---

---

From Leibniz to Lefty: Country Music as Theodicy

---

---

John Bernau, ABD
Emory University
ASA Summer 2017
Montreal, QC
Canada
___
___

ABSTRACT
This paper constructs a conceptual link between theodicy and country music. Theodicy, the theoretical attempt to reconcile earthly suffering with divine justice, played a prominent role in the work of Max Weber and Peter Berger, and cultural sociologists today have used this concept to understand meaning-making more broadly among secular communities. In this paper I examine country music through this theoretical lens to better understand the role of suffering in one of America’s most popular music genres. I explain 1) how the genre’s historical roots in rural southern culture led to the institutionalization of tragic lyrical themes, 2) how the canonization of figures like Jimmie Rodgers and Hank Williams perpetuated this archetype, and 3) how the juxtaposition of humor and suffering contribute to country music’s unique efforts at an aesthetic theodicy. The cultural need for a meaningful interpretation of suffering is not confined to religious communities or medieval theologians, and by looking at country music, this paper furthers our understanding of this important meaning-making process. This handout provides ideas for quantitative applications.

Keywords: culture, religion, media, theodicy, music, country music, aesthetics

---

---

Quantitative Directions…

---

---

Research Question

How can we describe the “mood” of country music lyrics? How has this changed? Do any variables predict a change in mood?

---

Theory

Olick, Jeffrey K. 2016. “The Poverty of Resilience: On Memory, Meaning, and Well-Being.” Memory Studies 9(3):315–24.

Horwitz, Allan V. and Jerome C. Wakefield. 2012. The Loss of Sadness: How Psychiatry Transformed Normal Sorrow into Depressive Disorder. Reprint edition. Oxford ; New York: Oxford University Press.

---

Hypotheses

H1: Country music lyrics have become “happier” in the last 25 years.
H2: This trend is led by non-Hall-of-Fame artists.
H3: The role of vocab diversity and poltical context may have some effect.

---

Data

More than 30,000 song lyrics from 1955-2015 were scraped from a country music lyric site (www.CowboyLyrics.com). Data was cleaned to check for date errors and paired with meta-data like album type, political climate, etc.

---

---

---

---

To get a better idea of the dataset, here are some descriptive graphs. First using only ‘Regular" albums and then with only ’Greatest Hits’ albums.

Sized by number of songs in the dataset, a word cloud gives an idea of the artists represented.

While not a perfect measure of the length of an artist’s carrer, we can calculate longevity using the years spanned within the dataset. Top-10 artists shown here.

This graph represents the average vocab diversity over time. While the yearly averages don’t seem to change much, at the song-level, this variable could be used to predict changes in sentiment.

Method

Sentiment analysis dictionaries provide lists of positive / negative words, which can be used to calculate a rough overall “sentiment” of a given song. This data can then be treated as a dependent variable in OLS or other statistical models.

---

Findings

The yearly sentiment averages presented here don’t show much variation, but if we look at the song level this variable may be of more interest.

Running a quick OLS regression doesn’t look very impressive…
More work to be done!

===============================================
                        Dependent variable:    
                    ---------------------------
                            SentimentGI        
-----------------------------------------------
year                         -0.001***         
                             (0.0001)          
                                               
lexdiv_s                     -0.039***         
                              (0.004)          
                                               
first_album                   -0.001           
                              (0.001)          
                                               
hof_dummy                    0.007***          
                              (0.002)          
                                               
py_rpres                      -0.002*          
                              (0.001)          
                                               
py_electyr                    0.003**          
                              (0.001)          
                                               
Constant                     1.569***          
                              (0.136)          
                                               
-----------------------------------------------
Observations                  26,241           
R2                             0.011           
Adjusted R2                    0.011           
Residual Std. Error     0.072 (df = 26234)     
F Statistic          50.105*** (df = 6; 26234) 
===============================================
Note:               *p<0.1; **p<0.05; ***p<0.01