Spotting a fake

Fake news is big news these days. Mostly, it's in the context of Donald Trump complaining about it, or alternatively being supported by it. There are millions of fake twitter accounts all tweeting away in support of something or other, or to try and scam something. Last week, Marc Owen Jones described his lengthy relationship with someone who was probably not who he claimed to be:

[Thread] As promised, today I want to tell you of how I became friends with a Twitter troll called Angus Gallagher. Angus recently had a sex/ethnicity reassignment operation. He is now called Jasmine, but we'll come to that a bit later. First though, say hi to Angus #StopTheCoup pic.twitter.com/z9cjTZxkxo

— Marc Owen Jones (@marcowenjones) August 31, 2019

I tried a quiz recently to see if I could tell which product reviews were real and which were fake; I did terribly, no better than chance. I had read that you might see too many technical details in fake ones, superlatives, or lack of personal detail, and still I couldn't pick the fake from the real. It's actually put me off online shopping a bit.

This is something that you'd think linguists ought to be good at. It's something that linguists sometimes end up working in, at least: forensic linguistics is the analysis of texts to detect or solve crimes, and it's all about looking for (in)consistencies, patterns and giveaways in the text to tell you who did or didn't write it. Therefore, linguists have turned their hand to various genres of fake news, reviews and tweets to determine how we can spot them. (I should note, you do need a postgraduate qualification to work in these areas, and/or experience - you don't just magically learn how to do it by being a linguist in general.)

The best I've done is with these emails that have been arriving lately, the kind that look like they come from someone you know. In my case, it's nearly always either David Adger or my Head of School, because I think they pick on someone who sounds like they're senior to you, and then the email says something like 'Are you free? I need a favour urgently'. But they're on to a loser here, because at the very least, I can recognise the writing style of people I know. I also know that these highly educated people would not make the kinds of mistakes that these emails contain, even if they were typing in the midst of great angst and favour-needing. Their emails might be short, lacking capitalisation, or be sentence fragments, but they wouldn't have odd exclamation marks, strange spacing or ungrammatical wording.

Subgenres of autogenerated text?

(NquiteSFW)

I was checking my spam folder earlier today, and it seems that on the 2nd of November I got a bit of an email flood, all somewhat missing what is presumably their target market:

Seems to be some text scooping going on - note the incorrect spacing and capitalisation of my name. Obviously I didn't open any of these, but gmail shows you a little preview of the text, and it's very odd. Here's a close-up:

Looks automated, as it doesn't make any sense. 'Nodded as you with some clothes', 'Mean anything else that long and ricky', 'Family was still on but as connie'? Strangely compelling. I wondered whether the text might not have been autogenerated from some kind of 'romance' novel because of the names, and certain phrases. You'd be able to check, probably, whether such words and phrases occur disproportionately often in certain genres and in conjunction with each other, and given a large enough sample of emails you could answer that question. I won't, though. I'll just be happy the way I am.

Forensic linguistics for spam detection

We're all getting quite good at spotting email spam now. Our filters are pretty clever and get rid of all the obvious stuff anyway, so that leaves us with just the non-malicious stuff from companies we've bought things from, and the ones that the filter can't quite detect. Most of us don't fall even for these ones, but some people obviously still do, or they wouldn't still be doing the rounds. Almost all of these tell you something about your account, either the email account you're using or some other account that they purport to be from, and ask you to verify your details. By doing this, you're giving your details to some scammer who will then use them for nefarious purposes.

Now I think that this is the chance for linguistics to come into its own, and simultaneously force everyone to learn how to write properly. The plan is twofold: