LOL: Research into texting runs out of credit, FML

15th June 2014 at 15:38

I'm always interested in research that generates findings that appear to be counter-intuitive. It's good to have one's sacred cows filleted from time to time, to remind oneself how fragile the bedrock of belief often is. So I was keen to poke my gloved hand into the entrails of a report this week that claimed children's use of texting could actually support and assist their spelling and grammar skills. Great Krypton.


Ths s mdnss


At first glance, this is madness, of course. If you've ever had the pleasure of reading a text message from a genuine, organically-farmed yoot (and not merely a careless adult), then you plunge into a terrifying waterfall barrel ride of phonetically-generated, abbreviated, truncated and hyper-compressed meaning. Words replaced by letters (C for see, U for you) and vowels dropped as often as pssbl (LTR for later). Whole common phrases blended and reduced to symbol chains (LMAO, ROFL, LOL). Add slang, idiom and neologisms to this concentrate and you have something that approaches the language of Babel. The evolution of these forms is fascinating, and has much to do with the platforms upon which they emerged. Texting with one finger was laborious, like asking a court stenographer to take long hand. Moreover, many young text users relied on top-up payments to buy phone credit, and in order to maximise text dollars, every character was rationed. Like the haikus of Twitter, a new form had been invented. 


The media about this was pretty excited: 


Texting 'can boost children's spelling and grammar' – BBC News Online


'Texting improves children's spelling and grammar' – The Telegraph


'Txting mks u spl btter- no really lol!' – The Mail Online



And so on. If you read these articles with anything less than ten-foot-long tongs, you'd be in no doubt that kids using text speak was not only a harmless part of their conversational repertoire, but potentially even an asset. FML.


But the reality is a little less conclusive. Taking a look at the research that informs these headlines, we discover that there are some obstacles with accepting the quotes on the poster. For a start, the research, conducted by the University of Tasmania in conjunction with Coventry University, involved a sample size of just over 160 pupils from the age of 8 to 16. This is important, of course, if we are to consider the results to have any meaningful sense of scale, and 160 children starts to feel like a snapshot rather than an aerial photograph. So before we've even begun the headlines look shaky.

"The researchers compared spelling and grammar in formal tests and in text messages, at the start of the project and again after a year. The results showed the most creative texters were among the best spellers. The children were asked to copy out all their text messages over a two-day period. They were also asked to do a range of spelling, grammar and cognitive tests. The process was repeated after 12 months." (BBC News)


And the exciting, counter-intuitive results?


"They found that for the primary age children in the sample, use of ungrammatical word forms and unconventional spelling in texts was linked to better spelling ability 12 months later. For secondary students, the use of word reduction when texting, was also associated with better spelling. For primary children, unorthodox punctuation and capital letters were linked with worse performance in the second set of tests but the reverse was true of secondary age pupils." (BBC News)



These results are, to be blunt, all over the place. Some kids showed a correlation to better spelling and some didn't? This is something that you'd expect in a sample as small as this. You'd be as well asking the inhabitants of a boxing ring who they fancied for the winner's belt, and affecting surprise at the answers. The authors of the research acknowledge this limitation on the last page of their paper:


"The first important limitation is that the sample sizes obtained within the individual age groups are quite modest, and caution therefore needs to be exercised when arguing that these data appear to suggest no
consistent link between poor attention to grammar when texting and the development of grammatical understanding. It is possible that the study design lacks sufficient statistical power to detect what could be some quite subtle effects."


It certainly is psbl. They even admit that the two text-speak samples, taken 12 months apart, might not be easily comparable:


"Another important limitation relates to the way in which ‘stability’ was explored in this paper. That is, a great deal can impact an individual user’s texting behaviour between two time points so far apart."


This is no small issue. People in a hurry text far less grammatically than those in, say, a science exam. Emotional states can influence your attention to precision; different relationships can engender different registers and detail. It's responsible of the authors to have conceded both these points. But they do rather overshadow the clarity of data interpretation. You can find the paper online here. It relies, in my opinion, on a strategy frequently employed in social science: research that supports the paper is quoted, and research that doesn't is not. 


"There is now evidence that textism use does not appear to harm children’s literacy (eg, Bushnell, Kemp and Martin, 2011; Coe and Oakhill, 2011; Plester, Wood and Bell, 2008) and may even support spelling development."

If you take a look at, for example, the first one (get it here), we see the same troubling sample sizes, and unusual indicators of 'success'. It is easy to read a paper and be impressed with the evidence that appears to predate and support it, but those foundations need to be sturdy themselves first of all. 


A number of other problems present themselves. We probably shouldn't be surprised that children with better grammar and general literacy will make better, more subtle texters. I know a number of good writers who tweet and text like Jedi Knights, as we would expect. And perhaps we should expect to see high text use correlate with good grammar and spelling in pre-teens (on the grounds that if they're writing anything frequently at that age, they probably accrue a benefit compared to their peers). 


It's your time you're wasting


Finally, and the point that I think most people would instinctively cleave towards, is the idea of opportunity cost. In other words, if a student is spending most of their time using text language, which by definition is simpler, almost free from grammatical and spelling conventions, and stripped down to the very basics of communication, then that is time spent not familiarising themselves with more conventional forms. I write a hell of a lot, and I have found – and I think I'm on pretty solid ground here – that the only way I improve as a writer is by writing a lot, and reading a lot. Pupils are unlikely to develop skills of discernment and subtlety, wisdom and nuance, if they invest their effort in forms that require none of these. 


I have taught many, many students who read and write a lot – this is the most active generation ever for those activities, I am told – but who struggle to form complete sentences, paragraphs or construct an essay because most of their word consumption/generation is in a form so simple that Orwell's proles would have thought it a bit childish. Text speak is simple and efficient, but cannot convey the complex and layered notes of the human symphony. It's like playing Mozart on a Stylophone, or Beethoven on a penny-whistle. You can approximate, but all subtlety is lost. 


This is a story with several villains, some of them very well intentioned: overambitious researchers, reaching for a conclusion; a press hungry for a headline. But at the end of this process are children, some of whom will be taught by people who read stories like this and wonder if, maybe, their next lesson should be in text, or that perhaps they shouldn't insist on standard English in written responses. And that would be very sad indeed. 


Research like this is often 2GD2BT. GIGO.