The Dangari tongue of choke and Machoke: tracing the proto-language of Shina enclaves in the Hindu Kush.
Article Type: Essay
Subject: Enclaves and exclaves (History)
Lexical phonology (Analysis)
Grammar, Comparative and general (Morphology)
Grammar, Comparative and general (Analysis)
Author: Liljegren, Henrik
Pub Date: 01/01/2009
Publication: Name: Acta Orientalia Publisher: Hermes Academic Publishing Audience: Academic Format: Magazine/Journal Subject: Social sciences Copyright: COPYRIGHT 2009 Hermes Academic Publishing ISSN: 0001-6438
Issue: Date: Annual, 2009 Source Volume: 70
Geographic: Geographic Scope: Pakistan; India Geographic Name: Hindu Kush Geographic Code: 9PAKI Pakistan; 9INDI India
Accession Number: 300652466
Full Text: 6.3. The four varieties compared

A comparison of cognates in all four varieties (Table 24, following page) shows a relative consonantal stability. The seemingly regular consonantal correspondences have mainly to do with: a) the presence/absence of word-initial h, w, or y b) the presence/absence of voiced aspiration, c) the presence/absence of word-final n and r, and d) the presence/absence of certain clusters. There are also a few nonsystematic differences, such as metathesis.

The most sonorant and vowel-like consonants (h, w, y) are those that seem most prone to change; A is in a sense the most minimal consonant (Trask 1996: 58) and is as such easily dropped, and more seldom inserted. That is most obvious word-initially. We have several examples of vowel-initial words in Klk, where the other varieties (also more distantly related) have an initial h, such as asiJ 'laughed', in 'is, are', im 'snow' (cf. Gilgiti Shina hin and SP hiimaal 'glacier'), aad 'bone' (NP had), an 'egg' (NP/SP hanoo), so it is reasonable to suggest A-loss in the case of Klk. However, while *h possibly disappeared altogether in Klk, it seems to have been reintroduced into the phonological system by means of recent loans, such as hukumat 'government' and har 'every'. In Sw, on the other hand, there are word-initial h occurring where it is missing in Pal: hangor 'fire' (SP angoor), huguroo 'heavy' (SP unguru), haaru 'peach' (SP ooru).

There are also examples of w-loss (in / 'water') as well as yaddition (in yek 'one' and yekos 'eleven') in Sw. As far as this w-loss is concerned it does not seem to be representative of any consistent wloss in initial position, rather conditioned by the following (high) front vowel. The y-addition seems to be its "mirror image" in that it occurs when there is a following front vowel, such as the example above, and also in yeeroo '(male) sheep' where it goes along with SP yiiru/yiiri, but not with NP iifi '(female) sheep' and Klk eer 'sheep'.

Perhaps the most salient feature of this comparison is the systematic presence and absence of voiced aspiration. As mentioned earlier, I have not been able to detect any such aspiration in my recorded data from Sw. Also, where there is voiced aspiration in the SP and NP items, there is no aspiration to be found in the corresponding Klk items. There may however be a connection between voiced aspiration in Pal and low-rising pitch in Klk and Sw, perhaps parallel to the optional breathy-voicing observed with low tone in Gawri (Baart 1997: 46). The question arises, however, whether we can be sure that aspiration always is lost in Sw and Klk whereas it is kept in SP and NP. It could also be attributed to the development of voiced aspiration in Pal after the other varieties split off.

For some of the cognates we may refute the latter alternative with a fair degree of confidence, as aspiration is documented in OIA or has been reconstructed; such is the case for item 10, ghota (Turner 1966: 4516), and we must draw the conclusion that original aspiration has been lost in Klk and Sw. For other words it is obvious that the aspiration was present in the OIA form, although originally associated with another segment, such as (item 12 above) gostha (Turner 1966: 4336), but has since been transferred from the voiceless plosive at the onset of the second syllable to the initial syllable in NP and SP. That such a "promotion" or left-shift of aspiration did not (or at least not always) take place in Sw is evidenced by beethu (item 9), which (contrary to Pal) has preserved the aspirate of the second syllable.3' In some cases, however, aspiration is not even a feature of the OIA forms, and the presence of it in the modern varieties can, for instance, be attributed to the influence of low-rising tone.

The n in NP, SP and Sw corresponds to n in Klk, regardless of its occurrence word-finally or intervocalically. The status of n as a phoneme in Klk is overall weak. The same is mostly true of r which occurs in the other varieties but corresponds to r in our Klk cognates. While r in the other varieties indeed corresponds regularly to Klk r, also in initial position (cf. third person singular pronouns ru, ras, rasi with SP am, aras, and arasfi), the phoneme r may be re-entering the variety, perhaps via Pashto loans such as laram 'scorpion'. In the other varieties, too, r is restricted in its occurrence; although the aforementioned third singular pronouns occur with a pronunciation w, ras, rasii in their shortened forms in SP, the more commonly heard pronunciation is with an initial I: 16, las, lasfi. The forms of these pronouns in Sw are la, lasee, leesi?1

As far as syllable structure is concerned, the two Pal varieties have preserved a number of clusters that have been fused or simplified in Klk and Sw. Excluding clusters that can only occur word-medially, the ones found in modern Pal (SP) are basically of three kinds: 1. Plosive + r. braam 'joint', utrapaanu 'is running', suutr 'thread'; 2. Nasal + plosive: paand 'path', ukhaandu 'is coming/going up', laang [ng] 'cross!'; and 3. Fricative + plosive: dhrfstu 'saw', ghraast 'wolf. Cluster types 2 and 3 show a considerable degree of intra-variety variation in the word-final position, often with a deletion or weakening of the last segment.34 The comparative Klk data for type 1 is of particular interest. Unfortunately, I do not possess any comparative data for the final position, but for the initial position assimilation with the second segment has taken place across the board: *kr, *tr > tr; *gr, *dr, *br > dr. (35) In Sw we witness a very different situation. Most of these clusters are preserved, except *tr which has fused into a voiceless lateral fricative /. (36) Its voiced counterpart dr is preserved in e.g. drac 'grape', whereas Pal yaandr 'mill' has Sw yaai, and Pal dhrfstu 'saw' corresponds to Sw darsoo. Buddruss notes that the dr cluster is "phonetisch eine sehr unfeste Gruppe" and gives examples of considerable variation in the development of this cluster, providing evidence that one and the same word is sometimes heard with a cluster pronunciation, sometimes with a vowel between the segments. However, the proto-form of yaandr/yaai, as the Sw form suggests, most certainly had the voiceless cluster *tr, with several parallels in Pal of a development *nt > nd, the voiceless plosive becoming voiced following a nasal. (37)

As in Pal, Klk pronunciation of what remains of the clusters of type 2 varies. For word-final ng, I hear only nasalization on the vowel when pronounced in isolation, whereas in the middle of a sentence I hear a velar nasal n. Intervocalically, there is an alternation between ng and n. If the latter prevails we would need to include the phoneme n in the consonant chart rather than regarding it as an allophone of n in front of g. (39) The same tendency of dropping the plosive element and leaving an assimilated nasal is observed for the cognates of Pal words including the clusters nd and ad, again making it necessary to include n as a phoneme and not as a mere allophone of n. While traces of clusters of type 2 remain in Klk, type 3 is lost altogether, in Klk as well as in Sw. Even in intervocalic position only the fricative segment remains in item 7. Where we have the development *st> s in Klk, as in dris 'saw'and as 'eight', we seem to have a seemingly unmotivated development *st > sin Sw: goos 'house' and as'eight'.

Item 16 in Sw is an example of metathesis, supported by the OIA form pandara (Turner 1966: 8047), while I am not able to say which of the Sw or the Pal forms of item 17 reflects an older order of the segments. Item 20 has a word-final b in Sw and Klk and a word-final p in the two Pal varieties. Considering that devoicing is common word-finally, it is more likely that the proto-form had a final b than a final p. The correspondence set w/0/m/m that item 14 ('name') gives us, seems straightforward, as we also have access to the OIA form naman (Turner 1966: 7067). As mentioned earlier, the raising of aa to oo in SP resulted in the assimilation/fusion of the final consonantal segment w with the preceding vowel. The Sw and Klk forms show that w in itself would have had to go through a lenition process from an original bilabial nasal: *m > w. (39)

The area of vowel development is more complex, and we will not be able to deal with it exhaustively. A few snapshots, however, will give a hint at what a reconstruction of the Proto-Dangari vowel system may look like. Not unexpectedly, the phonological outlook of the present-day vocabulary is the result of a general tendency of raising and tensing, but rather more clearly so in the Pal varieties than in the other two. Especially Klk has retained many of the vowel qualities in stressed syllables, although a dramatic "onslaught" in the form of apocope has made the vocabulary at large look rather different than the cognates from the other varieties. Virtually all final unstressed segments were lost some time after the variety split off from the rest.

In the Pal varieties we can witness a chain shift (Table 25), which could be described as a push chain rather than a drag chain (Trask 1996: 86-7), where the accented open syllable *a goes through tensing to aa (*tato > taatu 'hot MSG'), pushing the previous *aa through raising to oo (*taaro > tooru 'star'), pushing the previous *oo to uu (*ghooro > ghuuru 'horse'), where it simply merges with the existing *uu. Similarly, in the front realm, some previous *aa and *a developed through umlauting into ee (*tati > *taati > teeti 'hot F'; *chaali> cheeli 'she-goat'), pushing the previous *ee through raising to ii (*deesa > diisa 'villages'), merging with the existing *ii. As we saw earlier, the vowel system and its symmetry was largely preserved from Proto-Pal to the present day NP and SP varieties, although along with a near-extinction of the diphthongs ai and au.

[TABLE 25 OMITTED]

As we noted in previous section, in NP (but not in SP), all open syllables *6o, "old" (e.g. *ghooro) as well as "new" (e.g. *tooro), were raised to uu. In SP (but not in NP), the closed syllables *a, *aa and *ee were tensed/raised, just as the open ones.

In Sw, a similar development can be traced (Table 26), with the additional (and probably late) feature of phonological merge (due to loss of two quantity contrasts): *ii/*i > *i; *uu/*u > *u. No umlaut arose from *aa in the development of Sw, the reason why the *aa of *c/iaatias well as *taaro developed into o: choli 'goat' and torn 'star', respectively. However, a more recent umlaut process is under way with present-day aa, according to Buddruss resulting in the two allophones [[epsilon]:] and [e:]. I do not exclude the possibility that this may lead to or already has led to the establishment of a new phoneme.

[TABLE 26 OMITTED]

In Klk, relatively few changes took place in its development from Proto-Dangari, as far as stressed vowels are concerned. Alone among the four varieties, Klk kept the open syllable *aa as in: taar 'star' < *taaro; baal 'hair' < *baalo; draa 'brother' < *bhraa, the exception being those (particularly feminine) nouns that developed umlaut: pheep 'paternal aunt' < *phaapi; mees 'maternal aunt' < *maasi; cheel 'she-goat' < *chaali. Similarly, the Proto-Dangari *oo remains in goor 'horse' < *ghooto, and *ee in eer 'sheep' < *eero. As in NP, the closed syllable *aa is also retained: daak 'back' < *daak; draam 'village' < *graam; arjaar 'fire' < *angaar. However, a tensing of *a, although not entirely lining up with the development in SP and Sw, has taken place (in open as well as in closed syllables): baat 'stone' < *bat; kaan 'ear' < *kan; taat 'hot' < *tato.w The other vocalic changes are mainly to do with diphthongs fusing with or giving birth to other phonemes. The "new" phoneme aa, contrasting with aa, is perhaps the result of one or more diphthongs, one of them probably *ai, being monophthongized: daar 'belly' < *dhair; baan 'sister' < *bhain; daar 'husband's brother' < *dair. In open syllables, the *ai monophthongized into a short open front a: bat 'sat down' < *baito; ba 'we' < *bai, now (at least partly) contrasting with a back a.

7. Relatedness and the development of the Hindu Kush Shina varieties

There are a number of unsolved problems in the development of the vowel systems, and without more data it would be too speculative to try and sort them all out. Instead we will return to the question of relatedness and whether it is possible to group the varieties in a way that would give an approximation of how the four are interrelated. Questions that arise are how similar traits in the varieties can be explained; is it a matter of shared innovation, shared retention, or parallel innovation? In addition, language convergence and the amount of contact between speakers will need to be looked into as a factor in producing similarities. In order to group the varieties as closely related, we will primarily look for shared innovation, but also take non-linguistic data and lexicostatistics into account, the latter with a great deal of caution.

7.1. Phonological development

A summary of the more salient phonological innovations or developments discussed so far is found in Table 27.

The last three innovations do not shed a lot of light. The (incomplete) fusion of *tr into / is an innovation unique to Sw, and possibly it has come about under influence from Gawarbati. (41) In any case, it is a development taking place subsequent to any split off from the other varieties. The assimilation of various plosives in clusters with r is a similarly unique and isolated development in Klk. The loss of voiced aspiration in Sw and Klk is most likely not attributable to any shared innovation, rather a reflex of shared retention in the two Pal varieties. It seems, as far as Sw is concerned, to be a rather late development, as Buddruss in the 1950s with some hesitation still noted a weak voiced aspiration in some cognate words. (42)

This leaves us with vowel raising, umlaut formation and apocope, all of them interrelated. For one thing, the umlaut formation must predate the apocope in Klk, the latter certainly a development taking place after splitting off from the other varieties. However, since this particular umlaut formation is not found in Sw, it is a development taking place after Sw's splitting off from the other varieties. Since umlaut-formation is quite a common process in the languages of the region, it could very well be a parallel innovation in Klk, NP and SP. The raising of vowels in open syllables is shared by Sw, SP and NP, whereas the raising of vowels in closed syllables is shared by Sw and SP. This would point us in a direction where the closed syllable vowels in Sw and SP were raised after NP split off from the other two varieties, the latter thus sharing a retention with Klk. There is also the raising of *oo to uu, unique to NP to take into account. A possible scenario is to regard the vowel raising as an unconditional process in SP and Sw, separate from a conditional process (in two steps) in NP, as shown in Table 28 (following page). This means that the rather restricted vowel raising processes in NP can be described as independent from the more extensive vowel raising processes common to SP and Sw, whereas Klk has not been affected by any of these. As far as phonological development goes we are therefore able to suggest a separate subgroup consisting of Sw and SP, whereas NP and Klk branched off earlier.

The vowel system of proto-Dangari must have been something similar to the one we sketched in Table 11 for the Palula proto-language, involving five basic qualities {a, e, i, u, o), length contrast and two or more diphthongs. In all four descendant varieties, the diphthongs in most stressed syllables have become monophthongized and subsequently fused with other vowel phonemes, and at best they have survived in some unstressed syllables. This development is (partly) responsible for the "new" front-back contrast among the open vowels in Klk.43 In Sw the quantity contrast seems to have been lost, with subsequent fusion of for instance *u and *uu, whereas the open-close dimension is utilized with four degrees. Klk is by far the most archaic when it comes to the retention of vowel quality in stressed positions, while open accented syllables have been subject to tensing and raising in NP, and open as well as closed accented syllables have been subject to similar tensing/raising processes in SP and Sw. On the other hand, a general loss of final unstressed vowel segments taking place in Klk after having split off from the other varieties has radically affected the word structure in this variety.

The consonant inventory of proto-Dangari was probably not very different than those in the four descendant varieties, although a more recent influx of loans have added the, largely fricative, sounds f, z, x, y, g, all of them to some degree phonemicized in the receptor varieties.44 The phonemic status of [n] and [n] is uncertain; they may have been allophones of n preceding another retroflex and velar consonant respectively. The relationship between the sounds [r] and [d] is also somewhat unclear, whether they contrasted fully or stood in an allophonic relationship. Voiced aspiration was a feature of the proto-language, but whether it had already developed secondary aspiration before branching out is still to be determined. Voiced aspiration is retained in NP and SP, but has been (segmentally) lost in Sw and Klk.

Clusters of plosive + r were permitted both word-initially and word-finally, and a number of final clusters of nasal + plosive and nasal + fricative occurred. Many of these clusters have been reduced or broken up in Sw and Klk. Most /r-clusters in Sw have developed into a voiceless lateral fricative /, constituting a new consonant phoneme. A rather interesting type of assimilation can be observed in Klk where all plosive + r clusters have been dentally assimilated by the following r: *gr, *dr, *br > dr, and the same for the voiceless clusters into tr.

The raising and tensing processes in SP and NP (and possibly also in Sw) have resulted in paradigmatic vowel alternations. Umlaut formation is present in all four varieties, although it seems to have affected the lexicon of Sw less than that of the other varieties. It seems, however, that umlaut has arisen independently and at different stages in the varieties.

7.2. Morphological development

There are so far no strong phonological reasons to group NP together with SP/Sw vis-a-vis Klk, while there are many grammatical features shared by SP and NP alone. Two of these were mentioned earlier, the (de-construction and the (i)m-suffix. Both of them must be defined as grammatical innovations, rather than retentions, and I hold that these have resulted from a more recent (approximately the last two centuries) interaction between the speech communities in Biori (NP) and Ashret (SP). Basically the entire nominal and verbal systems have converged. The relevant grammatical categories in the nominal paradigm are number (singular and plural) and case (nominative, oblique and genitive), and the great majority of nouns belong to one of three main noun classes or declinations (see Table 29). The oblique case is a multi-purpose case, having locative function, being the case used with most postpositions and also when the noun occurs as an agent in ergative clause constructions.

The most obvious point where Sw differs from Pal is the absence of any inflections with an /n-element, whether number or case forming. We can otherwise spot very clear parallels to the noun classes in Pal, although the distribution looks different. Most Sw cognates of the m- and /-declensions in Pal form their plurals with -e. (45) The masculine subclass of the Pal a-declension with stems ending in (unaccented) -u has its direct parallel in a class of Sw masculine nouns forming the plural with -ee. However, in this Sw class we also recognize some Pal nouns (with singular -oo and plural -ee) from a small subset of the i-declension.

Another subset of the a-declension in Pal has its parallel in Sw (mainly masculine) nouns forming plural with -a.

Another similarity across these varieties is a set of four endings masculine singular, masculine plural, feminine singular, feminine plural--that are used as agreement suffixes on adjectives and verbs alike, identical to the noun endings of the (present-day) most dominant noun classes: Pal -u, -a, -i and -im, Sw -oo/u, -ee, -i and -e. Although it would be tempting to see a link between these masculine singular and feminine singular noun endings, respectively, and the OIA masculine -as and feminine -i or -s, the latter segments were early on subject to weakening and were eventually lost altogether in most NIA languages. The heirs of those we would actually find among the consonant-ending nouns (such as those in Pal i- and adeclensions), rather than among the nouns with "overt" gender in modern-day Pal and Sw (Masica 1991: 222). The modern masculine ending -0/-00 etc. is instead, according to Morgenstierne (1941: 15) and Buddruss (1967: 29), a weakened form of an OIA derivational suffix -aka, and the feminine ending -i' comes in a similar way from a feminine counterpart -ika. (46)

As far as the plurals are concerned, the -a of Sw goes back on a Prakrit form -ao, according to Buddruss, whereas -ee has developed out of *-aya (1967: 37). For Pal, it seems the two have largely fused into -a, only keeping an accented -ee distinct from -a in a small subclass of nouns. Thus -a has become the main masculine plural marker (also reflected in the agreement suffixes) in Pal, whereas the corresponding one for Sw is -ee. The characteristic feminine /n-plural common to NP and SP, is obviously a morphological innovation (Buddruss 1967: 37) that has come about after Sw split off from SP. Buddruss suggests that the Sw plural -e is derived from *-ya < *-iya (1967: 37), thus giving us a reconstructed plural formation: *anguriyaa. This plural morpheme was subsequently extended to other feminine nouns, such as the consonant-ending Jib (which historically lost its final vowel). In Pal, the unaccented -iyaa may very well have undergone a development -iyaa > -i, resulting in the loss of a segmental singular/plural contrast (since the singular form was already -i). Somewhere along the way, and to make up for the loss in number contrast, the -m of what had already become an oblique plural was being reinterpreted as a plural marker in this noun class.

Just as the -u and -a have become the main players as masculine singular and plural markers (and invariably as agreement suffixes) in Pal, the same can be said about the feminine singular and plural -i and -im, respectively. (47) This pattern has apparently become a system defining structural property for nouns (McMahon 1994: 103-4), and even relatively recently added non-native words conform to the pattern: nalyaci/nalyacim 'baking-board' (fr. Khowar).

It is less clear how the case inflections of Sw should be compared with the Pal ones; apparently the varieties have diverged more here than is the case with the plural inflections. With singular reference, two different case-marking forms occur: -ee and -o~ (Buddruss notes -a~ and -oo~, which I tentatively interpret as instances of -o~). Both of these markers are "multipurpose" as far as case is concerned. The -ee is used as genitive (of animates), oblique, dative (with or without a "to"-postposition), sometimes as locative, and invariably as the ergative case-marker. The -o~ is more limited in its distribution; it occurs as genitive and ablative of inanimates. The ending in -ee is certainly related to the Pal genitive (-fiZ-nm SP and -ii/-e in NP), but whether it is a question of several suffixes merging with the genitive in Sw or the genitive suffix being extended to other case-functions, I am not able to determine at this point. With plural reference there is only one "multipurpose" case marker -oo~/-u~ in Sw, used for virtually all nonnominatives. Interestingly, the forms -ee (in singular) and -oo~ (in plural) are also used as accusatives, marking definite objects, especially when preceded by a demonstrative pronoun, as in example (10), most likely due to influence from Gawarbati (11).

The information at hand on Klk nominal inflections is too scanty to make any meaningful comparisons. It is likely that a number of morphological distinctions that were present in an earlier stage of this variety were lost in connection with the disappearance of unaccented final vowels. Most nouns with plural reference in my data are, at least segmentally, identical to those with singular reference. However, I have noted what looks like a plural oblique in -um preceding a postposition in the phrase larkeer-um t(h)a 'to the girls', a form that very well may be related to the Pal plural oblique -am (accented -aam/-oom) and the Sw -oo~/-u~. An ergative marker, -a, may have developed recently, as it attaches at the end in a fashion similar to postpositions and does not seem to have any other allomorphs: cf. zamaan-a 'Zaman (agent)'; mees-a 'the man (agent); larkeer-a 'the girl (agent)'; larkoowm-a'the boys (agent)'. (48)

I propose a common origin of Pal plural oblique -am/-6om, Sw -00-/-UU- and possibly what looks like a Klk plural oblique -um. In the Sw variety the nasal segment is retained only in the nasalization of the preceding vowel. The distinct form of the plural genitive in Pal is probably also an innovation come about during the convergence phase of NP and SP. As a result of a transparency/uniformity pressure on the morphological system (McMahon 1994: 98), the relatively invariable singular genitive was transferred and analogically attached to the periphery of the already inflected noun: taapar'hiW, taapar-u GEN.SG, taapara NOM.PL, taaparam OBL.PL, taaparam-u GEN.PL. NP may even have been the primary agent of pressure in this case. The third person plural genitive of the pronouns that has the form (ha)tenume, unique to NP (Table 30) suggests an innovation with more transparent forms that started in this variety and spread to the noun paradigm of SP but did not quite make it into the innermost of the pronominal system. Of course, it still needs to be shown conclusively that the NP form is indeed the result of innovation rather than retention.

The reconstruction of proto-Dangari morphology is more challenging than the phonological one, especially as apocope in Klk resulted in serious loss of previous suffixation, and NP and SP have converged to such an extent as to make morphology almost identical in the two. The only meaningful comparisons made are those between Pal and Sw. Those comparisons point in the direction of the proto-language having three noun declensions, with three different plural formations, one declension made up of largely feminine nouns, one largely masculine, and one with both masculine and feminine nouns. I have not attempted to reconstruct the case system, as that would need further analysis and additional Klk data, but we seem to have a singular genitive ancestor form with the descendants -ii/-e in NP, -ii in SP, and -ee in Sw, and there is also evidence for a plural oblique (maybe including genitive functions) with a nasal element.

7.3. The development ofTMA categories

The most central grammatical distinction in the Pal verbal system (see Table 31) is aspect, clearly reflected in verb morphology, -aan (or -aand) for imperfective and mostly -il or -(i)t for perfective (along with a smaller number of verbs with an imperfective stem clearly different from a perfective stem, almost all shared by NP and SP).

Tense distinctions are clearly secondary, the only overt reflex being the de-construction already mentioned. In the perfective as well as in the imperfective, the verb displays gender/number agreement (-?/MSG, -a MPL, FSG, -im FPL, attached to the aspect morpheme), whereas in the future there is personal agreement (-um lSG, -er/ar 2SG, -e/a 3SG, -iia lPL, -et/at 2pl, -en/an 3PL). (49) The aspectually unmarked form is used with future reference or (when combined with the past tense marker de) for continuous or nonfinite actions in the past. Historically, the paradigm of this form goes back on the OIA present tense, although the second singular and first plural forms seem to be Pal-specific innovations (Morgenstierne 1941: 22).

The centrality and relative time-depth of the imperfective vs. perfective distinction is supported by the fact that it is primarily the perfective and the imperfective verb forms we recognize when broadening the comparison to include Sw and Klk. The imperfective-forming segment is found in Sw as well as in Klk (in the former, like in Pal, followed by gender/number agreement, whereas in the latter fused with gender/number agreement): bhesaanu (SP), beesaanoo (Sw), bisuun (Klk) 'is/am sitting (MSG)'. The reflexes of the perfective, -il and -t, are also found in both of these varieties, often with the same distribution as in Pal: nikaatu (SP/NP), nikhaatu (Sw), nikhat (Klk) 'appeared, came out' vs. mheeruu (SP), moriloo (Sw), maril (Klk) 'killed'. A perfective/imperfective stem alternation is likewise often seen with cognate verbs: pasaanu/dhristu (SP), pasaanu/darsoo (Sw), and pasuun/dris (Klk) 'is seeing/saw'.

Some of these forms found at the core of verb morphology in all four varieties are obviously old, although the exact origin is not always easy to fix, and further research will be necessary. First, the perfective forms belonging to what I will refer to as the T-forming class (forming perfective forms with -t or in some cases -d or -t) most certainly go back on a Sanskrit past (passive) participle -ta (Whitney 2002 [1889]: 952), representing an early development of a perfectivity category, contrasting initially with an aspectually unmarked plain verb stem. (50) The other perfective marker, -il-, found in an L-forming class, is younger, and can be traced back to the Prakrit -illa (Schmidt & Kohistani 2008: 140). (51) The T-forming class is clearly a kind of residual category in all four of the varieties under discussion, limited in number and unproductive, whereas the paradigm of the L-forming class (and its subclasses) has become the system-defining structural property for verbs, evidenced by the inclusion of recent loans: NP newesilu 'wrote' (from Khowar niwes-). (52)

The origin of the imperfective is more difficult to determine with certainty. Both Morgenstierne (1941:22) and Buddruss (1967:48) state that the -aan of Pal, as well as the virtually identical element in Sw, goes back on the Sanskrit present active participle -ant, even this with numerous parallels in other NIA languages (Masica 1991: 2701). What complicates the picture is a small class of verbs in Pal that form their imperfectives with -aandu. (53) We also have a problem posed by the imperfective verb forms of Klk, where the imperfective morpheme has three different allomorphs: -uun, -iin, -aan. I see it as very unlikely that the different vowel qualities would be straightforward examples of umlauts triggered by a now lost final vowel. (54) Instead, a possible origin is a gender/number agreeing auxiliary *hino or *hano 'is' attached to the aspectually unmarked verb stem. (55)

Turning to the third core category in Pal, the future, the corresponding form category seems rather marginal in Sw and Klk. In Sw, it is used as a subjunctive, but infrequently so. Only the first person singular -um/-om/-aam corresponds with any precision to SP -um/-uum/-eem. The form -iyee is used for third person singular, first and third person plural alike, and comes closest in resemblance to the Pal first person plural. A verb form built on the first person singular subjunctive, an zi-element and a gender/number agreement suffix (giving dumnoo, dumnee, dunrni, dumne 'will give'), is used as a future in Sw. This construction is obviously a Sw-specific innovation; it is likely that the combination n + gender/number agreement is related to the copula (hinoo, hinee, hini, Aine): dumnoo < dum hinoo. For imperfective or habitual past, the other instance where the future/subjunctive is used in Pal, (56) Sw uses a construction with the imperfective followed by a suffigated form of aalo 'was (MSG)' or one of its gender/number alternants: thaan-aloo 'was doing', etc. The ancient agreement pattern has therefore been given up almost exclusively in favor of the new gender/number pattern in Sw (Buddruss 1967: 46-54).

The near-absence of any forms in Klk related to the Pal future (and the OIA present) points in the same direction. Elicitation of future propositions tends to produce the same imperfective verb forms as for most present tense propositions. Only in a few examples there are forms with person agreement: -am/-um lSG (as in ma guwaa tham 'What should I do?'), -a 2/3SG, and -aan 3pl, which correspond closely to the Pal first singular, third singular and third plural agreement forms, respectively. Although form sharing between the second and third person singular of the aorist is also found in Kohistani Shina (Schmidt 2002: 39), a loss of contrast, probably as the result of sound changes in a more distant past, in Klk (and Kohistani Shina), is more likely than a split into second and third person in Pal. (57)

The other TMA-categories found in Klk are, as we saw, formed with a suffix added to the imperfective or perfective, thus giving them an overt past tense marking, a strategy we recognize from Sw and Pal. Considering the differences in construction and the usage across the varieties and also the peripheral position of such elements, these categories are considerably less stable and newer than the core categories (imperfective, perfective, future/aorist). A sketchy presentation of the strategy "core + tense extention" to form new TMA-categories is given in Table 32 on the following page.

Note that the Sw form seems to be marked for present as well as past tense (aorist + present + past); this may be diachronically true, but it is more likely that the -a (probably from 'is') was already a grammaticalized future marker when the past tense marker -aloo (from "was") was added to create a new, primarily modal, category. Obviously most of the tense markers are grammaticalizations of 'be'. (61)

As far as verb morphology and TMA categories are concerned, the formal distinction between perfective and imperfective, most likely was a feature of the proto-language, having two distinct inflectional forms with perfective meaning, -//and -(a)t, each representing a major inflectional class. The imperfective as well as the perfective finite verbs, originally based on participles, showed number and gender agreement. Person-inflecting verb forms probably played a more prominent role in proto-Dangari than in most of the descendant varieties. In Pal as well as in Sw, future reference is based on one or more of these forms, whereas in Pal they are also used, along with a tense marker, for imperfective past. A variety of strategies to indicate tense are used in the four varieties, and we must therefore regard tense on the whole as a newer and more peripheral category than aspect.

7.4. Lexical development

Lexically, the present environments of the varieties have exercised considerable influence. NP and SP have converged in the vicinity of e.g. Kalasha and Dameli, who have supplied the two varieties with new, shared vocabulary. Sw is, as we have pointed out before, entirely surrounded by Gawarbati speakers, and lexical influence from that language is inevitable. Klk, surrounded by Gawri, owes much of its present-day vocabulary to the latter. Further studies of these environments and their sociolinguistic dimensions may throw some light on why certain changes (and not others) have taken place, lexically, phonologically and morphologically.

Although lexicostatistics is highly questionable as the sole method of determining language relatedness (Trask 1996: 361-2), it may hint at how closely related these four varieties are to one another, or at least to what extent they have diverged (and later converged) since splitting off. In the SIL sociolinguistic survey that was referred to earlier, phonetic similarity counts were made based on a 210-item word list collected from all the locations under study. Those counts show a 95 per cent lexical similarity between NP and SP, 56 per cent between SP and Sw, 57 per cent between NP and Sw, and 32 per cent between NP and Klk (K. D. Decker 1992: 80; S. J. Decker 1992: 70). No counts were presented showing similarity between Klk and Sw or between Klk and SP. The aim of that study was to indicate to what extent speakers of the different varieties can be expected to understand one another, and not primarily to determine language relatedness. (62) For our purpose, I find it more helpful to count the number of shared cognates, and I have therefore used the same lists to obtain lexical similarity percentages between all four varieties, as can be seen in Table 33. (63)

This shows us is that NP and SP are lexically very similar, while Sw is considerably different from all of the other varieties, showing an almost equal similarity rate with NP and SP, respectively, and a slightly lower one vis-a-vis Klk. Klk, on the other hand, is equally different from all the other three varieties. If we were to take this at face value, Klk would have split off first from the other varieties, then Sw, and finally the two Pal varieties from each other. However incomplete that picture may be, we should keep it in mind for later.

8. Non-linguistic clues to historical developments

Looking for evidence in local history or oral tradition, such as memorized genealogies, the most extensive documentation of the kind is found among the SP speakers of Ashret. Somewhat simplified, the people of Ashret consider themselves as Shin, i.e., members of a tribe whose history goes back on one of the four castes inhabiting the traditional territory of Shina. (64) According to local history, the ancestor of the people of Ashret was a certain Choke son of Machoke, (65) who migrated to the present location from Chilas in the Indus Valley some 15-16 generations (or at least 300 years) ago, a scenario that has been convincingly corroborated by Alberto Cacopardo's ethnohistorical research (Cacopardo & Cacopardo 2001: 84-93). One source states how this Choke and two accompanying brothers arrived in Chitral, reached Drosh, and subsequently separated. One brother went to Kalas in the Shishi Koh Valley, one continued to Sau in the Kunar Valley, where he settled, and Choke himself settled in Ashret (Cacopardo & Cacopardo 2001:84). This would support a common origin of the speakers of SP and those of Sw. As pointed out earlier, some of my informants from Sau indeed acknowledge the people of Ashret as their "brothers". All of the inhabitants of the present-day village Kalas are now Khowar speakers, but according to information obtained in Puri (the only remaining Pal-speaking village in Shishi Koh), the people of Kalas used to speak Palula. An independent tradition among the Bozhokey in Laspur Valley (about 200 km northeast of the

Palspeaking area) also speaks of a migration from Chilas some 12-15 generations ago. Dr. Inayatullah Faizi, himself a Bozhokey, has documented this tradition, according to which the two brothers Choke and Machoke left Chilas after a power-struggle with their elder brother. After parting during their exile, Machoke arrived in Laspur where his elder son Laphur subsequently settled. The descendants in Laspur Valley have since been linguistically assimilated by their Khowar-speaking neighbors. Some Ashreti sources also claim that the Chilasi immigrants came to Ashret from the north, perhaps via Laspur, rather than through the Lowari pass and Dir Kohistan, thus lining up with the Laspuri tradition (Cacopardo & Cacopardo 2001: 85, 125-6).

What then about the speakers of NP? It has been explicitly pointed out to me by Ashretis that the people of Biori (i.e. the main bulk of NP speakers) are not descendants of Choke and Machoke, but are Kohistanis from Dir who have adopted the language of Ashret (cf. Strand 2001: 255; Saeed 2001: 296). The first part of the statement is probably true, as Biori genealogies lack any convincing links to the genealogies of Ashret, but I suspect the second part to be an overinterpretation on the part of the Ashreti informants. Although the ethnic composition of Biori Valley is more complex than that of Ashret, there is indeed a local tradition that connects a good section of the population with Dir Kohistan, (66) in particular with a village called Biyar, (67) and Cacopardo (2001: 111-18) holds that the Palula of Biori most likely came to the valley from Dir Kohistan, somewhat later than the Palula arrived in Ashret, and possibly to escape conversion to Islam in Dir Kohistan. However, present-day Biyar is a Gawrispeaking village, and we do not know anything for certain about any previous language spoken there, but it is not unlikely that the Klk speakers are a remnant of a previously more widely spoken Shina variety in Dir Kohistan, which had to give way to Gawri and Pashto. Kalkot is in any case not far from Biyar, both situated along the Panjkora River. Atiq Ullah, my main Biori informant, further claims that the people who came from Biyar were originally from Tangir, (68) one of the Indus side valleys west of Chilas, hence the designations Tangiri, Dangari or Dangarik. (69) While visiting Puri, an old man claimed that his village was founded by two brothers, Dooshi (diiusi) and Kanooshi (kanuusi), who came via Dogdarra in Dir Kohistan from a place in Tangir Valley called Dangeri Phururi (dangeri pnururj).70 As an explanation for the exclusiveness on the part of the SP speakers vis-a-vis the NP speakers, Inayatullah Faizi (p.c.) suggests that the ancestors of the Northerners may very well have been Shina speakers, but being Yeshkun rather than Shin would immediately have placed them in a non-kin category, regardless of their linguistic proximity. (71)

[FIGURE 1 OMITTED]

What this gives us are two possible migration routes from Indus Valley to Chitral (Figure 1). One would have originated in the Chilas area, taking the way over Shandur Pass to Laspur, continuing south through Chitral to Ashret Valley and later on branched out to Sau. The other one would have originated in Tangir or its vicinity, taking the way over Swat and Dir Kohistan, ending up in Biori Valley. Linguistically speaking, a variety of Shina spoken in the Chilas area was transplanted in Ashret Valley, later on producing an offshoot in Sau, developing into present-day Sw, much influenced by the locally dominant Gawarbati; another variety spoken in Tangir Valley would have been transplanted in Dir Kohistan, leaving a trace in Klk, with an offshoot in the Biori Valley and a few other places. The varieties that found themselves in the immediate neighborhood of one another in southern Chitral, i.e., NP and SP, would as time passed cross-fertilize each other and converge considerably, grammatically and lexically. Further research in Dir Kohistan would probably give us more clues, but for now I consider this a likely historical scenario.

9. Dangari and other Shina varieties

In the previous section we proposed that the 'Urheimat' of the Dangari speakers most likely is to be found in the present-day Diamer District in Indus Valley. Interestingly enough, Radloff (1992: 142-3) points out that some vocabulary items from a word list collected in Darel and Tangir set these speech varieties apart from other Shina varieties in Indus Valley while agreeing with Pal and Sw (Table 34).

A continued examination of our four western Shina enclaves, and further efforts to determine the features of their source speech, will be important for any future attempts at reconstructing proto-Shina. In connection with a large-scale comparison of Shina varieties, Schmidt (2002: 52) specifically points out a closer study of Pal along with research in the Shina of Astor as important keys in understanding the early developments of Shina at large. As far as lexical items are concerned, a comparison of all known Shina varieties would identify the two Pal varieties as representative of a number of archaic phonological features, such as retention of OIA clusters tr, dr etc., where many varieties have developed retroflex consonants c and z, and final clusters such as st, where most other varieties are left with a single segment s or t. Pal alone has also retained voiced aspiration. Such features would certainly be of importance when trying to tackle the question of tonogenesis or the source of the pitch accent system in Shina, a system shared by all of the varieties that have been subject to any deeper phonological analysis, including Gilgiti in the north, Astori, Guresi and Drasi in the east, Kohistani Shina in the south, and Pal in the west (Radloff 1999:57-107; Schmidt 2002:36-7). A comparative study would also be of importance when reconstructing the case system of early Shina, as well as its pronominal system.

References

Baart, Joan L. G. 1997. The Sounds and Tones of Kalam Kohistani: With Wordlist and Texts. Studies in Languages of Northern Pakistan, vol. 1. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Baart, Joan L. G. 1999. A Sketch of Kalam Kohistani Grammar. Studies in Languages of Northern Pakistan, vol. 5. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Buddruss, Georg. 1967. Die Sprache von Sau in Ostafghanistan: Beiirage zur Kenntnis des Dardischen Phalura. Miinchener Studien zur Sprachwissenschaft, Beiheft M.

Cacopardo, Alberto M. and Augusto S. Cacopardo. 2001. Gates of Periston: History, Religion and Society in the Hindu Kush. Rome: Istituto Italiano per PAfrica e l'Oriente.

Decker, Kendall D. 1992. Sociolinguistic Survey of Northern Pakistan. Vol. 5, Languages of Chitral. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Decker, Sandra J. 1992. Ushojo. In Sociolinguistic Survey of Northern Pakistan. Vol. 1, Languages of Kohistan, ed. Calvin R. Rensch, Sandra J. Decker, and Daniel G. Hallberg, pp. 65-80. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Fox, Anthony. 1995. Linguistic Reconstruction: An Introduction to Theory and Method. Oxford: Oxford University Press.

Jettmar, Karl. 2002. Beyond the Gorges of the Indus: Archaeology before Excavation. Karachi: Oxford University Press.

Liljegren, Henrik. 2008. Towards a Grammatical Description of Palula: An Indo-Aryan Language of the Hindu Kush. Ph.D. diss., Department of Linguistics, Stockholm University.

Masica, Colin P. 1991. The Indo-Aryan Languages. Cambridge: Cambridge University Press.

McMahon, April M. S. 1994. Understanding Language Change. Cambridge: Cambridge University Press.

Morgenstierne, Georg. 1932. Report on a Linguistic Mission to Northwestern India. Instituttet for sammenlignende kulturforskning, Serie C II-1. Oslo: Instituttet for sammenlignende kulturforskning.

Morgenstierne, Georg. 1941. Notes on Phalura: An Unknown Dar die Language of Chitral. Skrifter utgitt av Det Norske VidenskapsAkademi i Oslo, II. Hist.-Filos. Klasse, 1940, No. 5. Oslo: Det Norske Videnskaps-Akademi.

Radloff, Carla F. 1992. The dialects of Shina. In Sociolinguistic Survey of Northern Pakistan. Vol. 2, Languages of Northern Areas, ed. Peter C. Backstrom and Carla F. Radloff, pp. 89-206. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Radloff, Carla F. 1998. Folktales in the Shina of Gilgit: Text, Grammatical Analysis and Commentary. Studies in Languages of Northern Pakistan, vol. 2. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Radloff, Carla F. 1999. Aspects of the Sound System of Gilgiti Shina. Studies in Languages of Northern Pakistan, vol. 4. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Rensch, Calvin R. 1992. Patterns of language use among the Kohistanis of the Swat valley. In Sociolinguistic Survey of Northern Pakistan. Vol. 1, Languages of Kohistan, ed. Calvin R. Rensch, Sandra J. Decker, and Daniel G. Hallberg, pp. 3-62. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics.

Saeed, Ahmad. 2001. Descendants of Choke in Ashret. Part 2 of Appendix 3 (Palula voices on Palula history) in Gates of Periston: History, Religion and Society in the Hindu Kush, ed. Alberto M. Cacopardo and Augusto Cacopardo. Rome: Istituto Italiano per l'Africa e l'Oriente.

Schmidt, Ruth Laila. 2001. Compound tenses in the Shina of Indus Kohistan. In Tohfa-e-Dil: Festschrift Helmut Nespital, ed. Dirk W. Lonne, pp. 433-52. Reinbek: Inge Wezler

Schmidt, Ruth Laila. 2002. A grammatical comparison of Shina dialects. In Himalayan Languages: Past and Present. Trends in Linguistics, Studies and Monographs 149, ed. Anju Saxena, pp. 33-55. Berlin: Mouton de Gruyter.

Schmidt, Ruth Laila and Razwal Kohistani. 2008. A Grammar of the Shina Language of Indus Kohistan. Beitrage zur Kenntnis sudasiatischer Sprachen und Literaturen 17. Wiesbaden: Harrassowitz Verlag.

Strand, Richard F. 1997/2008. Nuristan: Hidden Land of the Hindu Kush. Website created 1997 and last updated 2008. http://users.sedona.net /-strand/ (Accessed 12 December 2008)

Strand, Richard F. 2001. The tongues of Peristan. Appendix 1 in Gates of Periston: History, Religion and Society in the Hindu Kush, ed. Alberto M. Cacopardo and Augusto Cacopardo. Rome: Istituto Italiano per l'Africa e l'Oriente.

Trask, R. L. 1996. Historical linguistics. London: Arnold.

Turner, R. L. 1966. A Comparative Dictionary of the Indo-Aryan languages. London: Oxford University Press.

Whitney, William Dwight. 2002 [1889]. Sanskrit Grammar: Including Both the Classical Language, and the Older Dialects, of Veda and Brahmana. Cambridge, MA: Harvard University Press.

Henrik Liljegren

Stockholm University

(1) I am consciously avoiding the non-linguistic and politically charged terms "dialect" and "language" as much as possible.

(2) All Kalkoti data discussed here (a word list recorded with two different speakers, a questionnaire focusing on verbs, another focusing on pronouns, and a few texts) were collected by Haider and Sager during a survey trip to Kalkot in 2006.

(3) Regarding Sawi I have mainly chosen to rely on Buddruss (1967) for this paper and only to a lesser degree on my own, rather scanty, field data.

(4) Purigal is the Khowar name form.

(5) The weakened status and vitality of Palula in Puri is due to intermarriage with Khowar speakers and a general trend of Khowar taking over as the first language.

(6) Another village, Ghos, situated on the mountainside east of Drosh, was also a partly Palula-speaking village, most likely of the Northern Palula type, until recently, but is reported to have gone through an almost complete language shift in favor of Khowar.

(7) The speech variety was referred to as Sauji (and never Sawi) in my own interaction with informants (cf. K. D. Decker 1992: 78). On one occasion Gaar Boti (not to be confused with Gawarbati) was claimed to be its actual name.

(8) A good portion of a refugee camp in Timargara in Dir District that I visited along with Ajmal Nuristani in 2000 was populated by Sawi speakers from Sau. Since then many are said to have returned to their home village in Afghanistan.

(9) Some evidence suggests that the proper name Kalkot is pronounced by the speakers of this variety with an initial aspirated plosive, khalkoot.

(10) I use Gawri to refer to the Kohistani varieties variously called Bashkarik, Kalam Kohistani or Dir Kohistani. Gawri seems to be a designation acceptable to speakers from Swat as well as from Dir Kohistan.

(11) Approximately 70 per cent of the population of Kalkot speaks Kalkoti and the remaining 30 per cent are speakers of a Gawri (Kohistani) variety (Muhammad Zaman Sager, p.c).

(12) According to some scholars they also belong to the same "Dardic" sub-branch.

(13) The verb forms given in the table are the imperfective and the perfective forms, respectively (the latter within parentheses).

(14) This opinion was reflected in the choice of the name Anjuman-e-Taraqqi-ePalula when a society for the promotion of Pal was formed in 2003 by representatives from all major Pal locations.

(15) Nouns with pitch accent on the last vocalic mora shift their accent from the stem to the suffix, a pattern observed also in Gilgiti (Radloff 1999: 90-98).

(16) Vowels in unaccented positions pose some further challenges that are outside the scope of the present discussion.

(17) But haata 'hands' < *hata, as will be predicted by the internal reconstruction in the following discussion.

(18) The elements in the tables and their mapping of proto-phonemes to phonemes in SP and NP should not be taken in the absolute sense, i.e. not all "aa developed into SP oo, only those in closed syllables.

(19) For the same reason I hesitate assigning the proto-form *soori to NP/SP suuri 'sun', since most of the attested OIA forms have a high back vowel (Turner 1966: 13574).

(20) An unpublished thesis written by Syed Ali Shah (Dept. of Archaeology, Peshawar University) with the title Descriptive analysis of Darag Dialect of Kalkot (Dir Kohistan) describes the Gawri (Kohistani) variety spoken in the same village.

(21) There seems to be a fluctuation between ta and ta 'to' and maybe even tha.

(22) Although I use single vs. double-written vowels in my transcription, this is based on what has become customary among Shina scholars (marking pitchaccent in an unambiguous manner) rather than reflecting any language-specific factors.

(23) At the moment I am unable to give any consistent account of the variation and possible conditioning factors.

(24) It could on the other hand very well be that further research would make it necessary to include two separate phonemes e and o, as short counterparts of e and o, to give justice to all possible contrasts.

(25) A tentative comparison of the tonal patterns of Gawri and Klk carried out by Baart (pc) indicated that Klk may have developed the same type of tonal system seen in Gawri (Baart 1997), possibly as a result of a historical loss of final vowel segments. This stands in contrast to the simple pitch-accent system found in most other Shina varieties (including NP and SP).

(26) Perfectives with /- or t-elements occur in many Shina varieties, e.g. Pal, Sawi (Buddrass 1967:50-1), Gilgiti (Radloff 1998:184) and Kohistani Shina (Schmidt 2001:444).

(27) I assume a separate masculine plural form but lack data to account for it.

(28) No examples of first and second person plural agreement.

(29) I have taken the liberty to regularize Buddruss' transcription to facilitate cross-variety comparison.

(30) Described as "sehr schwach und zweifelhaft" (Buddruss 1967: 22).

(31) [??] is also a phoneme in Gawri (Baart 1997: 18).

(32) That aspiration in earlier stages of Pa! was sensitive to placement of accent or stress is supported by paradigmatic aspiration contrasts noted in the conservative speech of Puri (a subvariety of NP): bhruii 'brother', brahu 'brothers'; gbaw'cow', gehfcows'.

(33) The sound r, as in the Sw form of item 7, is, according to Buddruss (1967: 16), in free variation with n.

(34) The word yaandr, item 18, exemplifies a combination of 1 and 2, not matched by any other example in my data.

(35) I don't have any examples of pr > tr, although that would be predicted; instead we have an example of metathesis pr> rp in item 2.

(36) An intriguing exception is the numeral '13' (B: truis, L: troo/ns).

(37) OIA hewanta > *heemanta > SP: heewanda 'winter' (obl); OIA vansanta > *basanta> SP: basanda 'spring' (OBI.)

(38) With the only reservation that its distribution would be limited to word-medial and word-final positions.

(39) Some other data, however, casts doubt over the in of Klk. (and possibly also Sw) always being the direct descendant of the OIA in. Also nuu/nuu/nain/nu 'nine' has an m in Klk, whereas the OIA form is nava (Turner 1966: 6984), suggesting that the Klk. in represents a word-final fortition: w > in. The same holds for naawu/naawu/nain/naawu 'new', also nava (Turner 1966: 6983) in OIA.

(40) A suprasegmental/tonal distinction seems to be maintained between aa < *aa and aa < *a in Klk, but I am presently unable to give a systematic and accurate account of it.

(41) Probably this is a matter of lexical rather than phonological influence.

(42) There may very well be other suprasegmental correspondences in Sw and Klk to the voiced aspiration in the Pal varieties, yet to be studied and described in detail.

(43) Partly, it may have arisen due to an influx of loans from the neighboring donor language Gawri, which already had this contrast.

(44) A number of Pashto and Urdu words are in their turn loans from Arabic, Persian, English, etc.

(45) All of the examples given by Buddruss (1967: 36) are feminine.

(46) Although in the latter case its development may also have been influenced by a restrengthened old feminine -i, Masica (1991: 222).

(47) It seems the feminine plural -im is still gaining ground, as it doesn't show quite the same consistency and distribution as the other three markers do. As an agreement suffix attached to an adjective, -im only occurs predicatively, whereas attributively -i'is used with singular and plural reference alike.

(48) Note, again, the plural oblique -um- in the last example.

(49) The presence of more than one set of agreement markers is not uncommon in IA languages at large (Masica 1991: 259).

(50) This element has numerous parallels in NIA at large (Masica 1991: 269, 272).

(51) Outside of Shina this mostly occurs in NIA languages in the eastern and the southern parts of the Subcontinent (Masica 1991: 270).

(52) Perfectivity allomorphy occurring within the L-forming class is quite easy to predict and the class as such has been subject to much more of leveling and innovation as compared to the older T-class (Schmidt, p.c).

(53) These are all frequent motion verbs: yhaandu 'is coming', whaandu 'is coming down', ukhaandu 'is going up', nikhaandu Ms coming out, is appearing'.

(54) Although umlaut is a feature of the historical development of Klk, we don't have any parallel cases where a or aa has developed into uu or ii elsewhere in this variety.

(55) The auxiliary may have dropped the initial h already when attached to the stem, and the first vowel may have been colored by the final gender/numberalternating vowel, leaving a trace even as the final vowel was lost.

(56) Along with the past tense marker de.

(57) However, the actual form of the somewhat "mysterious" second person singular in Pal is an innovation peculiar to SP/NP.

(62) K. D. Decker rightly points out that "looking at the word lists, it is easy to see that there is a historical cognate relationship between a greater percentage of the words than this chart shows" (1992: 80).

(63) With a few exceptions, such as when my own material suggests that the items in the survey may be incorrect or misunderstood, I have used them as they stand. In general I have only counted them as similar when they can be assumed to have developed from the same proto-form.

(64) The Shins were considered ritually cleaner than the other three castes, the Yeshkuns, the Kamins, and the Doms (Jettmar 2002: 17).

(65) In some versions of this tradition Choke and Machoke were brothers (Cacopardo & Cacopardo 2001: 85; own notes.)

(66) That is particularly the case in Bhiuri, the uppermost village in Biori Valley.

(67) This is pronounced bhiaar, which in a genitive, or perhaps an adjectival form, along with the regular sound change aa > uu gives bhiduri 'Biori' (lit. "from Biyar").

(68) However, as pointed out by Alberto Cacopardo (p.c), the narrow modern-day use of Tangir referring to this particular valley differs from a broader, past use of Tangir that refers to the whole region around Chilas, including many major Shina-speaking areas.

(69) Interestingly, while Ashreti (SP) people usually dislike the designation Dangarik, the people of Biori (NP) do not seem to mind.

(70) This probably corresponds to a place in Tangir rendered Phurori on some maps.

(71) This idea is disputed by Alberto Cacopardo (p.c), who holds that the awareness of such a caste identity in relatively recent times is out of question.
(10) (Sw)

mi      la     moonus-ee    darsoonoo
I.erg   this   man-acc      have.seen
'I have seen this man.'

(11) (Gawarbati)

mui      sa     maanus-a   maaritum
I. erg   that   man-acc    killed

'I killed the man.'


Abbreviations

acc    accusative

B      Buddruss

erg    ergative

f      feminine

fut    future

gen    genitive

IA     Indo-Aryan

imp    imperative

IPA    International Phonetic Alphabet

ipfv   imperfective

Klk    Kalkoti

L      Liljegren

M      masculine

NIA    New Indo-Aryan

nom    nominative

NP     Northern Palula

obl    oblique

OIA    Old Indo-Aryan

Pal    Palula

pfv    perfective

Pi     plural

prs    present

prt    particle

pst    past

S      syllable

sg     singular

SP     Southern Palula

Sw     Sawi

TMA    tense, mood, aspect

1      first person

2      second person

3      third person


Table 1: Lexical comparison between Pal (southern), Klk and Gawri
(for Gawri. Baart 1997:1999: Sager p.c.) (13)

Pal               Klk             Gawri

baabu             bab             bob             'father'
yeey              yi              yeey            'mother'
bhroo             draa            jaa             'brother'
bheen             baan            ispo            'sister'
maarnu            mool            moot            'maternal uncle'
kuri              treer           khamaniin, is   'wife'
suur              sur             susur           'father-in-law'
pres              irpas           cis             'mother-in-law'
Ainu (de)         in (aas)        thu (aas)       'be'
biaanu (guum)     buun (gu)       bacant (gaa)    'go'
bhaanu (bhflu)    buun (bil)      hoant (hu)      'become'
tha ami (thnlu)   thuun (thaal)   karant (kur)    'do'
akoos             akaas           ikaa            'eleven'
boos              baaS            baa             'twelve'
be                ba              ma              'we'
tus               tis             tha             'you PL'
aro               ru              ay              'he, that'

Table 2: Klk personal pronouns

           Singular

           Nom   Acc/Obl    Erg     Gen

1          ma    ma         mi      mi

2          tu    tu         t(h)i   t(h)i

3 distal   ru    ras        ra      rasi

3 remote   su    tas        ta      tasi

           Plural

           Nom   Acc/Obl    Erg     Gen

1          ba    asaa(~)    is      asi

2          tis   tusaa(~)   tis     tusi

3 distal         ranaa      rin     rani

3 remote   tin   tanaa      tin     tani

Table 3: Derived or compounded words vs. monomorphemic noun stems
in SP

Derived word                              Stem

gookh(u)ra     'bovine animals'           guu      'bull'
bhooriilu      'became deaf'              bhuuru   'deaf'
bhiaanmut      'willow tree'              bhioon   'willow'
bhraaputr      'brother's son'            bhroo    'brother'
seenboo        'side of string bed'       siin     'string bed'
deesneeour     'village hunt'             diis     'village'
kharamoos      'donkey meat'              khaar    'donkey'
yambaat        'mill stone'               yaandr   'mill'
corkuundu      'rectangle'                cuur     'four'
saurool        'house of father-in-law'   suur     'father-in-law'

Table 4: SP numerals, cardinals and ordinals

Cardinals                 Ordinals

cuur        'four'        corima       'fourth'
saat        'seven'       satuma       'seventh'
nuu         'nine'        no(y)ima     'ninth'
daas        'ten'         dasuma       'tenth'
boos        'twelve'      baasuma      '12th'
triis       'thirteen'    treesuma     '13th'
candiis     'fourteen'    candeesuma   '14th'
satoos      'seventeen'   sataasuma    '17th'

Table 5: Noninflected vs. Inflected word forms in SP

Noninflected   Inflected

basaand        basandi     'spring'

jhaat          jhati       'fur'

saar           sari        'lake'

dhaataar       dhaatara    'fire-place'

haal           hala        'plough'

raat           rata        'blood'

bhroo          bhraawu     'brother'

Table 6: SP future vs. present tense forms

Future 3sg   Present msg

udhiiwa      udheewaanu    'flee'

jhoona       jhaanaanu     'recognize'

maara        maraanu       'die'

Table 7: Vowel alternation and proposed historical sound changes

               Possible sound
Alternation        shift            Tentative pre-forms > SP

aa~oo         stressed aa > oo   * baas> boos 'twelve'
oo~uu         stressed oo > uu   * gooli > guuli 'bread'
ee~ii         stressed ee > ii   * seen > sun 'string bed'
a~aa          stressed a > aa    * khar > khaar 'donkey'
o~uu          stressed o > uu    * cor > cuur 'four'
au~uu         stressed au > uu   * saur > suur 'father-in-law'

Table 8: NP nominal paradigms with vowel alternation

Nom sg   Obl sg    Obi pl     Gen sg

angdur   anguura   anguuram   anguure   'fire'

dees     diisa     diisam     diise     'village'

kram     kraama    kraamam    kraame    'work'

Table 9: Comparative word list, SP and NP

        SP       NP

1     kraam    kram      'work'
2     anu      anu       'this' (m)
3     khaar    khar      'donkey'
4     daar     dar       'door'
5     mhaas    mhaas     'meat'
6     kaal     kaal      'year'
7     kaaku    kaaku     'big brother'
8     chaar    char      'water fall'
9     saar     sar       'lake'
10    baat     bat       'stone'
11    akoos    akaas     'eleven'
12    so       so        'he'
13    troo     truu      'three'
14    noo      naaw      'name'
15    soon     saan      'pasture'
16    phoo     phoo      'boy'
17    ghoost   ghoost    'house'
18    suuri    suuri     'sun'
19    kud      kud       'wall'
20    nuu      nuu       'nine'
21    cuur     caar      'four'
22    suur     saar      'father-in-law'
23    siluuk   silaak    'story'
24    ata      ate       'bring!'
25    chiir    chiir     'milk'
26    chiitr   cheetr    'field'
27    niilu    niilu     'blue/green'
28    bheen    bheen     'sister'
29    deeri    deed      'beard'
30    dees     deeri     'day'
31    be       be        'we'
32    ani      ani       'this' (F)

Table 10: Correspondence sets, SP vs. NP, including accent
and syllable structure

Open S: SP/NP          Closed S: SP/NP

aa/aa    ee/ee   i/i   aa/a    uu/aa
oo/uu    a/e     i/i   aa/aa   ii/ee
oo/oo    a/a     e/e   aa/a    ii/ii
uu/uu    o/o           oo/aa   ee/ee
ii/ii    u/u           oo/oo   u/u

Table 13: Sound changes producing vowel alternations in the NP
paradigm for kiam 'work'

Pre-proto
forms       * kram        * krama     * kram         * krama

Stage 1     * kram        * kraama    * kram         * kraama

Stage 2     * kram        * kraama    * kraam        * kraama

Output      kram          kraama      kraam          kraama

            NP (nom sg)   NP (infl)   SP (nom sg)    SP (infl)

Table 14: Sound changes producing vowel alternations in the SP
paradigm for saar 'lake, pond'

Pre-proto   * sar         * sari      * sar         * sari
forms

Stage 1     * sar         * sari      * sar         * sari

Stage 2     * sar         * sari      * saar        * sari

Output      sar           sari        saar          sari

            NP (nom sg)   NP (infl)   SP (nom sg)   SP (infl)

Table 15: Klk consonants

ph     th     (th)           kh
p      t       t              k     (q)
b      d       d              g
               ch     ch
      (ts)     c      c
                      j
       s       s      s      (x)    (h)
      (z)                    (y)
       r      (r)
       l
w                     y
m      n       n             n(?)

Table 16: Klk consonants exemplified

ph    pheep      'father's       ch   cheel      'goat'
                 sister'

P     paan       'path'          c    caam       'skin'

b     baal       'hair'          j    jib        'tongue'

th    theer      'hand'          s    saat       'seven'

t     taar       'star'          s    sis        'head'

d     daan       'tooth'         s    saak       'wood'

th    (ithyil)   'to stand up'   r    raat       'blood'

t     khatan     'short'         1    loon       'salt'

d     daak       'back'          w    waat       'came down'

ich   khaal      'to eat'        y    yaal       'to come'

k     kaal       'year'          m    maam       'mother's
                                                 father'

9     goor       'horse'         n    naam       'name'

ch    chiir      'milk'          n    cunil      'wrote'

c     dracum     'rieht'         n    an(g)aar   'fire'

Table 18: Klk vowels exemplified

a     siir    'sun'         aa    naam    'name'

i     sis     'head'        a     nam     'nine'

ee    treer   'woman'       oo    goor    'horse'

a     dar     'door'        u     dur     'dust'

aa    daar    'husband's    uu    duur    'far'
              brother'

Table 19: Klk imperfective vs. perfective forms with suppletive verbs

Imperfective
(msg)          Perfective

in             aas             'to be'

buun           gu (msg), gee   'to go'
               (f) (27)

pasuun         dris            'to see'

maruun         mur             'to die'

piluun         piil            'to drink'

duun           dit             'to give'

wuun           waat            'to get down'

nikhuun        nikhat          'to appear'

bisuun         bat             'to sit down'

Table 20: Klk verbs with person agreement (28)

1sg   dam 'I may give', thaw 'I may do', bum 'I may become'
2sg   china 'you may/will cut'
3sg   china 'he may/will cut', ya 'he may/will come'
3pl   chinaan 'they may/will cut'

Table 21: Verb forms in Klk (The English glosses represent the
translations of the sentences these forms are found in)

Ipfv                     Ipfv + pst             Pfv

b-uun, -aan, -iin    b-uun-s,             gu, gee
                     -aan-s,-iin-s

'is going, goes,     'was going, used     'went'
will go'             to go'

bis-uun, -aan,-iin                        ba-t
'is sitting'                              'is seated (sat
                                          down)'

pas-uun                                   dris
'sees, is seeing'                         'saw'

                     cun-uun-s            cun-il
                     'was writing'        'wrote'

Ipfv                     Pfv + pst

b-uun, -aan, -iin    gus

'is going, goes,     'went'
will go'

bis-uun, -aan,-iin   bas
'is sitting'         'had sat down (was
                     sitting), was
                     seated'

pas-uun              dri-s
'sees, is seeing'    'had seen'

                     cun-i-s
                     'had written'

Table 22: Sw consonants

ph     th     th        kh
P      t      t         k      q
(bh)   (dh)             (gh)
b      d      d         g
              ch   ch
       (ts)   c    c
                   j
f      s      s    s    x      h
       z                y
       r      r
       l
       l
w                  y
m      n      n         (n)

Table 24: Comparative word list, NP, SP, Klk and Sw

          NP          SP         Klk          Sw

1     kram         kraam      tram         kraam      'work'

2     ipres        pres       irpas        pras       'mother-
                                                      in-law'

3     sing         sing       sin / si~    sin        'horn'

4     dan(d)       daand      daan(d)      daand      'tooth'

5     mhaas        mhaas      maas         mos        'meat'

6     lhoon        lhoon      loon         loon       'salt'

7     NO DATA      bistiinu   bisiin       bisi~ru    'wide'

8     saareeni     saareeni   saran        saroni     'wife's
                                                      sister'

9     bhetu        bhetu      bat          beethu     'sat down'

10    ghuuru       ghuuni     goor         guroo      'horse'

11    akaas        akoos      akaas        yekos      'eleven'

12    ghoost       ghoost     NO COGNATE   goos       'house'

13    truu         troo       traa         lo         'three'

14    naaw         noo        naam         nom        'name'

15    muru         muru       mur          muroo      'died'

16    panaaru      panaaru    panar        paranoo    'white'

17    tiincuk      tiincuk    no data      tikuc      'scorpion'

18    yandr        yaandr     no data      yaal       'mill'

19    hansilu      hansilu    asil         hansiloo   'laughed'

20    jip          jip        jip          jip        'tongue'

21    hinu         hinu       in           hinu       'is'

22    wii          wii        wa           i          'water'

23    NO COGNATE   groom      draam        grom       'village'

24    bhruu        bhroo      draa         bro        'brother'

25    angaar       angoor     an(g)aar     hangor     'fire'

Table 27: Summary of phonological innovations in Sw, SP, NP and Klk

Innovation                                  Sw    SP    NP    Klk

Vowel raising in open syllable              YES   YES   YES   NO
Vowel raising in closed syllable            YES   YES   NO    NO
Umlaut of *aa                               NO    YES   YES   YES
Apocope                                     NO    NO    NO    YES
Loss of voiced aspiration                   YES   NO    NO    YES
Dental assimilation of plosive+ r-cluster   NO    NO    NO    YES
l-phoneme                                   YES   NO    NO    NO

Table 28: Front proto-forms to present-day forms ('twelve',
star', 'horse') in Sw, SP,NP and Klk

                                         Sw         SP

Proto-form                            * baas     * baas
Unconditional raising: aa, oo         bos        boos
Surface derivation                    bos        boos

Proto-form                            * taaro    * taaro
Unconditional raising: aa, oo         toru       tooru
Conditional raising 1: aa in open S   -          -
Conditional raising 2: oo in open S   -          -
Surface derivation                    toru       tooru

Proto-form                            * ghooro   * ghooro
Unconditional raising: aa, oo         * ghuroo   ghuuru
Conditional raising 2: oo in open S   -          -
Surface derivation                    guroo      ghuuru

                                         NP        Klk

Proto-form                            * baas     * baas
Unconditional raising: aa, oo         -          -
Surface derivation                    baas       baas

Proto-form                            * taaro    *taro
Unconditional raising: aa, oo         -          -
Conditional raising 1: aa in open S   *tooro     -
Conditional raising 2: oo in open S   tuuru      -
Surface derivation                    tuuru      taar

Proto-form                            * ghooro   * ghooro
Unconditional raising: aa, oo         -          -
Conditional raising 2: oo in open S   ghuuru     -
Surface derivation                    ghuuru     goor

Table 29: Noun classes in Pal (SP) and Sw compared (inflection,
gender, example words in sg/pl)

Pal                          Sw

-a    m/f   baat/bata        -a    m/f   baat/bata     'stone'

      m     kucuru/kucura    -ee   m     kuciiroo/     'dog'
                                         kucuree

-i    m     hanoofhanee                  andoo/andee   'egg'

      f/m   jip/jipi         -e    f     jib/jibe      'tongue'

-im   f     anguri/angurim               anguri/       'finger'
                                         angure

Table 30: Some pronominal forms compared (NP, SP, Sw and Klk)

          3sg obi        3sg gen       3pl obi       3pl gen

NP    (ha)tes           (ha)tesee   (ha)tenaam      (ha)tenume
SP    tas               tas/i       tanaam          tanii
Sw    tasee, tasi, ta   teesi       teeno, teena-   teeni
Klk   tas               tasi        tanaa           tani

Table 31: Main conjugational distinctions in Pal (NP forms of
'walk', 'come down' and 'find' exemplified).

                Ipfv (msg)   Pfv (msg)   Fut (1SG)

L-form, -aan    tilaanu      tililu      tilum

T-form, -aand   whaandu      whaatu      whaam

Stem change     Ihayaanu     laadu       lhaayum

                Imp (SG)     Converb

L-form, -aan    til          tili

T-form, -aand   wha          whai

Stem change     Ihay         Ihayi

Table 32: TM A categories in SP, Sw and Klk (Ext = Tense extension)

Core     Ext       SP           Sw          Klk

Ipfv     Pst   -            thaan-aloo    cunuun-s

PfV      Prs   muru hinu    thiloo-noo    -

         Pst   samoolu de   thil-aaloo    cuni-s

Aorist   Prs   -            khom-noo      ?

         Pst   tilum de     khom-n-aloo   ?

Core     Ext                      Usage

Ipfv     Pst   Sw: "was doing, used to do"; Klk: "was writing,
               kept on writing" (58)

PfV      Prs   SP: "died, has died"; Sw: "has done" (59)

         Pst   SP: "was built"; Sw: "had done"; Klk: "had written"

Aorist   Prs   "will eat"

         Pst   SP: "was walking"; Sw: "would eat, would have
               eaten" (60)

Table 33: Lexical similarity percentages (based on shared cognates)

Southern fatula

92                Northern Palula

67                67                Sawi

59                59                60     Kalkoti

Table 34: Words shared by Darel/Tangir and SP and Sw (for Pal
the author's own data, for Sw Buddruss (1967), and for the other
varieties Radloff(1992))

                'heart'       'girl'   'died'

Darel / Tangir  hiru          phu'I    mu~ro~

SP              hiro          phai     muro

Sw              hi'roo        pho'i    mu'roo

Diamer          'hio          mu'lai   mu~

Kohistan        hio/hali'li   mu'lai   mu~
Gale Copyright: Copyright 2009 Gale, Cengage Learning. All rights reserved.