FECL 53 (January/February 1998):
Language tests for determining the country of origin of aliens were first introduced in 1993 by the 'Language Section' (Språksektionen) of the Swedish Immigration Authority (SIV: Statens Invandrarverk). Until very recently, however, very little was known about the language tests, as a result of the remarkably low profile kept by the 'Language Section' which is now offering its services under the name Eqvator.
According to Thomas Öhrn from Eqvator, the unit has a permanent staff of about 20 employees. For translations, interpreter services and language tests, Eqvator has formed a pool of about 70-80 translators and so called "analysts" working on a free-lance basis. From 1993 to 1997, Eqvator operated an average 2,500 language tests per year (sometimes several tests concerning one single asylum application). Eqvator is also selling its services to foreign Immigration authorities in Denmark, Norway, Finland, the Netherlands, Switzerland and Germany.
The tests are based on at least 15 minutes of tape-recorded "free speech" of the test person. Thus, the test person never meets the "analyst". The test person is asked to speak freely of things such as his upbringing, childhood memories, eating habits, street names in his home town, the type of government in his country, and "people he particularly likes or dislikes".
Anders Sundquist is an immigration lawyer at Rådgivningsbyrån, a legal office advising asylum seekers, run jointly by a number of Swedish NGOs including Amnesty International and 'Save the Children'. Sundquist is an outspoken critic of Eqvator. He points to the fact that no particular professional, let alone academic, qualifications are required for becoming a language "analyst" and says that sometimes analysts are not even from the same country as the person subjected to the test. Moreover, the identity of the "analyst" is never revealed, allegedly on security grounds. This makes it impossible for asylum lawyers to check an analyst for possible lack of qualifications or bias (based for example on his membership in another ethnic or political group than the tested person). All this is of particular concern, Sundquist says, because immigration authorities tend to consider the test results as conclusive evidence. This is "sheer legal folly", says Sundqvist.
Critics also complain about Eqvator's extremely restrictive information policy which appears to conflict with Sweden's far-reaching freedom of information legislation. Under Swedish law, all public authorities are obliged as a rule to disclose all their records to anyone applying for disclosure. Exceptions to the rule can be made only on a number of detailed secrecy grounds and must be motivated in each case. Nonetheless, it proved very difficult to obtain any relevant information from either SIV or Eqvator regarding the language test practice and its actual effects on decisions concerning aliens' right to stay in Sweden.
Some staff members of Eqvator have tried to turn away curious journalists by describing Eqvator as a private business, detached from SIV, and therefore not bound by freedom of information rules. Pressed by your editor to answer whether Eqvator was a section under the public authority SIV or not, spokesman Thomas Öhrn somewhat evasively described Eqvator as an "independent unit under [SIV], with its own management board and budget". The unit's 1997 budget plan indicates that the unit is economically self-bearing, with a 0.6 per cent profit margin.
Eqvator uses its own test methods, developed since 1993. The unit was unable to provide any statistics showing the percentage of cases in which language tests are being carried out, or the proportion in numbers or percentage of asylum seekers' claims confirmed or invalidated by test results. According to Mr Öhrn, e.g. Iraqi nationality of test persons is most often confirmed by test results, while purported Liberian nationality is rather rarely confirmed.
Analysts are required to indicate the degree of certainty of their findings according to a 4-point scale ranging from "great certainty" to "no certainty whatsoever", but no figures are available either regarding the number of test results deemed "reliable" as against those considered less reliable.
Asked whether any regular external expert control and evaluation of the test methods and the individual analysts' competence was taking place, Öhrn first said there were "no external quality checks". In a later written reply he mentioned contacts with "colleagues at home and abroad", as well as with "customers", with a view to improving test methods. Evaluations of test results by external experts are only ordered, when this appears to "be called for". Öhrn stressed it was difficult to find trained language experts. "There are few academics who are acquainted with, for example, Liberian English". What is really important, Öhrn held, was that the analyst speaks the same native language as the test person.
Mr Öhrn was keen to point out that the language tests are "no verdicts" but merely a means among many others, to help asylum seekers lacking identity documents to confirm their statements.
A number of academic experts forcefully question the reliability of Eqvator's language tests. Last year, a family who claimed they had come from Afghanistan were denied asylum in Norway, on the grounds among others that Eqvator tests indicated that they were not Afghanis. The case ended up in court, where the family's lawyer presented two expert opinions regarding the language tests in question. In her opinion, Ruth Schmidt, a linguist and researcher at the Department of Eastern European and Oriental Studies at the University of Oslo, noted that "the transcription [from the recording] of the words in the text analysis is inconsistent, unscientific and unreliable", and that the tests in question were "too defective to permit an evaluation of [the family members'] linguistic affiliations". "The report of [Eqvator], appears to be a series of claims for which the linguistic evidence has not been provided in the report itself".
Another leading Norwegian linguist, Professor Even Hovhaugen noted a "lack of professionalism" that could be documented at all levels of Eqvator's testing activities, from "the very design of the tests to the "expert's" evaluation".
Commenting on the Court case, Ruth Schmidt told your editor that scientifically acceptable language tests must be carried out by trained linguists, specialising in the language concerned. Ms Schmidt emphasised that it is extremely difficult to determine the origin of a person from a border region and that the use of standardised test formats is impossible, particularly for countries with many regional languages and no universal education system. She also stressed that language habits in unstable regions with constant trans-border movements of refugees change very quickly, making it difficult even for trained linguists to keep up to date with language developments. "I might be able to carry out a test that would tell you in 1997, with a certain degree of reliability, whether a person has just come from Afghanistan and has not stayed in Pakistan for more than a month or two, or whether he was settled in Pakistan for a long time".
Two Swedish experts, Kenneth Hyltenstam, a professor at Stockholm University specialising in bilingualism, and Tore Janson, a professor of African languages at the University of Gothenburg, are just as unequivocal in their condemnation of Eqvator's methods. In a letter of 5 January to the Head of the Swedish Aliens Appeals Board (Utlänningsnämnden) the two distinguished scholars wrote: "We are of the opinion that these "analyses" [concerning an African family] are of no value whatsoever, among other things because of the complex language situation in the language areas in question". The letter goes on to note that "it is obvious" that the persons who carried out the specific "analyses" in question "do not have sufficient qualifications for carrying out a reliable linguistic analysis". Hyltenstam and Janson find it "shocking" that similar "analyses" have been used in a large number of cases. "It seems that [SIV] and the Aliens Appeal Board are basing a large number of decisions on completely unreliable investigations". In view of the above, the two professors "strongly advise" the Swedish immigration authorities to stop carrying out such "analyses".
Expert opinions as the above have not, however, not prevented Eqvator from selling its know-how to a growing number of foreign customers. In the meantime, Immigration authorities in Switzerland, the Netherlands and Germany have begun to set up similar units, in close cooperation with the Swedish pioneers.
"Switzerland hires on ethnologists to hunt 'bogus refugees'", read a head-line in the Swiss daily newspaper, Le Nouveau Quotidien", last October. Indeed, the Swiss Federal Office for Refugees (BFF: Bundesamt für Flüchtlinge) has set up its own unit for determining the country of origin of suspected "false" asylum seekers. From June to November 1997, the Lingua unit had already carried out 172 tests. According to BFF, 86 per cent of the tests enabled the determination of the country of origin "with certainty", and a further 8 per cent with "high probability".
The Swiss test method appears to be somewhat more sophisticated and lavish than its Swedish forerunner. As opposed to Eqvator, BFF hires on academic experts from various disciplines, including ethnologists, to carry out the tests on the spot. The expert communicates by internal phone with the test person, sitting in another room of the same building.
According to the head of Lingua, Florentin Lutz, some of the experts are renowned scholars, sometimes from abroad. "They are prepared to be flown in, not least because of our generous fees", says Florentin Lutz. Experts are often hurried in to interview asylum seekers held in the transit zone of one of Switzerland's international airports. In these cases, test results must be ready within the 48 hours a person may be turned away upon arrival.
Experts have developed various methods to get the test persons to say the truth about their origin. Among others, according to Mr Lutz, they deliberately "emotionalise" the test person concerned as a way to make them say the truth. Asylum seekers who claimed they came from Kosovo were asked to answer questions such as: "What does it smell like in the market in Pristina?".
Meanwhile, the Swiss Society of Ethnologists (Schweizerische Ethnologische Gesellschaft) has condemned Lingua's methods as a violation of personal integrity and as conflicting with "the ethics and the practice of ethnology".
Following the Swedish and Swiss examples, the German Federal Interior Ministry announced a similar project for Germany last October. DM 2.4 million have been allocated to carrying out 4,000 "scientific analyses" at a cost of DM 600 each. According to a Ministry spokesman, during a trial period, mainly "African asylum seekers" are to be subjected to the test.
The Ministry's plans have already drawn harsh comments from Günter Burkhardt, the secretary of the German refugee aid organisation, Pro Asyl. According to Burkhardt, trying to determine with certainty a person's origin through language analysis amounts to "reading in the coffee-grounds".
Sources: Our inquiries; Expert opinions by Ruth Schmidt and Even Hovhaugen, Oslo, June 97; Letter by Kenneth Hyltenstam and Tore Janson to the Head of the Swedish Aliens Appeals Board, 5.1.98; information provided by Beat Leuthardt, journalist, Basel, 6.3.98.