After seeing a data set containing over 18,000 different music reviews hosted on the site Pitchfork I was quite eager to write some code analyzing some of the trends that could be found within it. My initial idea was to see if there was a discrepancy between the scores that man-fronted acts are given compared to woman-fronted. This idea quickly developed into analyzing the different language used reviewing these groups.

One of the first issues that I faced was how I would determine the gender of each group. I decided that the most relevant way to address this was through the general attitude of the reviewer and which artist in the group was addressed the most. Therefore, I enumerated the pronouns that were used in each article and guessed the gender from there.

def guessGender(self):
	maj = [0,0,0]
    for review in self.reviews:
    	pronouns = review.countPronouns()
        for i,b in enumerate(pronouns): maj[i] += b

    if maj[0] > maj[1]: return('feminine')
    if maj[0] < maj[1]: return('masculine')
    else: return('neutral')

Then, I collected each adjective that was used in the article with a filter, and ran the totals for each occurrence.

def overallCommonWords(self,count=10,limits=['JJ', 'JJS', 'NN', 'NNP', 'NNS', 'RB', 'RBS', 'UH', 'VB', 'VBG', 'VBP']):
	reviews = []
    most = {}
    for review in self.reviews:
    	reviews.append(review.findCommonWords(limits))

	for review in reviews:
    	for word in review:
        	if word in reviews:
            	most[word[0]] += word[1]
            else:
            	most[word[0]] = word[1]
    return(sorted(most.items(), key=itemgetter(1))[-count:])

Each artist was given a page where you could see their average score, the most common adjectives that were used to describe them, and their guessed gender. There were some immediate observations that could be made.

Most of the time, acts referred to as feminine were described with words such as beautiful, soothing, or dreamy more than they were strong, exciting, or commanding. The numerical scores themselves didn't seem to have a statistically significant amount of difference, however woman led acts seemed to have a lesser number of reviews overall.