A contrastive (or antithetical) construction which makes simultaneous use of two separate particles is identified through a mainly corpus-based study as a typical feature of a number of lesser-described languages spoken in the Afghanistan-Pakistan borderland in the high Hindukush. The feature encompasses Nuristani languages (Waigali, Kati) as well as the Indo-Aryan languages found in their close vicinity (Palula, Kalasha, Dameli, Gawri), while it is not shared by more closely related Indo-Aryan languages spoken outside of this geographically delimited area. Due to a striking (although not complete) overlap with at least two other (unrelated) structural features, pronominal kinship suffixes and retroflex vowels, we suggest that a linguistic and cultural diffusion zone of considerable age is centred in the mountainous Nuristan-Kunar-Panjkora area.
This study is an attempt at generalizing patterns that were first observed when investigating a few items of exceptionally high text frequency in Indo-Aryan Palula [phl], a language spoken in a mountainous area in northern Pakistan bordering Afghanistan. The textual occurrence of two form-identical items, with remarkably similar functions and distributions, in neighbouring Indo-Aryan Dameli [dml], and what seemed like functional and distributional equivalents in two other Indo-Aryan languages, Kalasha [kls] and Gawri [gwc], also spoken in the area, further spurred the interest. It was found that each of the four languages makes a fairly consistent use of two clearly different particles to explicitly signal a paired opposition of the type: I went out, but he stayed at home, in which I is postposed by one of the particles (were it to be used in English), while he is postposed by the other. In (1) we see an example from Palula of a fairly typical use of such a construction.
It was further noted, when consulting grammars and other linguists, that at least two of the Nuristani languages, Waigali [wbk] and Kati [bsh]/[xvi], spoken in the vicinity of the aforementioned four, employ the same bisyndetic marking strategy, while constructions with similar functions in other languages of the surrounding region, are either monosyndetic (i.e., they use a single particle) or asyndetic (they do not use any overt marking at all). As the marking strategy itself is of an unusual type, this suggests that the present geographical distribution is the result of significant language contact rather than a common origin, a hypothesis further strengthened by the co-occurrence of other structural features in roughly the same geographical area, and their non-occurrence in closely related languages spoken further afield.
In Section 2, the area of diffusion is defined and briefly described along with its place in the wider linguistic context surrounding it. In Section 3, we outline the scope and method of our present investigation, and in Section 4 we describe and further exemplify the (antithetical) construction as it is found in the languages of this particular area. Along with it, some suggestions as to the origin and subsequent development of the individual forms are provided. This is followed, in Section 5, by a discussion of the possible areal implications of the findings, taking a number of other languages and two other linguistic features into account. In Section 6, we point to possible historical and cultural reasons behind the observed linguistic areality. In Section 7, we finally summarize the conclusions drawn and offer suggestions for further research.
2 The Central Hindukush Area and the Surrounding Region
While carrying out his pioneering research in the 1920s in the languages of present-day northwestern Pakistan, the Norwegian linguist Georg Morgenstierne remarked that “Lower Chitral is one of the most polyglot [sic] regions of Asia” (Morgenstierne, 1941: 7). It is indeed an area characterized by a high degree of multilingualism as well as of cultural diversity. Geographically, it is a land of high mountains and deep valleys, with a fair share of evergreen forests and pasture land. This is all equally true of the area just across the Pakistan-Afghanistan border. Our current focus area (see Map 1) comprises most of Nuristan Province (Afghanistan), the northeastern part of Kunar Province (Afghanistan), the southern-most part of Badakhshan Province (Afghanistan), the southern-most part of Chitral District (Pakistan) and most of Upper Dir District (Pakistan), both of the latter administrative divisions of Khyber Pakhtunkhwa Province. In this area, which we will refer to as the Central Hindukush Area (or chka) henceforth, 15–20 distinct languages with a reasonably long history in the area are spoken as first languages.1 The great majority of them belong to two different linguistic genera, i.e., Nuristani and Indo-Aryan, and a few (but nevertheless very influential) languages are Iranian.
Although the exact relationship between Indo-Aryan and Nuristani on the one hand, and Iranian and Nuristani on the other, still remains to be settled, suffice it to say that modern-day scholars almost unanimously consider these three as distinct branches within the larger Indo-Iranian branch of Indo-European (Degener, 2002; Strand, 1973: 298; Zoller, 2005: 13–15).
The Indo-Aryan languages of the region were often (and are still) lumped together under the grab-bag term “Dardic”, although it is highly questionable whether they collectively form a sub-branch of Indo-Aryan vis-à-vis other Indo-Aryan languages such as Sindhi or Lahnda (Bashir, 2003: 822; Morgenstierne, 1961: 139; Strand, 2001: 251). There is on the other hand a considerably higher level of consensus over the classification of these individual languages or varieties as belonging to lower-level groupings. Mainly following Bashir (2003: 824–825), the following groupings have been identified (the individual languages represented in the focus area in italics; individual languages outside the area but subject to regional comparison in this study in normal type): Pashai; Kunar Group: Dameli, Gawarbati; Chitral Group: Kalasha, Khowar; Kohistan Group: Gawri,2 Torwali, Indus Kohistani; Shina: Palula, Kalkoti, Sawi, Kohistani Shina, Gilgiti Shina; Kashmiri.
The Shina language enclaves in the focus area are relative newcomers. According to local history, the forefathers of today’s Palula speakers (possibly including closely related Sawi and Kalkoti) migrated from the Indus Valley (to the east of the focus area) a few centuries ago and settled in the area (Cacopardo and Cacopardo, 2001: 84–93; Liljegren, 2009: 54–58). Even Khowar, the contemporary lingua franca of Chitral District, has expanded into the lands south of Chitral Town, from their original homeland in the northern part of Chitral, in a relatively recent past (Morgenstierne, 1932: 46–51; Strand, 2001: 252), mainly at the expense of its sister language Kalasha, the latter whose speaker community of today is but a remnant of its past extension (Morgenstierne, 1973a: 188–190). Although nothing can be stated for certain about the presence of the languages themselves or the ultimate origin of the speakers’ ancestors, it seems plausible that the Dameli and Gawarbati communities, both treated as part of the little studied Kunar Group, have had a presence in the area well predating both the arrival of the ancestors of the Palula speakers and the Khowar speakers of southern-most Chitral (Cacopardo and Cacopardo, 2001: 168–171, 241–246; Strand, 2001: 252). The present homeland of Gawri is the upper reaches of the Panjkora and Swat Valleys, but historical evidence suggests that they were centred around the confluence of Dir and Panjkora only a few centuries ago, i.e., before a dramatic Yuzufzai Pashtun expansion took place in Dir and lower Swat (Morgenstierne, 1932: 28–29; Strand, 2001: 254). All other Kohistani languages, however, are spoken well outside of our focus area. Pashai, in the form of a number of distinct varieties and subvarieties (essentially “languages” and “dialects”), is spoken just south of the focus area and is therefore not included here, although it might be argued that it has a natural place in a diffusion zone common to Nuristan and the adjacent parts of Kunar and Panjkora Valleys. Kashmiri is due to its geographical location entirely outside the scope of the present study.
The Nuristani languages are spoken almost entirely within the limits of this area. Linguists (Degener, 2002: 104; Strand, 2001: 253) hold that there are five distinct Nuristani languages: Kati (or Kamkata-viri), Prasun (or Vasi-vari), Waigali (or Kalasha-ala), Tregami, and Ashkun (or Ashkunu-Sanu-viri).3
While Pashto and Dari (i.e., Afghan Persian) are languages of wider communication in the regions to the south and northwest, and as such have made inroads quite recently (and, at least in the case of Pashto, largely by means of aggression) in the more central parts of the focus area,4 the only truly indigenous Iranian language is Munji, spoken in the Munjan Valley to the north of the Nuristani speaking locations. However, interaction between the Munjan population and those of Nuristan or Chitral has probably been much more infrequent and less intimate than that with Persian-speaking Badakhshan to the north (Morgenstierne, 1938: 9–10). A closely related language, Yidgha, is spoken to the northeast of the main Munji area, across the mountains, in Lutkuh valley in western-most Chitral. Munji and Yidgha belong to a Pamir group of Eastern Iranian, most of whose other members are found in the valleys of western and southern Pamir, primarily in Tajikistan (Èdel’man and Dodykhudoeva, 2009: 773).5
A few important suggestions have prior to our own study been offered with regard to contact-induced areality (or rather sub-areality) in the Central Hindukush Area (Mørch, 1997: 106–119; Heegård and Mørch, 2004: 67–73; Di Carlo, 2011). We will have reason to return to those suggestions in our discussion in Section 5.
Widening the geographical scope, the larger region immediately surrounding the chka is equally multiethnic and multilingual. The mountainous region (elsewhere referred to as the Greater Hindukush (Liljegren, 2014)) covering northeastern Afghanistan, northern Pakistan and the disputed territories of Kashmir, counts around 50 distinct languages belonging to six different genera: Indo-Aryan, Iranian, Nuristani, Tibeto-Burman, Turkic, and the isolate Burushaski. While attempts have been made in the last half-century to define this particular region (or something approximating it) in areal-typological terms (Bashir, 1988: 385–421; 1996a; 1996b; 2003: 821–823; Baart, 2003; Èdel’man, 1980; 1983; Fussman, 1972; Tikkanen, 1999; 2008; Toporov, 1970) or in more general sociopolitical or cultural terms (Cacopardo and Cacopardo, 2001; Jettmar, 2002: 9–44), the overall picture remains one of typological diversity rather than unity, and it may be more helpful to focus on its transitional characteristics, situated as it is at the crossroads between the Indian Subcontinent, the steppes of Central Asia, and the Himalayan mountain system, linguistically as well as culturally.
3 Scope and Method of Investigation
In six languages of the Central Hindukush Area, viz., Palula, Dameli, Gawri, Kalasha, Waigali and Kati, there is evidence of what we will refer to as bisyndetic antithesis or a bisyndetic antithetical construction, here abbreviated ba. Functionally, the construction is used to contrast two opposed pairs or entities in many different discursive contexts, and as such it corresponds to similar and widely occurring “antithetical sentences” in other languages, in English normally expressed with but as a medial antithetical link (Longacre, 1996: 55–59; 2007: 378). Formally, what we have in our six languages, is an example of (emphatic) adversative bisyndetic coordination (Haspelmath, 2004: 4–6), i.e., a structure consisting of two contrasted items (each referring to an entity or a situation), usually occurring in two separate clauses, and each followed by a different particle or “correlative coordinator” (Haspelmath, 2007: 15–16).6 The two particles are also used each alone, often with various discourse functions, such as topic or focus marking.7 Importantly, the ba construction appears to be very frequently used in these languages, and where-ever it can be ascertained, the two particles are high-frequency items. Outside of the chka, the bisyndetic marking type either does not occur with adversative meaning, or, if it occurs, it occurs very infrequently, (see Section 4), although particles with functions related to the two particles, as used individually, are found in most languages of the Greater Hindukush.
The material used in our initial investigation mainly consists of annotated interlinear texts recorded during recent linguistic fieldwork in the region. Apart from Henrik Liljegren’s annotated Palula material, we were kindly supplied with texts by Emil Perder (Dameli), Ronald Trail (Kalasha) and Joan Baart (Gawri). These corpora, which were in the format used by Field Linguist’s Toolbox by sil International, consist primarily of oral narratives, but there are also dialogues, proverbs, as well as some originally written texts by native speakers. The number of words and different texts is shown in Table 1. For Nuristani Waigali, we only have access to a reference grammar by Degener (1998), although it is very extensive with ample descriptions and examples of discourse features, and for Nuristani Kati we rely on examples provided by Richard Strand in personal communication.8
The Field Linguist’s Toolbox does not have advanced searching capabilities (e.g., regular expressions), so the first step was to convert the data into a more usable format. This was accomplished by programming a Python application, which simply turned the material into one continuous spreadsheet for each language by omitting all information except the original sentences, the glossing, as well as the translations. Besides allowing for the use of advanced search functions of modern spreadsheet software, it also made the data much easier to scroll through and gain an overview of. Perhaps most importantly, it made it much easier to investigate sentences in their wider contexts (i.e., with the surrounding utterances actually viewable on the same page). This method was used in Svärd (2014), where this script was one of three used to analyze discourse particles (specifically those participating in ba constructions) in Palula, Dameli, Gawri, and Kalasha.
4 The Bisyndetic Antithetical Construction in the Languages of the Central Hindukush Area
What we found was that the ba construction in these languages (referred to in the previous section) is utilized to contrast two referents or two situations. This is done by placing two particles following two topicalized nominals, most typically nouns, pronouns, or determiners. The construction is often used in discourse to contrast two parts of previously mentioned whole or grouping, although it can also be used with new referents; it is typically found in narratives and proverbs. One example in Palula is shown in (2), where two sons are introduced and subsequently contrasted using a ba, in this case marked by ta and ba, glossed as atm1 (first antithetical marker) and atm2,9 respectively.
A similar example from Dameli is shown in (3), where a part-whole relation is stated explicitly (via maana ‘of them’), or implicitly as in the Waigali example in (4) in the context of traditional wine making.
However, the ba construction is not only used to compare parts or members of a previously mentioned whole or group. In the Kalasha example in (5), the particles ta and o are used to introduce two characters in the beginning of a narrative.
Following this pattern of contrast, it can be used, e.g., by a referent to contrast with the addressee, as in the Gawri example in (6).
Similar to this, in (7), an example from Kati, the high status of a third person is contrasted with the low status of the speaker.11
The ba construction can also be used to contrast two consecutive actions, as in the Waigali example in (8) where separation of the two actions is stressed, i.e., that it is not done at once, but in two steps.
Typically, at least as far as Palula is concerned, this is also a construction occurring frequently in local proverbs (Haider, 2012; Liljegren and Haider, 2015: 207–222), as in (9)–(11).
As the above examples show, the ba constructions in these languages are versatile with the same core function of contrasting two referents.
However, the ba constructions in the different languages have another interesting feature in common: each of the two particles involved can be preceded with a negation, Palula/Waigali na, Dameli ni, Gawri nä, and Kalasha ne (for Kati we do not have any comparable example of this), but in this case the referent, which normally precedes the particle, instead follows it. Functionally, it results in an emphatic negative coordination (Haspelmath, 2007: 17–19), a construction which Longacre terms a negated extremes paraphrase (1996: 79), where neither of the stated values or propositions holds true. Examples from Dameli and Waigali can be seen in (12) and (13), respectively.
Thus, the ba constructions of these different languages are functionally very similar, with the basic discursive function of contrasting two referents, for instance two different persons in a group, two parts of a whole, two separate participants in a narrative, the hearer versus a third person, the speaker versus a third person, two consecutive actions, or two situations that somehow exclude one another.
As evident from the examples, the ba constructions in all six languages are bisyndetic constructions consisting of two clearly different particles (see Table 2). At least some of these particles are also clearly related in form across the languages, with the most obvious similarity being the identical particles ta and ba in Palula and Dameli. In fact, the first particle in all of these languages is very similar, consisting of an alveolar stop and an open front or central unrounded vowel. With Palula, Dameli, and Kalasha ta, the similarities are immediately apparent, but Gawri tä is actually also equivalent since <ä> represents /a/ (in contrast to <a>, which stands for the phoneme /ɑ/). Waigali da and Kati de are somewhat different with a voiced stop, but it is not far-fetched to presume that they too are related.
Although we believe it to be very likely that the construction also occurs in the other Nuristani languages in the area, more research, and access to comparable textual data or grammatical descriptions, is needed to determine that for certain. The situation is the same when it comes to Gawarbati and languages closely related to it; the available data is not enough to draw any conclusions. As for Shina-related languages other than Palula, spoken within the chka, we likewise lack evidence to say anything for sure about the presence or absence of bisyndetic marking. For Kalkoti (spoken in Dir Kohistan, in close proximity to Gawri), there is simply too little text data available (Liljegren, 2013), and for Sawi (spoken in the Kunar Valley, not far from Nuristan), we find no evidence in the data at hand, which again might be related to the relatively small corpus of annotated data. Curiously, a particle bo (that might be related to Palula ba, while functionally non-overlapping) is used alone as a clause-final subordinator in Sawi conditional sentences,12 while a cognate of ta seems to be missing altogether; instead, subsequence is marked by a (y)e, cliticized to the dependent predicate (Buddruss, 1967: 59–61).
As argued in Svärd (2014: 54), the first particles of the antithetical constructions are probably cognates and derivatives of what in Sanskrit is reflected as tá-, occurring in words such as tátas ‘from there; there; then’, tátha ‘so, in that way’, tadaˊ ‘then’, and taˊt 13 ‘thus, then’ (Turner, 1966: 5639, 5646, 5649, 5753). The Sanskrit tá- is in turn derived from Proto-Indo-European *tod ‘this, that’ (neuter nominative of the pronoun *so ‘this, that’). This further means that it is related to words such as English then. Cognates of this particle is found in many Indo-European languages, and it is likely that its subsequential meaning expanded to include a contrastive or topical meaning as well (Svärd, 2014: 54). As for the second particles, no clear etymologies have been found, not least due to the variety of forms.
In all six of the languages where we have good evidence of a bisyndetic antithetical construction of the kind described above, the two particles also occur individually. Even in those uses, the textual frequency rates are exceptionally high. E.g., in both Palula and Dameli, ba and ta came out as frequency-wise top-ranked items in our investigation. The particles are very common in Gawri and Kalasha, too, although the corpora are too small to adequately quantify their rate of occurrence. For Waigali or Kati, we do not have access to any text corpora, the reason why we cannot say anything about their relative discourse frequency.
As a separate particle, ta/tä/da functions as either a topic or a subsequence particle (i.e., ‘then’), as can be seen in the Kalasha example in (14) where ta emphasizes the preceding topic ía ‘this one’ (i.e., the girl).
One of the core functions of Gawri tä is to contradict or challenge the listener’s presuppositions or add something new to the listener’s knowledge base (Baart, 1999: 160). Waigali da, which Degener (1998: 168) terms a “Fokuspartikel” (focus particle), highlights the immediately preceding constituent and indicates a contrast, although it often remains somewhat unclear with what it is being contrasted. Palula ta is only rarely used alone with the same distribution as Gawri tä or Waigali da; instead it is the most common clause-final subordination particle in the language, and as such mainly signalling subsequence (Liljegren, 2008: 312–314),14 thus semantically as well as syntactically corresponding to Gawri to (Baart, 1999: 149–151). While some of the functions of ta in Dameli await a more thorough analysis, it seems beyond doubt that signalling subsequence is an important aspect when occurring immediately following verbs.
The second-place particles of the ba construction are much less obviously form-related. Firstly, Palula and Dameli ba are most certainly related, probably as a borrowing from Dameli to Palula, both of which may in their turn owe something to the similarly shaped particle be in Waigali (Degener, 1998: 167–168), possibly with cognates in other Nuristani languages. The remaining three, Kalasha o, Gawri i, and Waigali ri, are not obviously related in a formal sense. Nevertheless, all of them function as topic (or switch-topic) particles when not part of the ba construction, as can be seen with Kalasha o in example (15).
While Gawri tä, as we saw above, mainly signals new information, the function of the other particle, i, is rather to remind the listener of something already known,15 but it “may also present new information about the topic”, although not in an unexpected way but rather as a conclusion from the preceding context (Baart, 1999: 161). In a strikingly similar manner, Waigali ri (classified by Degener as a coordinating conjunction) serves to connect different entities with one another, while at the same time signalling some sort of change. It is also used to clarify something about a newly introduced entity, as such corresponding to ‘as for’ (Degener, 1998: 162–163). These uses in Gawri and in Waigali correspond very closely to some of the many functions of Palula ba (Liljegren, 2008: 377–383), cf., for instance the first ba, glossed as top in (2), whereas, interestingly, some other functions of ba clearly overlap with the Waigali “Abtönungspartikel” (modification particle) be, such as expressing reproach or assuming an agreement with the listener that something is likely and thereby also excluding any potential contradictions, similar to some uses of English ‘however’ or German doch (Degener, 1998: 167–168). Curiously, Dameli shows, apart from the switch-topicality function shared with Palula, yet another use of ba, namely as a subordinator, either with a relativizing function or as a clause-final conditionality marker (Perder, 2013: 173). The latter is strikingly similar to some functions of bo/ba in Sawi, Gawarbati and Ashkun (as mentioned above), and to the conditional marker pe in Kalasha (Heegård Petersen, 2015: 71).
As will be discussed in Section 4, the ba construction as such is not to the best of our knowledge found outside of the chka. However, most languages of the Greater Hindukush have sets of discourse particles, some of which are functionally as well as formally parallel to the individual particles of the ba constructions. This is particularly evident with the first-place particle of the construction (ta/tä/da), since formally similar markers of topic or emphasis are found in many languages inside and outside of the area, e.g., to in Kohistani Shina (16) as well as tō in Urdu (Schmidt, 1999: 210) and te in Gojri (Wayne Losey, pc.).
In summary, the six languages Dameli, Gawri, Kalasha, Palula, Waigali and Kati have remarkably uniform so-called ba constructions, with a core emphatic contrastive function. In all languages, it is a bisyndetic construction with particles that even separately display similar cross-linguistic patterns in their use as discourse markers. As will be discussed in the next section, there is no firm evidence of the construction appearing in languages outside of this area, even in those cases when these languages happen to be genealogically more closely related to one or more of these six languages, thus pointing to it as an areally confined phenomenon.
5 The Bisyndetic Antithetical Construction, Language Relatedness and Areal Patterns
While the surface forms obviously differ, the essential marking features of the ba constructions are shared between at least a handful of the languages in the Central Hindukush Area, as described in previous section. More importantly, however, languages outside of, or more peripheral to, this focus area seem to lack these features altogether. For instance, Indo-Aryan Khowar, the language most closely related to Kalasha and spoken in the area immediately to the North of chka, does not, to our knowledge,16 display any bisyndetic marking in adversative sentences. Although there is a conditional (or subsequence) marker the (cf., Kalasha ta), placed at the end of a subordinate clause which occurs before a main clause, there are few examples of its use in the material at our disposal. In addition, there is no evidence of a correlative coordinator (similar to Kalasha o) working in tandem with it. More frequently, a very generic subordinator ki (cf., ki/ke in Urdu and Persian), serves (among other subordinating functions) as a subsequence/conditional marker, while antithesis tends to be asyndetic, i.e., zero-marked, as in examples (17) and (18).
Also immediately to the North, Iranian Munji is spoken. The material consulted has not suggested the presence of a construction akin to the ba construction in this language; instead it seems, just like in Khowar, that a single conjunction ke has a variety of different coordinating and subordinating functions (Morgenstierne, 1938: 166–167), and expressing contrast might very well be one of them. It is, however, a question that needs to be left open to a more thorough investigation.
For Shina spoken outside of the area of our focus, i.e., to the East, the apparent absence of bisyndetic marking, is based on much larger quantities of data of a comparable kind. Even in semantically identical, or at least analogous, antithetical sentences, that in Palula are expressed with ta and ba (see (9) and (11) above), the two contrasting parts are merely juxtaposed, as in examples (19)–(20) from Gilgiti Shina.
As for Kohistani languages other than Gawri, all spoken to the East of chka, there are few descriptions available, and we do not have access to any larger quantities of text, allowing us to compare antithetical constructions with any greater precision. Although Lubberger (2014) describes and analyzes a few particles with relevance for discourse structure in Indus Kohistani in great detail, these mainly have functions related to evidentiality, and while the emphatic use of a particle ta alone (often translatable as German ja), lines up the use of its ta-cognates elsewhere, there is obviously no corresponding “collaborating” second particle to be found (Beate Lubberger, pc). Neither does Lunsford (2001) mention such a construction in his description of Torwali.19
The same absence is confirmed by comparisons with Indo-Aryan languages further afield in the Greater Hindukush, such as Kashmiri; see e.g., Koul’s (1992) dictionary of proverbs, where no ba construction can be found even in prototypical adversative contexts.
Exactly where to draw the Southern boundary of the chka is a question we leave open at this point. However, we have not found any suggestions as to the presence of similar antithetical structures in Pashai (Morgenstierne, 1973b), the language, or rather the chain of language varieties, bordering the Nuristani languages to the South. Instead, Lehr (2014: 342) describes a type of adversative coordination that makes use of a single particle, either xo (from Pashto) or magar (from Dari), and negated contrasts seem to contain only the two negators alone, or a combination of a negator and a complementizer ke preceding each of the two thus contrasted items.
In addition, none of the Pashai proverbs with a typical antithetical reading appearing in Yun’s (2010) collection make use of a bisyndetic marking strategy. They are either juxtaposed or contain the Pashto-derived particle xo ‘but’, occurring between the two contrasted entities or situations.
The language isolate Burushaski as of today is not spoken immediately adjacent to the languages of the chka, but since it has appeared in discussions on previous substrata in the region (Tikkanen, 1988; 2008: 257–259; Èdel’man, 1980; Bashir, 1988: 408–419), it too should be mentioned in this regard. Even here, no construction akin to the one in the chka has been documented (Yoshioka, 2012: 201–203; Berger, 1998: 182–183; Munshi, nd: 36). To form an emphatic negative coordination, a generic subordinator ke is identically postposed to each of the two contrasted entities, while the negation is expressed in the verb itself (Berger, 1998: 183). Curiously, Burushaski has borrowed the Urdu topic marker tō (Yoshioka, 2012: 233).
The same non-occurrence of a bisyndetic construction can be noted for Iranian Pashto, one of the more important, if not the most important, contemporary lingua franca or trade languages inside and to the immediate South of the focus region. Comparing for instance popular proverbs with similar antithetical readings occurring in Palula as well as in Pashto, we find bisyndetic marking in the former (see (10) above) and mere juxtaposition in the latter. A single clause-linking adversative conjunction xo can also be used (Tegey and Robson, 1996: 196–197).20
Neither do we encounter any evidence for such a construction in Iranian Dari (i.e., Persian as spoken in Afghanistan), the other important lingua franca at the Western periphery of our focus area (Windfuhr and Perry, 2009: 468–470),21 nor in Urdu, the most important lingua franca of modern-day Pakistan at large, although in the latter there are individual particles that correspond semantically and distributionally more or less closely to some of the individual markers described above in the languages of our focus area (Schmidt, 1999: 210–227).
Having thus concluded that the bisyndetic antithetical (ba) construction has a rather narrow areal scope, the natural question to ask is whether there are other defining isoglosses related to chka. This question must probably be answered in the affirmative. At least two other areal features, both significantly overlapping with the geographical distribution of the ba construction, have been suggested by Di Carlo (2011): the presence of a certain type of pronominal suffixes and that of retroflex vowels. This is not the place to describe any of these in detail, but a brief recapitulation is nevertheless in place.
Although pronominal suffixes of various kinds occur quite extensively in the wider region, it is their exclusive occurrence in possessive constructions with kinship terms that set languages in chka apart from the rest of the region. In Kalasha, for instance, a pronominal kinship suffix (pks henceforth) simultaneously distinguishes the possessor of a possessive np in terms of person, and the possessee in terms of number (Bashir, 1988: 43–45; Heegård Petersen, 2006: 30; 2015: 39), as shown in Table 3, and exemplified in (23)–(25).
The presence of pks has been tentatively verified (Di Carlo, 2011) for Indo-Aryan Kalasha, Dameli and Gawarbati, and for Nuristani Kati, Waigali and Ashkun, while pronominal suffixes or clitics are missing altogether in nominal phrases in Khowar, Palula and in the Kohistani languages.
Regardless of their phonemic status, a significant rhoticity or retroflexion has been noted for vowels (rv henceforth) in a number of the languages of the area, while entirely absent or at the best marginal in any of the languages of the immediately surrounding region.23 The areal aspects of this rarissimum feature was first hinted at by Mørch (1997) in her study of retroflex vowels in Kalasha and their origin, and later reiterated by Heegård and Mørch (2004) and Di Carlo (2011). Kalasha has been analyzed as having (almost) complete sets of oral, retroflex, nasalized, and retroflex-nasalized vowels, each with five articulatory targets (Heegård Petersen, 2006: 28; Trail and Cooper, 1999: xxii–xxiv; Cooper, 2005: 76).
The phonemic status is more uncertain for Nuristani languages, but retroflexion or rhoticity does seem to be a feature of certain vowels in Waigali and Kati (Heegård and Mørch, 2004: 71; Di Carlo, 2011), possibly arising from assimilation with adjacent retroflex (and sometimes nasalized) approximants that occur in many of the chka languages (Strand, 2001: 255). Naturally, their actual inclusion in the phoneme inventory may very well be an artefact of analytical choices, as noted by Perder for Dameli, for which a retroflex approximant has been posited, although the phenomenon could equally well (but less economically) have been described with a set of retroflex vowels (2013: 30).
Plotting these three features (ba, pks, rv) discussed so far on a map of the Greater Hindukush Region as in Map 2 visualizes an interesting subareal pattern, where the chka area stands out, with Nuristan as a possible centre of diffusion.
This naturally begs the question what significant non-linguistic factors might have facilitated such an areal diffusion, a matter we will touch on in the next section.
6 Explaining Areality in Central Hindukush
In order to fully understand the mechanisms underpinning the suggested areal diffusion in Central Hindukush, as earlier defined, we need to take a step back in history. While many of the more dramatic cultural influences of today flow from the state-wide dominant centres, on each side of the international (Pakistan/Afghanistan) border, into the valleys of the area by means of formal education, media in the languages of wider communication (Urdu, Dari and English) as well as (mostly Pashtun-dominated) trade and religious instruction and practices (and, to a lesser extent, tourism) following the main roads leading from the lowlands, there is little evidence of any large-scale interaction going on between the lowland cultures of South Asia and the highlands of the Hindukush until relatively recently. Although it is true that the various mountain communities continuously have picked and chosen elements throughout its history from surrounding cultures and politically more dominant societies, the ties between the various local communities nevertheless remained much stronger, probably for centuries, than any of those that arose from time to time between individual communities and parts of lowland society (Cacopardo and Cacopardo, 2001: 26, 31).
Even if the non-Muslim section of the Kalasha community today, comprising less than 5,000 individuals (Parkes, 2000: 272; Heegård Petersen, 2006: 7–8; 2015: 13), is the sole carrier, in the entire mountain region, of practices and a worldview explicitly connected with a pre-Muslim religion, recent research has revealed that a very long time period passed between the very first inroads of Islam into the mountain region at large and the completed process of Islamization that we observe in the surrounding, now Muslim, communities. While Badakhshan to the immediate north of chka fell under Arab control as early as in the 7th and 8th centuries ad, most of the area of our immediate interest was still unconverted only two centuries ago, and even much later some of the communities, or sections of them, were characterized by “nimcha”, a nominal adoption of Islam (Cacopardo and Cacopardo, 2001: 32–39). Interestingly enough, there is a significant overlap between the habitats of the most recent converts (along with the still unconverted Kalasha) and the isoglosses discussed above. While, for instance, northern Chitral, i.e., the original homeland of Khowar, was subject to Muslim influence (from the North) perhaps as early as in the 16th century, most of southern Chitral (i.e., the forefathers of today’s Palula and Dameli speakers) was probably not fully Islamized until the middle of the 19th century (Cacopardo and Cacopardo, 2001: 45–54). Although Dir Kohistan, the eastern-most extension of chka, is one of the least known areas of the larger region, historically as well as linguistically, the picture that now emerges is very similar to the one of southern Chitral and Nuristan, i.e., one where the full penetration of Islam was accomplished at a considerably later time than in the surrounding, Pashtun-dominated, regions (Cacopardo and Cacopardo, 2001: 36–37).
As late-comers into the Islamic world, the Dameli, Palula, Gawarbati and, in particular, the Nuristani communities shared (and to some extent still share) a significant number of cultural features with Kalasha, possibly with a slight dominance of the Nuristani elements vis-à-vis the cultural practices of the other communities (Morgenstierne, 1992: 96; Jones, 1974: 168). There is for instance evidence of intense cultural influences and copying going on between the Kati of Bashgal and the Kalasha of Chitral, as late as in the second half of the nineteenth century, with the latter as the main recipients (Klimburg, 2008; Jettmar, 1975: 394). Many of these communities presumably continued to relate to one another more intimately than any one of them did with the surrounding, and by then fully Muslim, populations of Upper Chitral, Badakhshan, and the Pashtun-dominated areas to the South, also in line with earlier suggestions of Heegård and Mørch (2004: 172). Cacopardo and Cacopardo (2001: 249) describe these features as “part of an articulate and coherent system of meanings, associated with a particular form of human adaptation to the ecosystem” and their worldview as “a general cosmology that provides the bases for an ethics, a legal order, a political structure, an architecture of economic relations, a system of technological and ritual prescriptions”. In the description offered by Jettmar (1982: 256; 1975: 464), and subsequently by Parkes (1987: 649–652), it was a culture centred around goat-herding, with a fundamental opposition between pure (Kalasha ónǰeṣṭa) and impure (prágata), a dualism further reflected in the (gradual) opposition, or complementarity, between a male sphere and female sphere, the former associated with the higher regions, inhabited by the fairies and gods, transhumance and livestock, and the latter associated with the lower regions, i.e., the river beds and the village areas, human contamination (particularly as it relates to child birth and menstruation in Kalasha practices) and agriculture, with a strict division of labour as one of its explicit outcomes (Cacopardo and Cacopardo, 2001: 69–70, 249–250; Edelberg and Jones, 1979: 50). Other important, and probably ancient, ingredients were hunting traditions, witchcraft and so-called milk adoption (Jettmar, 1982: 256; Tucci, 1977: 69). Politically, there was a pronounced difference between the flat “brothers model” of power sharing and reciprocity governing relations and economy within the local communities of chka, as opposed to the “father model” characterizing the urbanocentric or “civilized” world surrounding (and gradually encroaching on) the area, the latter ruled by hierarchical structures, military force and centralized powers (Cacopardo and Cacopardo, 2001: 250).
Apart from similarities in worldview, social organization and material culture, we also have reasons to believe, with Alberto Cacopardo (2008: 445–446), that the chka communities were a great deal more intertwined with one another than is the case today. Both the Palula and the Dameli territories were for instance areas of ancient Kalasha settlement and dominance, and many of the forefathers of the modern-day Palula and Dameli speakers were most certainly Kalasha speakers (Alberto Cacopardo, 2008: 421; Augusto Cacopardo, 2008: 400; Decker, 1996: 165). In quite a few of the side valleys and villages, non-converted, i.e., Kalasha, moieties lived peacefully side by side with converted (or rather nimcha) moieties during the aforementioned gradual transitioning of these communities into the Muslim fold (Cacopardo and Cacopardo, 2001: 169–170; Cacopardo, 1991).24 Biori Valley, comprising three smaller villages, is now a culturally and linguistically homogenous community, with few native speakers of any other language than Palula. Its ethnic or genealogical composition, however, is everything but homogenous. While the lower village, Mingal, was founded only about a century ago by new Muslim converts (possibly having already shifted from Kalasha to speaking Palula), the middle village, Dhamaret, was probably part of a much older Eastern extension of Kalasha, whereas the upper village, Biori proper, was populated by a group of settlers from across the mountain passes escaping Islamic conversion in Dir Kohistan (Cacopardo and Cacopardo, 2001: 111–118). The latter were most likely speakers of a form of Shina related to the language spoken in today’s Kalkot. In addition, Nuristani components have been added to the mix throughout its history (pc. Muhammad Atiqullah). Parallel to a gradual conversion of the population to Islam, through a period of religious co-existence and bilingualism (particularly in Dhamaret), a linguistic convergence with the neighbouring Palula-speaking – and at least nominally Muslim – population of the Ashret Valley took place (Liljegren, 2009: 54–58). The Damel Valley displays a similarly heterogeneous origin, not the least reflected in the intriguing mix of Indo-Aryan and Nuristani elements in Dameli (Strand, 2001: 254).
Intermarriage between the groups, or at least across lineages, were also more frequent (Morgenstierne, 1942: 118), and even highly preferred in pre-Muslim times (Cacopardo and Cacopardo, 1995: 263; Parkes, 2000: 272–273, 291), either due to deeply rooted cultural preferences, or for the practical purpose of forming alliances, of which the historically important (but now very limited) interaction between the Palula and Dameli exemplify (Cacopardo and Cacopardo, 2001: 107, 135–136). While there is no evidence of any all-encompassing lingua franca being used throughout chka, many local (village-level and valley-internal) patterns of extensive bilingualism and multilingualism must have been maintained, perhaps over several generations. We also need to remember that language is not the only, and definitely not the most important, carrier of identity in this region; genealogy, belonging to a certain village, or ideology (being a non-Muslim, Sunni, Shia, Ismaili, etc.) are much more crucial criteria for inclusion or exclusion in the local context. It is for instance interesting to note the overall preference for referring to one’s own language by using a derivation of a village name as compared to a more inclusive term, exemplified in the use of the term Urtsuniwar by the converted (i.e., Muslim) speakers of Kalasha residing in the Urtsun Valley (Akhunzada and Liljegren, 2009: 5).
Even today, some villages in Southern Chitral, such as Kalkatak and Badrugal, bear witness of complex multi-language co-existence. Kalkatak, situated in the main Chitral/Kunar Valley, is another ancient Kalasha settlement that has experienced a gradual incorporation into the Muslim fold. As part of that process and greatly facilitated by its location on the main road leading from the district capital to the lowlands (of Pakistan as well as Afghanistan), other languages have gained a foothold in the village. A portion of the village is Palula-speaking, the language that the new converts started acquiring already by the end of the 19th century. It seems that it was competing with Kalasha for almost a century (Cacopardo and Cacopardo, 2001: 95), with the result that there were only a few remaining Kalasha-speakers by the end of the 20th century. The other main portion of the village is Khowar-speaking, i.e., speakers of the increasingly influential lingua franca of Chitral as a whole. This has partly come as a prolonged result of men (mainly native speakers of Palula) marrying Khowar-speaking wives from other areas. There was no clear preference for any of the two languages, Palula and Khowar, in the 1990s, but it seems clear that Khowar is taking the lead today. Pashto is another language of some importance. Although the number of native Pashto-speakers in the village is relatively low, its importance as a trade language, especially in dealing with Pashtuns and speakers of other smaller languages of southern Chitral, and as the main language of the province as a whole, should probably not be underestimated (Akhunzada and Liljegren, 2009: 10; Decker, 1992a: 47). Also Badrugal, another village situated near the main valley, just south of Kalkatak, displays a similar degree of multilingualism. The village was founded by a Nuristani group variously referred to as Jashi or Shekhani, who first came to the area from Gawardesh (with which they maintain links) just across the Afghan border a couple of centuries ago. Their “original” Nuristani language, a form of Kati, is still spoken, but they have increasingly acquired Palula, Pashto, and to a lesser extent, Khowar as second languages. Probably due to an extraordinarily high degree of intermarriage with non-Jashi women from the surrounding communities, especially Palula has become an important language, probably spoken actively in many of the homes (Cacopardo and Cacopardo, 2001: 173–175; Decker, 1992b: 144–147).
It might very well be that the situations in Kalkatak and Badrugal reflect a typical rather than anomalous language use pattern of chka, not only today, but also in the past: speakers of a small language interacting intimately with speakers of another small language. In fact, much more so than any stable monolingual situation, the multilingualism of those villages is the default case, especially when we consider e.g., the strategic role that marriage arrangements play. As pointed out by Decker, with special reference to Kalkatak, intermarriage is (and probably has been for a very long time) “both a means and a reason for language shift”, and the acquisition of yet another useful instrument of communication (1996: 165).
Assuming that interaction and cultural exchange between the communities of chka, as painted above, has indeed been as frequent and at time as intimate, as we suggest, language use will naturally have been effected, too, through such prolonged contacts, and that would have facilitated lexical and, ultimately, structural borrowing of the kind that the features identified in Section 5 point to. Morgenstierne (1941: 9–10; 1942: 146–147; 1973a: 188) points, for instance, to distinctly Nuristani lexical influences in Kalasha and Dameli, and vocabulary that is common to Kalasha, Palula and Dameli (e.g., kírik, kir, kiir ‘snow’; preṣpél, 25 prašpíili, prašpyeel ‘perspiration’), but not necessarily shared with other Indo-Aryan languages in the region.
Particles of the kind discussed here are in a cross-linguistic perspective actually among those lexical elements that are along with nouns most susceptible to borrowing. Not only do connectors score high in terms of “borrowability”, as compared to other grammatical categories, but those having an adversative semantics (i.e., ‘but’-connectors) are also more likely to be borrowed than simply coordinating conjunctions (i.e., ‘and’-connectors). There is likewise a very strong tendency for languages to borrow discourse markers (Matras, 2007: 54–62).
The spread of vowel retroflexion or rhoticity, could have come about in more than one way: by incorporation of the new sounds along with acquisition of lexical material from one of the other languages in the area, most likely from a Nuristani language such as Kati; by an adjustment of the pronunciation of already existing words so as to “sound” more like the speakers of the other language (e.g., Kalasha speakers imitating culturally dominant Kati speakers); or by a more direct replacement of sounds in the inherited vocabulary with a borrowed sound (Matras, 2007: 37). We suggest the truth to be closest to one of the latter two scenarios, as some of the items where the sounds occur are indeed part of the core vocabulary. As for such hypothetical Nuristani influences being behind the introduction of vowel retroflexion in Kalasha, it makes sense that only the western dialects of Kalasha, spoken geographically next to the habitats of Nuristani people, acquired the features, whereas the eastern, more peripheral dialects did not, as is evident from the comparisons provided by Mørch and Heegaard (1997: 62–65).
The mechanism involved in the acquisition of a new phonological feature, as suggested above, is similar to the development of complex tonal differentiations in Kalkoti, which is mainly explainable by a language-internal process, but one which is also greatly facilitated by other structural changes (such as apocope) taking place, in that case modelled on features of neighbouring Gawri (Liljegren, 2013: 134–144).
As for a possible “spread” of kinship suffixes, it is the feature, of those at hand, most difficult to explain. However, deserving a more thorough study is its possible relationship to the spread or borrowing of kinship nouns, as well as kinship differentiations, across languages, possibly as a result of frequent intermarriage. This feature might in fact point to an older stage of diffusion than the other two, most likely pre-dating the arrival of Shina speakers (i.e., the speakers of modern-day Palula, Sawi and Kalkoti) in chka.
7 Conclusions and Further Studies
In this study we have identified and described an antithetical construction formed with two separate (often clitical) markers or particles that are placed after the contrasted items. The construction, here referred to as a bisyndetic antithetical construction (ba), occurs in a number of languages confined to the mountainous Nuristan-Kunar-Panjkora area, regardless of their exact genealogical classification. Interestingly, this particular construction, i.e., one with the same formal characteristics as in the focus area, have not been found in functionally equivalent contexts in the languages in the larger surrounding region, not even in languages closely related to the ones under study. A striking overlap with the geographical distribution of two other structural features, pronominal kinship suffixes (pks) and retroflex vowels (rv), leads us to conclude that we are dealing with a linguistic and cultural diffusion zone of considerable age. This is paralleled, as previously described, by long-lasting patterns of multilingualism, frequent interaction in the past, a shared material culture as well as practices and narratives grounded in a pre-Muslim world view with significant similarities across ethnic and linguistic boundaries.
Further studies are likely to uncover other features with a significantly similar geographical scope. Apart from structural features, such as optional or conditional plural marking, loss of gender distinctions, differential patient marking and agent agreement (often in spite of or in competition with ergative case-marking), we consider focused attempts at identifying areal-specific lexical organization or other patterns of semantic areality (e.g., in terminology related to kinship or pastoral economy) as particularly fruitful. In further studies it will also be necessary to include Iranian Munji, all of the Nuristani languages, as well as the Pashai-speaking areas immediately to the south, in a fuller sense, in order to paint a more complete picture, and in the latter case in order to fully determine its relationship to or possible inclusion in a Central Hindukush diffusion zone. It will also be of outmost importance to relate any areal or subareal findings in this geographically confined area to ongoing areal-typological investigations in the larger mountain region covering Hindukush, Karakoram, Western Himalaya and the Pamirs.
This work is part of the project Language contact and relatedness in the Hindukush region, supported by the Swedish Research Council (421-2014-631).
Antithetical Marker (first)
Antithetical Marker (second)
Request (for explanation)
Baart Joan L.G. 2003. Tonal features in languages of northern Pakistan. In Baart Joan L.G. and Sindhi Ghulam Hyder (eds.) Pakistani languages and society: problems and prospects132–144. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.
Bashir Elena . 1996a. The areal position of Khowar: South Asian and other affinities. In Bashir Elena and Israr-ud-Din (eds.) Proceedings of the Second International Hindukush Cultural Conference167–179. (Hindukush and Karakoram Studies v. 1). Karachi: Oxford University Press.
Bashir Elena . 1996b. Mosaic of tongues: Quotatives and complementizers in Northwest Indo-Aryan, Burushaski, and Balti. In Hanaway William L and Heston Wilma (eds.) Studies in Pakistani popular culture187–286. Lahore: Lok Virsa Pub. House and Sang-e-Meel Publications.
Liljegren Henrik . 2014. A survey of alignment features in the Greater Hindukush with special references to Indo-Aryan. In Suihkonen Pirkko and Whaley Lindsay J. (eds.) On diversity and complexity of languages spoken in Europe and North and Central Asia133–174. (Studies in Language Companion Series 164). Amsterdam: John Benjamins Publishing Company. (8 December 2014).
Mørch I. and Heegaard J. . 1997. Retroflexe vokalers oprindelse i kalashamon i historisk og areallingvistisk perspektiv and Variation i sprogbeskrivelsen: vokallengde i kalashamon. En analyse af to emner i kalashamon-fonologien. Det humanistiske FakultetUniversity of Copenhagen.
Munshi Sadaf . nd. A grammatical sketch of Hunza Burushaski. Unpublished manuscript. http://www.ltc.unt.edu/~sadafmunshi/Burushaski/language/grammatical_sketch.pdf (3 December 2014).
Strand Richard F. 2001. The tongues of Peristân. Appendix 1. In Cacopardo Alberto and Cacopardo Augusto (eds.) Gates of Peristan: History Religion and Society in the Hindu Kush251–257. (Reports and Memoirs 5). Rome: Istituto Italiano per l’Africa e l’Oriente (IsIAO).
Tegey Habibullah and Robson Barbara . 1996. A Reference Grammar of Pashto [microform] / Habibullah Tegey and Barbara Robson. [Washington d.c.]: Distributed by eric Clearinghouse. http://www.eric.ed.gov/contentdelivery/servlet/ERICServlet?accno=ED399825 (7 July 2015).
Toporov Vladimir Nikolayevich . 1970. About the phonological typology of Burushaski. In Jakobson Roman and Kawamoto Shigeo (eds.) Studies in General and Oriental Linguistics Presented to Shiro Hattori on the occasion of his sixtieth birthday632–647. Tokyo: tec Corporation for Language and Educational Research.
The term “language” is not unproblematic in the absence of a longer literate history or an established association between a linguistic variety and a well-defined political unit. Our somewhat impressionistic definition is a negotiation between an established academic tradition, mutual intelligibility and a perceived shared group-identity. It should for instance be born in mind that we (along with many others) tend to treat geographically relatively unbroken continua of varieties, regardless of the lack of intelligibility between the end points, as dialects of one and the same language, whereas linguistic enclaves, even when they have close relatives elsewhere, are referred to as languages in their own right.
Gawri is also known in the literature as Bashkarik, Kalam Kohistani and Swat-Dir Kohistani.
Kamviri is sometimes treated as a language separate from Kati (Lewis, Simons and Fennig, 2013), but we follow Strand’s treatment of Kamviri as a dialect of Kati (or Kamkata-viri), on par with two other main dialects, or subvarieties, Katavari and Mumviri (1973: 298; 2001: 258).
Although inside the window displayed in the map, we don’t consider the lower Pashtun-dominated parts of the Panjkora Valley part of the chka.
Apart from Morgenstierne’s account (1938: 3–282), Munja and Yidgha are very little described.
Note that our use of bisyndetic henceforth is more restricted than Haspelmath’s use, as ours assumes non-identical shape of the two particles, whereas his does not (2007: 10–11).
We have chosen to use the term “particle” in a non-committal way for the markers in question, thus not taking any stance as far as their status as words, affixes or clitics is concerned. Most of our sources treat these markers, graphically, as separate units, and as such they are usually called conjunctions, particles or discourse markers. For semantically-based definitions and more fine-grained classifications of such elements as distinct parts of speech, see e.g., Schachter and Shopen (2007) and Weydt (1989).
Strand’s data is taken from the Kamviri subvariety of Kati (or Kamkata-viri, which is the term used by Strand).
This glossing is applied throughout when two different particles are used in ba constructions, regardless of previous annotations or other, language-particular, uses of each particle.
All Waigali examples from Degener (2002) are originally unglossed and translated from German by the present authors.
Apart from this clearly adversative or antithetical function, Kamviri de…di is also used conjunctively (Richard Strand, pc).
Buddruss (1967: 60) in fact points to functional similarities with, on the one hand, the auxiliary bo in neighbouring Indo-Aryan Gawarbati (Morgenstierne, 1950: 21) and, on the other hand, the particle ba in Nuristani Ashkun (Morgenstierne, 1929: 213, 216), but abstains from making any claims regarding borrowing. Schmidt and Kohistani (2008: 204) also mention a Kohistani Shina particle bo, signifying presumption, occurring clause-finally, postposed to a finite verb, in their example.
Turner notes that New Indo-Aryan forms of tāˊt cannot be distinguished from derivatives of tadāˊ (Turner, 1966: 5753).
The use is also governed by switch-reference. There is on the one hand a correlation between the use of ta in finite subordinate clauses and different-subject clause chaining, and on the other hand a correlation between the use of converbs (sometimes with ba, sometimes without it) and same-subject clause chaining, as described in Liljegren (2008: 310–313) and confirmed in our present quantitative investigation.
To be compared with Trail’s glossing of o as oi, meaning “old information”.
Based on data collected by the first author in collaboration with native speakers of Khowar.
In its original German translation: ‘Das Kind wird zu rennen verstehen, zu fallen verstehen wird es nicht.’
In its original German translation: ‘Der Dreck des eigenes Nackens wird nicht gesehen werden, der Dreck des Nackens (anderer) Leute wird gesehen werden.’
It is worth pointing out, that we have been told (in pc with Muhammad Zaman Sagar, fully bilingual in Gawri and Torwali) that the Gawri ba with tä… i corresponds to Torwali da … i. However, at this point we don’t have any data to substantiate that claim.
Interestingly, Pashto has a form-identical emphatic or “adverbial clitic” xo (Pate, 2012: 27), reminiscent of the distribution and function of some of the single discourse particles in chka (approximately ‘indeed, you know’ or Swedish ‘ju’).
While paired conjunctions occur frequently in Persian, they are form-identical, and as such similar to what is also found in many other languages in the wider region.
The transcription in this example has been adapted slightly to facilitate comparison with the above examples.
Some evidence seems to suggest that vowels with similar features occur marginally in Torwali, some of them diachronically corresponding to combinations of a non-retroflex vowel and a retroflex consonant (Lunsford, 2001: 27–29).
This is not to say that all interaction was necessarily peaceful, reflected in Robertson’s vivid late 19th century description of the Ashret and Damel valleys as “favourite hunting-grounds” for raiding Nuristanis from Bashgal (Robertson, 1974: 181).
Only in the Urtsun dialect (Trail and Cooper, 1999: 162).