Don’t ‘overdo’ it online, if you fear the fuzz A linguist explains the messy science that fuels Russia’s anti-extremism policing

The number of “extremist” crimes committed in Russia has been growing for the past several years. According to the “Sova” analytical center, police launched 563 extremism investigations in 2016 and 858 new cases in 2017. Most of the suspects in these investigations are ordinary Internet users who have expressed their political views on social media. Dozens of these people have appealed to the European Court of Human Rights, claiming that they are the victims of illegal persecution. Many of the sentences handed out for extremist crimes are based on linguistic analyses provided to the authorities, and these experts rarely contradict investigators’ charges. Meduza correspondent Irina Kravtsova spoke to Igor Ogorelkov, the head of the Moscow Research Center’s linguistics department, to find out what qualifies as online extremism. (It turns out that Russia’s experts are still working on that one.)

On April 19, Moscow’s Pushkin State Russian Language Institute hosted a workshop on conducting psycho-linguistic analysis for investigations into extremist crimes. The conference brought together professional linguists and representatives from research institutes to discuss the absence of an established method for carrying out these analyses. Staff from the Moscow Research Center also attended the workshop, and Meduza asked Igor Ogorelkov, the head of the organization’s linguistics department, to explain what experts do in extremism cases that attract public attention.

Why is the number of new extremism cases rising?

More people have started using the Internet. We used to analyze books, leaflets, and posters. When everyone moved online, we started looking at Internet content, and that’s where we stumbled into this uncharted territory.

Do your experts and the analysts at other centers in other regions share common criteria or algorithms that allow you to determine when something written online is extremist?

The Interior Ministry, the Federal Investigative Committee, the Justice Ministry, and various private expert groups all use their own methods. Unfortunately, they all differ, and to this day we don’t have a shared, interdepartmental approach to analyzing materials identified in extremism cases.

Also, with the development of the Internet, we got something new: polycode text — images with text. Right now, we’re spending most of our time studying this, because Internet users find it much easier to post ready-made images with text, and they do it a lot. Going forward, there will be even more of these images on the Internet. Today, we’re on the verge of a revolution in linguistics analysis. Developing a methodology for analyzing polycode content is a quite critical task today.

And how long will it take to create this methodology?

We’ll have a methodology once we’ve accumulated enough research experience. That will take at least another three years. Right now, experts are trying to develop an algorithm for polycode texts, but it doesn’t always work.

Can you give an example where an algorithm designed to analyze normal texts doesn’t work with a polycode text? We’re talking about memes and that kind of thing, right?

Ordinary texts are only analyzed for verbal components, but visual images can dramatically change the meaning of content.

They differ how exactly?

They’re very different. An image with text contains a combination of verbal and visual information. Two experts — a linguist and a psychologist — examine this content. The linguist analyzes the text to identify the meanings and values therein, while the psychologist determines if there is any conflict inherent in the content, and identifies its socio-psychological orientation.

Professionally analyzing polycode texts requires that the linguist and the psychologist work in tandem. All that being said, they don’t examine the personality of whomever authored the text.

Why develop these methodologies, if the authorities can just look to anti-extremism policing in practice?

We’re aware of the Criminal Code, and we know the various statutes against extremism and hate speech. We study the expertise from state and private institutions, and we’re designing our methodology with all this in mind.

Meduza recently published a story about Alexander Byvshev, a teacher in Oryol. He was convicted of extremism for his poetry. The investigators in this case first asked scholars at a local university to analyze his poems, and then the court curiously requested a second review from the Moscow Guild of Independent Experts. One report said the poems were extremist and the other didn’t. How do you explain that?

Experts specially trained in linguistics and psychology from both state enterprises and private organizations can carry out these examinations, but a private expert even with the highest scholarly qualifications might lack the necessary experience with psycho-linguistic analysis. Experts can only draw reliable conclusions if they practice a certified methodology and have experience and specialized knowledge.

Where’s the line between free speech and extremism?

The courts decide when a statement is extremist, while experts offer linguistic or psychological evaluations of whatever’s in the case materials. Criminal Code Article 282 says inciting hatred is illegal, but people have caught on and rarely anybody is so open about it now. Free speech, meanwhile, is the inalienable right of any Russian citizen, enshrined in the Constitution.

What if someone wants to write on their social media page that their country is at war and they support the people in the enemy country and don’t share the views voiced by their own government? This person would simply be sharing their opinions, without calling on anyone to fight. Is that extremism?

You’ve got to examine each case individually. There are no clearly defined marks or algorithms. You can’t just take a picture or some text and say, this one’s okay, but this one isn’t, because there are a lot of different factors.

What if someone writes online that they disagree with one of the government’s positions, calling on others to support them?

There are many factors to consider in these analyses. For example, where did you post this information? In a public space? In the news media? If you don’t want to attract attention [from the police], don’t overdo it when you write — especially on social media. You should be thinking not only about what you want to say, but also about how others might interpret it. It’s one thing to tell your mom something while sitting around the kitchen table, but it’s something else when you share it online.

So where’s the line people shouldn’t cross?

People need to take responsibility for what they do online and why they do it. I can’t define what’s acceptable and what isn’t.

Do you ever read in the news about someone convicted of hate speech and then shrug to yourself, thinking: how was that a crime?

I don’t have access to the specific evidence used against people convicted of hate speech. The federal list of extremist materials includes just their titles, not their content.

Okay, but you read the news and you see the context. For example, a woman in Krasnoyarsk was recently sentenced to two years’ probation for posting an image that she shared in a private photo album on Vkontakte, where she was the only one who could see it. It was almost exactly like a kitchen-table chat with mom: it was private, and there was no incitement to violence.

In my experience, I’ve never encountered any rulings that I felt were unfair.

Does that mean on principle that I can’t use my social media accounts as platforms to express my own opinions?

You can use social media to express your opinions, but it’s best not to go beyond the bounds of the law.

What if I hide all my Vkontakte posts from everyone but myself?

Can you be sure that no one else would see your posts? You know, it’s like getting a debit card from Sberbank and thinking you’re the only one who can use it.

You don’t think police agencies abuse Russia’s law against hate speech, and that the law itself is unjust because it allows the authorities to persecute people for self-expression?

I’m an expert and I analyze the evidence presented to me. Experts don’t offer legal input in these cases, and our political views here are irrelevant to what the courts decide.

But I’m not asking about your political views — I’m asking you to extrapolate your personal experience in the context of what’s been happening in Russia.

I have to be objective when analyzing the materials presented to me by law enforcement agencies.

Does it matter who publishes posts that incite people to violence or hatred? Like if it’s a person with liberal views or pro-government views?

When we analyze case evidence, we focus on the author’s communicative goals and the socio-psychological orientation of the materials. We draw our conclusions using established methodologies.

When TV pundit Dmitry Kiselyov talked about turning the United States into “radioactive ash,” was that an act of extremism?

As experts, we don’t determine whether a particular text or statement qualifies as “extremism.” Those decisions are made by the courts. Regarding the specific case you mentioned, there’s no incitement to burn anything to ash, but really you can’t analyze statements out of context, or else you risk making a mistake.

If the line is so fine, do you think there should be such heavy criminal penalties here? Basically, you’re sitting there and you can determine whether the incitement to something is direct or indirect, and the suspect is charged with criminal activity and could go to prison.

There needs to be shared, interdepartmental methodologies that state clearly which polycode texts contain the linguistic and psychological indicators of incitements to violence and hatred.

But you said yourself that this shared methodology doesn’t exist.

It doesn’t exist for polycode texts.

Then you, as an expert, admit that you can’t determine objectively in all cases what is legal or illegal.

Well, that’s true with any expert analysis. You talk as if linguistics were the only subjective field and everything else were empirically precise. Handwriting analysis also isn’t 100-percent accurate — it depends on the quality of the evidence. The same is true with us. Not everything is ideal.

Have you ever encountered any pressure from law enforcement agencies? Maybe they took an interest in the results of one of your studies?

Absolutely not. How do you see this happening?

Sometimes experts collude with investigators and...

Absolutely not! Every expert is a professional, and we value our reputations. What kind of expert are you, if you succumb to influence like that? We examine any case materials, and we can’t offer any assessment but our objective analysis because we value our names and the reputations of our research institutions.

How often do you determine that these case materials contain evidence of extremism?

In roughly half of all cases.