Linguistic issues in the evaluation of international asylum petitions

This talk surveys issues that arise as part of political asylum petitions in the fine-grained identification of individuals’ provenance based on their speech. Western European and North American governments receive vast numbers of applications for political asylum every year. Since members of persecuted minorities from war-torn countries are more likely than others to be awarded asylum, it is common for individuals to claim that they belong to such a group even when they do not (though they often belong to a closely related group, or the same group but in a neighboring and less embattled country). For example, Armenians from the Republic of Armenia seeking refuge in the Netherlands sometimes claim that they are from the Armenian enclave of Nagorno Karabagh inside Azerbaijan, which was ravaged by ethnic civil war from 1991-1994 and most recently in 2016 and therefore qualifies for special dispensation with regard to Holland’s immigration policies. (The Republic of Armenia, which is relatively peaceful, receives no such dispensation.) How can a government verify where an applicant for asylum is actually from? To date most western governments have employed native speakers of the relevant languages to interview the applicants and assess their place of origin to the best of their abilities. Unfortunately, the arguments and methods used by these evaluators are often unjustified, inappropriate, or misleading, with dire consequences for the applicants. Armenians claiming to be from Nagorno Karabagh, for example, are typically interviewed in Russian or Standard Eastern Armenian, and no effort is made to verify whether or not they can speak the Karabagh dialect, which is controlled by almost all Karabagh Armenians but extremely difficult for Armenians from outside the enclave to understand or emulate. Similarly, government evaluators frequently justify rejecting applications with arguments such as “the applicant used form hishum chem ‘I forget’, which is used in northern Armenia, therefore they are from northern Armenia, not Karabagh”, or conversely “the applicant does not use forms found only in Karabagh, therefore they are not from Karabagh”. Neither of these types of argument actually demonstrates that the applicant is not from Karabagh, as (i) the use of a form in region A (e.g. northern Armenia) does not imply that it is not also used in region B (e.g. Karabagh), (ii) it is physically impossible for an individual to produce every form characteristic of a region in an interview of finite length, and (iii) no human invariably uses every dialect form characteristic of his/her native region, especially in a formal setting where there is pressure to use standard/educated forms. In this talk I survey the methods and arguments employed by government-hired language experts (typically teachers of the language in question or interpreters, but sometimes random bilingual speakers), and examine which of these work and which do not. I then suggest alternative techniques based on work in dialectology and linguistics that produce more reliable evaluations and reduce the risk of legitimate refugees being unjustly deported.