Do you remember tag clouds? They started during the heyday of blogging: we all had a small one in the sidebar of our blog and oftentimes you’d have this small java or Flash applet spinning and showing a measure of what were your favorite subjects — pets, cooking, norwegian black metal, you get the idea; actually, you could get the idea, if you cared to have a look at the tag cloud (which almost nobody did).
The idea of showing quantitative information linked to the relative font size dates to way before that: maps have been showing city names in different sizes relative to the city population for a few centuries now, and a few more applications of the concept have popped up here and there, but it’s safe to say that the “tag cloud” concept was picked up during the “first” web 2.0 era and quickly became ubiquitous (talk about too much of a good thing). The main drawback was that, besides being cute for visualisation purposes, as far as site navigation goes — eh. Delicious and Flickr were the pioneers; so much so that in their 2004 Webby acceptance speech, Stewart Butterfield from Flickr was quoted apologising for it.
Sorry about the tag cloud.
And that was about it. But the fact it’s not the all-purpose tool it was initially thought to be doesn’t mean the tag cloud is completely useless: it’s still an excellent way to show very quickly what the general gist of a text is about, moreso if you’re trying to decode large amounts of content which has been natively tagged for context. Like, tweets, or Instagram posts or comments. A #hashtag cloud can show very quickly what are the ongoing themes in a vast pool of online conversations and sometimes also an empirical measure of the general sentiment.
Enter emojis: from IMs to the web, the addition of little faces (and then all the few hundreds of them) dates back to the texting habits of japanese teens in the mid-1990s — we were just about getting used to “ 🙂 ” and “ 🙁 ”; from there, it took off rapidly and with the explosion of smartphones and mobile IMs now we’re getting to a point where emoji usage has expanded beyond words. In fact, the Oxford dictionary, by way of SwiftKey data, recently announced that this is the Word of the Year for 2015, with its president quoted:
It’s not surprising that a pictographic script like emoji has stepped in […] — it’s flexible, immediate, and infuses tone beautifully. As a result emoji are becoming an increasingly rich form of communication, one that transcends linguistic borders.
So, when we’re analysing the ongoing conversations in social media, it’s only normal to expand the analysis to further levels of meaning by taking into consideration not just the wording of the content, but its pictographic form as well, and while running a rather ordinary frequency analysis on the wording of a few thousands Instagram comments, we noticed the impressive amount of emojis and decided to plot them in a “cloud”, sized by frequency.
The quick-and-dirty visualisation shows a lot of the general sense of the comments, and their sentiment (pay no attention to the tiny Munch-esque screaming face at the top center) as well, so much so that it speaks volumes without even bringing up the associated numbers.
We need to take into consideration quite a few factors before implying that this is completely meaningful in its own right, though — and most of the caveats that apply to language analysis and visual communication hold here. Firstly, does the same emoji mean the same thing across the world? Not really. Did you know that the “nail polish” emoji means “don’t care” in UK usage?
Quantitatively, emoji users are prone to repetitions. So, if you’re using three consecutive red hearts(tap-tap-tap) does this mean that you’re loving what I say or -sigh- me three times (or rather 2.718) more than a single red heart? (Whatsapp addressed this: one single heart is larger, and animated; two or more are smaller and static).
Emojis have entered mainstream communication a few years ago, which is quite a while (it’s measured like dog years, I’m told), but as far as content analysis goes, they’re still in their infancy. We’re starting to look into them as a complement to regular analysis, and they’re showing more than a little promise.