1.1.1 Brahui is primarily spoken in Balochistan Province of Pakistan, in a belt running through the Brahui Hills from near Quetta through Kalat and up to Las Bela, and in adjacent areas of Afghanistan and in pockets in eastern Iran, as far as the Marv oasis in Turkmenistan. Today, it is also spoken in Quetta, Karachi, and most of the cities of Sind Province (Elfenbein 1997b: 797; 1998: 388-89). Brahui is nowhere dominant and is always embedded in another language, usually Balochi, but any language of the area is possible. Multilingualism is the norm. See Elfenbein (1990) for ethnography and history with an annotated bibliography. The Brahuis are traditionally transhumant pastoralists who move with their flocks from the hills in summer to the plains of Sind (or elsewhere) in winter, and back again. They are also agriculturalists, and often combine or exchange the two lifestyles. The land they inhabit is semiarid and typical of the Iranian plateau. Brahuis traditionally straddle a cultural divide, spending the summer on the Iranian plateau and the winter in South Asia.
1.1.2 The Brahuis are members of a tribal confederation, usually the Brahui Confederation, but other Balochi confederations occur as well. Since the term Brahui can refer to a speaker of the language or a member of the Brahui Confederation, and the two are by no means coterminous, there is great confusion in the population counts. Reasonable estimates are from 4,000,000 in Pakistan, 200,000 in Afghanistan, and 20,000 in Iran (Lewis and Simmons 2013) to a tenth of that (Elfenbein 1998: 388-89). Similar problems of terminology are found in dialect numbers and boundaries.
1.1.3 Dialect variation is not major in Brahui and is limited to a few phonological and grammatical markers. Terminology is fluid, and on the face of it, details can appear contradictory; note pronunciation of /h/ in Bashir (1991: 4) and Elfenbein (1997b: 798). This work will follow Bashir (1991: 4; 2003: 3), which has more details. Traditionally, only two documented subdialects are based solely on location: Kalati (a subdialect of Sarawan) and Nushki (a subdialect of Chagi). Kalat was the capital of the former Khanate, and Kalati is the standard dialect. Brahui's major dialects are based on traditionally migrating groups, consisting of Sarawan, north and east of Kalat (using the Bolan Pass to the Kacchi plain), and Jhalawan. south of Kalat (using the Mula Pass to lower Sind). To this, Bashir (1991: 4) adds Chagi in the west of Balochistan and in Iran.
1.1.4 While there are a small number of written works including poetry, Brahui is not commonly written. Elfenbein (1998: 391) confirmed Bray's estimate that there were fewer than 6,500 literates in Brahui and probably never have been more, but more recent educational efforts may have increased this. Brahui lacks a standard script, and none of the variants of the Perso-Arabic script is satisfactory. The standard grammar (Bray 1909), excellent for its period, uses a Latin script, as does an updated grammar (Andronov 2006). While minimally written, Brahui is sophisticated, with extensive borrowings from Persian and more recently from Urdu in Pakistan. It was a minor language of the court of the Khan of Kalat.
1.1.5 Notwithstanding the previous paragraph, Brahui has a literary (and poetic) style with differences from everyday usage. Andronov (2001) has considerable discussion and examples of poetic usage. Urbanization is causing another split in style between urban and rural dialects. The urban dialects have more complex verb forms, particularly for complex aspects, and innovate rapidly; see Bashir (2010). Rural dialects are more conservative.
1.2.1 Elamite was spoken in what is now southern Iran, in the Zagros Mountains and adjacent plains. Elam was essentially contemporaneous with neighboring Sumer in the development of civilization and writing, usually following by a generation or two, but sometimes leading. Writing began in Elam around 2,900 B.C. out of the same use of tokens and bullae that were used in Sumer. The oldest script, Proto-Elamite, is quite distinct and separate from Sumerian. It has never been deciphered, and it is only by association that we believe it to be Elamite. This may be mistaken. Another innovation out of this tradition, called Linear Elamite, came into use around 2,500 B.C. and was used for several centuries. It has a single bilingual and a very tentative reading in Elamite. Around 2,300 B.C. the Sumero-Akkadian cuneiform script was borrowed intact from southern Mesopotamia. It is a lexico-syllabic system with complex rules for interpretation. Fortunately, Elamite largely ignored or discarded most of this and is mostly a syllabary based on (C)V(C) syllables. This formed the basis for all further writing in Elamite and is well understood.
1.2.2 Cuneiform Elamite can be divided into four periods or dialects. The first, Old Elamite (OE), was used in a few documents from Susa from the third millennium and early part of the second millennium B.c. After a long gap in the attested record with documents in Akkadian, the reemerging Elamite kingdom used Middle Elamite (ME) during the second half of the second millennium B.c. This was centered on Susa in the plains next to Mesopotamia. It is attested primarily on inscribed bricks, which are numerous, but very repetitive. In the first part of the next millennium B.C. another variety, called New Elamite (NE), was in use around Susa in the plains. A major dialect change is found in Achaemenid Elamite (AE), where Elamite functioned as the major administrative language of the Achaemenid Persian Empire, whose ceremonial capital was at Persepolis in the mountains. We now know that the site of Anshan, the second great center of Elam (besides Susa) is at Malyan, not far from Persepolis. There are two usages in AE, the royal inscriptions and bureaucratic records. The first are public documents, often bi- and trilingual, and include the great trilingual inscription of Darius I at Bisitun (DB). The administrative records range from periodic summaries (like journal entries), through the common detailed record tablets (pay and warehouse receipts and disbursements), to temporary finger texts (like sticky notes). The last Elamite inscription comes from the last years of the Achaemenid Empire.
1.2.3 The best, and readily available, overview of the Elamite language is Stolper (2004), which summarizes and cites all of the sources. The most detailed grammar with extensive examples is Khachikian (1998). Other reliable sources are Diakonoff (1967) and Grillot-Susini (1987). Grillot-Susini (1998) also gives an excellent short summary of the grammar. There is no standard grammar. The standard source for archeology and history is Potts (1999).
2 ZAGROS IAN HYPOTHESIS
2.1.1 The underlying working concept for this paper is that the Proto-Elamo-Dravidian group, which I have renamed Proto-Zagrosian, split into Elamitic and Dravidian subgroups, and that Brahui belongs to the Elamitic subgroup. Starting with the tree diagram in McAlpin (2003: 543), I have relabeled PDr2 as Proto-Peninsular-Dravidian (PPD), PDr, as Proto-Dravidian (PDr), and PDr0 as Proto-Zagrosian (PZ), following an earlier suggestion in McAlpin (1975: 109). Proto-Zagrosian then adds Proto-Elamitic (PEI), which splits into Elamite and Brahui branches; see Figure 1. In other words, Proto-Elamitic has been added to PDr0 and Brahui has been attached to it, completing the migration of Brahui from North-Dravidian. My hypothesis is that Brahui is Elamitic and not properly Dravidian, but still cognate. This is the primary point of this paper.
3 PHONOLOGICAL ENVIRONMENT
3.1.1 All of the languages involved use variations on the same vowel structure, which is seen in detail in PPD. There are three primary vowels in all positions, /i, a, u/, with two more, /e, o/, that have a more restricted basis and may be lost in Elamitic. The root structure is basically initial (C)V(C) with stems optionally adding +V((C)C). Stems regularly show contrasting vowel length while roots do not; i.e., roots can have derived stems with both long and short vowels. There are patterned vowel losses in Elamitic while Dravidian stem vowels tend to be very stable. Stress tends to be on the stem, first or second vowel, but in Brahui it may be lexically marked on any vowel. This pattern of stem stress matches what little can be inferred on stress from Elamite; see Khachikian (1998: 10).
3.2.1 Consonant clusters are restricted, with no initial clusters in Proto-Dravidian. Brahui has developed initial clusters due to vowel loss, i.e., #CVC > #CC, and borrowing. The cuneiform script makes initial clusters difficult to determine for Elamite; i.e., Elamite shows no initial clusters, but that could be due to either script or phonology. The noninitial clusters are normally homorganic due to assimilation. There are regularly four primary points of articulation: labial stops, coronal stops, alveopalatal affricates, and dorsal stops. Coronals and dorsals both tend to have a three-way split, usually involving coarticulated tongue-root features: coronals into dentals, alveolars, and postalveolars (retroflex for Dravidian), while dorsals have palatal, velar, and uvular stops. There are regular and common simplifications and mergers from this underlying arrangement. The result is patterns with five to seven contrasting points of articulation. There are bilabial and coronal nasals, with homorganic variants with all the obstruants. There tend to be multiple liquids. Semivowels are restricted to /w/ and /y/, and pattern as consonants. Dravidian languages tend to have no diphthongs or vowel clusters; Brahui has them; Elamite does not. Brahui and Elamite have /h/. Details of the phonemes for each language are given in Appendix 1.
3.3 Special Phonological Patterns
3.3.1 The coronal and dorsal obstruents commonly...