Europe is in the middle of its biggest refugee crisis since World War II. As hundreds of thousands of migrants come to seek a better life, European governments have to determine who to let in and who to keep out.
In order to be granted asylum as refugees, they have to convince border agencies they are fleeing serious danger or persecution. There is also a growing suspicion that the few documents refugees do provide are forged—and most don’t have any papers anyway. European countries have “safe” lists of countries (pdf) that they won’t accept refugees from, such as the Balkan states and many migrants trying to get in from those places are disguising themselves as Syrian in hopes of getting asylum.
To determine which asylum seekers are telling the truth about their country of origin, governments use what is called “language analysis.” The test is pretty simple: Do they speak like where they say they are from?
Some European countries, such as Germany, Switzerland, and Belgium use in-house government departments. But Sweden, the UK, Denmark, and other countries around the world use private companies to carry out this delicate work.
One Swedish company, Verified, has conducted over 24,000 analyses and insists their services provide robust “expert testimony” on an “individual’s linguistic background.” These experts include analysts who are native speakers of the language being analyzed and supervising linguists.
“The way we speak is shaped by our past experiences,” Roderick Martin, CEO of Verified, tells Quartz. “In cases where no documents to support the identity are produced, attributing a dialect to the claimant can greatly help assessing the veracity of the residential history given.”
A sample Verified report (pdf) shows how it works. An asylum seeker claims to be from Syria, in the Aleppo province. A native speaker and a linguist are brought in to test for this claim. The native speaker interviews the asylum seeker for around 20 minutes, asking a number of questions, while using their “over-all intuition” to identify particular dialectical traits.
The analyst then compiles a report that looks at different aspects of the language: phonology, morphology, syntax, and lexicology. The report will then arrive at a conclusion, in this case “the language analysis shows with certainty that the results obtained are clearly consistent with the linguistic community as stated in the hypothesis.”
It sounds good in theory. But Verified’s rival, Sprakab, also from Sweden, has been embroiled in a range of controversies.
Last year, one of its key language analysts was accused of being a convicted drug smuggler that lied about his qualifications. Sprakab was also criticized by the UK Supreme Court for providing “wholly inappropriate” advice to the British Home Office, which may have contributed to the wrongful deportations of hundreds of asylum seekers. A Swedish immigration tribunal had also cast doubt on the work by a Sprakab analyst.
Sprakab denies the allegations, insisting the analyst in question’s work is “flawless.” In a statement sent to Quartz, it argues “a drug conviction 21 years in the past, which was erased 17 years back, has no implications of the work ea20 [the analyst] does at Sprakab.” As for claims on quality control, Sprakab says that several professors have failed its tests to become an analyst.
The British government has since dropped Sprakab as their main supplier for language analysis, replacing it with Verified. A Home Office spokesperson tells Quartz: “Language testing is just one piece of evidence used when making the asylum decisions and does not replace the requirement for detailed questioning at a full asylum interview.”
The Home Office said its language analysis processes have been repeatedly supported in court judgments. It didn’t comment on Sprakab specifically, but its language-analysis instruction (pdf) shows that Sprakab will continue to be used in “a secondary capacity.”
The Swedish Migration Board continues to use Sprakab as its main language firm, with Verified as the second provider. “When we assess an asylum claim the burden of proof is on the applicant to basically prove who she or he is and where he or she comes from,” Carl Bexelius, a spokesperson for the Board, tells Quartz.
Bexelius says the Swedish Migration Board is aware of the UK Supreme Court decision regarding Sprakab, saying “language analysts are experts in evaluating language, not experts on the circumstances of a country.” He insists the language analysis is there to “help the applicant,” not hinder them, but says the Swedish government is aware of the limitations of language analysis and is “currently assessing how we use language services.”
Does it even work?
“The notional science behind it relates to an assumption, which does not hold up, that the language that people speak can be traced in its origin to a specific nation state,” Dr. John R. Campbell, a researcher from SOAS Centre for Migration and Diaspora Studies, tells Quartz. He argues that any asylum seeker “with a non-standard national language is going to have problems.”
He points to Eritrean asylum seekers as an example. As a result of war between Eritrea and Ethiopia between 1998 and 2000, Ethiopia deported around as many as 75,000 Eritreans across a war zone.
“A lot of the people they deported were Ethiopian nationals, but were seen as ethnic Eritreans,” Campbell says. ”These people’s first language through the Ethiopian education system was Amharic, which is not an official Eritrean language.” He says when a lot of these people began to appear in asylum systems in Europe and North America, they were deemed through these language tests to be Amharic speakers and their cases were refused.
Prof. Monika Schmid, from the Department of Language and Linguistics at the University of Essex, suggests there’s a “sell-by date” on matching someone’s language to their country of origin. For the reports to be reliable, she said, language analysis has to be done “within the first three years for an adult and the first few months for a child.”
She has looked at a number of different reports and from what she’s seen, she doesn’t believe many of these analysts have had linguistic training, saying “even my first-year PhD students would be able to do this better and more consistently.”
An international group of linguists released language analysis guidelines in 2004, which have been endorsed by several linguistic organizations, but many academics weren’t happy with it. In response to the criticism, the International Association of Forensic Phonetics and Acoustics (IAFPA) adopted a resolution in 2009, which stated language analysis should be done by “linguists and trained native speakers with the latter working under the guidance and supervision of the former.”
Verified said it has incorporated these new recommendations, as well as IAFPA’s emphasis on transparency. Martin also says Verified wants to “have an open dialogue with scholars practicing in relevant fields.”