FEATURE1 June 2010

Words ain't what they used to be

Researchers are anxious to make the most of blogs, forums and social networks. But, Rosie Campbell asks, are we equipped to analyse the new language of online communication?

More and more research is happening online, and the trend shows little sign of abating. Indeed, the growth is exponential and arguably a terrific thing for the research community.

Or is it? I am concerned about the rush to sell MROCs, panels, SMS surveys, communities, listening platforms, e-ethnography and buzz mining (which, incidentally is more or less what we used to called ‘desk research’). There are core issues that we are not addressing. What are we making of all this online and electronic data? Is it deepening our understanding or are we just accumulating access points?

At one level these prolific data streams are useful and enriching. When did you last visit a new client or start working with a new agency without checking out their website, their blog, their LinkedIn profile? We have ever-widening access to samples and methodological opportunities for research. The amount of data we can scour, collect and deliver seems endless. What could be better in a world where data is our bread and butter?

I see two clouds that could rain on this parade.

First is the worrying tendency to conflate ‘data’ with ‘insight’. We hear this everywhere, especially in the media and the blogosphere. We read about “continuous insight feeds from social media” (that’s from an industry grandee) and about how forums “are such a rich source of insight”. There may well be insight to be sourced through engagement with social media communities and groups, but most often what is referred to is simply what we used to call verbatims. We used these to illustrate and help to evolve insight for our clients, but we didn’t offer them up as the ‘findings’.

Surely by now it is established that the material of our research – the language, comments, words of all kinds – are just the clay from which we create answers for, and with, our clients. It is not the answer in itself.

A second concern about the development of online and mobile research methods is rooted in the issue of analysis. Much of the data we are talking about comes from social networks and online communities. Frankly this is a new and very different language (or languages). We cannot behave as if we know what it all means. We cannot use old models of garnering meaning. We cannot rely on the surface of the words. We cannot assume a simple ‘transmission’ model of communication.

Furthermore, we do not understand the context in which it has been created. We increasingly collect this data in real time – there is no day or night, no geography, no boundaries. And yet on the horizon are text analytics and capturing systems which may mean a team of bots can do our research work for us while we sip a glass of Merlot and watch Mad Men.

Any researcher worth their salt knows that taking words, comment and expressed beliefs at face value never got us to the breakthrough understandings that our clients pay us for. Qualitative researchers are inclined to take the view that language is about constructing as well as delivering – ‘messages’ are not exactly the same as that which is sent or that which is received. The reality is constructed in the transaction, the words and phrases selected, the understandings created.

As researchers, when we hear a string of ‘mights’, ‘maybes’ or ‘other people coulds’, we know from experience of analysing language in the moment that we’re getting a polite rejection of what we’re talking about. In other words, we have always acted – wisely – as if language tells us far more than the simple facts and has the potential to reveal worlds beyond the words. We find evidence of the cultural discourses that run through society, the stories that inform thinking in particular market areas and the hard-wired belief systems of individuals.

But online, in social networks and certainly in textspeak, whole new worlds of language use are emerging.

Are you the same person on Twitter, Facebook, LinkedIn? Are you the same person in any virtual environment as in your real world? Isn’t some of the point of ‘playing’ in these spaces about creating new identities? Significant cultural discourses are created by our ‘semi-avatar’ beings re-tweeting a link whose purpose might be to tell a story about the kind of intellectual life we want others to see us living, or, in gathering more friends or followers, increasing our ‘popularity’.

And what about the language we use in these new forums? Take this example of a snippet from my 20-year-old daughter’s Facebook page a couple of months ago (which I have her permission to reproduce, I should add):

?Jo Allen massive lolz ed
Jo Allen wow wa wee wa
Jo Allen me julie best mum
Jo Allen buy tickets mandem. wick

“?All this obfuscation helps to place her within her peer group, define the group’s boundaries – and limit her mother’s stalking”

Visible for perhaps two or three days, the page contained statements and exchanges about my daughter, her friends, her age, her lifestyle and many encoded ideas. Are we equipped to analyse this kind of language in the way we might have analysed the ‘maybe’ comment in a focus group or the unspoken meanings in an observed ethnographic conversation?

The qualitatively inclined researcher might be better placed to explore the meaning of, for example, the almost compulsory use of lower-case letters. This indicates a facility with the medium, is modelled on the single-finger mobile phone texting process (so it speaks of an on-the-go world), and heralds a knowledge of highly contemporary culture, especially when used for first names (‘ed’ is a friend). Or how about the obfuscatory abbreviations – what does that choice of words signify? ‘Lolz’ is a third-level reference, firstly to the original ‘lol’ (the well-established internet abbreviation for ‘laugh out loud’), secondly Americanised via the z in the plural, and thirdly it’s most likely ironic anyway, with no actual laughing involved. All this obfuscation helps to place Jo within her peer group and, like spray-marking by animals, helps to define the group’s boundaries, keep out the uninformed and limit her mother’s stalking.

The use of a slang much employed by white middle-class youth is evident in the reference to a line from a rap song and the word mandem indicating the group of friends (ironic given that the request is to buy air tickets for a ski holiday).

Some of this I know because, as the author’s mother, I am semi-included. Some of it comes from cultural exposure, and some I understand because I am positing new language analysis which is appropriate to this new kind of use.

I am not an expert but I can glean far more understanding from far less content than any text analytic gizmo out there – essentially because I have the intellectual lubricant of human emotion which, as neuroscience increasingly confirms, is far more useful in understanding than any capturing technology can ever be, however many social network sites it scours.

What I am certain about is that we need far more exploration in this arena. The nuanced ironies of ‘public’ communications online, the way responses, statements, comments and attitudes are expressed through language and in the more mannered and abbreviated SMS world need to be seriously and thoughtfully considered as we accumulate ever more research data from such sources. We may need to look to academia to develop theories.

If the upsides of the new forums are access and breadth, I think it is timely and sobering to remind ourselves in the research community that technology won’t do us out of a job. Let’s use more of our famous intellectual curiosity and historical adaptability to take on the challenge of researching the new language frontiers’ culture rather than risk drowning in the rising seas of data.

Rosie Campbell is a director of Campbell Keegan. She has been a qualitative research consultant for over 25 years, working for major FMCG companies, service industries and government departments. She focuses on communication and language and is currently developing this in coaching practice

1 Comment

13 years ago

Really interesting article Rosie. As someone who works for a company that prides itself on it's automated sentiment recognition engine, we face this problem every day. While our engine is 88% accurate with "normal" language, slang always becomes a grey area. The hardest part is that different slang means different things to different people depending on their demographics (age and location being the two main ones). While it's nice to see that our system distinguishes positive and negative sentiments, it's still always good to give the results a human once over. Our system allows for users to change the sentiment rating if they disagree with our engine (like if someone uses the Michael Jackson version of "bad" to mean "good"). We're constantly trying to learn and teach our system to do the same, but with slang changing and evolving everyday it's not the easiest of tasks. I wish I could say that there was a quick fix for this problem, but there is not. Until there is, I would always recommend that people use humans to check over their computer generated info. Thanks for the interesting read! Cheers, Sheldon, community manager for Sysomos

Like Report