Keeping It One Hundred: Urban Dictionary, Mining Black Internet Culture and Why “Fleek” Still Matters
When I want to find out what a new phrase in African-American Vernacular English (AAVE) means, I am most likely to search for it on Twitter (now that Vine is gone). On Twitter, I can track the original memes, observe the adoption of the phrase and trace linguistic innovation in realtime. On Twitter, I can get a sense of what this new phrase means.
Making sense of what a phrase means on Urban Dictionary is different. Urban Dictionary is a profitable, bootstrapped company which relies heavily on Black internet culture’s content production and linguistic innovation to remain relevant. In this post, I offer a critique of the paper, “Emo, love and god: making sense of Urban Dictionary, a crowd-sourced online dictionary”. This is part of my work on culturally relative artificial intelligence, digital ephemera and orality.
Who, what and where is Black Twitter?
Although the phenomenon’s inflection point differs depending on whom you ask, respondents in this research described Black Twitter’s development as a space in which black people discuss issues of concern to themselves and their communities — issues they say either are not covered by mainstream media, or are not covered with the appropriate cultural context. For these users, Black Twitter allows everyday black people to serve as gatekeepers for the news and information needs of a plurality of black American experiences — with coverage, perspective and consideration not found elsewhere.
A recent report by the Knight Foundation delves into how several Twitter subcultures, including Black Twitter, interact with the news. They found that these online communities “give voice to issues that mainstream media don’t cover.” One community member who was interviewed shared, “Black women, black feminists, black gay men — they’re basically invisible communities outside of Black Twitter.” Other findings relating to trust, media framing, control over intellectual property, online harassment and news impact unpack the relationship between several marginalized communities and news media. This report provides pertinent, nuanced and robust insights. It is accessible, it delicately manages the overlapping interests of the different Twitter subcultures it looks at (e.g. Black Twitter and Feminist Twitter) and it is a wonderful case study in how to combine qualitative and quantitative methods in ways that make sense.
What about UK Black Twitter?
In my work on Grime music I have been grateful to encounter Dr Joy White’s research on Urban Music and how it is defined in the UK. It has also been my great joy to discover Francesca Sobande’s work on Black British YouTubers. Both researchers point to the conversation across the Atlantic between Black Brits and African-Americans and how the internet facilitates that conversation. This is particularly interesting as Grime music has played a substantive and at times substantial role in injecting Multicultural London English (MLE) into parts of AAVE.
A popular multicultural and millennial lifestyle podcast in the UK, #TheReceiptsPodcast, has received recognition from prominent members of Black Twitter in the US. Well-established Black British Vloggers regularly meet with their American counterparts, and more recently we’re seeing US Black Twitter make their way over here. UK Black Twitter is a subculture within global Black Twitter, and it’s important to recognise how these diaspora relationships translate into the digital realm; the scope of this work is the linguistic exchange between these two groups.
How “Urban” is Urban Dictionary?
Urban Dictionary was founded in 1999 by Aaron Peckham. In an interview about the early days of the site, he recounts its original purpose, “I started the site in 1999 as a parody of dictionary.com. I had just started college, and I thought it would be fun if my friends and I were the ones who wrote the dictionary. I also felt that people take the dictionary too seriously! In some ways, Urban Dictionary is the opposite of a regular dictionary — the definitions are opinionated, not researched, and sometimes even misspelled. But it can be useful and often hilarious.”
This week, in a new paper entitled “Emo, love and god: making sense of Urban Dictionary, a crowd-sourced online dictionary” researchers from the Oxford Internet Institute and the Alan Turing Institute task themselves with making sense of what they describe as a dictionary where “[u]sers contribute by submitting an entry describing a word and a word might, therefore, have multiple entries.” The paper goes on to explore three features of UD; growth, coverage and types of content. The paper is peppered with unsubstantiated inferences about formality of “slang” (which in many instances is code for AAVE). Bizarrely, it makes claims such as “the decentralized and often unmonitored environment of [large-scale collaborative] projects make them susceptible to low-quality content.”
Despite asserting a desire to “study language innovation” because “language is constantly changing”, the paper doubles down on Draconian ideas of what constitutes formal language;
“Another concern affecting such collaborative dictionaries is the question of whether their content reflects real language innovation, as opposed to the concerns of a specific community of users, their opinions, and generally neologisms and new word meanings that will not last in the language.”
In all the interviews available online, UD’s founder Aaron Peckham presents a harmless recounting of the almost accidental beginnings and happenstance success of the platform. In a 2010 interview with Terry Heaton’s Pomo Blog, there are some attempts made to acknowledge the more insidious aspects of the platform.
“We have definitely received e-mails from people asking us to remove definitions of common offensive words. But removing an offensive word from Urban Dictionary doesn’t remove it from the world’s vocabulary — and when somebody uses that word in the real world, we want its definition to be available in Urban Dictionary.”
In no uncertain terms, Clio Chang in The New Republic reports that while it is a successful site, “Urban Dictionary is also really, really racist.”
Offensiveness and the erasure of AAVE
There are two glaring omissions in the Urban Dictionary paper. Firstly, it makes zero attempts to address race, AAVE and the history of technology companies like Urban Dictionary, Genius and Giphy drawing on Black internet culture to simultaneously grow their platforms and profit without attribution. Secondly, the research design is lacksadalsicle, offering grand claims about the nature of large-scale collaborative projects on the web and language innovation without asking whether Urban Dictionary offers different “wisdoms” for different “crowds”.
The paper focuses of three features; Urban Dictionary’s growth, coverage and types of content. Growth is measured by the number of new entries added each week, where “headwords” refer to definitional terms. Coverage is not given a definition in this paper. The content types categories are proper nouns, opinions and familiarity. The offensiveness of these content types is also analysed.
The abstract and introduction of the paper make several demanding claims while the manual observations are not substantive. I empathise with some aspects of the methodological approach of this paper; researchers used computation (the researchers crawled UD, parsed Wiktionary, sampled headwords “according to the number of their entries”) and collected annotations using CrowdFlower. I appreciate that the researchers “marked the crowdsourcing tasks as containing explicit content, so that the tasks were only sent to contributors that accepted to work with such content.”
This paper examines the frequency with which headwords appear in different content type categories in UD and compares these to Wiktionary. The term “threshold” refers to the threshold of offensive for individual definitions. The researchers asked the crowd-workers to indicate how offensive they found definitions. In the example below, the ratings were “reversed to give an intuitive presentation of the results”.
The framing, however, leaves much to be desired. There are several holistic claims made which misrepresent the applicability, scope and impact of the findings. The work on offensiveness fails to capture some simple facts about UD as a platform; earlier we saw that Aaron Peckham had intended to build the site as a parody of Dictionary.com. Parody, satire and humour lend themselves to offensiveness in qualitatively different ways than their more earnest antonyms. On Urban Dictionary, you can find offensive words as well as offensive definitions. This uniqueness is not sufficiently teased out by the researchers. This is acknowledged, “UD not only contains offensive entries describing the meaning of offensive words, but there are also offensive entries for non-offensive words (e.g. a definition describing women as ‘The root of all evil’).” Definitions can be intended to be funny yet deeply problematic; the framing of offence in this paper does not fully take into account humour, context or intention. The approach to resolve this tension, is lacklustre:
To investigate how offensive content is distributed in UD, we ran a crowdsourcing task on CrowdFlower (see Data and methods for more details). Workers were shown three definitions for the same headword, which they had to rank from the most to the least offensive.
The essence of offensiveness it that it has a harm-prone or harm-intended capacity and is contingent on being able to take the perspective of people who might be offended. What is unclear throughout the paper is how both the crowd-workers and the researchers were coding offence.
These misrepresentations have actual consequences. Given that many of the manual observations in this paper have a bias towards formal language register, there is a distinct likelihood that neither the researchers nor the crowd-workers had experience coding AAVE. This opens up the possibility that a huge amount of the data which was crawled was mis-categorised as either offensive, informal or random. By creating a crude ranking of offence, where there are only three possible options, and not taking into account, for example, that different races view offence differently, the paper creates a problematic methodological chasm for racial slurs to slip through without interrogation.
Overexposure can also be a form of erasure. One criticism levelled at internet research in elite academic institutions is that the path from publication to media proliferation of research findings is relatively seamless. This is problem because it means that there is little time to gather feedback from the academic community at large and even fewer mechanisms for scholarly accountability and scrutiny. An added problem is that if you get things wrong, the marginalised groups who you are researching tend to trust you less. This seems to be the case with journalists and their relationship with online subcultures as explored in the Knight Foundation report mentioned earlier in this post. The Urban Dictionary paper has garnered a lot of press interest as a direct result of the scope of the claims it makes.
It is shameful that the researchers chose the term “fleek” and its variations to use it as academic clickbait and media coverage without attributing any single citation or reference to Kayla Newman AKA Peaches Monroee, who created the word and is the source of its virility.
For now, I am keen to pay homage again to the folks at the Knight Foundation because their report used a brilliant mixed-methods approach. In the journal article version of this post, I unpack the methods used in both documents and evaluate what we can learn; I make recommendations and provide a comprehensive literature review for internet researchers and journalists covering Black internet culture.
Who does the media consider an expert on Black culture online?
In the summary finding of “How Black Twitter and other social media communities interact with mainstream news” Deen Freelon (University of North Carolina at Chapel Hill), Lori Lopez (University of Wisconsin-Madison), Meredith D. Clark (University of Virginia) and Sarah J. Jackson (Northeastern University) explain a few things pertinent to the media reporting of the Urban Dictionary research paper. An interview with a member of Black Twitter brings to light how journalists perceive the timeline as a shortcut to gaining access to the community without “going into the physical representation”. The report also highlights how this harvesting by journalists (and in this instance, academics) “exposes [Black Twitter users] to potential online harassment, threats or violence that they otherwise might not have faced had their tweet not been promoted on a larger platform.”
“Active participants did not like having their tweets harvested by journalists for story permission, citing two major concerns: lack of control over intellectual property and the potential for online harassment.”
At the time of writing this post, it has been five days since the Urban Dictionary research paper was published. So far it has appeared in British and New Zealand press. In every press article and video coverage, there are no Black people referenced, featured or even seen. In the video above, the term “fleek” is erroneously associated with Kim Kardashian.
Is collegiality failing Black academics?
I am not a professor. I do, however, spend a lot of my time reading work from Black women academics and wondering, if the academy is a safe and encouraging space to do my best work.
Being a black professor at a predominantly white university can be just as uncomfortable as — if not more so than — being a black student at one.
An article in the Atlantic, Adia Harvey Wingfield traces the “Plight of the Black Academic” and how “for universities that see no real reason to change their existing practices, traditions, and organizational cultures, bringing in a critical mass of faculty of color is often a stated goal that never materializes.”
There is a convention in academic circles that one ought not to critique the work of peers (or faculty members) in public. The idea of collegiality is seductive and would make sense if everyone in the department had equitable access to the fruits of labour. Black women do not have equity in the academy, so it seems particularly absurd to expect us to “take one for the team”. We are few and far between and if we have the privilege of not being on a #TenureTrackHustle, we can shed a light on areas of research which erase Black experiences online.
The OII are hiring for new senior faculty positions. I would love to see Black women in these roles. I would also like for their work to be as respected, supported and as venerated as their non-Black peers.
Liked this article?
You can buy me a book or a cup of coffee to fuel my research.
If you’d like to share this article or use parts of it for something you are working on, please make sure to cite me (@natalieisonline). This article is a draft for one I’m working on for publication in an academic journal; for now, you can cite is a blog post. It might seem absurd to raise this given the nature of this article, however, unfortunately from activists to Pulitzer Prize winners, the appropriation of intellectual property from Black women is commonplace online. Rachel Kaadzi Ghansah summarizes:
It takes nothing to cite people. I cite billions of people constantly in my writing. And yet, this is what happens to black women over and over again as we try to do this work. We get devalued or wiped out.
Equally, if you see my work being used without attribution, I would really appreciate you flagging this. I have a newsletter where I talk about what I’m working on, how many yoga classes I’ve missed this week and where you can find me presenting my work.