A grammar for Balochi: challenges of describing a language that maybe is not one

International audienceWhile it has generally been acknowledged that Balochi (an Indo-European language of the Iranic branch) is historically one language, the synchronic variation of the dialects, spread over large areas in Iran, Pakistan, Afghanistan etc., is bewildering. There are three major dialect groups (S, W, E), plus one outlier: Koroshi, spoken in Fars province (historically an offshoot of Southern Balochi). The differences are found on all levels of the grammar, and tend to be given in lists of what the respective author regards as salient criteria. Furthermore, the fact of being spoken in different countries exposes the dialects to superstrate influences not coinciding with dialect boundaries. These differences are not usually mentioned in the (few) works on Balochi dialectology. For instance, any Bal. dialect spoken in Pakistan is under the influence of the national languages Urdu and Pakistani English; depending on the region, Brahui, Pashto, Sindhi and other Indo-Aryan languages also play a role. In Iran, the influence of Persian is strong, now ever more so due to TV etc. In Turkmenistan, Balochi encounters Russian and Turkmen while the varieties in Oman are under heavy pressure from Arabic etc. However, for speakers of Balochi, the most important obstacle for communication is the lexicon, a major part of which comes from Persian in Iran and from Urdu and Pakistani English in Pakistan, thus not comprehensible to Baloch from the other country. As for other levels of grammar, the Balochi phonemic system is of an Indo-Aryan type (having merged the inherited fricatives with stops, and acquired retroflexes). In Pakistan, IA influence may help to preserve ergativity and non-finite subordination, while Persian influence results in the use of prepositions and contributes to a weakening of ergativity and to the adoption of modal and aktionsart constructions calqued from Persian, as well as (in recent decades) in loanwords pronounced with fricatives instead of replacing these by stops. Persian influence also results in a major refashioning of the case system in some of Balochi spoken in Iran, while Balochi spoken in Turkmenistan has elaborated a combination of endings into a regular locative case, to the concept of which contact with Turkmen and Russian could have contributed. Another non-trivial problem are the major discrepancies in the available data and descriptions. The only works summarising what was known about Balochi at the time are the substantial chapters by GEIGER 1901 (in the Grundriss der iranischen Philologie) and GRIERSON 1921 (in the Linguistic Survey of India), who had access only to Balochi spoken in British India, and JAHANI & KORN 2009, who include the by then known data from Iran, but not Koroshi, which has been investigated only recently. Conversely, available data from Eastern Balochi chiefly consists of material collected in the 19th century, noted according to the standards of the time -- owing to the political situation, access to the area has been difficult or impossible for foreigners. The only group which is rather well described is Western Balochi, but exactly these varieties are somewhat less interesting from a historical and typological perspective, as they pattern heavily on Persian. The question thus arises how to write a grammar of Balochi, or whether one should refrain from doing so at all and rather consider writing separate grammars for some six (or more) forms of Balochi. -- In this paper, I present thoughts and problems from my work-in-progress on a Balochi grammar. For the sections where this seems feasible, I apply a diachronic approach, starting from Proto-Balochi, from which the attested dialects are derived. I also make use of the concept of Common Balochi, which is an abstract entity describing traits that Bal. varieties typically show