I just discovered a fellow WordPress blog called RapMETRICS. It’s an amazing site.
The site’s sole credited writer, rapmetrics, who I’m fairly sure is Liban Ali Yusuf, aka RapGenius (FaceBook here) posts detailed statistical analyses of hip-hop lyrics–end rhymes, internal rhymes, rhyme density, meter, “simple” word usage vs. “novel” word usage, syllables, and so on.
You can get a sense of the site’s approach to raw statistical analysis by clicking on RapMETRICS comprehensive table analyzing Eminem’s Marshall Mathers LP. More interesting than the raw data, of course, is the interpretations RapMETRICS make of the data–the table linked to above appears in a post asking the following charmingly specific question: “Dropoff Effect: Do Emcees Experience a Dip Statistically in Concept Songs?”
I find the site’s analyses not only jaw dropping in their level of geekery, but also actually illuminating in terms of what makes for interesting compelling listening. Perhaps RapMETRICS just like the same kind of lyrical content and flow that I prefer (e.g., any metric they have is going to praise MFDoom, an emcee with a stream of consciousness approach and a steady output of records under various names since the early 90s, above all). That said, the standards the site has for analysis and the questions they ask about language and performance are pretty impressive. For example, RapMETRICS measures the percentage of “perfect”/exact rhymes and argue that the highest achievement as an emcee is actually to use as low a percentage as possible because, for RapGenius/Yusef, exact rhymes are too easy and thus a low percentage of exact rhymes is a measure of a creative rhymer. It also puts a premium on novelty and specificity, and thus, an implicit premium on concrete and detailed narrative. I like, too, how RapGenius/Yusef focus on emcees from not only the present day, but also from the 00s, 90s, and 80s.
I have found that the site is fairly impossible to navigate in any kind of systematic way. There’s no glossary of terms, no tagging of content, every post is categorized under they helpful category “uncategorized,” etc. The only organizational principle is a chronological arrangement of posts. But the quality of the analysis in the first couple of posts I read on RapMETRICS made me willing to do some work as a reader to find out more about the ideas that he refers to and introduces in earlier posts. And I eventually did find that most of the methodology is introduced on the RapGenius site, including a paper (.pdf) Yusef presented at the 2009 International Society for Music Information Retrieval Conference in Kobe, Japan. Best to start with these resources and then read the blog, I suppose, though I found the initial detective work kind of fun.
So, what do you think of this approach to analyzing lyrics and performance?