Skip to main content

Language tests: The Norway model – a success story

Press Release   •   Nov 18, 2014 12:36 CET

Language tests used in the asylum procedure in Sweden are carried out by a contractor using inadequate methods. This was stated November 13th, 2014 on prime time in an hour long documentary broadcasted on Swedish Television.

The contractor has previously supplied such tests to Australia, Canada and the UK. Australia and Canada has discontinued language testing altogether, while the UK has changed supplier. The alleged malpractice risks dogmatic critics being rendered undue credibility. The facts of the situation, and what is at risk with a sentiment-driven debate, is outlined in the following.

The use of language in the asylum procedure

Many asylum claimants lack means of identification, such as passports. In Sweden, for instance, this is true for 9 out of 10 applicants. Unable to document their identity, paperless claimants are offered a second chance by certain, but not all, EU members, among those Sweden – to corroborate their origin by displaying native level use of the dialect that matches their claimed residential history.

The success of using language tests

The refugees seem to be aware of this and make their choice of refuge country accordingly. Sweden, who offers the claimant the added chance of a language test, receives 8 times as many claims per capita than the average within EU-27. In stark contrast to the refugee assessment are socio-linguists arguing that language tests are bad for the claimants. The arguments are not based in experiments or in observable data. A practice that would not be accepted if followed by an expert witness in court.

(Source: Eurostat downloaded November 12, 2014) (Source: UKSC 2013 0202)

A question that naturally presents itself is whether the asylum applicants are ill informed and make choices not in their best interest and that instead the conjectures of socio-linguists, such as professor Peter Patrick in the UK and professor emeritus Kenneth Hyltenstam in Sweden, are right in their advice. This question can be settled by simply inspecting facts.


The UK has a language test practice similar to that of Sweden, Germany uses language tests to a limited extent, while France has refrained from using this device altogether. It turns out that the asylum approval rates in first instance are considerably higher in the countries using language tests. In UK and in Sweden they are, respectively, 35,4% and 39,3%, to be compared to the EU-27 average of 28,2%.

Germany, with its conservative practice, weighs in just above average, at 29,2%.

France, abstaining, is lagging behind at a mere 14,4%.

It would seem the facts support the asylum claimants standpoint, while the socio-linguistic conjecture will have to be abandoned.

(Source: European Migration Network: Ad-Hoc Query on the use of language analyses in the removal process, 2011).

(Source: Eurostat downloaded November 12, 2014)

Closer look at a success story

If Germany can be said to have a conservative procedure, Norway has arguably a modern language testing method, totally overhauled a few years back. Interestingly, Norway can boast an outstanding first instance application rate of 48,8%. This is more than 60% more than the EU-27 average!

In Norway, several man years – experts on the asylum procedure and linguists joining forces - were dedicated to analyzing existing practices and to defining a method were everything that was needed for legal transparency and precision in testimony was included.

The novel method gives analyses that are making the best use of the native ear as well as the tools of linguistics and each analyses being safely anchored to the academic literature. It is worth noting that the UK Supreme Court has pointed out exactly these criteria in a recent decision: the native ear to be used for a narrowly defined native dialect and for all distinguishing traits that observations are intended to exemplify, to be supported by references to exact pages in the academic literature where the trait is as such described.


Is Norway then the dream of the asylum claimant, while the system is a burden to the country in general? This seems not to be the case, but rather a genuine win-win situation to have been established. In Norway 11,6% of first instance decisions are reverted at appeal. The EU-27 average is 19,1%. It is a good thing that there is a right to appeal, but it is also a good thing for claimant and receiving country alike that a correct decision is arrived at with precision and clarity at the first go.

Whether Sweden would get the same approval rates as Norway if the Norwegian method was adopted is a matter for speculation and difficult to address with available facts. But if so, this would translate to about 8 claimants a week having their applications erroneously rejected.

(Source: UKSC 2013 0202)

(Source: Eurostat

The way forward

Holding the above facts in regard should be a necessary, but also sufficient, requirement for entering the debate. It is important not to settle with what happens to be the best at the present time, but to strive to further improve also that. In doing so, huge costs in human suffering and administrative efforts can be avoided. In order to achieve this, contributions from all stakeholders are welcome, indeed needed, be that authorities or courts, human rights organizations or lawyers, linguists or statisticians. Only then can we be satisfied that we are making an honest attempt to shape the language tests of the asylum procedure into the tool best suited for the task.

For further information, please contact:
Per Lunqe, Press Officer
+46 735 106010

Verified provides expert testimony regarding linguistic behaviour, in particular relating to an individual´s linguistic background. These services are required by police, migration authorities and immigration attorneys for trials in court.  We also assist non governmental organisations.

Comments (0)

Add comment


By submitting the comment you agree that your personal data will be processed according to Mynewsdesk's <a href="">Privacy Policy</a>.