Quinn XCII Sentiment Analysis Project

Introduction

Mikael Temrowski, or Quinn XCII as he’s more commonly known by his 6 million monthly listeners on Spotify is a 26 year-old musician. Quinn XCII sings an extremely wide range of music from pop to hip-hop to soul to electronic and everything in between. Participating in collaborations with artists like Jon Bellion, Elohim, Ashe, and even Louis the Child, Quinn XCII is an extremely versatile artist. In his most recent album, “From Michigan with Love”, Quinn XCII did something he’s never done before in an album. According to this interview with “Hollywood Life”, a blog on everything Hollywood by Bonnie Fuller, Quinn XCII decided to use his voice and audience following to experiment with mixing genres of music, and even content of what he sings. This led him to begin singing about his own struggles with mental health. Mental health amongst milennials has also been a huge issue in recent years, and through this article by Forbes, there’s been a 33% increase in diagnosed cases of depression since 2013 (in 2016, a 6% rate amongst women and 2.8% amongst men). With this in mind, Quinn XCII is speaking to a wide audience through his lyrics and speaking about real issues that people are facing today. This brings me to the topic of my sentiment analysis today, which is uncovering the sentiment behind Quinn XCII’s words in “From Michigan with Love”.

Hypothesis

For my hypothesis, I believe that Quinn XCII’s newest album, “From Michigan with Love” has a negative sentiment with afinn levels below 0. As stated in my introduction, his album focuses highly on mental health and that has a general negative sentiment in today’s culture. Speaking out and normalizing these disorders, although not always a happy topic, is important nonetheless.

Methodology

Through studying the frequencies of most popular words in his songs, boolean sentiments in his album through the “bing” lexicon, and an in-depth afinn sentiment analysis of each song in his album, I plan to put my hypothesis to the test.

Packages I’m Using

1. Tidytext
2. Tidyverse
3. Dplyr
4. Ggplot2

Procedure

1. I began first by gathering the song lyrics. Although there is already a Genius API for Quinn XCII, I meticulously copied and pasted lyrics for every song off of google, added them to a spread sheet with lyrics under the column, “songLyrics”, and made another column titled “songTitle”. In “songTitle” I sifted through each line of lyrics and matched them with their corresponding songs. I nestled this data under a variable called quinnLyrics: #quinnLyrics

2. After that, I filtered the lyrics to put one word per line and to remove stop-words (word like a, it, the, etc.), which is under the variable #quinnWords. I finally created a new column to count the frequency of words and put them in most-least used order, which is under the variable #quinnWordCounts.

   {r}quinnWords <- quinnLyrics %>% unnest_tokens(word,"Song Lyrics"){r} {r}quinnWordCounts <- quinnWords %>% anti_join(stop_words, by = "word") %>% count(word, sort = TRUE){r}

3. After, I made 3 ggplots: one to show general most used words, and then two more to show most popular positive sentiment words, and another to show most popular negative sentiment words. This boolean sentiment is called “bing” and it involves classifying words either negatively or positively. The code:

   {r}quinnWords %>% anti_join(stop_words) %>% count(word, sort = TRUE) %>% head(25) -> quinnTopWords{r}

   {r}ggplot(quinnTopWords, aes(reorder(word, n), n)) + geom_col() + coord_flip() + xlab("Lyric") + ggtitle("Quinn Word Sentiment"){r}

   {r}get_sentiments("bing") quinnSentimentbing <- quinnWordCounts %>%inner_join(get_sentiments("bing")){r}

   {r}quinnNegative <- quinnSentimentbing %>% filter(sentiment=="negative") %>% head(25){r}

   {r}quinnPositive <- quinnSentimentbing %>% filter(sentiment=="positive") %>% head(25){r}

   {r}ggplot(quinnNegative, aes(reorder(word, n), n)) + geom_col() + coord_flip() + xlab("Lyric") + ggtitle("Negative Quinn Word Sentiment"){r}

   {r}ggplot(quinnPositive, aes(reorder(word, n), n)) + geom_col() + coord_flip() + xlab("Lyric") + ggtitle("Positive Quinn Word Sentiment") {r}

4. I proceeded to make a world cloud to visualize the words Quinn XCII used the most in that album. I included the first 10 words in the original table of words with highest frequencies. The code:

   {r}quinnWordCounts <- quinnWords%>% anti_join(stop_words, by = "word") %>% count(word, sort = TRUE){r} {r}quinnWordCounts{r} {r}library(wordcloud2){r} {r}wordcloud2(quinnWordCounts){r}

5. For the second part of my methodology, I used sentiment analysis with afinn, nrc, and the boolean lexicon “bing” to identify the general sentiments of his entire album. The code is below (visualization in results):
   1. The code for afinn:

   {r}get_sentiments("afinn"){r} {r}quinnSentimentafinn <- quinnWordCounts %>% inner_join(get_sentiments("afinn")){r}

   2. The code for nrc:

   {r}get_sentiments("nrc"){r} {r}quinnSentimentnrc <- quinnWordCounts %>% inner_join(get_sentiments("nrc")){r} {r}quinnNRC <- quinnSentimentnrc %>% head(25){r} {r}ggplot(quinnSentimentnrc, aes(x=sentiment, y=n)) + geom_bar(stat="identity") + ggtitle("Quinn XCII NRC Album Sentiment Analysis") + theme(axis.text.x = element_text(angle = 90, hjust = 1)){r}

   3. For individual sentiments:

   {r}quinnAnger <- quinnSentimentnrc %>% filter(sentiment %in% "anger"){r} {r}quinnAnticipation <- quinnSentimentnrc %>% filter(sentiment %in% "anticipation"){r} {r}quinnDisgust <- quinnSentimentnrc %>% filter(sentiment %in% "disgust"){r} {r}quinnFear <- quinnSentimentnrc %>% filter(sentiment %in% "fear"){r} {r}quinnJoy <- quinnSentimentnrc %>% filter(sentiment %in% "joy"){r} {r}quinnNegative <- quinnSentimentnrc %>% filter(sentiment %in% "negative"){r} {r}quinnPos<- quinnSentimentnrc %>% filter(sentiment %in% "positive"){r} {r}quinnSadness<- quinnSentimentnrc %>% filter(sentiment %in% "sadness"){r} {r}quinnSurprise<- quinnSentimentnrc %>% filter(sentiment %in% "surprise"){r} {r}quinnTrust<- quinnSentimentnrc %>% filter(sentiment %in% "trust"){r}

   4. The code for bing:

   {r}get_sentiments("bing"){r} {r}quinnSentimentbing <- quinnWordCounts %>% inner_join(get_sentiments("bing")){r}

6. For the last test of my hypothesis, I used afinn as a numerical scale to quantify the sentiment of each song in Quinn XCII’s album “From Michigan with Love”. I took the average “afinn” of each of his songs. The code is below:
   1. “Holding Hands” (first song in the album) {r}holdingHands <- quinnLyrics %>% filter(songTitle == "Holding Hands"){r} {r}holdingHandsWords <- holdingHands %>% unnest_tokens(word, songLyrics) %>% anti_join(stop_words, by = "word") %>% count(word, sort = TRUE){r}

   {r}get_sentiments("afinn"){r} {r}holdingHandsafinn <- holdinHandsWords %>% inner_join(get_sentiments("afinn")){r}

   {r}hhAfinnMean <- mean(holdingHandsafinn$score){r}

7. I repeated this same exact code with all 12 songs in that album: Autopilot, Life Must Go On, U & Us, Werewolf, Tough, Matches, When I Die, Abel & Cain, Sad Still, Good Things Go, and Right Where You Should Be.

Results:

The word cloud below has the words feel, yeah, pill, tough, day, time, and more as the ones that stand out the most. With an emphasis on mental health, Quinn XCII’s album has a big emphasis on his feelings and on resilience. We definitely see the theme of keeping on going, despite struggles through words like pill, but also through seeing the word life in big letters. It shows that although mental health is a bigger struggle, the most important thing is to keep going.

“From Michigan with Love” Most Commonly Used Words

Quinn Word Frequencies

The graph below which is another way of displayed the most frequent words in his album is an even better way of gaging what’s the most frequent. The word feel is recognized as most frequent, followed by the word yeah as more of a filler word, and then the word pill. This is referring to medications for illnesses such as anxiety and depression.

Quinn Graph Frequencies

In this next plot, I began testing directly for sentiments. I took this graph to bring up negative sentiments right away to see if my hypothesis, is in fact supported, and the overall negative hypothesis is backed up

Quinn Negative Sentiment

This plot demonstrates that there is an overall high frequency of negative-sentiment words in this album. The word “sad” had the top number of times used, followed by “hate”, “die”, and “begging”. With the context of mental health themes in this album, these songs bring imagery to the true feelings of Quinn behind his lyrics.

Quinn Positive Word Frequencies

On the flip side, there were also positive words, as seen in the graph above. “Tough” and “love” are both the most popular, tough seemingly strange shows the resiliance of someone battling mental health struggles, but also being able to admit when things aren’t ok and pushing away that barrier. And finally “love” is an important positive sentiment as that is what has ultimately kept him going. Love for music, his fans, and family.

The next steps of my process involved doing sentiment analysis of the album as a whole with NRC and afinn. Below is the graph for the NRC album sentiment analysis.

NRC Album Sentiment Analysis

For the album as a whole, negative sentiment dominates, with positive sentiment following closely behind. After that is sadness. This refers to the negativity in regards to Quinn’s mental health, but also the positivity in how he’s battling these issues. Below is the first sentiment, anger, and interestingly enough in his entire album, the words “hate”, “words”, and “deserve” emote the most anger. “Hate” is the only one of these words that really carries a negative connotation, and words as well as deserve could speak more to his resilience.

Anger

The biggest word that stands out with anticipation is “time” and that’s because of describing the time he has spent stuck in his head absorbed in his thoughts.

Anticipation

Joy speaks a lot more to how he has learned to accept himself, and as a word occurring 11 times in the album, love is huge. Love for his fans, family, friends, but also more importantly for himself.

Joy

The sentiment fear speaks a lot more to the negative impact of mental illness: abuse to oneself through physical harm and even suicide. The most impactful word to me is “surrender” because it personifies mental illness completely taking over when someone is truly sick.

Fear

Disgust speaks more to just negative words in general with the word “hate” being the most popular, followed by “bad” and weirdly enough, “feeling”. These words all reflect the underlying negative sentiments of this album.

Disgust

The most commonly used word in trust is “pill”, which could refer to the trust involved in taking medications and not exploiting drug usage. Quinn’s song “Sad Still” involved medications for depression and anxiety and how they offer the “cure” but in reality users such as himself are still sad. In this sense, it could refer to trust in the pill’s success in it helping mental illness.

Trust

Under surprise, feeling is the most frequently used word. This word falls under many under sentiments as well showing how many different usages it has. For this in particular, it could express how unpredictable and uncontrollable mental health is.

Surprise

Under positive, “pill”, “love”, and “ay” are most commonly used. Using the word “pill” could refer to the positive of medicating to help mental illnesses. Ay is just a catchy filler word and sounds positive, and love expresses resilience towards these sicknesses.

Positive

“Tough” is the most frequently used negative word and expresses the fronts that people put up around mental illnesses. Despite seeming tough and confident, there are struggles.

Negative

The last album-wide sentiment analysis that I did was with afinn. For this one, I took the 15 most frequently used words in the album and analyzed them quantitatively. In the first 15 songs, 8 of them are negatively assorted, demonstrating that the most commonly used words on average are perceived negatively.

Quinn Afinn Analysis

After looking at the album as a whole, I completed a sentiment analysis of the individual songs in the album. After seeing the clear negative sentiment among songs, I wanted to make sure that wasn’t just isolated to one or two songs. I made a table with two colums, one of which incldued the mean afinn sentiment of each song in the album, and the other column was the song title to serve as an identifier.

{r} afinnAvgs = c(rwysbMean, gTgAfinnMean, sadStillAMean, abelAndCainAMean, whenIdieMean, matchesafinnMean, toughafinnMean, werewolfafinnmean, UandUsafinnMean, lmgoAfinnMean, autopilotAfinnMean, hhAfinnMean) {r}

{r} songTitles = c("Right Where You Should Be","GoodThings Go", "Sad Still", "Abel and Cain", "When I Die", "Matches", "Tough", "Werewolf", "U & Us", "Life Must Go On", "Autopilot", "Holding Hands"){r}

Quinn Sentiments by Song Graph

After one glance, the average sentiment for every song in this album is negative except one song, Matches. Even after reading the lyrics from this song, it doesn’t even seem clear why this is seen as positive. The first line of this “positive” song reads “Why am I so obsessed with self-destructiveness”. Thus, while classified as positive, Quinn’s admitance towards being self-destructive is more negative in nature. The negative sentiment carries through in 92% of the songs in the album showing its prominence.

Conclusion

Through studying word frequencies, overall album sentiment analysis, and individual song afinn averages, one thing is very clear. My hypothesis proves true, and that is that there is a negative overall sentiment with Quinn XCII newest album, “From Michigan with Love”. Between most common words including “feel”, “pill”, “life”, “hate”, and more, mental health is a huge underlying theme and shows through.