Last Thursday, I wrote an article for Slate about The Anthology of Rap, officially published this week by the Yale University Press. In that piece, I pointed out the errors I'd found in reading the book's transcriptions of rap lyrics. (I also noted how tricky and difficult such transcription can be.) Inspired by my investigation, the blogger Jay Smooth launched one of his own and found more mistakes in the anthology, which he published on his site on Monday. Commenters on both Slate and on Jay Smooth's site, Nil Doctrine, noted that many of the mistakes found in The Anthology of Rap also appear on the Online Hip Hop Lyrics Archive, a collection of lyric transcriptions posted by hip-hop fans that has been around for more than a decade.
For example, I pointed out in my original article that The Anthology of Rap incorrectly quotes Ghostface Killah, in his track "Daytona 500," as saying "voice be metal like Von Harper." The line is actually "voice be mellow like Vaughn Harper," a reference to the smooth-talking radio personality. The "metal like Von Harper" mistake is also present in the version of "Daytona 500" posted at OHHLA.
This is not to say that every error in OHHLA's transcriptions made it into the anthology. The editors do not repeat another OHHLA mistake later in the same line from "Daytona 500." (OHHLA has "radio bubble"; the anthology correctly has "radio barber.")
Still, readers of my piece and Jay Smooth's were struck by how many of the errors discovered so far in the anthology are also apparent in the OHHLA transcriptions. Here is a chart listing errors found in my Slate piece and in Jay Smooth's Nil Doctrine post and indicating whether the same errors show up in the OHHLA transcription:
In 15 of the 18 instances listed here, the error also appears on OHHLA. While not proof positive, these numbers strongly suggested that OHHLA was a source for the anthology transcriptions. Further bolstering this impression was a post in the comments section at Nil Doctrine. A commenter posting under the name Julian Padgett wrote:
As one of the undergrads who helped [Anthology co-editor] Adam Bradley with the transcriptions, I'd like to personally apologize for some of the errors in the book, although fortunately I didn't do any of the songs in this post haha.
In a later post, responding to questions from Jay Smooth about the transcription process (and whether it relied on lyrics from OHHLA or similar sites), Padgett wrote:
Yeah, we were encouraged to find online lyrics and correct their errors as we formatted the line breaks to coincide with the measures of the track's beat. Sometimes there wasn't any option but to start from scratch though. Jean Grae's lyrics, for example, usually didn't have any online lyrics to use as a starting point.
A Julian Padgett is thanked in the acknowledgements section of The Anthology of Rap for his transcription work (along with several others). I located a Julian Padgett who is an undergraduate at the University of Colorado, Boulder—where Adam Bradley teaches— but he did not respond to my e-mail request for further comment.
I also contacted the editors of the anthology, Bradley and Andrew DuBois, both English professors, and asked them about their process for transcribing lyrics. I was surprised by the examples suggesting they may have relied on online transcriptions. In the introduction to the anthology, the editors write:
The most significant effort to record rap lyrics to date is not in a book but online, in the numerous user-generated lyric databases launched in recent years. The OHHLA is the most comprehensive. One can find nearly every rap song imaginable with a few keystrokes. However, these transcriptions are often so flawed that they are of limited use for anything more than a casual perusal.
That's a rather backhanded compliment, and it does not leave the reader with the impression that OHHLA played a significant role in the creation of this book. The editors are aware of the mistakes that plague online transcriptions and seem to have set out to provide more accurate ones in their book. So I asked them to describe in detail what their transcription process had been.
Bradley got back to me late Tuesday and requested that I publish his responses to my questions in full. Click
(1) Listen to each song multiple times, typing out an original transcription with the song itself as the primary "text."
(2) Pass that preliminary transcription on for checking by another set (or sets) of ears.
(3) Check that transcription against a range of other transcriptions, including online resources such as OHHLA and AZ Lyrics as well as print resources such as liner notes and lyrics published in books.
(4) Contact the rights holder and whenever possible the artist him- or herself to review our transcriptions. Nearly thirty artists did just that.
(5) Submit our transcriptions to the publisher for copy-editing.
(6) Subject every copy-edited song to another process of listening and correction.
(7) Review the page proofs by reading through each lyric again, revisiting problematic passages.
That's a painstaking, labor-intensive approach to transcription. Using the original song as the primary text—and not a potentially corrupted transcription—seems prudent. When you work from an existing transcription, it can be too easy for your ears to start hearing what your eyes are seeing on the page. Due to rap's heavy reliance on ever-changing, regionally inflected slang, it also seems smart to consult as many sources as possible, as the editors say they did.
The process Bradley describes is as thorough and careful as you'd expect of an academic volume. And its thoroughness also tracks with the editors' seriousness of purpose as expressed in their introduction to the anthology. They write that their book "treats rap as a body of lyrics that responds to transcription, explication, and analysis as poetry. The lyrics included offer a kind of laboratory of language for those interested in the principles of poetics." Later on, they add that they "have endeavored to include those indispensable lyrics that define the body of rap at the present time, lyrics whose formal virtuosity, thematic interest, or both merit preservation and sustained study." Given these mission statements—this book treats rap as poetry and endeavors to preserve exemplary lyrics for study—you'd expect great attention to have been paid to the transcription. Getting the lyrics right in a book like this isn't a bonus feature; it's the book's raison d'être.
What remains puzzling, however, is how so many errors made it into the anthology, and why so many of those errors repeat mistakes present in the OHHLA transcriptions. I asked the editors how they accounted for so many of the errors that have been caught so far also appear on OHHLA and other online lyrics sites. Their response to this question was less satisfying:
Like most rap fans, we've gone online over the years to find transcriptions of songs that interested us. As we describe above, we considered consulting these sites an important element—but only one element—in an exhaustive process.
That doesn't really answer the question. Naturally, there are certain instances in which a line is very tricky—because the words are garbled, because the beat is drowning out the vocals—and two different transcribers, working independently, could produce the same mistake. In other instances, however, it's hard to figure how multiple, independent transcribers could arrive at the same mistake. Consider the example of 50 Cent's "50 Shot Ya," which I mentioned in my original piece. Listen to this clip:
The anthology renders the line "Semiautomatic, Luger trey revolve 'em." I have listened to the song over and over, trying to hear the word Luger here, but to my ear, it's simply not present. (I believe the line is "semiautomatically or trey-eight revolve 'em," a trey-eight being a .38 handgun.) OHHLA's transcription of "50 Shot Ya" features the same Luger mistake as the anthology.