A post from Shady Characters

Emoji, part 6b: steps in the right direction

This is the seventh in a series of thirteen posts on Emoji (😂). Start at PART 1, continue to PART 8 or view ALL POSTS in the series.

The emoji enlightenment dawned in August 2015. As we saw last time, that was the month in which the Unicode Consortium published “Emoji 1.0”, a document that listed all available emoji characters and, crucially, described how to create new emoji by combining existing symbols.1 It was a big change to the status quo, and it was done with one overriding aim in mind: to allow emoji to become more representative of the people who used it. So what did Unicode do with that newfound freedom? We’ll find out over the next two parts as we follow emoji’s journey from Emoji 1.0 right up to the present day.

With Emoji 1.0, the Unicode Consortium introduced two ways to create new emoji. First, they gave an existing Unicode character called the ZERO WIDTH JOINER the power to fuse emoji together to make new symbols (👩 + ZWJ + ❤️ + ZWJ + 👩 = 👩‍❤️‍👩 ). Second, they created five new “Fitzpatrick modifier” symbols, each one representing a different skin tone (‘🏻’, ‘🏼’, ‘🏽’, ‘🏾’, and ‘🏿’), each of which modified an immediately preceding emoji to change its skin colour (👶 + 🏾 = 👶🏾).2 It was the latter of these that made the first, biggest splash.

Several months before Emoji 1.0 had been formally published, in the spring of 2015, the Atlantic, the Washington Post, the Guardian and a number of other news outlets reported excitedly that Apple was planning to introduce skin tone support to the emoji available on its devices.3,4,5 It could easily have been a non-story: an Apple news website called 9to5Mac had discovered some new emoji in the beta version of a minor update to Apple’s desktop operating system, OS X, and said as much in a perfunctory news story about the release.6 In an world in which emoji were not alienated from a large proportion of their users, that would have been the end of the matter. But dissatisfaction with emoji had only grown more pronounced in the wake of Apple’s admission in March 2014 that yes, emoji were not representative of many of the people who used them.7 Though the final version of Apple’s software would not be made available until April 2015, and though the official publication of Emoji 1.0 was still months away, Apple was eager to show itself to be a company in tune with the times — and in turn, the media was happy to jump on the story.8,9,2 The headlines said it all: “Finally, Emoji People of Color” (Robinson Meyer, the Atlantic3); “Finally, Apple emojis reflect America” (Dean Obeidallah, CNN10); “Diverse thumbs up! Emojis with different skin tones finally here” (BBC Newsbeat11).

Apple had adopted wholesale Emoji 1.0’s draft recommendations for the handling of skin tone. Chiefly, this meant adding support for the five skin tone modifiers, but it also meant changing the default apperance of the emoji to which they could be applied. In drafting Emoji 1.0, Peter Edberg and Mark Davis had suggested that “When [realistic emoji] are shown in color, they should preferably use a non-realistic skin tone (such as yellow-orange)”12 — that is, the absence of a realistic skin tone should be as obvious as its presence. As such, Apple modified its previously white-skinned human-form emoji — GIRL (👧), BOY (👦), WOMAN (👩), MAN (👨) and so on — to use the same bright yellow skin that had long been used for more stylised “smileys”. In doing so, they exchanged one problem for a brace of new issues.

White people in Twitter's 2014 emoji
White people in Twitter’s 2014 smileys. (CC-BY 4.0 images taken from Twemoji version 1.4.0.)

First, and perhaps most obvious, was the clumsy choice of yellow as a “neutral” skin colour. As Caitlin Dewey of the Washington Post explained in an article entitled “Are Apple’s new ‘yellow face’ emoji racist?”:

The problem arises, of course, when [yellow skin] is applied to a more humanoid face — particularly as part of a big “multicultural” rebranding. Now Apple is asking us to see emoji not as icons, only, but as actual representative stand-ins for real human people. And as many a Twitter pundit has pointed out, when real human people are called “yellow-faced,” it’s … explicitly racist.4

White people in Twitter's 2016 emoji
White people in Twitter’s 2016 smileys. (CC-BY 4.0 images taken from Twemoji version 2.0.0.)

And there was another problem. When applying Emoji 1.0’s skin tone modifiers to each of their newly–yellow-skinned emoji, Apple simply reused the underlying artwork so that, for example, ‘👩’ became ‘👩🏻’, ‘👩🏼’, ‘👩🏽’, ‘👩🏾’ and ‘👩🏿’. Unfortunately, this is not how ethnicity works: a black person is not just a white person with darker skin, and a white person is not just a black person with lighter skin. Apple had not created diverse emoji so much as a graphic warning of the dangers of the tanning salon. Paige Tutt of the Washington Post pithily summed up the problem: “These new figures aren’t emoji of color; they’re just white emoji wearing masks.”13 To be fair, Apple had merely implemented the Unicode Consortium’s own guidelines — guidelines that recognised skin tone but not race — but in doing so they had made themselves a lightning rod for criticism.

By 2016 the novelty of emoji skin tones had worn off but the controversy had not. Writing for the Atlantic, Andrew McGill found that in the USA, the first five of emoji’s six skin tones were used on Twitter in roughly the same proportions, despite Twitter’s user base being far more white than black. (The very darkest skin tone, EMOJI MODIFIER FITZPATRICK TYPE-6, or ‘🏿’, was used considerably less often than each of the others.) In interviewing Twitter users for his article, McGill found that some white users felt so uncomfortable using a mechanism that had been designed for non-white people that they considered an emoji with a light skin tone to be tantamount to a “white power” gesture. But in falling back to the default yellow emoji, McGill suggested, those same white users were essentially claiming that supposedly neutral skin tone for themselves — and scoring an own goal in the process.14

Complaints of racial insensitivity enabled by Apple’s new emoji were not confined to the opinion pages. Mere days after the new emoji had been launched, Clorox, a brand of cleaning supplies, tweeted a collage of new emoji in the shape of a bottle of their bleach. The picture was captioned:

“New emojis are alright but where’s the bleach.”15

The intent, so Clorox hastily explained, had been to lightheartedly call out the absence of a “bleach” emoji, but the implication of metaphorical whitewashing did not sit well with many of Twitter’s users.16,15 Also on Twitter, the Grey’s Anatomy actor Ellen Pompeo and fashion model Kendall Jenner, both of them white, were criticised in 2016 and 2017 respectively for using non-white skin tones in tweets.17,18

In light of all the problems surrounding skin-tone emoji, it is tempting to wonder: what if Unicode had simply recommended a different “neutral” skin colour? Could a patently inhuman colour without yellow’s negative connotations — cyan, say, or bright green — have obviated the need for realistic skin tones entirely? We’ll never know, of course, because Unicode did not do that. Moreover, for all the ensuing missteps — Unicode’s creation of a mechanism for emoji diversity that made it possible to be even more racist online; Apple’s over-hasty rush to implement it; and our own use (and misuse) of the resulting emoji — the Unicode Consortium held firmly to its chosen course. Google and Facebook, both voting members, added skin tone support in 2016 and 2017 respectively; Twitter, an associate member, in 2016.19,20 Like them or not, the ethnically-diverse icons of Emoji 1.0 are here to stay.

The next tectonic shift in the emoji landscape came late in 2016 as part of Emoji 4.0. (For the most part, Emoji 2.0 and Emoji 3.0 had been concerned with codifying which emoji supported skin tone variations, but more on these two emoji updates will follow in a subsequent post.) This time round, the changes were motivated by a paper entitled “Expanding Emoji Professions: Reducing Gender Inequality” that had been submitted to the Unicode Consortium in May 2016. In it, Rachel Been, Nichole Bleuel, Agustin Fonts and Mark Davis made a two-pronged case for expanding the range of professions represented in emoji and for letting users choose from male and female versions of each one.21 The four (all Googlers, though Davis also moonlighted as Unicode’s president) opened their proposal with a quote from an NYT op-ed entitled “Emoji Feminism”, published in March 2016. Amy Butcher, its author, pulled no punches:

Where, I wanted to know, was the fierce professor working her way to tenure? Where was the lawyer? The accountant? The surgeon? How was there space for both a bento box and a single fried coconut shrimp, and yet women were restricted to a smattering of tired, beauty-centric roles? […] Men were serving on the police force, working construction and being Santa. Meanwhile, on our phones, it was Saturday at the Mall of America — women shopping while men wrote the checks.22


To get round the sclerotic dual-track processes of the Unicode Technical Committee and the ISO/IEC, Been et al turned to the ZERO WIDTH JOINER as the means to implement their new emoji. By gluing together existing characters, rather than proposing new, standalone symbols, the Google quartet gained access to a fast-track standardisation process aimed at helping emoji become more diverse, more quickly.23 As an example, WOMAN plus ZWJ plus WRENCH might produce a female mechanic: “👩 + ZWJ + 🔧 = ‍👩‍🔧”. “Expanding Emoji Professions” suggested thirteen such formulations, each of which provided for both male and female versions:

👩 + ZWJ + 💼 = ‍👩‍💼
👨 + ZWJ + 💼 = 👨‍💼
👩 + ZWJ + ⚕ = ‍👩‍⚕️
👨 + ZWJ + ⚕ = 👨‍⚕️
👩 + ZWJ + 🔬 = ‍👩‍🔬
👨 + ZWJ + 🔬 = 👨‍🔬
👩 + ZWJ + 🎓 = ‍👩‍🎓
👨 + ZWJ + 🎓 = 👨‍🎓
👩 + ZWJ + 💻 = ‍👩‍💻
👨 + ZWJ + 💻 = 👨‍💻
Industry (e.g. factory worker)
👩 + ZWJ + 🏭 = ‍👩‍🏭
👨 + ZWJ + 🏭 = 👨‍🏭
Industry (e.g. mechanic)
👩 + ZWJ + 🔧 = ‍👩‍🔧
👨 + ZWJ + 🔧 = 👨‍🔧
👩 + ZWJ + 🚜 (proposed); 👩 + ZWJ + 🌾 = ‍👩‍🌾 (ratified)
👨 + ZWJ + 🚜 (proposed); 👨 + ZWJ + 🌾 = ‍👨‍🌾 (ratified)
Food service
👩 + ZWJ + 🍳 = ‍👩‍🍳
👨 + ZWJ + 🍳 = 👨‍🍳
👩 + ZWJ + 🏫 = ‍👩‍🏫
👨 + ZWJ + 🏫 = 👨‍🏫
👩 + ZWJ + 🤘 (proposed); 👩 + ZWJ + 🎤 = ‍👩‍🎤 (ratified)
👨 + ZWJ + 🤘 (proposed); 👨 + ZWJ + 🎤 = 👨‍🎤 (ratified)

Eagle-eyed readers will count only eleven professions here. Been and her co-authors suggested separate “doctor” and “nurse” healthcare emoji, but only a single, generalised version was ultimately approved. A proposed “high-tech industry” profession, formed by 👩 + ZWJ + ⚡️, was also rejected.21 Worst of all, the joyless technocrats of the Unicode Emoji Subcommittee rejected “👩 + ZWJ + 🤘” as the sequence for “singer”, or “rock star”, in favour of the crushingly literal “👩 + ZWJ + 🎤”.

But if Been, Bleuel, Fonts and Davis lost those minor battles, they won the war convincingly. When Emoji 4.0 was released in November 2016, it took the premise from “Expanding Emoji Professions” and ran with it, adding not thirteen new professions but sixteen,* all available in male and female versions. More significantly, it used the same ZWJ-based mechanism to make essentially all human emoji available in gendered variants. HAPPY PERSON RAISING ONE HAND, for example, which was nominally genderless but usually rendered as female, could now be combined with ‘♀️’ or ‘♂️’ to create explicitly female (🙋‍♀️) and male (🙋‍♂️) versions respectively. Emoji’s 4.0’s sweeping changes to gender rocked emoji just as surely as the skin-tone variations of Emoji 1.0 — and, speaking of which, Emoji 4.0 also codified skin-tone support for its new professions and newly-gendered emoji.24,25

All this led to a combinatorial explosion of emoji. Depending on how you count (and, for a variety of technical reasons, counting emoji is not at all easy26), Emoji 4.0 introduced somewhere north of 750 new glyphs, leading to a cumulative total of more than 2,300 emoji from which users could choose.27 Emoji platform vendors — Google, Apple, Facebook, WhatsApp and the like — scrambled to add emoji search boxes and other mechanisms to make it easier to apply skin tones, gender changes or both.28,29,30

There was another, subtler side-effect. To render the male version of HAPPY PERSON RAISING ONE HAND with EMOJI MODIFIER FITZPATRICK TYPE-5, for instance, a typical denizen of Emoji 4.0, requires five separate Unicode characters: “🙋 + 🏾 + ZWJ + ♂︎ + VARIATION SELECTOR-16 = ‍🙋🏾‍♂️”. There’s a lot going on here, so let’s examine each character in turn:

  1. 🙋 — our base emoji. Though it is normally rendered as a woman, it has no specific gender.
  2. 🏾 — a skin-tone emoji that modifies the emoji preceding it to create ‘🙋🏾’.
  3. ZWJ — the “glue” character that combines the preceding “🙋🏾” with the male symbol that follows.
  4. ♂︎ — joined to the ‘🙋🏾’ that precedes it by the intervening ZWJ so that it displays as the male variant of the emoji.
  5. VARIATION SELECTOR-16 — another invisible character, like the ZWJ, that modifies the visible characters around it. VARIATION SELECTOR-16 converts certain characters normally drawn in a plain, typographic style into their emoji counterparts. In this case, it is applied to ‘♂︎’, to convert it into its emoji-fied variant, ‘♂️’, so that it correctly modifies the ‘🙋🏾’ emoji that precedes it.

That’s five Unicode characters, totalling 17 bytes of storage, to render a single emoji. (Just as it is hard to count emoji, working out the number of bytes occupied by a string of Unicode characters is not always intuitive. Blame UTF-8, Unicode’s clever but convoluted scheme for turning “code points”, or individual characters, into the smallest possible number of bytes.31) By way of comparison, most characters that can be typed on a standard QWERTY keyboard require only a single byte of storage,32 but even so, in a world of 4K screens, 4G connections and unlimited streaming budgets, 17 bytes per emoji is not unreasonable. On the face of it, at least.

For some platforms, however, 17 bytes per character is entirely unreasonable. The ubiquitous but ancient SMS text messaging service, for example, designed in 1985 to secrete its messages in an under-used part of the mobile phone network protocol, managed to carve out space for a mere 160 seven-bit characters per message.33 Use an esoteric character such as ‘é’ or ‘ø’ and this drops to 140 eight-bit (i.e. one byte) characters per message; use a multi-byte character from the depths of Unicode, such as an emoji, and the budget drops to just 70 characters. Composite emoji, therefore, such as the five-character ‘🙋🏾‍♂️’, are positive space hogs.34

Twitter, too, imposes a character limit on each post, although unlike SMS its 280-character cutoff is an artificial limit imposed to keep messages short and to the point. In common with SMS, however, emoji on Twitter had always been more costly than normal characters, and composite emoji costlier still. In 2018, then, to avoid penalising users for adding skin tone or gender to their emoji, Twitter modified its character counting logic so that all emoji were considered to occupy two of those 280 characters, whatever their actual length in bytes.36,37 It was a small step towards a more equitable emoji world, but a notable one: no longer would the minutiae of character encoding schemes be permitted to influence how one might express ethnicity or gender. All emoji were now equal in Twitter’s eyes.

All this aside, Emoji 4.0 and its gendered professional emoji received a far more relaxed reponse than had Emoji 1.0 and its support for skin tones. Headlines were either neutral or tended towards the positive: “New Female Career Emojis Are Coming!”;38 “Finally: Female Emojis Will Have Actual Occupations Other Than ‘Dancing Lady’”;39 “Unicode’s next emoji update focuses on gender and jobs”.40

Unicode’s drive to diversify emoji was finally having the desired effect. Next time round, we’ll take a look at just how far emoji have come since then — and how far they still have left to go.

Jeremy Burge, “Emoji Version 1.0”, Emojipedia
Mark Davis and Peter Edberg, “Unicode Technical Report #51: Unicode Emoji (version 1.0)”, Unicode.Org, 2015. 
Robinson Meyer, “Finally, Emoji People of Color”, The Atlantic, February–2015. 
Caitlin Dewey, “Are Apple’s New ‘yellow face’ Emoji Racist?”, Washington Post, February–2015. 
Jess Zimmerman, “Racially Diverse Emoji Are a Nice Idea. But Will Anyone Use Them?”, The Guardian, April–2015. 
Zac Hall, “More Diverse Emoji Characters Likely Coming With OS X 10.10.3”, 9to5Mac, February–2015. 
Joey Parker, “What Does Apple Think About The Lack Of Diversity In Emojis? We Have Their Response.”, MTV Act, March–2014. 
“OS X Yosemite 10.10.3 Update”, Apple Support, 2015. 
Mark Davis and Peter Edberg, “Proposed Draft Unicode Technical Report #51: Unicode Emoji (version 1.0)”, Unicode.Org, 2014. 
Dean Obeidallah, “Finally, Apple Emojis Reflect America”, CNN, April–2015. 
“Diverse Thumbs Up! Emojis With Different Skin Tones Finally Here”, BBC Newsbeat, April–2015. 
Peter Edberg and Mark Davis, “L2/14-172R: Proposed Enhancements for Emoji Characters: Background”, 2014. 
Paige Tutt, “Apple’s New Diverse Emoji Are Even More Problematic Than before”, Washington Post, April–2015. 
Andrew McGill, “Why White People Don’t Use White Emoji”, The Atlantic, 2016. 
David Goldman, “Clorox Apologizes, Deletes Tweet After Racial Uproar”, CNN Money, April–2015. 
Clorox, “Wish We Could Bleach Away Our Last Tweet. Didn’t Mean to Offend - It Was Meant to Be about All the [toilet] [bathtub] [wine Glass] Emojis That Could Use a Clean Up.”, Twitter, April–2015. 
Laura Beck, “Ellen Pompeo Criticized for Using Black Emoji, Responds by Getting Super Defensive”, Cosmopolitan, December–2016. 
Ikran Dahir, “People Are Calling Out Kendall Jenner For The Emoji She Used In A Tweet”, BuzzFeed, August–2017. 
Jeremy Burge, “Twemoji 2.0 Emoji Changelog”, Emojipedia, December–2015. 
Unknown entry 
Rachel Been et al., “Expanding Emoji Professions: Reducing Gender Inequality”, 2016. 
Amy Butcher, “Emoji Feminism”, New York Times, March–2016. 
“Submitting Emoji Proposals”, Unicode.Org, 2019. 
“113 New Unicode Emoji (plus Skin Tones)”, The Unicode Blog, November–2016. 
Jeremy Burge, “What’s Planned for Emoji 4.0”, Emojipedia, August–2016. 
“Unicode Emoji Chart Format”, Unicode.Org, 2019. 
“Emoji Versions, v12.0”, Unicode.Org, 2019. 
“Use Emoji on Your IPhone, IPad, and IPod Touch”, Apple Support, 2018. 
“Using Emoji”, WhatsApp FAQ, 2019. 
Cameron Summerson, “How to Search for Emoji and GIFs in Android’s Gboard Keyboard”, How-To Geek, December–2016. 
Francois Yergeau, “RFC 3629: UTF-8, a Transformation Format of ISO 10646”, ed. Network Working Group, November–2003 November–2003. 
Mathias Bynens, “UTF-8 String Length & Byte Counter”, mothereff.In
Mark Millan, “Why Text Messages Are Limited to 160 Characters”, Los Angeles Times, May–2009. 
“Character Sets and Coding”, in 3GPP TS 23.038 V15.0.0 (2018-06), 2018, 19-24. 
Casey Newton, “Twitter Just Doubled the Character Limit for Tweets to 280”, The Verge, September–2017. 
Andy Piper, “New Update to the Twitter-Text Library: Emoji Character Count”, Twitter Developers, 2018. 
Jon Porter, “Twitter Now Counts All Emoji Equally, Regardless of Gender or Race”, The Verge, December–2018. 
Katie Rosseinsky, “New Female Career Emojis Are Coming!”, Grazia, July–2016. 
Megan Friedman, “Finally: Female Emojis Will Have Actual Occupations Other Than ‘Dancing Lady’”, Marie Claire, July–2016. 
Jessica Conditt, “Unicode’s Next Emoji Update Focuses on Gender and Jobs”, Engadget, August–2016. 
In addition to the eleven mentioned above, Emoji 4.0 added the following five new professions at Apple’s request:

👩 + ZWJ + 🎨 = ‍️👩‍🎨
👨 + ZWJ + 🎨 = 👨‍🎨
👩 + ZWJ + 🚀 = ‍👩‍🚀
👨 + ZWJ + 🚀 = 👨‍🚀
👩 + ZWJ + 🚒 = ‍👩‍🚒
👨 + ZWJ + 🚒 = 👨‍🚒
👩 + ZWJ + ⚖️ = ‍👩‍⚖️
👨 + ZWJ + ⚖️ = 👨‍⚖️
👩 + ZWJ + ✈️ = ‍👩‍✈️
👨 + ZWJ + ✈️ = 👨‍✈️


Depending on your browser, this may or may not actually work. 
Twitter increased its character limit from 140 to 280 as recently as 2017. The original limit had been chosen to allow users to interact with Twitter via SMS.35 

2 comments on “Emoji, part 6b: steps in the right direction

  1. Comment posted by dtw42 on

    Regarding default skin tones (and the difficulties in picking one that won’t prompt outcries) – blues and greens surely were never going to take off – they’d make everybody look ill or like aliens. Isn’t grey (dull-looking though it is) the best solution? If someone’s rendered in mono, you can’t tell what colour they are…

Leave a comment

Required fields are marked *. Your email address will not be published. If you prefer to contact me privately, please see the Contact page.

Leave a blank line for a new paragraph. You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>. Learn how your com­ment data is pro­cessed.