The decision

Upper Tribunal
(Immigration and Asylum Chamber)

RB (Linguistic evidence – Sprakab) Somalia [2010] UKUT 329 (IAC)


Heard at Field House
Determination Promulgated
On 26 February 2010









For the Appellant: Mr G Davison instructed by Corbin & Hassan Solicitors
For the Respondent: Ms P Whipple QC instructed by the Treasury Solicitors

1. Linguistic analysis reports from Sprakab are entitled to considerable weight. That conclusion derives from the data available to Sprakab and the process it uses. They should not be treated as infallible but evidence opposing them will need to deal with the particular factors identified in the report.
2. Recordings of all material derived from the appellant and used as material for linguistic analysis should be made available to all parties if the analysis is to be relied on in the Tribunal.
3. Sprakab linguists and analysts are not to be required to give their names (as distinct from their identifiers, experience and qualifications) unless there is a good reason particular to the individual case.
1 - 2 Introduction
3 - 8 Procedural history
9 - 22 Linguistic reports in the present case
23 - 27 Anonymity of Sprakab personnel
28 - 82 The claimant’s claim in its factual aspects
83 - 98 Sprakab
99 - 116 The individual linguistic evidence in the present appeal
117 - 119 Submissions
120 - 123 Other evidence
124 - 142 Post hearing evidence
143 - 153 Findings and Conclusion: the appellant
154 -169 Findings and conclusion: Sprakab
170 - 174 General guidance on linguistic analysis evidence
175 Conclusion

1. It is a matter of common experience that it is sometimes possible to tell where people come from by the way they speak. In asylum cases, the question of where the claimant comes from is often of crucial importance. First, a claim falls to be assessed against the claimant’s nationality, so it is important to be able to establish what that nationality is. Secondly, even within one country the risks may be different for those from different parts, or with different tribal, clan or cultural backgrounds. The question is the extent to which analysis of the way in which the claimant speaks can be used in the executive and judicial determination of those issues, when they are relevant.
2. In what is now a considerable number of cases, the Home Office has commissioned analysis from Skandinavisk Språkanalys AB (Sprakab), a privately owned company in Stockholm, Sweden. It has relied on Sprakab’s analysis to a greater or lesser extent in reaching its conclusions on the claim, and has invited immigration judges to take account of the analysis if the claimant appeals to the Tribunal against the rejection of his claim. Such a case is that of the appellant. In this determination, to the writing of which we have all contributed, we give our assessment of the processes adopted by Sprakab, and the manner in which Sprakab analyses ought to be taken into account in asylum appeals.
Procedural History
3. The appellant arrived in the United Kingdom on 20 June 2007, and applied for asylum, stating that she was from Somalia. She was briefly interviewed on that date. During the course of that interview the interviewer referred her case to Sprakab, who conducted a telephone interview with her, and concluded that she was not Somali. The result of that analysis was communicated to the appellant. She was asked to return for a fuller interview on 5 July.
4. She did so, accompanied by a member of the firm of solicitors then representing her, and brought with her a written witness statement, dated the previous day. A full interview was conducted.
5. The respondent’s reasons for refusal letter, dated 20 July, adopts the findings of the Sprakab analysis, but offers a number of other reasons for the respondent’s disbelief of the appellant’s claim. Asylum having been refused, the respondent made a decision to remove the appellant as an illegal entrant. That decision carried a right of appeal, which the appellant exercised. Her appeal was heard by Immigration Judge Pugh on 31 August 2007. The appellant appeared in person: she was not represented. The Presenting Officer was prepared to cross-examine only on a limited range of topics. The Immigration Judge thought that the appellant needed a better opportunity to present her case, and so took her through the reasons given in the refusal letter. The Immigration Judge concluded that the appellant's claim was not credible: in particular, the appellant was not telling the truth about her origins. The Immigration Judge accordingly dismissed the appeal.
6. The appellant sought an order for reconsideration. The Tribunal made no order, but on renewal of the application to the High Court, Black J ordered reconsideration. She said this:
“It is plain that the Immigration Judge took great care over this case and dealt with it with considerable sensitivity. I am quite satisfied that there is no arguable error of law in the majority of her decision for the reasons that the Senior Immigration Judge gives.
The one issue that has given me concern is the question of language for the following reasons.
1 There was a linguistic analysis to be found in the Home Office bundle which indicates that the Appellant is Kenyan. The Immigration Judge accepts the analysis as very careful and detailed and relies on it, together with what she was entitled to conclude was the Appellant's lack of consistency and credibility in her evidence, to reach the conclusion that the Appellant is an educated Kenyan woman tried to pass herself off as a Somali (see paragraph 22 and paragraph 29 particularly). The Appellant complains in the grounds of application for reconsideration that the identity and expertise of the expert was not established and the identity of the expert was not revealed to her or to the court. It seems to me that the Appellant was entitled to know these matters and to consider whether to challenge the credentials and/or to call her own linguistic expert.
2 It appears to have been accepted that the Appellant gave evidence in Kibajuni. (Paragraph 7) which contradicts part of the Respondent's case (paragraph 15) which was (as the Immigration Judge recorded) that she "does not speak the variety of Swahili known as Kibajuni and found in the area from which she claims to come". I cannot be clear from the decision whether the suggestion of the Respondent was (1) that the Appellant's form of Kibajuni was that found in Kenya rather than that found in Somalia (which would not pose a difficulty for the Immigration Judge's findings on language) or (2) that the Appellant did not speak Kibajuni at all (which would be inconsistent with paragraph 7).
It seems to me that a material error of law may have been made in this area and given that the Immigration Judge took into account the question of language in reaching her overall conclusions, there must be a real possibility that the AIT would decide the matter differently on reconsideration.”
7. There were two interlocutory hearings. At the latter, SIJ Batiste declined to accept the argument put on behalf of the respondent that the Immigration Judge’s conclusions were sound even without regard to the linguistic evidence. He observed that the linguistic evidence appeared to have caused the Immigration Judge to take no account of the appellant’s having given her evidence through a Kibajuni interpreter; he concluded that no specific findings of fact could be salvaged from the Immigration Judge’s determination. He made a number of directions in relation to evidence, and adjourned the reconsideration for oral evidence to be taken and the appeal to be re-determined. Subsequently, although the respondent had expressed reservations about this course of action, the present appeal was selected as an appropriate one in which to look in more general terms at linguistic evidence and the Sprakab approach.
8. The matter came before us on 24, 25 and 26 February 2010. At the end of the hearing, the respondent sought leave to put in further evidence (which was awaiting translation), subject to an opportunity for the appellant to comment on it. We agreed to that. After the hearing, we became aware that the Immigration Law Practitioners Association (ILPA) had recently sent a circular to its members including statements about Sprakab analysis. We circulated the material to the parties, inviting any further comment. Both parties made submissions on that material.
Linguistic reports in the present case
9. It is convenient at this point to summarise the conflicting reports on the appellant’s language, which have been compiled since her screening interview at the Home Office.
10. The first ‘Linguistic Analysis Report’ on the appellant is dated 24 June 2007, and is based on a tape recording made on 20 June 2007. This report is cited in the ‘reasons for refusal’ letter of 20 July 2007, and is said to provide “strong evidence” that the appellant does not come from Somalia and is not a Bajuni. The analyst is ‘EXP 249’, who is said to originate from Kenya and to have completed the analysis “in co-operation with Sprakab linguist 03.” The conclusion is that the appellant knows quite a bit about the Bajuni islands but speaks a variety of Swahili spoken in Kenya. Although she does use some Kibajuni words, her pronunciation, intonation and grammar are typical of Kenyan Swahili, indeed with a level of grammatical rectitude which shows her to be highly educated.
11. A different conclusion was reached by Margaret Kumbuka, who had been instructed by those representing the appellant. She interviewed the appellant on 11 July 2008 and compiled a report on 20 September 2008. She concluded that the appellant “is a Somali Bajuni from Koyama”.
12. On 3 December 2008 Ms Kumbuka produced a report for the appeal hearing at Surbiton, based upon the CD and transcript of the Sprakab interview of 20 June 2007 with which she had been provided by Corbin & Hassan. She is very critical of this interview, saying of the interviewer that he “spoke broken Swahili with a very heavy Kikuyu accent”, which probably caused the appellant to adapt her own speech towards his. Thus she did not use much Kibajuni, as opposed to the way she spoke to Ms Kumbuka, who herself used Kibajuni at her interview with the appellant in July.
13. Almost simultaneously with Ms Kumbuka’s second report, a second Linguistic Analysis Report on the telephone interview of 20 June 2007 was completed by Sprakab on 4 December 2008. It is more detailed than the first, the analysis being conducted by analyst EA 19 and confirmed by analysts EA 14 and 249, with approval by linguist 04. The qualifications and expertise of all four are listed, and the conclusion is that the appellant “speaks a variety of Swahili with certainty not found in Somalia” but “with certainty found in Kenya, nearest in the country’s coastal regions, such as Lamu.” She is said to sound “as if she is trying to alter her speech in order to sound like she speaks Bajuni.”
14. The recording of the appellant’s half-hour interview on 20 June 2007 is the subject of yet a third Linguistic Analysis Report completed on 10 March 2009, and this time involving analysts EA 19, 14, 249 and 20, and linguists 4 and 1. All of their credentials are listed in great detail, and the report itself is more detailed than the two previous ones. But the conclusion is the same. The appellant has some knowledge of Koyama and Somalia, but speaks a variety of Swahili found “with certainty” on the coast of Kenya, e.g. Lamu.
15. While the three Sprakab reports so far were all based on the interview of 20 June 2007, the appellant was interviewed again over the telephone on 2 February 2010, and this resulted in a fourth Linguistic Analysis Report dated 15 February 2010. This time the interviewer is said to speak Somali Bajuni, while the analysts (EA 19 and 24) are said to hail from Kenya and the Bajuni island of Chuvay respectively. Their analysis is reviewed and approved by linguist 04. This report is particularly detailed. It identifies, with examples, nine phonological features of her speech corresponding neither with Somali Bajuni nor standard Swahili, four used inconsistently, a failure to stress polysyllabic words in the Bajuni way, and seven phonological features of her speech that are typical of the variety of Swahili spoken on the Kenyan coast. It notes inconsistent and incorrect use of Bajuni vocabulary and ignorance of the meaning of certain common Bajuni words. The Sprakab linguist and analysts’ assessment of this 20-minute interview is that the appellant tries to use linguistic features typical of Somali Bajuni, but often gets it wrong. She is in fact a speaker of coastal Kenyan Swahili, who has not mastered Somali Bajuni.
16. Finally we come to Sprakab’s response (prepared by analysts EA 19, 20 and 249 and linguists 01 and 04) to Ms Kumbuka’s two reports. She is criticised for not having visited any Bajuni-speaking area in Kenya or Somalia, and for lacking academic qualifications in linguistics. The “Bajuni refugees” whom she had met in the UK might not have been Bajunis at all, and no recording is available of the interview which led to Ms Kumbuka’s first report on 20 September 2008.
17. The Sprakab writers do admit that the appellant displays some knowledge of Bajuni life and culture, but note that the appellant makes use of words which are Kenyan Swahili as well as Kenyan Bajuni, and doubt whether the appellant would have picked up all those words while living in a refugee camp for Somali Bajuni in Mombassa. As a woman staying with family members, her opportunities for intercourse with outsiders would have been restricted.
18. The methodology of Ms Kumbuka’s interview is criticised as falling short of professional linguistic standards and failing to adhere to the Guidelines for the Use of Language Analysis, which are summarised below in paragraph 121.
19. As for Ms Kumbuka’s second report of 3 December 2008, she is taken to task for supposing that the interviewer on 20 June 2007 spoke “broken Swahili with a very heavy Kikuyu accent.” The interviewer spoke a variety of Kenyan Swahili close to standard Swahili, and cannot speak Kikuyu at all. This called in question Ms Kumbuka’s ability to carry out a proper language analysis. During the interview, the interviewer spoke Swahili, with only a little Bajuni, lest the appellant have difficulty understanding him. (If the appellant was not Bajuni, she might have been unable to understand questions framed in Bajuni.) The appellant herself used Kenyan Swahili and some Standard Swahili during the interview, but only a few words in Bajuni.
20. There is also what might be called a report on the reports. Pia Fernqvist, the manager of Sprakab, compiled a report for the hearing before us, in which she describes the expertise and qualifications of ‘analysts’ and ‘linguists’, and says of analysts EA 19, 20 and 249 that they have all worked or resided at some time in Bajuni-speaking areas. EA 19 is said to have “mastered Bajuni to a very high level”, while 249 speaks Bajuni “to the level of a mother tongue speaker.” Another analyst, EA 14, is said to be a native Bajuni speaker.
21. Ms Fernqvist goes on to describe the interview of 20 June 2007. The interviewer (240) is said to be a native Swahili speaker with mother-tongue competence in the Bajuni dialect, but no knowledge of Kikuyu. Hence there are no Kikuyu influences in his speech, contrary to what Ms Kumbuka thought. The reason why the interviewer spoke only standard Swahili to the appellant was that she would thereby be able to understand him whether she was Bajuni or not, whereas if the interviewer had spoken to her in Bajuni, she might not have been able to comprehend him if she was not Bajuni herself. Ms Fernqvist ends by saying that it is unusual to have a second, far less a third, report on an interview, as happened in this case. All the analysts and linguists who were involved in the first and second reports contributed to the third report.
22. It is not Sprakab’s policy to disclose the identity of its analysts (hence the use of numbers) because of threats which have been directed at analysts in the past by disgruntled claimants. Indeed, EA 14 has stopped working for Sprakab, because of such threats.
Anonymity of Sprakab personnel
23. That observation by Ms Fernqvist takes us directly to the first of the two grounds upon which Black J ordered reconsideration. The respondent proposed to call oral evidence from a number of those involved in preparing the Sprakab reports in this case, but asked on Sprakab’s behalf that the identity of the individual analysts and linguists be not disclosed. We therefore needed to deal with this issue at the outset of the hearing.
24. Sprakab’s policy is not to make the names or personal details of its analysts public. It feels that analysts’ independence may be compromised, and their personal safety may even be endangered if it is known that they are producing analyses, some of which may be adverse, for governmental authorities. Sprakab does, however, provide each member of its staff with a unique identifier, and the language background and training, and any other relevant experience, is associated with the identifier. Thus, it is possible to see the qualifications and background of the individuals involved in a particular analysis, and it is possible also to see whether the same or a different analyst has been involved in different pieces of work. The only thing that is lacking is the analyst’s personal details: name and address and so on.
25. Sprakab’s request for anonymity extended in the present case not merely to its employees as authors of the analyses, but in addition to some of those who provided written evidence or appeared before us as oral witnesses. During the hearing, we were provided with the names of all oral witnesses, which were not disclosed to the appellant or his representative; but the names themselves have no bearing at all on our decision. Although it is of course exceptional for witnesses to give evidence anonymously, it appeared to us that that was the appropriate course to adopt in this case. In view of the potential of threats to Sprakab personnel, we concluded further that it was proportionate to give a direction prohibiting the disclosure to the appellant of the names of those involved. (It is fair to say that Mr Davison raised no reasoned objection to this course of action in the circumstance of the present case.) We accordingly directed that we be given the names of those giving oral evidence before us, but ordered under paragraphs (1) and (2) of Rule 14 of the Tribunal Procedure (Upper Tribunal) Rules 2008 that the names should not be further disclosed.
26. If a witness chooses to remain anonymous then it may be difficult or impossible for his or her curriculum vitae to be checked and cross-examination on that issue may be handicapped. Below we endeavour to explain with some care Sprakab’s methods. We did not expect the evidence from Sprakab to be so undermined by the anonymity of the team members that the evidence would be valueless, and it was not.
27. Generally speaking, however, the position is simply that there is a Sprakab report, available in writing, authenticated not by a signature but by one or more unique identifiers of Sprakab employees. Given the information that is associated with the identifier, it seems to us to be virtually inconceivable that anybody is disadvantaged by not knowing the name or address of the individual concerned. It might perhaps be that in some particular case there would be a proper reason for enquiring whether a named individual had been involved in the analysis of a sample. If it was necessary to ask that question, it could be directed to Sprakab, and a Tribunal might in due course have to decide how to deal with whatever answer was given. But in the general case the reports are available on the authority of Sprakab itself, with full information about the qualifications of those who have contributed to them. That is sufficient.
The claimant’s claim in its factual aspects
28. As this case is essentially a case about evidence and its interpretation, it is appropriate to set out the evidence from both sides in some detail.
Screening Interview
29. The appellant applied for asylum on 20 June 2007. She attended a screening interview the same day. She indicated there that she wished to be interviewed in the Swahili language, although the record makes a tantalising reference to “see notes” which makes no sense to us.
30. The appellant said that she was born in Koyamo in Somalia. In 1993 she went to a camp in Mombassa in Kenya. In 1996 she married in that camp, but separated from her husband in 2004.
31. She was asked about her travel to the United Kingdom. Her answers were not entirely clear but she appeared to be saying that she left Somalia for the Yemen, arriving in Yemen in April 2007, and staying there for two months before travelling from Yemen to the United Kingdom. She claimed to have arrived in the United Kingdom by airline at “around dawn”. She could not identify the airline on which she travelled. She never had a national passport. She used a red passport provided by an agent to facilitate her travel to the United Kingdom. She passed through immigration control without difficulty. An officer inspected and appeared to be content with her passport. In answer to the question “Have you ever been to or sought leave to enter any other country?” the appellant indicated that she had been to Yemen for about two and a half months in 2007 on her way from Somalia. This does contrast with an earlier answer where she said she last saw her mother in Koyama in Somalia. She then said “I then went to a camp in Mombassa Kenya”. She said she came to the United Kingdom because she was from the Bajuni clan and feared the Darood and Hawiye clan.
32. After a break for the first language analysis interview, which concluded that she was a Kenyan national, it was put to her that she was not telling the truth. She insisted she was not a Kenyan national but was from Somalia and had not been telling lies. The appellant then explained that she learned English in the Jomvu camp in Kenya. The appellant said that “we were there from 1993–1998 and then we went back to Somalia”.
33. She was also asked supplementary questions about the colour of the aeroplane on which she travelled and details of the uniform of its staff and any motifs on that uniform. She gave answers to the questions which may be described as unilluminating but were not shown to be wrong.
Statement of 4 July 2007
34. The appellant set out her case in a statement dated 4 July 2007.
35. She repeated there that she was a Somali national and that she was born in Koyama Somalia. She had married another member of the Bajuni clan but they had now separated. She said she had had an Islamic education. She had also learnt to write some Swahili when she was in the Jomvu refugee camp in Kenya. She gave details of her family and her last address in Somalia.
36. She gave details of some of the traditions of the Bajuni minority clan and the likely employment of clan members. She said that people identified them as Bajuni because they spoke the Kibajuni language. She then identified the clan enemies. She explained how Somalis started attacking the Bajunis in 1991. She gave details of diverse and nasty acts carried out on Bajuni people; she included an allegation that she had been raped.
37. In 1993, she said, things became worse. She lost contact with her parents and on the advice of her uncle left Koyama and fled to the Jomvu refugee camp in Kenya. While she was there she was married, again, on the instigation of her uncle, and gave birth to a son. They remained in Jomvu until 1998. The Kenyan government led them to believe the situation was now peaceful in Koyama and closed the camp in Jomvu. She returned to Koyamo with her uncle, his wife, her husband and son.
38. The situation had not improved. In 2004 she was raped for a second time. She said that as well as the mental trauma of being raped she was left with lasting physical damage. The village was under constant attack although it was not always easy to say if it was being attacked by members of the Darood or the Hawiye clans. The situation continued to deteriorate and was made worse by Islamic insurgents imposing values that she found objectionable.
39. Her uncle then decided that they had to leave and they joined other people from Koyama on 1 April 2007 to travel to Yemen. She travelled under the supervision of her agent. Her uncle told her she must obey the agent to ensure the agent’s complete support.
40. She finished the statement by saying it had been read to her in the Kibajuni language.
41. The appellant attended an interview on 5 July 2007 which was conducted in the Kibajuni language.
42. She said that her last address in Somalia was Koyama Island, Koyamani Street, Somalia. She gave details about her clan membership and, when asked where she came from, said: “I am coming from the island called Koyama”. Her village was called Koyama Island and she had lived there until she moved to the Jomvu camp. She returned to Koyama Island. She named other islands that she said were nearby. She claimed never to have made the journey to the mainland but understood from those who had that it was a three to four hour journey
43. She described the island itself. She said that it is not possible to see the mainland from Koyama, but it was possible to see it from nearby islands. She answered questions about buildings on the island. She said there were two mosques there and about 70 or 80 houses. She guessed that between 700 and 800 Bajuni people still lived in her village. She said that she was unemployed, but that her father worked as a fisherman. She named people who she said were the local leaders. She described her home and life before the civil war.
44. She claimed not to be able to answer questions about the history of her clan:
“Unfortunately I cannot give you the history of my own clan simply because I lost my granddad who was the person I could get more history about our clan.”
45. She spoke about the village being attacked by Daroods in 1991. She gave details of how she was attacked and how she had left the Jomvu camp in 1993. She was asked about the camp and about her marriage and her son being born. She repeated her claim that in 1997 the Kenyan government had said that her home area was safe and how “we decided to move back home”. She confirmed that her husband left her in 2004 and that she left Yemen in 2007. She talked about the events that had led up to the decision to move.
46. She was asked about languages spoken in her village. The answer is not entirely clear from the manuscript record of the interview. It appears that she said:
“We used to speak Kibajuni and I say some words Kibajuni if when they are there say Kibajuni here say Bajuni language and some Arabic language we used to speak in my area). But we do not speak so much Arabic language we do not speak any of the other Somali language. We know some Somali words 2, 3, 4 something like that.”
47. The appellant said she could speak Somali. However it would be difficult for her to be interviewed in Somali because she only knew a few words. When the village was attacked it was attacked by people who spoke their own language. Sometimes people could speak Somali.
48. When, towards the end of the interview, it was put to her that language analysis results appeared to show that she was Kenyan, she insisted that she was not Kenyan. She attributed her accent to the time that she spent in Kenya in a refugee camp. She then went on to speak about the state of her health.
Statement of 12 September 2008
49. The appellant made a further statement dated 12 September 2008. It was clearly based on her statement of 4 July 2007 but included additional material.
50. She repeated her claim to be a Somali national born in August 1975 in Koyamo in Somalia. She said she was a Bajuni clan member who had married another Bajuni clan member and was now separated from her husband. She gave details of her family and of cultural practices which were distinctive to or at least typical amongst Bajunis. For example, she referred to the Kirumbizi dance and the Msondo dance. She said that Bajuni men tended to work as fishermen, house builders and shell gatherers. She said she had never been to mainland Somalia.
51. She repeated that her problems began in 1991 when women in her clan were attacked. She gave details of the horrors that she had experienced and of going to the Jomvu camp. There is no need to repeat the details here. The things that she described were very serious and, she said, led to lasting physical damage as well as her husband leaving her.
52. She said her interview was conducted in the Kibajuni language and that she had no difficulty understanding the Kibajuni interpreter throughout the 4 hour and 25 minutes in which she was interviewed. She maintained it was wrong to dismiss her claim to be Bajuni when she had demonstrated her proficiency in Kibajuni which is the mother tongue of the Bajunis.
Oral Evidence
53. The appellant gave evidence before us. She was assisted by an interpreter who was chosen because the appellant had indicated she wanted a person who could interpret the Kibajuni and the English languages. We made it clear to the interpreter in clear terms that it was his duty to speak to the appellant in the Kibajuni language and no other language. We are not in a position to determine whether he carried out that instruction. It was not, however, suggested to us that our understanding of the substantive content of what the appellant said had been affected by any difficulties with interpretation.
54. Mr Davison reminded the appellant that she had made statements and had been interviewed. She said there was a mistake in the interview record. According to the record (question 72) the appellant said that her uncle was killed in 1990. The appellant said that her uncle did not die in 1990. Her aunt died sometime in the 1990s. The appellant said that she told the interviewer that he aunt had died in the 1990s and the reference to her uncle dying in 1990 was the result of mistranslation. She confirmed that with these corrections her claim was contained in the documents summarised above.
55. Cross-examined by Ms Whipple, the appellant confirmed that she was 34 years old and that she was born in 1975 in Koyama an island off Somalia. She said both parents were Bajuni and she spoke Bajuni at the home as a child. She said Bajuni was her native language and she was speaking Bajuni now. She confirmed that her father was a fisherman and that she had two brothers and one sister and that her father sold fish at the market. She was one of the large family. She had many cousins who were Bajunis.
56. She said that problems began in 1991 when her aunt was raped and killed. It was put to her that in her interview she had said that her uncle had been beaten and killed as well. She said that was a mistake. She was shown the record of her answer to the question, “what happened when the Darood came to your village?” as follows:
“In 1991 they came to our village they started raping women they raped me too, they beat us. They took everything from our house. They beat the brother of my dad until he died, they raped my aunt they killed my dad’s brother, they killed my uncle in 1990 and raped my aunt and killed her in 1990. They made sure that if they raped a woman she would become (?) completely handicapped could (?) all they were doing was to destroy us completely so that we can’t ever move away from our own area.”
57. She said that was a mistranslation or transcription. She said the cassette, to which she had listened, was very clear. They did not beat the brother of her dad until he died. They beat her aunt. Her aunt was killed not her uncle. Ms Whipple pointed out that the appellant had not said earlier that there was a mistake in the record. The appellant said that she had made the point to the previous judge and then said that she realised the error when she listened to the recording of the interview.
58. The appellant confirmed that she left the island in 1993 and went to the Jomvu camp in Kenya with her family, that is with her uncle and her aunt. It was, she said, a different aunt. She was not confusing her with the one who had died. They went with a lot of other people from Koyama Island. They travelled together. She was not sure how many went. It was a boatful. When pressed she said it could be about twenty people.
59. While in the camp for four years she stayed with Bajuni people, and as well as their own language they were speaking Swahili. Pressed, she insisted that they did stay with people in the camp. They did register with camp officials. She said it was her uncle who registered her and her aunt. She did not register herself.
60. In 1996 her uncle arranged the marriage to a Bajuni man. She said they did not make much of the wedding. There was a small group of people, herself, her uncle, her aunt, her husband and his friends and siblings. Her husband’s parents did not come. She did not know their whereabouts.
61. The appellant confirmed that she left the Jomvu camp in 1998 and went back to the islands. She went with her uncle and aunt and her husband and child. The camp was closed. Other Bajuni friends and relatives went back although very few people came back to the island. When they got to the island they found that their house was destroyed.
62. An uncle had remained there throughout. He supported them when they returned to the island. Her uncle and her husband went fishing. They did not own their own boat. She did not know exactly how the arrangements worked but her uncle and husband worked on fishing boats owned by other Bajunis. They also went to the coast and worked for traders. Five people lived in her uncle’s house. She said they were poor.
63. The appellant said the raids continued and she was raped again in 2004. Her husband had gone and took the child with him. She claimed not to know if he had left the islands but she had not seen him. She tried to find him. She wanted her child back. The family supported her. She said that she was still with her uncle in 2007. Her uncle’s wife was attacked, beaten and raped and died afterwards. This was another aunt. She was not confusing her with an aunt who had been raped and beaten and died on an earlier occasion.
64. It was put to her that it was not clear from her statement or interview that she had had two aunts who had been so badly treated that they died. The appellant said that many of her family died in the war and she could not mention all of them. She confirmed that the aunt she had most recently been discussing was the one she had lived with for many years.
65. She was asked if she could explain why she did not leave when she was raped in 2004. She said she could not make the decision: she had to do what her uncle said. She could not remember exactly when her uncle’s wife died. She was attacked at the same time as the appellant was attacked. Her uncle suggested the war was escalating and it was time to get out. She was not consulted. She was told.
66. The appellant said they gathered their things and went. It was not a good house. It was not fit to sell. They went by boat to Yemen where they stayed for two months, then her uncle introduced her to the man who took her away. She had family jewels and she used them to pay. She was asked if she could explain why the jewels were not stolen by the intruders. She said they were buried. Her uncle had buried them somewhere. She claimed not to know when he had buried them. It was suggested to her that she knew when she stopped wearing them. She said he came back from the jungle and collected all their valuables.
67. She was asked it if it really was her case that she took her jewellery to the Jomvu refugee camp and her uncle took it to bury it on the way back. She confirmed that that was her case. The wedding jewellery was an Islamic dowry. There were very strong cultural pressures to give gold at the wedding. Her husband gave a dowry as did her mother. It was kept by her uncle. She said they all went together to the Jomvu refugee camp and then back to the island. There was enough money to pay for a false passport and the flight to Europe. She had since lost contact with her uncle.
68. She said that she arrived in England in June 2007 and claimed asylum straight away. She confirmed she had arrived wearing a sheepskin coat. She was asked from where she had obtained that and she said it was from a neighbour on the island. It was, she said, given to her by someone she had not met before. It was a gift.
69. Asked about her languages, she confirmed that she spoke Bajuni with a little bit of Arabic and Somali. She said she had learnt to write some Swahali in the refugee camp. She said that on her two interviews she spoke the Bajuni language. Bajuni was the only language that she spoke. She said that she knew Arabic and Swahili but had forgotten Somali and she spoke Bajuni. Bajuni is her native language and she used it in both interviews. She knew some Swahili and understood it but she did not speak it. When they were in the Jomvu camp they spoke Kibajuni between themselves but used Swahili when talking to others.
70. The appellant was asked about particular words that the linguistic analysis reports considered. She said that she did understand the Bajuni word /ikoto/ but missed it because she was under pressure. Similarly the word /jabiya/. She understood it but she was under pressure. She was uncertain of the meaning of the word /mkuru/.
71. She did know about /vehundu/. She said she was familiar with the word. She denied telling the interviewing officer that she had little knowledge of the Bajuni culture because her grandfather had died. She said that was a misinterpretation. She said she had explained sufficiently that the Bajunis had descended from Bantus and Arabs.
72. She accepted that she had told the Immigration Service that she was unemployed. She said there is a tradition that women do not go to work although her mother did collect and sell shells. She then qualified that to say that her mother collected shells but her father took them to sell.
73. It was put to the appellant expressly that she was just not telling the truth and that she was in fact from Kenya. She insisted she was Bajuni. It was similarly put to her that she did not speak Bajuni as a native language but she spoke Kenyan Swahili. The appellant denied this. She said that she spoke Kibajuni. She spoke Swahili sometimes when she lived in Kenya. She got some Swahili words.
74. Ms Whipple reverted to the Sprakab report and took her in detail to her own responses in relation to the four words mentioned above. The report said that the appellant was not familiar with the word /jabiya/ which the report described as a construction for gathering and storing water. According to the report it is a well-known construction, part of everyday life in Koyamo, and it is not something located in the ocean which is how the appellant described the word to the Sprakab interviewer.
75. The report also showed that the appellant was not familiar with the word /ikoto/. She thought the word was connected with hair. According to the report, in Koyama the word /ikoto/ described a woman who was not a virgin: that is a woman who had been married or was married. However in Kenya the word /igoto/, which was a variation of the same word, refers to hair extensions believed to be from the hair of virgins.
76. The word /mkuru/, according to the report, refers to a non-religious leader, being an older man elected by the population of the island, who has an advisory role. It is a role fulfilled by one person only rather than several simultaneously, and it was quite contrary to the report writer’s understanding of life in Koyama for there to be three mkurus at any one time but, according to the report the appellant named three people as mkuru in Koyama.
77. Similarly the report said that the word ‘vehundu’ was used in Koyama to describe a red (meaning a lighter) skin tone. According to the report the name is used for part of the Bajuni ethnic group deriving from the Somali Bajuni island Koyama. These persons have lighter skin compared to the rest of the population. According to the report the appellant did not recognise this word and pronounced it in an untypical way.
78. The appellant said that she did not know why she had not been able to explain these things at interview. The appellant agreed that they were ordinary Bajuni words. She had no reason for not describing them properly except that she was under pressure. It was put to her if she had discovered their meaning since and was trying to change her position. She said that was not the case.
79. In answer to questions from the Tribunal, she said at first that she had been living in the United Kingdom for nearly four years, and then corrected herself to say less than three years. She had been in Kenya for five or six years. She accepted that on her figures she had been in Koyamo for 27 years. She confirmed that she had been in Koyamo both before and after living in the camp in Kenya.
80. She said that she had not spoken any Swahili until she went to the camp. She said she was somewhere between 18 and 20 when she last had contact with her father. It was not the habit amongst the people on the island to care too much about age. She had remembered her father and remembered he had worked as a fisherman.
81. She claimed not to be sure if he had his own boat but she knew he was fishing. He did not go very far off the coast. Sometimes she saw him working. It was about twenty minutes walk away from their home. She said that if he got a lot of fish he took them to market, otherwise they were used solely for food at home. They had two markets. One was in Koyama, you had to take a boat to the other..
82. When her mother collected shells her father took them for sale. The shells were used for decoration.
83. Sprakab (whose full name Skandinavisk Språkanalys means “Scandinavian Language Analysis”), is now a privately owned company, though we understand it was originally set up by the Swedish government. It is based in Stockholm. Its work is linguistic analysis, which it undertakes for two primary reasons: the identification of place of origin of asylum seekers, and the matching of examples of recorded speech to the speech patterns of a known individual for forensic purposes. We are primarily concerned here with only the first of those fields of work.
84. The company has a large number of clients. They include the immigration services or equivalent, of Canada, Ireland, Sweden, Australia, Estonia, Norway, Austria, the Netherlands, and the United Kingdom. Sprakab says that it has conducted over 40,000 linguistic analyses since the year 2000. At the hearing we enquired whether Sprakab works for private individuals. As we understand the position, the company will not take instructions from a private individual in circumstances where it already regularly acts for the relevant national authority. We were told of one case where an individual sought a language analysis in order to prove his country of origin. Sprakab referred him to the national authority, who agreed that the analysis should take place. We understand further that, despite the fact that the analysis supported the individual’s claim, he was required to pay for it. That is the only example we were given of the company accepting instructions from a private individual.
85. As well as the manager, Pia Fernqvist, from whom we heard oral evidence, and other administrative staff, the company employs linguists, who have university qualifications in various relevant disciplines, are members of the International Association for Forensic Phonetics and Acoustics, and are subject to regular evaluation.
86. The company has a pool of analysts, which it employs as and when necessary. In general they speak the language they are asked to analyse at the native level, and are taught at Sprakab to “think critically and analytically regarding language”. They are required to keep their knowledge of their languages up to date.
87. Analysts are recruited within Sweden, using what is apparently an informal network of education institutions and recommendations by existing analysts. They are subject to evaluation before they are appointed, and to regular evaluation during their work for Sprakab.
88. Linguistic analysis at Sprakab is a two-stage process. It starts with an analyst listening to a specimen of speech, which is recorded. Typically, the analyst converses with the person whose speech is to be analysed. Using his developed skills as a listener, the analyst notes features of the subject’s speech which appear to be of interest. Those features may be phonological, that is to say related to the precise sounds used in speaking, and to the intonation and stresses of both words and sentences, morphological, that is to say related to inflection, grammar and syntax, or lexical, that is to say related to the speaker’s choice of words.
89. The analyst then discusses the features that have been identified with a linguist. Together, the analyst and the linguist decide whether the features are indeed diagnostic in the sense of assisting to decide the speaker’s origin. A report is produced, giving a judgement, usually in one of the four following forms:
“The person speaks a variety of [language or dialect] found
- with certainty in (or certainty not in):
- most likely in:
- likely in:
-possibly in:”
90. Sprakab’s account of the measure of these phrases is as follows:
“If the analyst is certain that the analysed language is found in a specific region, he/she uses the degree of certainty “with certainty in”. The degree of certainty decreases if the speaker mixes in elements not normally found in use in the region he/she maintains his/her linguistic background in. Rationale for a determined degree of certainty is ordinarily explained in the report.”
91. The person conducting the interview with the claimant, which is saved onto CD, may also be the analyst, in which case a preliminary result can be given at the end of the interview. Analysts are given extensive training by the linguists, so that they will be on the lookout for certain distinctive features of the language or dialect spoken by the claimant. Indeed, before analysts are taken on by Sprakab, they must undergo an evaluation process, in which they must listen to a dozen or so sample recordings from Sprakab’s collection and get the language or dialect right. Then they are asked to analyse for themselves another twenty or so recordings which have already been analysed by an experienced analyst, and only if they show themselves competent are they appointed to work for Sprakab.
92. The contribution of the linguists is different. They may well not speak the language of the claimant, but will check that the analyst has identified linguistically significant features of the claimant’s speech which enable a conclusion to be drawn with the requisite degree of certainty.
93. In her oral evidence, Pia Fernqvist confirmed what she had said in her written report about the training of analysts and linguists, and explained that the interviewer on 20 June 2007 had used Swahili because 97% of those claiming to be Bajuni turned out to be from Kenya. That a second and third linguistic analysis had been carried out on that interview was not because of doubts about the first analysis, but because the Home Office had requested them. Ms Fernqvist assured us that her analysts made due allowance for dialect and language mixing, but agreed that the “referral details” about the appellant given to Sprakab by the Home Office were parsimonious. They said merely that the applicant for asylum claimed to come from the Bajuni islands in Somalia, and that the applicant’s first language and other languages spoken there comprised Swahili, Kibajuni and Somali.
94. In cross-examination, Ms Fernqvist agreed that linguistic analysis cannot determine nationality (although it helps) and that Sprakab only works for governments. A report would normally be compiled for the Home Office, she said, after an interview lasting 20-30 minutes. The analyst would run through the recording of the interview three times before producing a draft report and discussing it with a linguist. There would then be a linguists’ meeting, before the report was typed up. How was it, if that were so, that after a telephone interview on the same day as the screening interview (20 June 2007), the appellant was told by an immigration officer that it had been concluded that she hailed from Kenya? Ms Fernqvist explained that, on this occasion, the analyst had felt able to give the Home Office a preliminary indication straight away.
95. Mr Davison returned to what Ms Fernqvist had said in-chief about 97% of claimed Bajuni being Kenyans, and asked whether a Swahili-speaking Kenyan had conducted the first interview because it was assumed that the appellant was also a Swahili-speaking Kenyan. Ms Fernqvist insisted that the islanders can all understand Swahili in any event. Alluding to Sprakab’s claim that boasted that analysts spoke “to mother tongue level” the language of the interview which they were analysing, Mr Davison noted that EA 19 held himself out as speaking Swahili, Luhya and Luganda to mother tongue level, but not Kibajuni. Ms Fernqvist replied that he could analyse Bajuni, having worked among Bajuni refugees. The first interview was actually conducted by interviewer 240, who does speak Bajuni at mother tongue level. The analysis was then carried out by EA 249, and verified by linguist 03. This method of working was, Miss Fernqvist asserted, in keeping with the resolution passed in 2009 at the annual general meeting of the International Association for Forensic Phonetics and Acoustics, which recognized the contribution to be made by:
“Linguists and trained native speakers, with the latter working under the guidance and supervision of the former.”
96. Ms Fernqvist went on to describe how Sprakab kept abreast of changes in the languages with which they regularly deal, such as newly formed words. She gave the example of northern Iraq, from which one of her analysts had recently returned with an update on the changes going on there. Would there not have been changes in Kibajuni since the breakdown of the Somali state in 1991?, asked Mr Davison. Ms Fernqvist did not think there had been much change, the islands being isolated, but she acknowledged that people leaving the islands would be likely to pick up words from elsewhere.
97. In response to questions from the Vice-President, Miss Fernqvist said that Sprakab carried out around 4,000 analyses per year. She was asked if she could indicate how often Sprakab supported an appellant’s claims. She said that she did not have statistics to hand but her “feeling” was that Sprakab supported applicants in about “60%” of the cases in which they were involved. She was asked if she meant that Sprakab supported appellants more often that it rejected appellants’ claims and she unequivocally replied in the affirmative. She acknowledged that it was quite normal for a screening interview to be interrupted for a telephone interview with Sprakab, followed by a provisional analysis, but she insisted on the accuracy of the conclusion in the instant case. The analyst who conducted the second interview in February 2010 (EA 24) was born on the Bajuni island of Chuvay and lived in the Bajuni area for 20 years. He spoke to the appellant in Bajuni, and was sure that she could not speak Bajuni.
98. Miss Fernqvist went on to describe how a database of recordings has been built up at Sprakab, with the language of each labelled for the use of the linguists, who do not themselves have to speak the language on which they are collaborating with the analyst. The recordings, she acknowledged, are not available for peer review, but she was certain that the identification of the languages and dialects on the database was accurate. There could not be anything wrong with the database. If there were, Sprakab’s customers would be dissatisfied.
The individual linguistic evidence in the present appeal
Margaret Kumbuka
99. We have summarised above the written reports prepared before the hearing. It was expected that we would hear oral evidence in support of the appellant’s case from Margaret Kumbuka. Unfortunately, as we were informed on the first day of the hearing, she had died a short while previously. We therefore have her evidence only in written form.
100. Margaret Kumbuka said in her report of 20 September 2008 that she was born in Tanga in Tanzania. Her parents were born there and lived there. Her mother tongue is Kiswahili and she grew up speaking Kiswahili and Kizigua. She learned English at school and graduated as Bachelor of Education at the University of Dar Es Salaam in 1975. Her main subjects were English and Swahili. She built a career in broadcasting and teaching and in 1999 became a “lector” in Swahili in the African Department at the School of Oriental and African Studies at the University of London. She had a particular interest in teaching and studying Swahili.
101. She described Kibajuni as a dialect of Swahili and said that she became familiar with the different forms of Kibajuni spoken in Kenya and Somalia because of contact with people from those areas that began in her childhood in the 1960s.
102. She interviewed the appellant in July 2008. The appellant answered correctly several questions about the customs and traditions of Bajuni people in Somalia.
103. She then turned to the appellant’s language. During the interview the appellant used Kenyan Kibajuni rather than Somali Kibajuni words to for “fisherman”, “keep quiet”, “bird”, “aeroplane”, “uncle” and “auntie”. Ms Kumbuka suggested that the appellant had learned these words when she lived in Mombassa. The appellant gave Somali words for “sea”, “fish”, “boat”, “house”, “food”, “water”, “hair”, “eggs” and “door”.
104. Ms Kumbuka then gave three examples from the interview of the appellant using Somali Bajuni rather than Kenyan Bajuni phrases in answer to questions. Two of the three chosen phrases only show one different word between the two examples. However information about the main work of the Bajuni people, “Men are doing the work of fishing and women pick shells” was given as “Vanaume vafanya kadhi ya utoniya na vanavake vachunda kauri” whereas, according to Ms Kumbuka, a Kenyan Bajuni speaker would have said “Vanaume ni vavuvi na vanaokota makombe”. This is clearly quite a different phrase and, it might be thought, is not as easy to attribute to dishonesty as simply using a single Somali Bajuni word rather than a Kenyan Bajuni word.
105. She was satisfied that the appellant is a Somali Bajuni from Koyama.
106. In her letter of 10 November 2010 Ms Kumbuka confirmed that opinion. The appellant showed a good knowledge of Somali Bajuni culture and used Somali rather then Kenya Bajuni words.
107. In her report of 3 December 2008 she was critical of the proofreading of the transcript of the appellant’s recorded interview. She also said that the interviewer spoke “broken Swahili with a very heavy Kikuyu accent.” She thought it was “obvious” that the appellant had avoided Kibajuni words so that the Swahili speaking interpreter could understand her.
108. Ms Kumbuka criticised the quality of the English translation of the taped interview and indeed the quality of the language that he used. There was also an “era” (error?) in the translation where there was talk about the appellant’s marriage.
Analyst EA 19
109. Analyst EA 19, whose name was disclosed to the panel in accordance with the agreed arrangement, but who otherwise remained anonymous, gave oral evidence. He confirmed that he had worked among Bajuni islanders both in Mogadishu and in the Jomvu camp near Mombassa, and was familiar with both the Kenyan Bajuni and Somali Bajuni dialects. He was sure that the appellant spoke coastal Swahili, rather than either of those dialects. He described how he was tested by Sprakab before becoming an analyst, and how his analyses were carried out, requiring consultation with a linguist. In respect of the interview conducted by interviewer 240 on 20 June 2007, the appellant was speaking Swahili throughout, and although she knew something of Bajuni life and culture, the Bajuni words which she did use had evidently been learned rather than acquired naturally, and were not used appropriately.
110. During her interview with Margaret Kumbuka the appellant was asked to translate some words from Swahili into Kibajuni, and she translated them into the Kenyan dialect of Kibajuni rather than the Somali dialect. Miss Kumbuka supposed that the appellant must have acquired Kenyan Kibajuni while in Mombassa, but EA 19 scotched the notion, saying that the police did not allow the Somali Bajuni who had taken refuge at Jomvu to go outside the camp. He admitted that the appellant had answered in Kibajuni five questions put to her by Miss Kumbuka, but the answers were very short, and were not enough to show that the appellant was a Bajuni islander.
111. Although her Bajuni had dramatically improved at the interview in February 2010, she was still not using Bajuni words appropriately. Referring to the ‘Knowledge Assessment’ section of the subsequent report, EA 19 gave further information about the words listed there, which the appellant had misused : /jabiya/, /ikoto/, /mkuru/ and /vehundu/.
112. In cross-examination, EA 19 admitted he had not been to the Bajuni islands himself, and acknowledged that the language might have changed since 1991. He was unsure whether the language had changed because of the presence of Marehan clansmen on the islands, but he did not think that the appellant was accommodating her language to that of the interviewer in the first interview, or that the incidence of Bajuni words in a language sample which was predominantly “Kenyan coastal Swahili” indicated that the appellant exhibited the phenomenon of ‘language mixing’.
113. EA 19 confirmed to Ms Whipple that his fellow-analyst after the interview on 2 February 2010 (EA 24) hailed from Chuvay, an island near the appellant’s claimed birthplace of Koyama. In response to questions from the Vice-President, he said that the dialect spoken in Jomvu camp was island Bajuni, although he and the other helpers spoke Swahili. Clarification was sought about certain words, particularly /ikoto/, which among the islanders means a woman who is not a virgin but is unmarried, and on the coast means a hair extension.
Linguist 01
114. The next witness was linguist 01, who told us he had been with Sprakab since 2004 and had presented a paper to the annual conference of the IAFPA in 2008. He was familiar with linguistic analyses of Bajuni claimants, and with the kind of traits that should be found in Kibajuni speech, such as the unvoiced /bo/ found in the bilabial position. He would check unusual features with the analyst. Margaret Kumbuka, he said, would not have been taken on by Sprakab, as she was not a qualified linguist, and could not be an analyst either, coming from Tanzania and not from the Bajuni area (Kenya and Somalia).
115. In cross-examination, it was put to linguist 01 that EA 19, like Miss Kumbuka, had not been to the islands. Yes, was the reply, but he had been to Jomvu and Somalia. 01 admitted that he did not speak Swahili himself, and that if the analyst had made a mistake, he would make the same mistake. In response to questions from the Vice-President, 01 admitted that he depended on the analyst not just for phonetic features but for lexical items and syntax in the language being analysed. As a linguist, however, he was familiar with recognising the distinctive features of a language and their modification.
Linguist 04
116. The last witness was linguist 04, who confirmed that, like linguist 01, he was familiar with the features of Bajuni which had been identified by the analysts as distinguishing that dialect from Swahili, and could compare the appellant’s speech with numerous other recordings built up by Sprakab. He was confident that the analysts were right in pinpointing features of the appellant’s speech which showed her not to be a Bajuni islander. The Vice-President observed that in the report on 15 February 2010 the absence of the devoiced plosive /bo/ did not feature, as it had in the report of 24 June 2007. But 04 was able to point to many other features of the appellant’s speech which were not characteristic of Bajuni.
117. In her closing submissions, Ms Whipple highlighted a number of inconsistencies and implausibilities in the appellant’s account which cast doubt upon her credibility, before turning to the Sprakab evidence. She characterised this as professional and of a high quality, in contrast to the reports provided by Margaret Kumbuka. The interview in July 2008 had not been recorded, and so was not available for analysis, but the criticisms made of it by Sprakab were cogent. The appellant did know some Kibajuni, but it was minimal, and Miss Kumbuka had to find an excuse for the Kenyan Swahili features which were so prominent in the appellant’s speech. The appellant also knew something of Bajuni life and culture, but that could be easily researched, and Sprakab had shown up inaccuracies in her grasp of that too. While, as the Home Office guidance itself made clear, the Sprakab analysis should never be the sole determinant of a nationality question, it was a reliable pointer to the resolution of that question.
118. The trouble was, said Mr Davison in his submissions, the Home Office and immigration judges did treat Sprakab analyses as if they were determinative. In the instant case, Sprakab had started from the assumption that the appellant was not Bajuni, and by interviewing her in Swahili, they made it very likely that she would adapt her speech to that of the interviewer. Sprakab had also underestimated the effect that Marehan occupation of the islands would have had upon the Bajuni dialect, and had discounted language change as an explanation for the absence from the appellant’s idiolect of some traditional features of Kibajuni. The first report on the appellant, written on 24 June 2007, was not of a suitable quality for production before the Tribunal, and two more attempts were needed to get an analysis that the respondent was happy with.
119. Mr Davison defended the quality of Margaret Kumbuka’s reports, the second of which had picked up mistakes in the transcript of the first Sprakab interview. He reminded us that the asylum interview had been conducted in Kibajuni, which was also the language used by the appellant before Immigration Judge Pugh. The appellant had also given good evidence about Bajuni culture. While acknowledging that language analysis does have a role in status determination, Mr Davison insisted that it had to be of a high standard, as in the Sprakab report of 15 February 2010. The Home Office ought to tell Sprakab more about the claimant when commissioning a report. In the instant case, Sprakab were not informed initially that the appellant had lived in Kenya from 1993 to 1998.
Other evidence
120. Before the hearing began, we were aware of further documentary evidence in relation to the issue of language and Kibajuni in particular.
121. “Guidelines for the Use of Language Analysis in Relation to Questions of National Origin in Refugee Cases” were laid down in June 2004 by an international group of linguists. These emphasize such matters as the need for language analysis to be done by qualified linguists and for the data on which the analysis is carried out to be useful and reliable. Also emphasized is the difficulty of gauging a person’s origins from linguistic evidence alone, since in many parts of the world multilingualism is the norm, and one speaker may produce a complex array of ‘language mixing’, ‘code switching’ and ‘style shifting’.
122. Appearing on the UNHCR website ‘Refworld’ is some information on the Bajuni folk compiled by the Immigration and Refugee Board of Canada in November 2005, which states that a Bajuni can completely understand someone speaking standard Swahili, but that a standard Swahili speaker might have difficulty understanding two Bajuni speakers who are conversing in their own dialect. The Bajuni folk themselves are said to number between 15,000 and 20,000 souls, and to dwell on the islands and coastal strip stretching from Kismayu in Somalia to the Lamu archipelago in Kenya.
123. As we have indicated, at the close of the hearing it became apparent that there was further material of which we might wish to take account.
Post hearing evidence
124. At the close of the three-day hearing, there were three matters on which it was felt that more information would be useful. We had not seen what (if anything) the Immigration Directorates’ Instructions had to say about language analysis. We had seen Sprakab’s website saying that “Sprakab Language Analysis is commissioned by immigration bodies and police authorities who wish to establish a person’s language background, e.g. to support an asylum investigation.” We wanted to know whether Sprakab ever worked for anyone other than governmental bodies. Thirdly, mention had been made during the hearing of a recent report on the Bajuni by the Country of Origin Information Centre in Norway (‘Landinfo’), but an English translation was not available. It was agreed that the Home Office would file and serve the relevant material, and that both representatives could send in written submissions based upon them.
125. It turns out that there is no actual ‘IDI’ on language analysis, but rather ‘Asylum Process Guidance’ to UKBA caseworkers. Neither side wished to make submissions on this guidance, but we shall reproduce section 4, entitled ‘How the Direct Analysis works’, for what it says about the Sprakab interview and report.
“There is a telephone interview between the asylum applicant and the Sprakab analyst [as we have heard, the interviewer may not be the analyst] (who will speak the language to be analysed at mother-tongue level). This is recorded by Sprakab, and usually lasts 20-30 minutes. The applicant is asked a variety of questions, limited to obtaining information appropriate for the language analysis. This will include information relating to linguistic and geographic information, as well as specific knowledge of the applicant’s supposed country or region of origin.
A preliminary result will be communicated to UKBA at around 15 minutes after the end of the interview if the case is straightforward.
A more detailed analysis will then be conducted by the analyst, taking fully into account phonological, morphological and lexical phenomena.
If the analyst judges that the asylum seeker does not speak in a manner consistent with the area from which they claim to originate, the case will be sent to a second analyst with a background from the area which the asylum seeker is said to originate (sic).
A language analysis report will be produced by a linguist working alongside the analyst.
Sprakab will produce the language analysis report (translated into English), which includes details of the analyst/linguist’s credentials (for security reasons, only their number codes will be revealed) and three CD copies of the interview. For security purposes, the analyst’s voice will be altered on the CD, and they will be identified in the written material by ID code rather than by name.
An electronic copy of the report is usually provided to UKBA within 72 hours of the language analysis interview, but the hard copy and CDs may take a little longer to arrive.”
126. We should add that section 11 of the Guidance begins “A decision must not [emphasis in original] rely solely on the direct language analysis report or an applicant’s failure or refusal to undergo language analysis.”
Sprakab’s clients
127. As for Sprakab only working for governments, their position is that “Sprakab will not act for a person/company if it is already engaged to work for the government in that country, because it would in that case perceive there to be a conflict of interest.” The respondent has made no comment on this, but Mr Davison submits that this goes to show that “they are not totally impartial/independent.” We do not think that follows. The fact that 60% of the asylum applicants interviewed by Sprakab are assessed as speaking the language which they claim to speak suggests that Sprakab is not tailoring its results to suit its clients.
Landinfo report on the Bajuni Islands
128. The Landinfo document, published on 16 February 2010, has now been translated by Sprakab, and describes itself as a ‘Topical Note’ on the Bajuni Islands. Landinfo’s information comes from interviews with Bajuni islanders in Nairobi and Mombassa during fact-finding trips in 2002, 2007 and 2008. The respondent’s comment on it, transmitted through the Treasury Solicitor, is brief. The report is said to be “consistent with existing objective evidence”, in particular confirming that there is a difference between the Kenyan (or “coastal”, as it was referred to at the hearing) and Somali varieties of Kibajuni, and that the majority of Bajunis do speak Somali. The latter point is said to tell against the appellant, who can only speak a few words of Somali.
129. We think that, at the hearing, “coastal” was used to describe a variety of Swahili spoken more widely along the coast of Kenya than just by the Kenyan Bajuni, but the Landinfo report itself is rather contradictory on this topic. On the one hand, the Kenyan Bajuni who met with Landinfo in Mombassa in 2008 said that, although there was a dialect difference between Kenyan and Somali Kibajuni, the difference was so small as to be imperceptible to Swahili speakers from other parts of East Africa. On the other hand, Somali Kibajuni is said to have incorporated “loanwords, expressions and pronunciations” from Italian and Somali, while the relative isolation of the Somali Bajuni from the Kenyan Bajuni in recent years, because of the security situation, has led to “greater dialect differences than one might expect.”
130. Landinfo’s information about the familiarity of Somali Bajuni with the Somali language is also somewhat contradictory. According to interviews with Somali Bajunis in Mombassa in 2008, “the inhabitants of the Somali mainland have also traditionally been far more skilled in Somali, Italian and Arabic than the islanders, who have had a very limited linguistic repertoire.” That is certainly what one might have expected in the past, but the report goes on to say this.
“According to several Bajuni representatives (interviews in 2002, 2007 and 2008), the majority of the Somali Bajunis speak Somali. Older persons may have poor Somali skills, but the younger generation speaks Somali. Even those with poor speaking skills in Somali are still able to understand it (interviews in Nairobi in 2007). This has to do with the fact that the island population was moved to the mainland in the mid-1970s. The main justification given by the authorities at that time was the need for development and schooling, and many children and young people attended schools on the mainland where the instruction was given in Somali.”
131. It surprised us to learn that the familiarity of the islanders with the Somali tongue did not result from the islands being swamped by majority clansmen after the fall of Siad Barre’s regime in 1991, but from forced displacement to the mainland in the mid-1970s. The report has more to say about this.
“The Bajuni population on the Somali Bajuni islands of Koyama, Chula/Mdoa and Chovai was forced to migrate to the mainland village of Kudai in 1975 (interviews in Nairobi 2002 and Mombassa 2008). The name Kudai was later changed to Kulmis. When most of the local communities on the coast were attacked by the local Somali clan militias in the late 1980s, many Bajunis sought refuge in the places they originally came from – the islands. Starting in 1997, a large number of Bajunis were repatriated to the islands under the auspices of UNHCR, but the repatriation programme was halted when the security situation in the region deteriorated in June 1999. Several hundred Bajunis also returned to the islands (UNCT 1999) when the refugee camps on the coast of Kenya were closed in the late 1990s. Somalis also live on the islands today and, even though there are instances of marriages between Somali men and Bajuni women on the islands, it has been claimed that the local Bajuni population is being exploited by Somali businessmen (interview with international aid organisation, Nairobi 2008).”
132. The Treasury Solicitor is certainly wrong to describe this as “consistent with existing objective evidence”. The principal source of information about the Bajuni in this jurisdiction has been the report of the joint British, Danish and Dutch fact-finding mission which was in Nairobi from 17 to 24 September 2000 and “met with a group of elders from the Bajuni community in Nairobi.” From them the delegation learned about the raids on the islands by Somali clan militias as the Barre regime collapsed in 1990-1991, and the flight of most of the Bajuni islanders to Kenya during 1992. The Bajuni émigrés who returned to the islands after the closure of the Jomvu camp in 1997 found that majority clansmen had seized their property and that it was very difficult to make a living on the islands, where they were still liable to attack by the militias.
133. There is no mention at all in the 2000 report of a wholesale displacement of the island population to the mainland in 1975, and a return to the islands at the end of the 1980s in order to avoid attacks by the majority clans. A footnote to the Landinfo report says that the story of the forced migration to the mainland is supported by an Oxford House report on minorities in Somalia written by Abdi Abby and published in October 2005, but this actually talks about the forced collectivisation of the Bajuni fishing operations in the mid-1970s, when the Marxist regime of Siad Barre seized the fishing boats and fishing gear of the islanders. The Landinfo report implies that many Bajuni did not go back to the islands at all after the forced settlement on the mainland, with the result that “many Bajunis born after 1975 have little knowledge of the islands.”
134. If the Landinfo report is correct, it is unlikely that the appellant would have stayed on the island of Koyama from when she was born in 1975 until she went to a refugee camp in Kenya at the age of 18. The 1975 migration is not the only respect in which the Landinfo report differs from the received wisdom about the Bajuni. The description of those Bajuni who live on the islands now being “exploited by Somali businessmen” seems rather mild compared to the impression of rape and pillage by marauding militiamen created by the Anglo-Dutch-Danish report. That report, on the other hand, also gives the impression that the Bajuni folk stop short at the border with Kenya. “The Bajuni elders”, we read, “stated that the Bajuni have no close links with the people in Kenya who live immediately below the Somalia-Kenya border on the coast and islands.” The elders also informed the delegation that “their language is very different to the Swahili dialect spoken in the areas of Kenya immediately below the Kenya-Somalia border, including the islands that continue from the Somali border down along the coast towards Mombassa, although there are some common words.” But that is precisely the region inhabited by the Kenyan Bajuni, who speak a variety of Kibajuni rather than Swahili (or even ‘coastal’ Swahili).
135. In his submissions on the report, Mr Davison does not mention these problems, but complains that only a translation of the report has been provided, not the original, and that the translation has not been “signed by the translator to certify that the translation is accurate”, as was required by rule 52(1) of the Asylum and Immigration Tribunal (Procedure) Rules 2005. This means that, under rule 52(3) the AIT would have been under no duty to consider the report at all. Mr Davidson did not draw our attention to any equivalent provisions in the Tribunal Procedure (Upper Tribunal) Rules 2008. In any event, he finds passages in the report which assist his client, such as the “very limited linguistic repertoire” enjoyed by the islanders as opposed to the Bajuni living on the mainland, and the confirmation of what the appellant said at interview about the two villages, the Koranic school and the old buildings on the island of Koyama, and about her membership of the Naufali sub-clan (which appears as ‘Nofaly’ in the report).
ILPA mailing
136. Before we could deal with this further evidence we heard of more. In March 2010 the monthly mailing of the Immigration Law Practitioners Association to its members included three brief ‘reports’ on Bajuni matters, all dated February 2010, by Professor D. Nurse, Emeritus Professor of Linguistics at St John’s University, Canada and a specialist in Swahili dialectology. He has visited Somalia for only a week, but has lived for six months in the Bajuni villages of North-Eastern Kenya, and describes the coasts and islands stretching from Kismayu in Somalia to Lamu in Kenya as home to a Bajuni folk whose language and culture are virtually homogenous.
137. In the first of his reports, Professor Nurse presents an ‘overview of Sprakab linguistic analyses of Bajuni refugee claims 2004-2010’, during which period he has read some 50 Sprakab reports. He criticises the brevity of these Sprakab analyses and the certainty of the conclusions reached, emphasizing the fluidity of Bajuni language and society amid the upheavals of recent years, and the ‘language mixing’ between Kibajuni and Swahili which many individual Bajunis now display. In the second report, ‘Overview of Sprakab telephone interviews of Bajuni refugee claimants’, Professor Nurse says that he has listened to over 20 recordings of such interviews, and makes sharp criticism of such features as the acoustic quality of the telephone link, the format of the questions, and the brevity of the interview. He makes the point that the interviewer’s use of Swahili, the language of prestige and power, “virtually removes any possibility” of the interviewee replying in Kibajuni, “the language of an impoverished and disempowered minority.” No conclusions are contained in this second report, but the first concludes with the view that it would be unwise to use a Sprakab report as a basis for any legal decision on whether an applicant is or is not a Somali Bajuni.
138. Third comes a ‘Review of the Bajuni sections of the report on minority groups in Somalia, by a joint British, Danish and Dutch fact-finding mission to Nairobi, Kenya, 17 to 24 September 2000’. Professor Nurse takes issue with what the Bajuni elders told the mission, in particular about most Bajuni being able to speak some Somali and about the difference between the dialects spoken north and south of the border between Kenya and Somalia. He regards the elders as out of touch with the language situation on the islands. The refugee camps in Kenya were “porous”, and so exposed were the Bajuni to speakers of ordinary Swahili that, on return to the islands after the camps closed, many young Bajuni spoke mainly Swahili.
139. Not surprisingly, Mr Davison finds support for his client in Professor Nurse’s reports, which show that she might well be able to speak both Bajuni and Swahili while not being able to speak Somali. The criticisms of Sprakab telephone interviews and subsequent reports made by the Professor could equally well have been made of the Sprakab report in the instant case following the telephone interview on 20 June 2007. But Mr Davison acknowledges that better quality reports – “suitably detailed, precise, sourced and compiled by qualified experts” – may have a part to play in an immigration context, bearing in mind that they can establish what language(s) or dialect(s) an appellant speaks, but cannot establish nationality.
140. Ms Whipple, on the other hand, submits that the Tribunal ought not to admit the Nurse reports at all. She makes the point that Professor Nurse “was not present at the hearing, has not heard the extensive evidence already given, and has not seen the documents in this case. He is simply not qualified to give evidence, as matters currently stand, on the matters raised in this appeal.” Ms Whipple appends a rebuttal of the Professor’s criticisms by one of the Sprakab linguists who gave evidence before us, but contends that it would be unfair of the Tribunal simply to read the rival documents and come to conclusions on their weight and cogency ‘on the papers’. “Either Professor Nurse’s evidence is in”, she insists, “as part of the evidence in the appeal, in which case he must be called to give evidence, or it is not.”
141. We do not think it would be right to shut our eyes to Professor Nurse’s views, once we became aware of them, and we had to afford the parties an opportunity to comment upon them. But we cannot treat Professor Nurse as an expert witness in this appeal. There are numerous points of factual dispute which would need to be addressed by way of live evidence and cross-examination. For example, Professor Nurse says that young Bajunis from Somalia who have lived in refugee camps such as Jomvu now speak Swahili as well as, or even in preference to, Kibajuni. The porous nature of the camps allowed intercourse with Swahili-speaking Kenyans. On the contrary, analyst EA 19 told us that Jomvu camp, where he worked, was kept sealed off from Kenyan society, and that the Bajuni refugees were not given the opportunity of mingling with ordinary Kenyans. We cannot prefer the former view to the latter, without hearing directly from Professor Nurse as we have heard from EA 19.
142. The problem of post-hearing evidence is illustrated even more vividly by the surprising account in the Landinfo report about the wholesale displacement of the Bajuni islanders to the Somali mainland in 1975, which is quite contrary to the received notion that they lived on the islands relatively undisturbed until, as Professor Nurse puts it, “in 1991 ethnic Somalis flooded onto the islands, bringing chaos, violence, rape and death with them.” The latter view is enshrined in the Anglo-Dutch-Danish report of the fact-finding mission in September 2000, but in truth the only evidence which the delegates had for it came from the Bajuni elders whom they met in Nairobi. The information in the Landinfo report similarly comes from Bajuni people whom the Norwegians met in Kenya, though perhaps from a greater number, since there were three trips in all to Kenya (in 2002, 2007 and 2008). The parties have not addressed us on this discrepancy, which we can neither resolve entirely satisfactorily nor ignore but which forms part of our overall assessment of the appellant’s claim.
Findings and Conclusion: the appellant
143. We are very aware of the difficulties faced by the Bajuni people. We accept that the appellant says that she had been the victim of the kind of persecution that Bajunis have experienced and continue to experience. If we believe the appellant then the appeal should succeed. We have also reminded ourselves of the low standard of proof applicable to these proceedings and that, sadly, there is an inherent plausibility about the appellant’s story: that is to say, the events she has described are of the sort experienced by many Island Bajunis.
144. Nevertheless we disbelieve the appellant and we dismiss the appeal.
145. There are several reasons for this finding. Some of them are on the periphery of our decision and have not been given much weight. Even so we do find it incongruous that a woman living in the dire circumstances claimed by this appellant would have been able to have secreted sufficient gold to have financed her trip to the United Kingdom and to have benefited from the gift of a sheepskin coat.
146. We also disbelieve that appellant’s claim that the explanation recorded at interview for not knowing about the history of her clan, namely her grandfather’s death, was the result of mistranslation. There is no obvious reason for an apparently competent interpreter to become inept when dealing with this particular answer. We accept that the appellant gave some sketchy details about her clan history but clan history is a defining element of Somali culture and a Somali must understand his or her clan history in order to function in society. We find it far more likely that attributing ignorance to her grandfather’s death is an attempt to head off questions that the appellant might find hard to answer because she is not telling the truth, than a reason for her being ignorant of something that she would know.
147. We have also considered the evidence in the Landinfo report. It is not contradicted by the Anglo-Dutch-Danish report but rather draws its information from a wider source. We find it extremely unlikely that the claim that the Bajuni islands were depopulated by the Barre regime is made up. It follows that it makes no sense for the appellant not to have mentioned anywhere in her evidence her removal from the islands whilst in her infancy, her growing up on the mainland and her return to the islands. This is a further indicator that she has not told us the truth.
148. We do not believe either her accounts of the fates of her relatives, or her explanations for the differences between her most recent account and earlier ones, or for her failure to mention earlier the fate of the other aunt.
149. We do find the appellant’s use of the phrase “Vanaume ni vavuvi na vanaokota makombe” (see paragraph 123) to be of some significance. It has not been explained properly and is a pointer in the appellant’s favour. But it is greatly outweighed by evidence counting against her.
150. One of the reasons for considering that the Immigration Judge might have made an error of law was said to be in her failure to consider the evidence that the appellant appeared to have ‘given evidence in Kibajuni’. We suspect that the position before the Immigration Judge was that it was before us: that is to say that a Kibajuni interpreter was used. Similarly, the appellant was interviewed in Kibajuni.
151. Kibajuni is a form of Swahili. The appellant’s ability to deal with questions in Kibajuni is not really the issue. The question is how well she understands Kibajuni, and the way she speaks it herself. These are issues on which we are dependent upon expert evidence. We have no basis for any proper assessment of the general linguistic features of her oral evidence.
152. We do, however, attach considerable weight to deficiencies in the appellant’s linguistic knowledge disclosed in her answers to questions in cross-examination. The appellant did not know the word for water store /jabiya/. She did not understand the role of the village elder /mkuru/. She did not know the word for a sexually experienced woman /ikoto/ but, significantly, mistook it for a similar sounding Kenyan word; and she did not know a word used to describe a skin colour and therefore ethnic group /vehundu/. The appellant admitted that she had been asked about ordinary words. She did not suggest that the words were not ones that she should have been expected to recognise. Rather she attributed her failure to understand them to being under pressure. We do not believe that “pressure” caused the appellant to forget the meaning of ordinary words. We find that she was not familiar with the words because she has not lived in the Bajuni islands. That conclusion is also, as it happens, entirely consistent with the Sprakab evidence, which we accept as reliable.
153. We conclude that the appellant is not truthful and not Bajuni, and shall dismiss her appeal.
Findings and conclusion: Sprakab
154. We are satisfied that Sprakab is a bona fide organisation that is linked to international linguistic organisations. Its general manager, Pia Fernqvist, has an academic reputation to enhance and protect and we are quite satisfied that it has devised and refined a system for analysing language that requires interaction between several employees. It is thus, to an extent, self checking and its mechanisms minimise the opportunities for one person’s incompetence or indifference leading to a false result.
155. We find it relevant that the manager makes no attempt at anonymity for herself. We see no way of avoiding a conclusion that the anonymity of some Sprakab employees could theoretically impact adversely on Sprakab’s reliability. For example an anonymous employee could deservedly have a poor reputation that his anonymity suppressed: but we regard this as a theoretical rather than real objection. Ms Fernqvist cares about her and Sprakab’s reputation and so cares about whom it employs. Much more importantly, no one person’s opinion is decisive; with the result that incompetent or corrupt employees could reasonably be expected to draw attention to themselves by taking positions very different from their colleagues. Further the opinions produced by Sprakab are reasoned and explained and can be checked and criticised without knowing the identity of each person involved in the decision-making process.
156. We reject the criticism that Sprakab is not independent. Presently it only works for government agencies. This is not immediately a sign of objectivity but we remind ourselves that applicants and decision makers have different functions. The applicant wants a result and some applicants will seek to achieve it by deception. A government based decision-maker, at least in a United Kingdom context, is charged with making a right decision. It has an inquisitorial role that applicants do not. Clearly Sprakab could not give a report to a government agency in a case where it had previously given a report to assist an applicant. If it did make its services available to all comers it could find that it was neutered by a swift and dishonest applicant who, in effect, commissioned not a Sprakab report but Sprakab’s silence. In our experience there is no competing organisation to which a government agency could turn for a report and against this background Sprakab’s fear of a conflict of interests makes more sense.
157. We think it inherently unlikely that a government agency would instruct Sprakab in a straightforward case where an applicant had made a good impression yet we are told, and accept, that Sprakab gives a report that assists an applicant in most of the cases in which it gives advice. We do not find any merit in the suggestion that it is inherently biased against applicants. One the contrary, in most contested cases it supports them.
158. We might have taken a different view if we understood that Sprakab would not give a report in cases where it was instructed jointly by an applicant and an official decision maker but there was nothing before us to suggest that this would be the case.
159. Sprakab does not claim to be infallible. It offers four categories of opinion (see paragraphs 89-90) and this opinion evidence must be evaluated with the other evidence in the appeal. The kind of linguistic analysis offered by Sprakab is a serious step towards independent and verifiable opinion but it does not claim to have the reliability of, for example, fingerprint evidence. It may that linguistic analysis of the kind used by Sprakab is a developing discipline and Sprakab will become subject to more peer review. It would certainly be in accordance with our understanding of Sprakab that it will constantly seek to refine and improve its methods. The evidence before us shows that Sprakab provides an honest, serious and useful guide to establishing the location where a person learned to speak.
160. The biggest weakness we see in the Sprakab system is that it depends at its core on an alleged expert in the language saying, for example “I do not think people who come from there talk like that”. However that opinion has to be explained. Reasons for holding it are given. Patterns of speech, for example, are recorded in ways that trained linguists can understand, reproduce and check. Sprakab is sensitive to the changes in language use and notes them and keeps in contact with those of a like minded interest. Further no final conclusion is reached just on one person’s opinion.
161. Sprakab’s methods contrast with those of Margaret Kumbuka who, we accept, was acting honestly and who had an academic reputation to cherish. However her reputation was not as a linguist. She did not record her main interview and so, unlike the Sprakab report, her opinion could not be checked by any other person. Ms Kumbuka noted that the appellant used some Bajuni words but Sprakab went further and noted that the appellant appeared to be putting on an accent (see paragraph 138). Ms Kumbuka did not have the degree of familiarity with the language to note such things. Sprakab did. Many of us who have tried to amuse ourselves and others by assuming an accent for the purpose of telling an anecdote (for example) will understand exactly how the assumed accent can so easily slip away and let the natural accent reveal itself. Unusual words can be learned. Distinctive speech patterns are very much harder to copy and even harder to abandon.
162. We do not think that Sprakab made up this fault by the appellant but identified it because she could not keep up a pretence.
163. Ms Kumbuka speculated that the appellant had learned phrases from mixing with Kenyans but Sprakab indicated that the appellant would not have had much contact with Kenyans if she had lived where she claimed to have lived.
164. Taken as a whole we find Sprakab’s approach to be altogether superior to that adopted by Ms Kumbuka and we regard Sprakab’s conclusions as much more reliable than hers.
165. Professor Nurse criticises Sprakab’s methods in relation to the cases he has examined, but for two reasons it does not appear to us that these criticisms are sufficient to cause doubt about Sprakab’s process or its expertise as a whole. In the first place although Professor Nurse found that the interviews were sometimes brief, sometimes over unsatisfactory telephone lines, and sometimes used questions of a form or in a language he did not approve, there is no proper basis for saying that in those cases, or because of those alleged defects, the opinion given by Sprakab was in fact wrong. Some cases will be clearer than others and will require less investigation before an opinion is reached.
166. Secondly, Professor Nurse’s criticisms do not purport to be an impartial survey of Sprakab reports and conclusions as a whole. He says that the 50 reports forming the basis of his ‘Overview of Sprakab linguistic analyses of Bajuni refugee claims 2004-2010’ all included a conclusion that the applicant ‘comes not from their alleged place but from some other place’. His ‘Overview’ therefore relates to negative reports only. We do not know in what proportion of claimed Bajuni cases Sprakab supports the applicant, but it seems to us that any valuable criticism of Sprakab’s methods as a whole would need to take into account its positive as well as its negative reports. Of course, it might be the case that all Sprakab’s positive reports were reliable and all its negative reports unreliable, but that would seem inherently unlikely.
167. It might be said (although it was not said in the present appeal) that Sprakab reports should be ignored or treated as of less weight because they do not comply with Practice Directions relating to expert evidence. We do not think that would be right. In the first place the reports do indicate, within the limits of the anonymity we have held to be justifiable, the identity and qualification of the writers. Secondly, unlike many “expert reports”, they come in essence from an organisation rather than an individual, and it is the characteristics and methods of the organisation that are of importance in this context. Thirdly, they are typically prepared for a decision-maker and not for an appeal, and so it is not appropriate to impose on them rules relating to evidence prepared specifically for use in litigation.
168. It seems to us that we have been given no substantive reasons for distrusting Sprakab’s reports either in this case or in general. In our judgement, because of Sprakab’s underlying library of data and the process by which it produces its reports, Sprakab evidence is of high quality and its opinions are entitled to very considerable weight.
169. This is particularly so when there is available such a detailed report as that of 15 February 2010 in the present case; but even where less detail is available Sprakab’s reasoned opinion should be regarded as carrying weight because it comes from a reputable and apparently reliable body with the characteristics and methodology we have described. Any evidence opposing the Sprakab evidence will need to deal with the reasons Sprakab has given for its view; and it seems to us highly unlikely that an opposing report based purely on anecdotal reasoning or personal opinion will give a proper basis for rejecting the Sprakab opinion.
General guidance on linguistic analysis evidence
170. We close this determination with three matters of general guidance in relation to appeals based on linguistic analysis in general and Sprakab reports in particular.
171. First, we note that it is said that a decision as to a person’s background or origin should not be based solely on linguistic analysis. We have heard and seen nothing enabling us either to endorse or doubt that advice. But where there is clear, detailed and reasoned linguistic analysis leading to an opinion expressed in terms of certainty or near-certainty it seems to us that little more will be required to justify a conclusion on whether an applicant or appellant has the history claimed.
172. Secondly, the conclusions we have reached about Sprakab’s reports do not, of course, mean that Sprakab or any other linguistic analyst is infallible. A decision-maker or judge must be alive to the possibility of error, whether or not the particular level of certainty expressed by the report leads one to expect it. Where there is linguistic evidence in a particular case it is important that all parties have a proper opportunity to submit it for expert assessment, and it is equally important that all the evidence be taken into account in deciding the questions in issue according to the appropriate standard of proof.
173. The parties must have an opportunity to challenge any linguistic assessment opposing them. That means that a sound recording of any interview of or discussion with an appellant that forms the basis of such analysis must be made available to the other party in good time before any substantive appeal hearing. In some of the early cases in which the Secretary of State relied on Sprakab reports the Tribunal had to direct that the CD be served on the appellant’s representatives. We are glad to know that such service is now routine. But it seems to us that the principles apply to both sides. In the present case we have not taken any adverse account of the fact that the appellant’s conversation with Ms Kumbuka was not recorded. But it should have been, and the recording should have been made available to the respondent. We would expect for the future that where linguistic analysis is in issue, no party should seek to rely on an analysis based on examples of the appellant’s speech that all parties have not had the opportunity to analyse.
174. Thirdly, we have given our reasons above for acceding to Sprakab’s request for anonymity for its linguists and analysts, subject to details being given of their background and qualifications. Those reasons are of general applicability and we would expect the same decision as to anonymity to be made in any case in which Sprakab evidence was tendered, unless there was some very good reason for departing from this practice.
175. For the reasons we have given we have no hesitation in concluding that the appellant is not an island Bajuni. Her claims have no basis in credible evidence. Her appeal is dismissed on all grounds.

Vice President of the Upper Tribunal,
(Immigration and Asylum Chamber)