregate-per-year-for-given-reference.xsl > selected_ref_2.text
$SAXON ref='1 Kor 13,13'   table-all.xml aggregate-per-year-for-given-reference.xsl > selected_ref_3.text
$SAXON ref='Matt 28,18-20' table-all.xml aggregate-per-year-for-given-reference.xsl > selected_ref_4.text

gnuplot < plot_references.gp
gnuplot < plot_selected_references.gp

This yields two graphs

• refs_per_year.pdf
• selected_refs_per_year.pdf

Aggregating references by year for verse only

I then repeat that using bible references in verse (songs, psalms and poems) using aggregate-poetry.sh.

#!/bin/bash

source "parameters.sh"

$SAXON poetry-table.xml aggregate-per-year.xsl > aggregated-references-per-year.text

$SAXON poetry-table.xml aggregate-per-year-for-given-reference.xsl > selected_poetry_ref_1.text
$SAXON ref='1 Mos 2,7'   poetry-table.xml aggregate-per-year-for-given-reference.xsl > selected_poetry_ref_2.text
$SAXON ref='Joh 6,63'   poetry-table.xml aggregate-per-year-for-given-reference.xsl > selected_poetry_ref_3.text
$SAXON ref='Ordsp 4,23' poetry-table.xml aggregate-per-year-for-given-reference.xsl > selected_poetry_ref_4.text

gnuplot < plot_selected_poetry_references.gp 

The resulting time series is presented in a plot

• selected_poetry_refs_per_year.pdf

Making cluster analysis and a cladogram

Here I use the counts of individual bible references to calculate a similarity between years and through a similarity matrix and from that plot a cladogram.

$SAXON table-all.xml clustering_data.xsl  > clustering-data.text

The clustering-data.text contains one line per year and 4637 columns, one for each bible location Grundtvig referred to. The values in the table refer to the number of times he referenced that locations. Most entries are zero, obviously. The do-cluster-analysis.r does the calculations. See Altuna Akalin (2020) Computational Genomics with R

#!/usr/bin/Rscript


yearly_quotes <- read.table(file = "clustering-data.text",
	      	     row.names=1,
                     sep = "",
                     quote = "\"",
                     dec = ".",
                     fill = TRUE,
                     comment.char = "#")

# yearly_quotes



pdf( "cladogram.pdf", width = 15, height = 10 )

# d=dist(yearly_quotes,method="euclidean")
d=dist(yearly_quotes,method="minkowski")
hc <- hclust(d,method="complete")

hcd <- as.dendrogram(hc)

plot(hcd,xlab="Years")

The final result is in the cladogram.pdf

Number of works per year by genre

• Transform all files using text-types-by-year.xsl using run_text-types-by-year.pl collecting all data in a single file, e.g.,
• counts-of-text-by-type-and-year.xml. Use that to aggregate per year using breakdown-per-year.xsl
• plot using plot_words_per_year.gp

See words_per_year.pdf