Language tests may screen asylum seekers in Britain

An article in the Guardian on October 22nd indicates that language tests are to be used to screen migrants who seek asylum in Britain. There has already been a pilot test. This was announced by Beverley Hughes, the immigration minister:

She said that a pilot scheme of tests had been successful in proving that one in five of those claiming to be fleeing from Somalia had in fact come from another country.

These tests have been used in Australia and are very much criticized by forensic linguists, partly because so many people have moved around and do not speak according to whatever rules the testers establish.

There’s a report by language experts on the Australian procedure here.

They say that many people are unaware of the limitations of their own understanding of language.

Language is very much more complex than is often realised, and many statements about language can only be made with appropriate hedging. Many points that an ordinary person considers to be ‘obvious facts’ turn out under linguistic analysis to be half truths or worse. Consider for example the idea that ‘a noun is a word for a person, place, or thing’, ‘the word ‘cat’ is made up of three sounds’, or ‘acoustic analysis can create a voiceprint which identifies a person in a way similar to a fingerprint does’.
Many people would consider these to be truisms but in fact each of them has very serious limitations in linguistic analysis.


Consequently, linguistic research shows that a person’s nationality, ethnicity and/or place of origin normally cannot be determined solely on the basis of a few words in his or her speech. However, according to the RRT cases we examined in which details of the Agency LingID are given, many determinations in these reports were made precisely on this basis. For example, on the basis of one applicant using some “typical” Pakistani words and Iranian words, it was determined that he lived some time in these countries (N20). Another applicant was deemed to come from Pakistan on the basis of his use of one Urdu word, one Iranian word, and two words (Afghanistan and dollar) spoken with an Urdu accent (N7), another because of one Urdu word and pronouncing some words with an Iranian accent and some with an Urdu accent (N2), and yet another because of one Urdu word, pronouncing several words with a Pakistani accent, and using two Pashtu words and two English words (N28).

(Information from the Forensic Linguistics mailing list: International Association of Forensic Linguistics.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.