Home Affairs Committee

Subject: The Use of Sprakab Language Testing in Country of Origin Analysis, Particuarly for the Somali Bajuni

Tiffy Allen is the National Coordinator of the City of Sanctuary movement—a network of groups committed to helping make cities and towns more welcoming and inclusive, a haven for those seeking safety. There are city of sanctuary groups in over 30 cities and towns across UK, and the movement is growing rapidly, with interest in several European countries and beyond.

Tiffy, a fluent Swahili speaker, lived and worked in Tanzania and Kenya for over 20 years. She has been involved in working with refugees in a variety of roles, including interpreting English/Swahili. Her background is in English teaching, and it was the needs of her asylum-seeker students that led her to found Welcome to Leeds, a college based drop-in where new arrivals could make friends, play games and chat in a relaxing and accepting environment. She later started RETAS Leeds, a charity focussing on the education, training and employment needs of refugees, before moving to Dublin for a few years. She completed an M Phil. at the Irish Schools of Ecumenics, where her dissertation, ‘Welcoming the Stranger’ researched Ireland’s response to new immigrants, especially during the Celtic Tiger years. In Dublin she worked on employment projects for immigrants, working at the Africa Centre as well as other small charities in Dublin.

Tiffy’s current role of helping City of Sanctuary groups to get started, form networks, and find a grassroots expression right across society draws on this background, and she believes the City of Sanctuary model has the credibility, appeal and sustainability to empower voluntary, faith and other groups.

Tiffy has met scores of Bajuni asylum seekers and has personally listened to recordings of over 20 Sprakab interviews. She has attended a CORI (Country of Origin research and information) conference and worked closely with a number of experts and lawyers including linguist Dr Derek Nurse, whose work is referred to in the report. Tiffy has written this report in consultation with her husband Brian, a retired Bajuni nationality expert who has interviewed over 400 Bajuni for nationality testing.


When asylum applicants claim to be from a country or ethnicity which the Home Office deems doubtful, they are frequently subjected to a telephone call lasting between 20 and 30 minutes—usually conducted by a Swedish company Sprakab—which claims to analyse language spoken over the telephone. This system, implemented by the previous government and perpetuated by the present one, is deeply flawed, unprofessional, flies in the face of expert opinion, and infringes the Human Rights of large numbers of vulnerable individuals.

This report will demonstrate that:

  • The practice of testing nationality on the basis of a phone call is regarded by linguistic and nationality experts as being inaccurate at best.
  • Many Home Office decisions made on the basis of such analyses are overturned at appeal, at great cost to the Home Office.
  • In the case of the Somali Bajuni, there is considerable evidence that the methodology, the analysts and the integrity of the SPRAKAB system as a whole is deeply flawed, and that it has caused confusion and misery to many genuine refugees, as well as excessive cost to the Home Office.

Factual Information for the Home Affairs Committee to Consider

Having listened to over 20 of these tests in detail and analysed the reports written, and having consulted extensively with Dr Derek Nurse who has analysed over 100 such reports, I would like to raise the following concerns about them and propose that this is not a just, accurate or professional way of testing asylum applicants claiming to come from Somalia.

1. The Analysts and Linguists

In listening to recordings of interviews and reading Sprakab reports, we have come across two interviewers who do the analyses. Sprakab claims that the analyses are counter checked by more senior linguists, but these people do not speak any African languages; moreover, as examples below will show, many decisions seem to be made on the spot.

The local languages they speak (in addition to Swahili), as mentioned on reports are Luyia (sometimes spelled Luja or Luhiya), Kikuyu, Luo and Acholi. Kikuyu, Luyia and Luo are all tribal languages spoken extensively in and around the areas of Nairobi, Nyeri, Eldoret, Kakamega and Kisumu of Kenya. Acholi is a language spoken mainly in Uganda but is very similar to Luo.

These languages are all spoken in areas of Kenya far from the coast where the Bajuni and related tribes live, Nairobi and beyond. They are not similar to any of the coastal dialects of Kenya, Tanzania or Somalia, nor is the Swahili spoken in the mentioned areas similar to Swahili spoken along the Kenyan coast, across Tanzania or in the Bajuni speaking areas of Somalia. The analysts in the Swahili they speak make occasional grammatical and lexical mistakes typical of speakers of the languages mentioned above.

On occasions Swahili mistakes made by analysts lead to a communication breakdown. On some of these occasions the analyst becomes insistent and overbearing as he repeats the wrong word and demands an answer. (Examples from recordings of incorrect words used in this way, mnala, (should be the word mnara meaning ‘tower’); thamana, (should be the word thamani meaning ‘value’). Mistakes made here are typical of the parts of Kenya where the analysts seem to come from.

It appears from the reports and from the recordings on the CDs that neither of these people are speakers of the Kibajuni dialect spoken by the Bajuni people of Kismayo and the islands of Southern Somalia. There are numerous instances on the recordings where the applicant says something in Kibajuni and the interviewer changes the subject or fails to respond. On the recordings the interviewers sometimes mention two or three Kibajuni words and ask what they mean. It would appear that they have been given these words as a means to conduct a test, but they do not pronounce them in a typical Bajuni way and the test is ineffective as the applicant often does not understand what is being required.

The reports do not indicate that the analysts are Bajuni or even speak Kibajuni, and this is further backed up by the fact that they conduct the entire interview in a Swahili, often making mistakes typical of upcountry Kenyan Swahili speakers. It is submitted therefore that these analysts are not in a position to determine whether the applicant is Bajuni or speaks the Kibajuni dialect. To ensure any level of accuracy, an analyst of the Bajuni ethnic group or someone fluent in the dialect should have been employed to conduct the analysis. It is submitted that comments such as, “He does not speak a variety of Swahili spoken by the Bajuni ethnic group”; “It sounds like the person is using Bajuni words but does not sound natural”; cannot be made with authority by a person who does not speak Kibajuni himself.

I further submit that many of the applicants given a Sprakab test have had asylum interviews conducted through Kibajuni language and using a Kibajuni interpreter. This simple fact seems to have been entirely ignored when the Sprakab analyst (who seems not to speak Kibajuni) claims that the person does not speak Kibajuni.

Dr Nurse and other experts have also pointed out that Swahili is the dominant language on the East African coast and that a Bajuni, when asked a question in Swahili, will feel compelled to answer in Swahili.

2. The Methodology

I submit that the methodology of analysing language through a telephone call lasting up to 20 minutes is flawed. Some of the applicants are given this test immediately upon applying for asylum. In several cases we have come across it was the applicant’s first time in their life to have a telephone conversation. The voices of the applicants on many of the recordings seem afraid and confused. Many of the recordings indicate that the conversation was disturbed or interrupted, often repeatedly. In one recording the interviewer gets involved in an argument about a key with someone in the same room while the interview is proceeding. In other recordings I have heard disturbances from traffic, telephones, typewriters and people talking and laughing, all while the interview was being conducted. Other recordings indicate that the phone line itself was of poor quality, and it is obvious that the applicant and interviewer sometimes have difficulty hearing each other. Dr Nurse and others have compained that sometimes the CD recording can hardly be heard.

In many of the recordings, the interviewer gives an assurance that the applicant’s speech would be carefully analysed by someone who is an expert on the Bajuni spoken in the place they come from. There is, however, no indication whatsoever on written Sprakab reports that we have seen that this assurance is honoured. At the end of one recording we hear someone telling the Home Office person that the preliminary report will be ready in ten minutes. The next thing heard is a conversation between that person and the interviewer in another language where the person says ‘Is she a Kikuyu?’ and the interviewer says, ‘She is from Mombasa’. I have heard two recordings where two people seem to discuss an immediate judgement on where they are going to say the applicant is from. It appears on these two occasions they forgot to turn off the recorder when they put the phone down.

3. ‘Degree of Certainty’

We submit that the reports we have seen have all claimed that the applicant is ‘with certainty not from Somalia’ and then claim that the applicant is ‘with certainty’ from Kenya or Tanzania. We submit that this degree of certainty is impossible in the context of a short telephone conversation or when the main criterion is language and pronunciation.

We refer to a document from an international meeting of linguistic experts, “Guidelines for the use of language analysis in relation to questions of national origin in refugee cases” (June 2004) quoted below:

LINGUISTS’ DEGREE OF CERTAINTY Linguists should have the right and responsibility to qualify the certainty of their assessments, even about the country of socialization. It should be noted that it is rarely possible to be 100% certain of conclusions based on linguistic evidence alone … This is because this kind of language analysis does not lend itself to quantitative statistics such as are often found in some others kinds of scientific evidence.1

4. Language Mixing

The Sprakab reports often pick up on the fact that an applicant mixes Swahili with Kibajuni, and takes this as an indication that the person is not of Bajuni origin. We draw attention to the findings of the same report as detailed below:

It is unreasonable in many situations to expect a person to speak only one language variety in an interview or other recording, for the following reasons:

(a) Sociolinguistic research shows that multilingualism is the norm in many societies throughout the world.

(b) In many multilingual societies, it is common for two or more language varieties to be used on a daily basis within a single family. In such families, it is also common for the speech of individuals in one language variety to show some influences from other varieties spoken in the family.

(c) Many bilingual or multilingual speakers use more than one language variety in a single interaction: this use of ‘code switching’ or ‘style shifting’ is very complex, and often subconscious.

(d) Further, there is variation in all language varieties, that is, more than one way of saying the same thing.

(e) It can often be hard for linguists to determine the difference between variation within a single language variety, and code-switching between related varieties. It is also important to note that while linguists distinguish these as separate varieties, their speakers often do not.

(f) Another factor which complicates this issue is that language varieties are always in the process of change, and one of the most influential sources of change is the vocabulary and pronunciation of related language varieties.

(g) A further complicating factor is that interviews may be done several years after an asylum seeker has left their home country, and their language variety/varieties may have undergone change in the interim.

(h) While linguists are devoting a great deal of research to language mixing, they have been unable to determine the extent to which an individual can consciously control the choice of language variety or of variables.

The Kibajuni dialect is a case in point for these observations, made by a panel of international linguist experts.2 Code-switching and language switching is typical among the Bajuni people, whose lives have been constantly disrupted since 1991, and many of whom spent several years in refugee camps such as Jomvu in Kenya. In particular, it is very difficult for Bajuni applicants who have had little or no formal education to analyse their own code switching, or to explain (as they are often asked) the theoretical differences between Swahili and Kibajuni. This difficulty is increased greatly by the circumstances mentioned above: a telephone conversation with interruptions or distractions, a poor line and an interviewer clearly speaking Swahili in an accent they are unfamiliar with.

An additional consideration, ignored in all the Sprakab reports we have seen, regarding language mixing as well as acquired accents, is that many Bajuni applicants have spent time in Kenyan refugee camps such as Jomvu. In many cases this is the first place where they spoke Swahili without the Kibajuni dialect and it is perfectly feasible that a Kenyan accent might be picked up during this time

5. Objective Comments on Sprakab Reporting

We also refer to the following documentation in relation to the companies which carry out language tests, including Sprakab, for consideration.: i) A document by Daly, Paul, The Age, 27 July 2002, How tapes sent to Sweden alter thousands of lives. http://www.theage.com.au/articles/2002/07/26/1027497412021.html states that:

“Since December, 1999, Eqvator and a smaller Swedish company, Sprakab, have analysed the language patterns of about 2500 asylum seekers for the Australian Government after being sent recorded interviews with the asylum seekers. The analyses have cost the Federal Government about $2 million, including $500,000 this year…But in Sweden, Eqvator’s critics say that in a number of cases, its analysis has been dramatically flawed. Sometimes, it has resulted in Sweden deporting asylum seekers to countries they were later proven not to have come from. In 1998, an internal Swedish Government evaluation, obtained by The Age, found that of 50 asylum seekers deported from Sweden, largely on the basis of language analysis, nine were sent to the wrong country. Even Eqvator’s managing director, Connie Lantz, admits there are problems. “Like all analyses... ours are not always 100% reliable,” she recently told Swedish television.

This criticism is further backed up by the following comments from FECL, an organisation which offers news, analyses and comments on European developments in the fields of liberties and human rights, public order and security, policing, justice, data protection, immigration and asylum. Particular attention is paid to EU Justice and Home Affairs cooperation and Schengen policies. FECL provides a forum for mutual information and critical debate among experts, activists, scholars and practitioners throughout Europe:

“Asylum and Immigration authorities in Sweden, Switzerland, the Netherlands and Germany have engaged in a new form of cooperation aimed at determining the country of origin of asylum seekers whose national origin is uncertain through language tests. Linguists and ethnologists reject the tests as unreliable and immigration lawyers say the tests are being used in an arbitrary way to turn down asylum applications.”

Another source of criticism comes from an article: “Should a refugee be judged by what he says or how he says it?”3

A group of Australian linguists published a report last spring that questioned the findings of Swedish companies like Eqvator and its rival Sprakab. The report found that the reports of these companies contradicted the applicants’ claims in 48 of 58 cases. But when those 48 applicants appealed, 35 of them were granted asylum. In some cases, the judges considering those appeals expressed concerns about the accuracy of the tests.

6. Country and Culture Analysis

There is no evidence to suggest that the Sprakab analysts have any qualifications or experience in this area so that they can carry any weight. Clearly it does not fall within the remit or the expertise of these analysts
to make unsubstantiated and entirely subjective claims regarding asylum claims based on spurious, ill-planned and often inappropriate questions that come in the guise of ‘culture and country testing’.
An examination of transcripts or a listening to recordings confirm that the analysts are unqualified to conduct country or culture tests. Questions asked are typically incohesive, random and beg single word answers. On many occasions questions are asked which assume a certain level of education. The information sought could easily be supplied by someone who has read a geography book about Somalia. In addition, questions are frequently repeated at various and unconnected points during the interview, suggesting that the interviewer just asks whatever question comes into his head. In one recording we heard, the client was asked four times whether she was married, and three times what age she is.

On the other hand, there is no indication of an attempt to draw out the applicants or look for subjective descriptions of things that they have seen and experienced within their place of origin or culture. A sample of questions heard on recordings illustrates this point:

  • Where are you from?
  • What religion are you?
  • If I draw a map of Somalia and place a big cross on it, which side of the cross is Koyama?
  • How many brothers do you have?
  • Are there any cars on Koyama?
  • Are you married?
  • Are there any tall buildings on your island?
  • What is the population of Somalia?
  • Name the neighbouring countries of Somalia?
  • Do you know any parliamentarians from Mombasa?
  • How many mosques are there on your island?
  • Are there any Christians in your place?
  • Who was the last president of Somalia?
  • Who is the president of Koyama?
  • Are there any mountains, cows, lions, buffaloes or cats on Chula?

In addition, the analysts often adopt a tone that is condescending, patronising and sometimes almost bullying towards the applicant. Examples follow:

  • A big person like you should stop saying you don’t know (on applicant saying he did not know which airport he was taken to).
  • You say they beat you because you’re a Bajuni, why (laughing) are Bajunis not human beings?
  • What do you mean you have no parents, how on earth did you get into the world in the first place?


In conclusion it is submitted that language analysis conducted by SPRAKAB on Bajuni applicants cannot be taken as reliable evidence. We would submit that these services and reports should no longer be used and if any language testing is to take place it should be conducted face to face, by established Bajuni speakers, and in the context of a proper country and culture test.

Moreover, significant numbers of these refusals, based on Sprakab analysis, have been overturned at court or further down the asylum process. The cost to the Home Office is huge, the cost to an asylum seeker who may have become destitute in the process, is immeasurable.

Further Information


John Brick


Allen, Brian The Bajuni People of Southern Somalia and the Asylum Process in The Researcher, Vol. 3, Issue 1, February 2008 http://www.unhcr.org/refworld/pdfid/4a545b330.pdf (Accessed 14 January 2010)

Australian House of Representatives (27 February 2006) Questions in Writing Asylum Seekers Question 1246
http://parlinfo.aph.gov.au/parlInfo/genpdf/chamber/hansardr/2006–02-27/0175/hansard_frag.pdf; fileType%3Dapplication%2Fpdf
(Accessed 14 January 2010)

The Age (27 July 2002) How tapes sent to Sweden alter thousands of lives http://www.theage.com.au/articles/2002/07/26/1027497412021.html (Accessed 14 January 2010)

Independent Race and Refugee News Network (21 July 2005) The use and abuse of language analysis in asylum cases http://www.irr.org.uk/2005/july/ak000011.html (Accessed 14 January 2010)

The Independent UK (16 June 2009) Languages test for suspect asylum-seekers http://www.independent.co.uk/news/uk/home-news/languages-test-for-suspectasylumseekers-1706172.html (Accessed 14 January 2010)

Eades, Diane, Testing the Claims of Asylum Seekers: The Role of Language Analysis, Language Assessment Quarterly, 6: 30–40, 2009

Eades, Diane, Applied Linguistics and Language Analysis in Asylum Seeker Cases, Applied Linguistics, 26/4: 503–526, 2005

Eades, Diane and Arends, Jacques, Using Language Analysis in the determination of national origin of aslum seekers: an introduction, Speech, Language and the Law 11 (2), 2004

Erard, Michael, Nov/Dec2003, Should a refugee be judged by what he says or how he says it?, Legal Affairs Magazine http://www.legalaffairs.org/issues/November-December-2003/story_erad_novdec03.msp (Accessed 14 January 2010)

Federal Magistrates Court of Australia Decisions WAIO v Minister for Immigration [2003] FMCA 114 (9 April 2003) http://www.austlii.edu.au/au/cases/cth/FMCA/2003/114.html (Accessed 14 January 2010)

Federal Court of Australia M17/2004 v Minister for Immigration and Multiculturaland Indigenous Affairs [2005] FCA 86 (16 February 2005) http://www.austlii.edu.au/au/cases/cth/federal_ct/2005/86.html (Accessed 14 January 2010)

Immigration and Refugee Board of Canada (14 November 2005) Somalia: Information on whether Kibajuni is commonly referred to as Bajuni; whether a Bajuni who speaks Kibajuni is considered to be speaking Kibajuni or Swahili; whether someone who speaks Kibajuni understand Swahili and vice-versa; whether an interpreter, translator or linguist would refer to Kibajuni as Swahili; information on the differences and similarities between Kibajuni and Swahili and where the two languages are spoken in the world (November 2005) SOM100785.E http://www.unhcr.org/refworld/docid/45f1480520.html (Accessed 14 January 2010)

Language and National Origin Group (June 2004 ) Guidelines for the Use of Language Analysis in Relation to Questions of National Origin in Refugee Cases http://www.upf.edu/enoticies/0809/_pdf/Guidelines.pdf (Accessed 14 January 2010)

Patrick, Peter L. (16–17 April 2009) Sociolinguistic issues in Language Analysis for Determination of Origins http://www.nomadit.co.uk/refuge/refuge2009/panels.php5?PanelID=562 (Accessed 14 January 2010)

Refugee Appeal No. 73545/02, New Zealand: Refugee Status Appeals Authority http://www.nzrefugeeappeals.govt.nz/PDFs/ref_20021011_73545.pdf (Accessed 14 January 2010)

Refugee Appeal No 73663, New Zealand: Refugee Status Appeals Authority http://www.nzrefugeeappeals.govt.nz/PDFs/ref_20040730_73663.pdf (Accessed 14 January 2010)

Refugee Review Tribunal of Australia Decisions N04/48762 [2004] RRTA 701 (1 November 2004) http://www.austlii.edu.au/au/cases/cth/rrt/2004/701.html (Accessed 14 January 2010)

Sunday Tribune (8 March 2009) Immigration officials under fire for phone ‘language tests’ http://www.tribune.ie/article/2009/mar/08/immigration-officials-under-fire-forphone-languag/ (Accessed 14 January 2010)

United Kingdom Home Office (undated) Language Analysis http://www.bia.homeoffice.gov.uk/sitecontent/documents/policyandlaw/asylumprocessguidance/miscellaneous/guidance/languageanalysis.pdf?view=Binary (Accessed 14 January 2010)

United Kingdom Home Office (21 July 2009) Country of Origin Information Report Somalia http://www.ecoi.net/file_upload/1226_1248264530_somalia-210709.pdf (Accessed 14 January 2010)

Tiffy Allen

April 2013

Prepared 11th October 2013