Descriptions in Twitter profiles


When Twitter tells me I have new followers, I can see their name and self-description before I can see their location (if they've given any). So I play a little game of 'guess which country they're from' before I click through to see it.  I seem to be good at picking out the Americans (or at least North Americans--the Canada/US distinction is hard to make here--sorry Canadians), based on the style of the name and self-description. To be fair, I'm probably guided by the photos too. (Pick out the Americans at the airport is another fun and not-too-difficult game. There, you can see the red maple leaf patches on all the Canadians' bags, which save them from the lumping-together.) But I'm a linguist, so I like to think it's the language I'm sensitive to.

(A postscript on 27 Oct 2014: In the comments, Dorothy Bishop reminds me of a post she did three years ago that was in the back of my head when I started this, but I failed to find it in my preliminary search. I didn't want to take the chance of you missing that if you don't read the comments. So, if you like this post, you'll love this one.)

Because I probably should have been doing something else, I decided to try to test 'what marks an American (versus British) Twitter profile'. Here's my method:
  • I worked backwards from recent followers using the 'Who Unfollowed Me' (Pro) list of followers whom I don't follow back. I used this because it does the opposite of Twitter: it shows me the location, and I have to click through for the description.
  • For each follower who (a) was a person, not a company, and (b) unambiguously listed their location as being in the US or UK, I recorded:
    • country: I only included people whose locations were unambiguous, so no London-Islamabad-Hong Kong multiple locations and no indications of internationality in the descriptor, such as An American in London
    • gender: by name/photo/description (female, male, unknown/other)
    • Twitter handle: does it reflect their name? Three possible values:
      • Yes/name: the handle is some version of their name or their name + numbers, e.g.  @lynne_murphy, @LynneM34, @Lynney, etc.
      • Mixed: part name/part descriptor, e.g. @LynneLinguist, @LynneEdits, @LordLynne, @CrankyLynne
      • Not name: e.g. @poltroonish, @LinguistYay, @subjunctiverobot
    • Number of self-descriptors: this is the tricky one. Basically, I counted nouns or verbs that constituted separate descriptions of the person, so:
      • Writer, teacher, blogger, linguist, parent, feminist, Scrabble player:  counts as 7.
      • Loves cooking, dreaming, whittling, singing: counts as 4.
      • Teacher of ESL, EFL and Intercultural Communication counts as 1--the main identity is 'teacher'.
      • Dreams are sometimes songs: counts as 0, since if it's label(l)ing the person, it's very indirect. 
      • An empty description also counts as 0, but I had a separate sub-category in which I distinguished the content-ful and content-less zero scores.
    •  Caveats: I also kept track of who said things like "RTs are not endorsements" or "All views are my own", but there were some in each nationality and not enough of these to warrant further analysis.
  • I did this for two notebook pages for each nationality, which totals 64 from each country. More would have been better, but I can only justify a certain amount of procrastination per day.

A big caveat here is that I'm only sampling my own followers, which is to say it's probably people with an interest in language or US/UK issues, possibly a bit older and educated than average.  So this might not be generali{s/z}able to US/UK Twitter users generally. The other caveat, of course, is that I'm equating location with culture. For all I know, half of the people who listed their location as 'London' are exchange students who aren't mentioning that they're from Iceland. But I'm working with what I've got, and we can only hope that the rate of 'false positives' in one country's data is matched in the other country's data.

All examples in this post are made up, mimicking profiles I read. I don't want people to feel like I'm giving any individuals a hard time. Or to [orig. AmE] out them to their friends and family as someone who follows me.  If it so happens that I've made up a handle or a profile that actually exists, that's accidental.

So this is what I found:

Gender
Overall the sample had more women than men, which is to be expected because 62% of Twitter users are female (according to one study).  (For this sample it's 56% female. The table below has raw numbers.) The gender breakdown was similar across nationalities, so whatever we see here is more likely to be a national effect rather than a gender effect.


gender
      F     M    O

US     37     26     1
UK     34     25     5
Total        71     51     6


Handle = name?
I was interested in the name versus pseudonym issue because, as we've seen before, Americans introduce themselves by name earlier in conversation than Brits do. So, I wondered, are the British more careful about giving out names on Twitter? The caveat for this result is that I have no way of knowing whether the handles people use are their real names. If someone's name was presented as 'Gemma Thornton-Baker' and her handle is @gemmatbkr, then the fact that the handle matched the name meant that I counted this as a name, rather than a pseudonym.  If their name was presented as 'Hunky Cloud' or their handle was @rottenweather, then I took it for granted that their name wasn't really Rotten Weather.

But after all this preamble, I'm not even going to bother to put together an HTML table of results because the numbers were exactly the same for US and UK.  So, hypothesis that Brits would be less apt to use their name as a handle was not supported.

Self-descriptions
On to the meat of what I wanted to look at. Remember, I'm not testing word-count of the self-descriptions, but the number of separate descriptions given--a single description may be one or ten words long. So, this isn't about how much one says, but how many different things one says. My hypothesis was that Americans list more different things, divulging more about themselves.

The result favo(u)red the hypothesis, in that Americans listed, on average, 3.58 descriptors and the British 2.78.  The range was exactly the same: 0 to 14 descriptors. The US median was 3 and the UK median was 2.

But although the numbers were in the right direction for the hypothesis, they are only significant at p=.100, which means, basically, that there's a 10% likelihood that the difference is down to chance. We'd probably have a better answer if I'd looked at more than 64 people per country. Which is why I'm going to point this out to our students who are currently looking for research projects to do...
Read more

Henny Penny, Chicken Little, Chicken Licken

While writing the other day, I wondered whether it would be widely understood if I used Chicken Little as a metaphor for a certain kind of language peever. It felt right, but I also knew the name Henny Penny (of the main character in the story--see comments for variations), both from my American childhood and from my child's English childhood. Then I got an email informing me that my Survey Monkey subscription had been auto-renewed for the next month. Which is to say, I had failed to notice the note in my (BrE) diary/(AmE) planner on Tuesday that said "UNSUBSCRIBE FROM SURVEY MONKEY". At that point, I decided to get my money's worth from this unintended subscription, and so I devised something called the Famous Chicken Survey. Because I'd read another name, Chicken Licken, on Wikipedia, I threw that into the survey.

(Now I know, with a bit more research that Hen-Len is another name, found for instance in a UK-published version from 1849.  For that and more, see this site, which catalog[ue]s Aarne-Thompson-Uther type 20C folktales.)

156 people from the US (n=80) and UK (n=76) had answered by tonight (and small numbers from other places, not to be analy{s/z}ed here), and 146 of those answered the key question:

So there we have it, Chicken Licken (orange) is BrE, Chicken Little (blue) is AmE.  While there's some Chicken Little in the UK answers, that's only 9 people on that blue bar. They might have been affected by the Disney films by that name (1943 and 2005).

I thought that perhaps Henny Penny was old-fashioned, but it's found across the age groups. That one may just depend on which book you had in your house (or your preschool). I don't have a historical corpus for BrE at home (and I doubt I have a big enough one for this job), but the Corpus of Historical American English has 2 Henny Pennys between 1880 and 1909, and 26 Chicken Littles, so that's clearly not a very new name.

It would not be surprising to find that Chicken Little is a corruption of Chicken Licken, since all of the story's other names rhyme: Cocky-Locky, Goosey-Loosey, etc.  It also would not surprise me if the Little corruption and the alternative Henny Penny arose from a Victorian desire to avoid the association with licking. At least, that's what I'd want to avoid, since Chicken Licken sounds like a (BrE) dodgy (orig. AmE) fast-food joint to me. But that might be because it is a fast-food (orig. AmE) chain in South Africa, where I used to live. Not to mention that the Victorians wouldn't have heard of it.



The Wikipedia page for the South African Chicken Licken funnily enough refers to the Henny Penny Corporation (USA), which supplies equipment to chicken-frying businesses. I can see why these companies wouldn't want little in their names, but they're clearly not worried about associating their businesses with muddle-headed paranoia.

Read more

The book!

View by topic

Twitter

Abbr.

AmE = American English
BrE = British English
OED = Oxford English Dictionary (online)