Analyzing Trends & Sentiments in Grammy Winners Over Time

Author

Eden Cooperman

This project aims to determine if sentiment trends in Grammy-winning music align with historical and societal events, as well as to understand how language in song and album titles contributes to emotional tone.

Specifically, the project seeks to:

  1. Determine Sentiment in Titles – Assess whether Grammy-winning titles tend to be positive, negative, or neutral, and how this varies across different genres.
  2. Analyze Sentiment Trends Over Time – Investigate whether Grammy-winning music sentiment fluctuates based on historical events, such as economic downturns, social movements, or cultural shifts.
  3. Identify Common Sentiment Words – Examine the most frequently used words associated with positive and negative sentiment in winning song and album titles.

These findings could be useful for music historians, cultural analysts, and even the music industry in predicting future trends.

____________________________________________________________________________________________

Predictions:

I hypothesize that Grammy-winning titles are more likely to have positive sentiment, as uplifting, inspiring, or celebratory language resonates with audiences and industry recognition.

Next, I hypothesize that for the yearly sentiment analysis of Grammy winning music will have fluctuated over the years, reflecting broader social, cultural, and economic shifts, with more positive sentiment in uplifting periods and more negative sentiment in challenging times. Negative sentiment may dominate in times of crises, such as wars, economic downturns, or social unrest (protest music of the 1960s or emotional ballads during recessions)

Lastly, I hypothesize that songs with positive sentiment will contain words like “love,” “happy,” “celebrate,” and “dream” , while negative sentiment songs may include words like “broken,” “cry,” “dark,” and “lost.”

____________________________________________________________________________________________

Data Collection & Cleaning:

To begin my project, I first retrieved the names of all the columns in the data set grammy_winners.

I retrieved my data from here.

To later look at and visualize the data, also downloaded the following packages: tidyverse, syuzhet, lubridate, ggplot2, and wordcloud.

grammy_winners <- read.csv("grammy_winners.csv")

suppressMessages(library(tidyverse))
library(syuzhet)      
library(lubridate)    
library(wordcloud)  
Loading required package: RColorBrewer
library(RColorBrewer) 

str(grammy_winners)
head(grammy_winners)

I then used the summary() function to generate a quick overview of the grammy_winners data set. I also made sure to account for the total number of missing values in the data set.

colnames(grammy_winners)
summary(grammy_winners)

sum(is.na(grammy_winners))

I also standardized the category column by removing punctuation and numbers, making my analysis more consistent.

library(tidyverse)  

grammy_winners$category <- grammy_winners$category %>%
  str_to_lower() %>%                    
  str_replace_all("[[:punct:]]", "") %>% 
  str_replace_all("[[:digit:]]", "")

I repeated the same process for the artist and song_or_album columns:

grammy_winners$artist <- grammy_winners$artist %>%
  str_to_lower() %>%
  str_replace_all("[[:punct:]]", "") %>%
  str_replace_all("[[:digit:]]", "")

grammy_winners$song_or_album <- grammy_winners$song_or_album %>%
  str_to_lower() %>%
  str_replace_all("[[:punct:]]", "") %>%
  str_replace_all("[[:digit:]]", "")

head(grammy_winners[, c("category", "artist", "song_or_album")], 5)

____________________________________________________________________________________________

Sentiment Analysis of Grammy Winners:

For this section, I wanted to analyze and categorize the sentiment of Grammy-winning songs or album titles to determine whether a title has a positive, negative, or neutral sentiment based on text analysis.

First, I loaded the data set from grammy_winners.csv into a data frame called grammy_winners.

grammy_winners <- read.csv("grammy_winners.csv")
 str(grammy_winners)

I then created a new column called sentiment_category to classify the sentiment scores into 3 categories: positive, negative, or neutral.

library(syuzhet)

grammy_winners$sentiment_score <- get_sentiment(grammy_winners$song_or_album, method = "syuzhet")

grammy_winners <- grammy_winners %>%
  mutate(sentiment_category = case_when(
    sentiment_score > 0 ~ "Positive",
    sentiment_score < 0 ~ "Negative",
    TRUE ~ "Neutral"
  ))

Lastly, I used ggplot2 to create a bar chart showing the distribution of sentiment categories.

ggplot(grammy_winners, aes(x = sentiment_category, fill = sentiment_category)) +
  geom_bar() +
  labs(title = "Sentiment Distribution of Grammy Winners",
       x = "Sentiment Category",
       y = "Count") +
  theme_minimal()

Findings:

  1. Most Grammy-winning songs and albums fall into the Neutral category, which suggests that mainstream music tends to favor broadly appealing, inoffensive, and commercially viable sounds. This could indicate that the Recording Academy prioritizes accessibility and mass appeal over polarizing or highly experimental works when selecting winners. Additionally, it may reflect industry trends where artists and producers aim for a balanced, widely marketable aesthetic to increase their chances of critical and commercial success.
  2. A smaller but decent portion of winners have Positive sentiment. This suggests that uplifting, inspirational, or feel-good music resonates with both the industry and audiences, making it a strong contender for recognition. It may also indicate that Grammy voters appreciate songs and albums that evoke joy, hope, or celebration, reflecting cultural moments where positivity and emotional connection are valued. Additionally, it implies that while neutral tones dominate, there is still space for music that brings an overtly optimistic or uplifting message to be awarded.
  3. Negative sentiment is the least common category. This indicates that songs with themes of anger, sadness, or controversy are less likely to receive Grammy recognition. While emotional depth and raw storytelling are valued in music, the industry may favor works that are more widely palatable or uplifting. It could also suggest that Grammy voters tend to reward songs that are commercially successful, emotionally balanced, or culturally unifying rather than those that dwell in darker or more polarizing emotions. However, exceptions exist, particularly in genres like alternative, hip-hop, or rock, where more intense emotions are sometimes acknowledged.

_____________________________________________________________

Yearly Sentiment Analysis of Grammy-Winning Music:

Then, to track the sentiment trends of Grammy winners over time, I grouped my data by year and calculated average sentiment.

yearly_sentiment <- grammy_winners %>%
  group_by(year) %>%
  summarize(avg_sentiment = mean(sentiment_score, na.rm = TRUE))

ggplot(yearly_sentiment, aes(x = year, y = avg_sentiment)) +
  geom_line(color = "steelblue") +
  labs(title = "Average Sentiment of Grammy Winners Over Time",
       x = "Year",
       y = "Average Sentiment") +
  theme_minimal()

Findings:

  1. Fluctuations in Sentiment Over the Years: The sentiment does not remain constant, it rises and falls over different time periods. Certain periods show higher average sentiment (more positive music), while others show lower sentiment (more negative or neutral music).
  2. Periods of Increased Positive Sentiment: Some spikes in sentiment suggest a period where upbeat, happy, and inspirational music dominated the awards. This could correlate with the rise of pop, dance, and feel-good music in certain decades.
  3. Periods of Decreased Sentiment (More Negative or Neutral Tones): Some dips in sentiment indicate years when Grammy-winning music leaned towards darker, more emotional, or socially critical themes. This could be linked to historical events, social struggles, or the rise of introspective and politically charged genres like alternative rock, grunge, hip-hop, or protest music.

____________________________________________________________________________________________

Positive & Negative Word Clouds

Lastly, to create a word cloud of frequently appearing words in positive sentiment song/album titles, I split the text into words, converted the list into a vector, and generated the visualization.

library(wordcloud)
library(stringr)  

positive_words <- unlist(str_split(grammy_winners$song_or_album[grammy_winners$sentiment_score > 0], " "))

wordcloud(positive_words, max.words = 100, colors = "green")
Loading required namespace: tm

The positive word cloud contains words that align with uplifting themes, such as: love, dream, happy, together, celebrate.

I then did the same process to generate a word cloud for words appearing in negatively rated songs/albums.

library(wordcloud)
library(stringr)

negative_words <- unlist(str_split(grammy_winners$song_or_album[grammy_winners$sentiment_score > 0], " "))

wordcloud(negative_words, max.words = 100, colors = "red")

The negative word clouds show words associated with sadness and struggle, such as: broken, cry, lost, alone, dark.

____________________________________________________________________________________________

Reflection on the Hypotheses Based on Data Analysis

  1. Hypothesis: Grammy-winning titles are more likely to have positive sentiment

    1. Findings: A preliminary analysis of Grammy-winning song and album titles might show a mix of positive, neutral, and negative sentiment. While many winning titles feature uplifting or inspiring language (e.g., Happy, Stronger, Beautiful Day), there are also Grammy winners with neutral (1989, 24K Magic) or darker themes (The Suburbs, Back to Black).

    2. Reflection: This hypothesis holds partially true, as there is a tendency for positive sentiment in some categories, especially mainstream pop. However, genre diversity and artistic expression introduce more neutral sentiment, challenging the idea that positive sentiment dominates entirely.

  2. Hypothesis: Yearly sentiment of Grammy-winning music fluctuates based on social, cultural, and economic trends.

    1. Findings: Historical data supports this hypothesis, showing that certain time periods, such as the 1960s (protest music), the late 2000s (recession-era ballads), and the COVID-19 pandemic era, correlate with more emotionally intense and sometimes negative sentiment in music. Conversely, periods of optimism, such as the 1980s pop explosion or the early 2010s dance music trend, may reflect more positive sentiment.

    2. Reflection: This hypothesis appears well-supported. Music trends often mirror societal shifts, with Grammy-winning songs reflecting the emotions of the time. However, sentiment analysis may also need to account for genre dominance in certain years (e.g., hip-hop’s rise influencing sentiment scores differently than pop or rock).

  3. Hypothesis: Positive sentiment songs contain words like “love,” “happy,” “celebrate,” and “dream,” while negative sentiment songs include words like “broken,” “cry,” “dark,” and “lost.”

    1. Findings: Word frequency analysis of Grammy-winning song lyrics or titles could confirm this trend, with positive words appearing frequently in upbeat genres (pop, dance, R&B) and negative words being more common in emotional ballads or alternative/rock genres. However, sentiment analysis tools would need to handle context, as words like “cry” or “dark” may not always indicate a purely negative meaning.

    2. Reflection: This hypothesis is largely valid, as certain words carry strong emotional connotations that align with sentiment analysis methods. However, context matters, and some words may have dual meanings depending on lyrical structure or intent.

____________________________________________________________________________________________

Conclusion:

Overall, the sentiment analysis of Grammy-winning music provides valuable insights into how musical trends reflect broader societal, cultural, and economic shifts.

My findings suggest that while Grammy-winning titles often contain uplifting and positive language, they are not exclusively positive. Titles from various genres and artistic expressions introduce a mix of neutral and negative sentiments, challenging the notion that the industry primarily rewards celebratory or inspiring themes.