This is the second installment of the lyric sentiment analysis series. Previous article So, we conducted a sentiment analysis on the lyrics of "The Beatles", but this time we will focus on the lyrics of "Queen".
Below, we will analyze 138 songs recorded on albums from 1973 to 1991, when vocalist Freddie Mercury was still alive. (Instrumentals without lyrics are excluded.)
Sentiment Analysis Report (Sentiment Summary)
As with ``The Beatles'', we are referring to a summary of the results of sentiment analysis.
Those painted in an orange-ish color mean positive, and those painted in a blue-ish color mean negative. If you look at the emotion score histogram (Score of Documents with New Sentiment) in the upper right, you can see that there are many positive songs.
In this report, we extracted emotional words from the lyrics of the famous song "Bohemian Rhapsody" and scored them. If you look at the Document Sentiments column at the bottom right, you can see that "love(s)" and "easy" are extracted as words that indicate positive emotions, and "poor" is extracted as a word that indicates negative emotions.
Calculate the emotional score considering the length of the lyrics
When considering sentiment scores for lyrics, I think it's fair to consider the length of the lyrics and calculate the sentiment score for each word in the lyrics. Therefore, we calculate the emotional score per word (Overall Score by Word Count) using the formula below.
Overall Score by Word Count = (Positive Sum( Total positive word score ) + Negative Sum( Total negative word score )) ÷ number of words
The histogram and summary statistics are shown for the emotional scores of songs created using this formula.
Since we are only targeting songs that had some kind of positive or negative emotion, 30 songs were excluded from the 138 songs targeted for analysis, resulting in an N number of 108.
There are 13 songs with a score less than 0 (Negative), which is 12% of the total. The mean score is 1.22 and the median is 0.72. Looking at the histogram, you can see that many songs belong to the 0 to 2.5 class. Overall, it can be said that there are many positive songs.
Which of the four members has the most emotional ups and downs?
At the time of this analysis ( 1973 ~ 1991 ) Queen members include Freddie Mercury, Brian May, Roger Taylor, and John Deacon. Four Despite their names, all of them are involved in writing and composing songs, and each of them has released hit songs into the world.
Therefore, we will use a violin plot to represent the distribution of songs whose lyrics are clearly known, divided by member.
*As was seen in a scene from the movie "Bohemian Rhapsody," late Queen songs are credited with the joint names of all four members. Therefore, it is a pity that the exact author of the lyrics is not known.
Looking at this graph, you can see that the range of values for Freddy is larger than for the other members. Looking at this result alone, you could say that the emotional ups and downs were intense.
However, whether this emotion is high or low depends on the repeated use of a specific emotional expression. The three songs with high scores and legends on the graph use the word "Love" frequently. The song with the highest score, "Get down, Make Love," is so embarrassing when you look at the lyrics that I can't explain the feeling (lol).
Also, "Bring Back That Leroy Brown", which has the lowest score, has a lot of "Bad" ratings.
As was the case in the analysis of "The Beatles," songs that use emotional words repeatedly or refrains tend to have higher or lower scores.
Therefore, is there any meaning in analyzing the emotions of the lyrics? Although this question comes up, I think that the fact that the word ``emotional'' is used repeatedly can be considered to indicate strong emotions, so I think it makes some sense.
I often see small children in toy stores yelling "buy it" and "buy it" over and over again when begging their parents for a toy.
Compared to “The Beatles”
Now, let's compare the emotional scores of "The Beatles" and "Queen." To do this, we will run JMP's "Fit Y by X" (bivariate relationship), plot the scores for each band, and use a nonparametric test (Wilcoxon test) to compare the distribution of scores.
I mentioned earlier that Freddie's emotional ups and downs were wild, but if you look at the scores, the Beatles' variance is much greater. However, looking at the results of the Wilcoxon test, there is no significant difference at the 5% significance level.
The median value for the Beatles is 1.04 and for Queen it is 0.72. Both bands generally express positive emotions, but could it be said that the Beatles have a greater range of emotions?
Next time, we will use JMP Pro 16's new feature ``Term Selection'' (word selection analysis) to extract words frequently used by two bands and create a predictive model to distinguish between the two bands.
by Naohiro Masukawa, JMP Japan
