4.1 Consonant inventories

Applying the same values as in W1A, the typical consonant inventory of Indo-Aryan in the Hindukush region is a large one (Table 3). While the average consonant inventory world-wide is in the lower twenties, the average of the present sample (in this case covering the totality of the region’s Indo-Aryan languages) is as high as 36, with nine of the languages falling within the “moderately large” category (26–33 consonant phonemes) and 22 within the “large” category (34 or more).

Without exception, all the languages contrast five major places of articulation (bilabial, dental, retroflex, postalveolar/alveolo-palatal, and velar) and added to that, individual phonemes representing additional places (such as uvular and glottal) occur. The presence of a large number of plosives is characteristic, and to a varying extent the set involves contrasts both in voicing and aspiration. In a number of these languages, the same contrasts are extended to affricate and fricative sets. This general characterization holds even in the face of alternative analyses and counts (particularly the treatment of aspirated series is an issue which is not always entirely straightforward). In some cases, the distribution and manifestation of aspiration (at least when co-occurring with voicing) has led linguists to apply a suprasegmental or a cluster analysis to some of the individual languages rather than positing an entire set of e. g. aspirated voiced units.

The average Indo-Aryan language of this region has a phoneme inventory that is at least as large as that of Indo-Aryan languages in general (Masica 1991: 106–107). Many of the former have larger affricate/fricative sets, and make more use of the retroflex place of articulation than e. g. Hindi-Urdu. On the other hand, only a few of the languages of the region display a four-way contrast by combining the features +/-voice and +/-aspiration. The number of individual phonemes contrasting in voicing differs greatly between the Indo-Aryan languages of this region. While the plosive set shows almost complete consistency, this is much less so with the fricatives and affricates. The interaction between aspiration contrast and voicing contrast is intriguing and in need of further, more detailed, study. Also in many Indo-Aryan languages in general, there is a firm voicing opposition in the plosive set, but no corresponding contrast in the (mostly minimal) fricative set. When there is a contrast, it is often between s and z, where z has been introduced via Persian loans. It should be noted that there were no voiced fricatives in Old Indo-Aryan, only three voiceless fricatives.

Most of the genealogical subgroupings show a slight variation in relative size, but there is a tendency for the languages with the largest inventories to be found in the northern half of the region, and those with relatively small inventories in its southern half. Looking at non-Indo-Aryan languages in the region, all of them (regardless of their phylogenetic identities) also have large or moderately large inventories, with two notable exceptions, namely Dari and Hazaragi, two Iranian languages spoken at the region’s western periphery.

4.2 Vowel quality inventories

Most languages worldwide have an inventory of 5–6 vowels which is also the case in Indo-Aryan in HK. What sets it apart from the global distribution (W2A) is the virtual lack of small systems, as can be seen in Table 4. Dameli is an apparent exception with its 4-vowel contrast.

There is a general problem drawing conclusions from analyses of individual languages, as some descriptions focus on quality while others focus on quantity. It is also difficult to know whether that reflects a language-particular focus or rather an analytical preference of the scholar in question. A tendency that can be seen in e. g. a number of Kohistani languages is the development of a front vs. back contrast for open vowels, essentially expanding from a basic 5-vowel system to a 6-vowel system. Especially noteworthy is also the presence of rounded front vowels in Kundal Shahi and the presence of a contrast between unrounded and rounded back (or central) vowels in Kashmiri. Both of those are situated in the southeastern part of the region.

Generally, there is more group-internal variation with vowel inventories than is the case with the consonant inventories. Even Indo-Aryan-wide there is a great deal of modern-day variation also outside of this region, ranging from five to at least nine qualitatively defined vowels (Masica 1991: 109–113), clearly contrasting with the ancestral system of a basic 5-vowel contrast.

4.3 Retroflexion

The display of retroflexion is not discussed as a dedicated feature in WALS. However, since it has turned out to be highly relevant for the characterization of the phonological systems of the region, and more specifically the extent to which the feature occurs in various phoneme subsets, a new feature was introduced, with the following classification values: a) no retroflex consonants, b) retroflex plosives but no fricatives, and c) retroflex plosives and fricatives. This does not mean there are no retroflex segments beyond those classes; there are indeed, and such instances will be touched on briefly in the discussion to follow. In order to arrive at a global distribution (for this and for some of the subsequent features that are not covered in WALS), data was extracted from UPSID, the UCLA Phonological Segment Inventory Database (with its 451-language sample), through an interface provided online. ^[3] The relevant distribution can be seen in Table 5.^[4] All of the languages in the HK region have retroflex plosives, and the large majority of them have retroflex fricatives in addition to the plosives.

In Table 6 the inventory of retroflex consonants in Gilgiti Shina is displayed as an example.

As for the geographical distribution, all the Indo-Aryan languages spoken in the central parts of the region have retroflex fricatives as well as plosives (Map 2). Retroflexion as a feature is pointed out as typically Indo-Aryan and also as a pan-South Asian areal feature, well beyond Indo-Aryan (Masica 1991: 131). What singles out this particular region from the rest of South Asia, however, is the prevalence of fricative (and affricate) retroflexion (Tikkanen 2008: 255–257; Hock 2015: 122–124).

Map 2:

Retroflexion in the Hindukush-Karakoram.

4.4 Affricates

The occurrence and size of an affricate subset is another relevant phonological feature that is not part of WALS. The classification values set up for this feature are: a) no affricates, b) only palatal affricate, c) only palatal and dental affricates, and d) palatal, dental and retroflex affricates. It is obvious from the results shown in Table 7 that the majority of the Indo-Aryan languages in HK have a tripartite contrast in its affricate set. Another four languages have affricates in two places of articulation, while the remaining seven only have palatal affricates.

The languages with affricates in all three positions seem to be clustered in the central parts of the region (Map 3). An example of such a language is Khowar, whose inventory of affricate consonants is displayed in Table 8.

It should be noted that affricates were not part of the Old Indo-Aryan system, and that the precursor of the modern-day palatal affricate was part of the plosive set. While the addition of dental affricates (not seldom a further development from a palatal pronunciation) is quite widespread in Indo-Aryan in general, the retroflex affricate seems to be a peculiarly HK development (Masica 1991: 94–95), and is as such a highly marked contrast (Tikkanen 2008: 255–256).

Map 3:

Affricates in the Hindukush-Karakoram.

4.5 Aspiration

Another important phonological feature for the region is the presence vs. absence of a contrast between aspirated and unaspirated consonants. The following values were set up: a) a contrast absent, b) a contrast present for voiceless consonants only, and c) a contrast for voiceless and voiced consonants. As shown in Table 9, only a few languages lack this contrast altogether, whereas the majority of them display a contrast between aspirated and unaspirated voiceless consonants (as in Table 10), and a smaller portion of the sample maintains a contrast for voiceless as well as voiced consonants, i. e. a four-way contrast,/p, pʰ, b, bʰ/, which is the normal pattern in Indo-Aryan in general (Masica 1991: 101).

Although classification may seem relatively straightforward – and for virtually all Indo-Aryan languages in the region, phoneme inventories are available – it remains a difficult and somewhat puzzling feature. There is often conflicting evidence in one and the same language, and even when the feature is not contrastive in the modern languages, there are frequently traces of an earlier distinction in the form of tonal distinctions.

Generally, aspiration contrasts seem to be weak or waning in the western-most parts of the region (perhaps with new phonemes arising, such as in Gawarbati where there is evidence of a development f<ph). In other cases (e. g. in Palula), problems have to do with the choice between analyzing the occurrence of aspiration as clusters with e. g. a plosive and h, alternatively as distinct phonemes.

4.6 Uvular consonants

While the presence of uvulars globally is a highly marked feature (W6A), they are present in the phoneme inventories of about half of the Indo-Aryan languages in the HK. Most of them only have uvular stops, but at least three of them also have uvular continuants (Table 11).

For some of the languages, an unvoiced uvular plosive is a marginal phoneme, susceptible to a great deal of intra-community variation. It often represents a high-prestige pronunciation of certain words of Perso-Arabic origin. However, particularly in the northern part of the region, such sounds are not only part of borrowed vocabulary – which is the normal situation in Indo-Aryan in general (Masica 1991: 105) –, but are equally found in basic vocabulary without any obvious loan origin (Bashir 2003: 844; Tikkanen 2008: 253), such as in Khowar /ɖɑq/ ‘boy’ or /qɑf/ ‘claw’.

4.7 Vowel nasalization

While in the world-wide sample (W10A), nasalization contrasts are absent in the majority of languages, the opposite holds for Indo-Aryan in HK, among which most languages display a phonemic contrast between oral and nasal vowels (Table 12).

However, the exact contrasting function is hard to pin down. Some of the languages obviously have no nasalization beyond pure assimilation with the nasality feature of an adjacent nasal consonant (e. g. in Khowar), others have a nasalization contrast but no full set of nasalized vowels (e. g. Palula). Yet others appear to have a fully developed contrast with a set of oral vowels and an (almost complete) set of corresponding nasalized vowels (e. g. in Gawri, see Table 13). There is a tendency for nasalization to be of less importance in the western part of the region, and while it is a pervasive feature in languages in the Kohistani group and in Kashmiri, there is more internal variation within the Shina group.

Vowel nasalization is also a prominent feature of Indo-Aryan in general (Masica 1991: 117), but in that case stronger in the western parts of the Indo-Aryan belt than in its eastern parts. Outside of Indo-Aryan we find it only in some Tibeto-Burman and Munda languages and in Iranian Baluchi (Masica 1991: 132).

4.8 Tones

Most languages in the world-wide sample (W13A) have no tones, whereas the majority of Indo-Aryan in HK show some indication of making tonal distinctions. It is in this case always a matter of relatively simple tone systems, i. e. at the most involving distinctions between high and low as far as relative tone levels are concerned, and always as the feature of a word as a whole rather than of each syllable (Table 14).

This is possibly an eastern or central feature, prevalent in languages belonging to the Shina and Kohistani groups, as well as in the languages more closely linked to the main Indo-Aryan languages of the subcontinental plains. Tonal distinctions have previously been reported also for Indo-Aryan languages, such as Punjabi, Bengali and Rajasthani.

In this case, we may be helped by a more fine-tuned characterization (Table 15) based on a taxonomy suggested by Baart (2014) with special reference to languages of this region.

The 2-way contrast, often described as pitch accent, appears to be the most common type, found throughout the region, with the system described for Gilgiti Shina as a typical representative of it. This may be an inherited system rather than one emerging as a response to segmental changes. This 2-way contrast is illustrated in Figure 1, showing f0 graphs of two Palula lexical items that are segmentally identical but contrast in terms of their respective fundamental frequencies.

Figure 1:

In Palula, the two lexical items/raát/‘night’ (left) and/ráat/‘blood’ contrast only in terms of a rising vs. a falling pitch pattern. The two items are pronounced by the same male speaker, occurring mid-sentence in an identical frame with the meaning ‘For this we say ___’.

The 3-way contrast seems to be related to historical loss of voiced aspiration and is typical of Hindko and some of its relatives within the Punjabi continuum (Bhatia 1975).

There is a possible correlation between the relatively complex Kalami-type tone and reduced syllable structure complexity in combination with the loss of voiced aspiration and apocope, diachronically added to the inherited contrast also reflected in the Shina-type tone. This complex system is found in Gawri (aka Kalami), its neighbour Torwali and in Kalkoti, the latter a Shina language spoken in an area overlapping with Gawri (Liljegren 2013).

4.9 Syllable structure

The majority of languages in the world-wide sample (W12A) are moderately complex (with CCVC as its most elaborate syllable), considerably fewer have complex patterns (with more consonants in the onset and/or coda than CCVC), and even fewer have a simple syllable structure (basically nothing beyond CV syllables). In the Indo-Aryan languages of the HK, however, the languages with moderately complex and the languages with complex structures are more or less equal in number, with a slight overweight for the complex ones (Table 16).

Perhaps by showing certain typical co-occurrence patterns we would get a more complete picture: Initial C+r-clusters are common as are final nasal+C clusters. Only a few languages permit e. g. initial s+C or final sibilant+stop. It appears that greater restrictions on clusters are typical of the centrally located languages; this becomes particularly clear if comparing the Indus Valley with the Kunar Valley or Kashmir. The examples from Torwali, Khowar and Brokskat, respectively, in Table 17 illustrate this tendency. Torwali is a typical example of an IA language spoken in the central parts of the HK region showing a diachronic loss of clusters. Apart from a few surface clusters, such as /kʰwami/ ‘in the feet’ >/kʰu/ ‘foot’+/a/ pl+/mi/ ‘in’ (Lunsford 2001: 33), no unambiguous consonant clusters occur at word boundaries. Khowar, spoken in the West, has preserved clusters, especially in coda position, while there are considerable restrictions as to what consonants can co-occur word initially. In Brokskat, spoken in the extreme East of our proposed region, consonant clusters are infrequent in coda position while it is unusually generous as far as initial clusters are concerned. In general, Indo-Aryan languages in this region have fewer co-occurrence constraints than are observed in most of the main Indo-Aryan languages of the Subcontinent (Masica 1991: 125–127).

The same diversity and subareal tendencies seem to hold for non-Indo-Aryan languages in the region. However, Tibeto-Burman Balti, in the East, allows very complex structures, as does Iranian Pashto, in the West.

5.1 Occurrence of nominal plurality

Whether nominal plurality is explicitly indicated varies between languages (W34A). Typically, at least some types of nouns can be or have to be plural marked, as shown in Table 18. The type represented by the largest group of languages is the one that indicates plural and where such marking is obligatory for all nouns. That is also the case in our sample, but interestingly there is a sizeable group that have a plural marking strategy but for which its use is optional, or at least variable. For another two languages (Khowar and Kalasha), plurality is a feature only of human nouns, and it is optional, and in one other language (SE Pashai), plurality is applicable to all nouns but is optional for inanimates.

Optionality is largely a feature of the western part of the region, particularly prominent in the Chitral, Kunar and Pashai subgroupings. As shown in example ‎(1), Khowar nouns referring to non-human participants are not explicitly marked for plurality; the contrast between a singular (a) and plural (b) referent is instead reflected by the verb agreeing in person. Nouns referring to human participants (c) and (d), can, and often do, occur with a plural suffix.

Even Indo-Aryan languages in general display a fair amount of diversity in this respect, but it is primarily in the Eastern languages that plural marking, often by use of relatively new agglutinative material, tends to be optional (Masica 1991: 225–229).

5.2 Coding of nominal plurality

Although a number of coding strategies are attested in the world’s languages as far as nominal plurality is concerned (W33A), suffixation is the overall most common way of coding it. That is also the case in the Indo-Aryan languages of the HK, as seen in the Kohistani Shina and Dameli examples in Table 19.

There are, however, a few notable exceptions. In one language (Kashmiri), stem changes are instead the method of coding, and in another language (Torwali), plurality is primarily indicated with tone. In another two languages (Gawri, Indus Kohistani), there is not one single strategy that could be said to be typical. Gawri applies tone as well as stem changes (Table 20).

Even for a number of the languages that do add a suffix to code plurality, the scope of that is limited. For instance, in Kundal Shahi and in the Chitral languages, a morphological contrast between singular and plural is largely restricted to non-nominative cases – a phenomenon noted also in a number of Indo-Aryan languages outside of this region (Masica 1991: 229). For some of the lesser studied languages better contrastive data is also needed to conclusively decide to what extent tone plays a role in plural marking.

5.3 Number of cases

Languages can be grouped based on the complexity of their case systems. Globally (W49A), there is a great deal of diversity. The largest grouping consists of languages without any morphological case marking at all, but conversely, languages with extensive case-marking systems are not unusual (Table 21).

As for Indo-Aryan in HK, variation occurs, but within a smaller range. None of the languages are entirely void of case marking; instead it varies between two (as in Kalasha, see Table 22) and nine (as in Kohistani Shina, see Table 23), with only a few languages having more than four cases. All three of the languages with 6–9 case categories are spoken in the northeastern part of the region.

When counting case-making categories, strict criteria were applied, and therefore only case forms that animate nouns can take were included, whereas e. g. locative cases that are clearly restricted to inanimate nouns were not counted. However, as also noted before, for some of the lesser studied languages better contrastive data would be needed to decide to what extent tone plays a role in case differentiation.

Another complicating factor, pointed out by Masica (1991: 230–231) for Indo-Aryan in general is the tendency for case marking to appear in successive layers in one and the same language. This makes cross-linguistic comparison particularly challenging.

5.4 Number of genders

Another grammatical category investigated in WALS is gender. In one of the articles (W30A), the focus is on the number of genders in each language. While gender differentiation (as evidenced by agreement) is altogether absent in a sizeable group of languages, the largest grouping with gender are languages with a two-gender system, in most cases making a masculine-feminine differentiation. The latter category is by far the most frequent one in the present sample (Table 24), thus representing an inherited system with many reflexes in the Indo-Aryan world in general, and there is not a single language, of those for which there is data to support it, without any gender differentiation at all. The situation in the Indo-Aryan world in general is one where gender is lacking in the northeast, a three-gender system prevails in the south, whereas a two-gender system dominates in the central and western parts of the Subcontinent (Masica 1991: 217–223).

Ushojo, in ‎(2), is a typical example of a two-gender system, where the inherent gender of a noun is evidenced in agreement patterns; the finite verb agrees with the noun oóš ‘wind’ in feminine gender, whereas the masculine noun šídal ‘coldness’ triggers masculine verb agreement.

However, somewhat hidden behind these figures lie some intriguing systems that strongly diverge from the more typical masculine-feminine ones. That is the topic of the next feature to be discussed.

5.5 Sex-based and non-sex-based gender

Another fundamental issue pertaining to gender is what the individual systems are based on, and the primary distinction is whether they are sex-based or non-sex-based. As can be seen in Table 25, the more typical one is the sex-based, but there is also a fair number of non-sex-based systems. Often, animacy plays a crucial role in such systems. In the present sample, the sex-based system is the prevailing one, having a two-way, female vs. male, differentiation at its core.

Significantly, however, the inherited sex-based system is altogether missing in two languages, Khowar and Kalasha, both belonging to the Chitral group (Map 4). Here, we find instead a two-way differentiation based on animacy, where animates (including humans and higher animals) are treated differently from inanimate nouns by at least one agreement target. In another six languages, all spoken in the western-most part of the region, the general part of the region where Khowar and Kalasha are also found, animacy differentiation occurs alongside sex-based differentiation (hence the 3 and 4-gender systems in Table 24), although sex-based differentiation is regarded as the primary distinction. An example of such an overlapping system is seen in ‎(3). In SE Pashai, the main verb agrees in masculine or feminine gender with the direct object (‘boy’ in [a], ‘prayer’ in [b], and ‘cup’ in [c]), while an auxiliary agrees with it in animate or inanimate gender.

Inherited feminine-masculine two-gender systems akin to that found in many Indo-Aryan languages exist in Nuristani as well, but are relatively rare in adjacent Iranian languages in that part of the region. Gender is not a general feature of Tibeto-Burman or Turkic, two of the other non-Indo-Aryan genera represented, whereas Burushaski, curiously, has a four-gender system not altogether different from what was noted in the languages in the region with overlapping sex-based and non-sex-based gender.

Map 4:

Number of genders and their bases in the Hindukush-Karakoram.

5.6 Alignment of case marking (full nouns)

Alignment is the topic of the next few features to be covered. First, alignment (W98A) that can be observed in the case marking of full nouns (as opposed to pronominal case) can take a number of values, of which three stand out as frequent in the global sample: namely a neutral one (i. e. no case marking of A, S or P arguments), a standard nominative-accusative one (different case marking of P vis-à-vis A and S), and an ergative-absolutive one (different case marking of A vis-à-vis S and P). The other three are relatively rarely occurring. As for Indo-Aryan in HK, we see a distribution markedly different from the global one. First, the overwhelmingly most common alignment type, displayed in 20 of the 27 languages included in this sample, is the ergative-absolutive (as exemplified with Indus Kohistani in ‎(4), with the A argument in (b) receiving an ergative case marking vis-à-vis the absolutive form of ‘man’ in (a)), while only one language, Bateri (and even that needing more careful research), displays a neutral system, two languages – once again, Khowar and Kalasha – have nominative-accusative alignment, and four display (at least conditionally) a tripartite type of alignment (i. e. differentiating A vs. P vs. S), namely Gawarbati, Wotapuri-Katarqalai, Sawi and Hindko (Table 26).

It should be clarified here that even within an individual language, alignment principles may be differently applied depending on e. g. the tense or aspect category that is expressed. In order to operationalize the distinction and make it more comparable, it was decided to categorize a language according to its maximum display of case differentiation, which in most instances were in a perfective or past tense-aspect category.

A similar distribution and relative diversity can be seen in Indo-Aryan in general (Verbeke 2011; Deo and Sharma 2006), where ergative case marking (applied in the perfective) represents the system inherited from earlier stages of Indo-Aryan, whereas languages (such as Bengali and Oriya) that consistently apply nominative case marking usually represent a later development (Masica 1991: 343–344).

5.7 Alignment of case marking (pronouns)

Defined identically to the previous feature, but now applied to case marking of pronouns, we find a similar global distributional pattern (W99A), with neutral, nominative-accusative and ergative-absolutive as the most frequently occurring patterns, especially the first two mentioned. But again, we find an intriguingly different distribution as far as Indo-Aryan in HK is concerned. The majority pattern (represented by 14 languages) is tripartite, another eight languages are ergative-absolutive, and two are nominative-accusative (see Map 5) (Table 27).

What we see is that the languages that display a tripartite differentiation with full nouns remain tripartite in their pronominal differentiation, but another nine languages (Gawri, Indus Kohistani, Torwali, Dameli, Grangali, Shumashti, Kalkoti, Palula, Pahari-Pothwari) that display an ergative (short for ergative-absolutive) differentiation with full nouns make a tripartite differentiation in their pronominal system (as in example ‎(5)), and, interestingly, also Bateri, which displays neutral alignment with full nouns, is tripartite as far as pronouns are concerned. I should hasten to add, that it is, once again, the maximum principle that has been applied: if a language displays tripartite differentiation for third person but only ergative for first person, it has been classified as tripartite.

Map 5:

Alignment of case marking (pronouns) in the Hindukush-Karakoram.

The tripartite differentiation can in many ways be explained as a result of two partly overlapping features, both with clearly geographical distributions (Liljegren 2014: 162–167). The first one is direct object marking. While object marking is altogether absent in four of the languages in the eastern-most part of the region, it is either restricted to pronominal objects or to definite objects in the rest of the region. In none of the Indo-Aryan languages investigated does it occur obligatorily with full nouns (Table 28).

The second feature is agent marking. While it is altogether absent in two languages, both located in the northwestern part of the region, it is for the most part conditional, mostly meaning that it occurs with A arguments in the perfective only. Only in four of the languages, all spoken in the eastern-most part of the region, i. e. in the vicinity of Tibeto-Burman, does it appear regardless of the tense-aspect categories expressed. A similar tendency for agent marking to spread from the perfective to the imperfective has been noted for Nepali, another Indo-Aryan language spoken adjacent to Tibeto-Burman (Verbeke 2011: 163) (Table 29).

5.8 Alignment of verbal person marking

Clearly related to, but yet morphologically distinct from, the immediately aforementioned features, alignment of verbal person marking is also part of the WALS set (W100A). In this case, the globally most frequent type is the accusative, followed by neutral alignment. All other alignment types are relatively infrequent, including ergative alignment or alignment types displaying ergativity in a subsystem of the language. Among the Indo-Aryan languages of HK, there is an even division between languages with accusative alignment and languages with a split alignment (Table 30).

It should be particularly noted that accusative alignment of verbal person marking is not only a feature of languages with accusative case marking, such as Khowar, see ‎(6). Instead, we find languages, both in the eastern-most part and in the western-most part of the region that have a consistent accusative alignment of verbal person marking while simultaneously displaying ergative case marking in one or all of their tense-aspect categories, such as Gilgiti Shina, see ‎(8). However, in Palula, see ‎(7), as in many of the other languages of the region, ergative case alignment lines up with ergative verbal agreement (in the perfective).

While the typical pattern in Indo-Aryan in general is the split ergative one (such as in Urdu-Hindi), there are other examples of Indo-Aryan languages that either combine non-nominative case marking of transitive subjects with nominative-accusative verbal agreement (e. g. Nepali), or that display verbal as well as nominal accusativity (e. g. Bengali) (Deo and Sharma 2006: 374–382).

5.9 Verbal person marking

Elaborating on the immediately aforementioned feature, the present feature (W102A) specifies the number and identity (whether A or P) of the person-marking on the verb. What is most significant here is that the value which stands for “person marking of the A or P but not of both”, is the least favoured in the global sample, whereas it occurs in as many as nine of the 25 Indo-Aryan languages included in this sample (Table 31).

This alternation is related, yet again, to a (primarily) aspectual split, not uncommon in Indo-Aryan in general, displayed also in a major language such as Urdu-Hindi, so that the verb agrees with the A in the imperfective, while the verb agrees with the P in the perfective.

However, contrary to the situation in Urdu-Hindi, where differential (non-nominative) case marking of an object neutralizes verb agreement, the majority of the Indo-Aryan languages of our target region display object agreement (in at least one of their tense-aspect categories) regardless of the case marking of the P argument (Table 32). In Sawi, as can be seen in ‎(9), the verb agrees in masculine gender and plural number with the direct object pronoun teeno ‘them’, itself occurring in an oblique form (the corresponding nominative form is se ‘they’).

Again, the various patterns observed in the Indo-Aryan languages of HK have been noted for Indo-Aryan outside of the region, although it seems that it is primarily in the northwestern part of the Subcontinent where simultaneous A and P agreement can be found (Masica 1991: 343–346).

6.1 Basic word order

The order of subject, object and verb in transitive, declarative clauses is considered a fundamental typological feature. Globally (W81A), the two types SOV (subject, followed by object, followed by verb) and SVO dominate greatly, while the remaining types are considerably less frequent. In the sample under study, the languages are almost exclusively SOV. The only notable exception to this pattern is Kashmiri (see ‎[10]), which, at least superficially is SVO, but is in more detailed studies better described as a V2 language (Table 33).

In this respect, the Indo-Aryan languages of HK are in line with the typology of much of the rest of Indo-Aryan as well as with that of the languages of the entire surrounding macro-area (Masica 2001: 240–243).

6.2 Adpositions

The feature defined as the order of adposition and noun phrase has two globally frequent values, namely postpositions and prepositions (W85A). As for the present sample, postpositions are exclusively favoured. Postpositions are often preceded or mediated by an oblique (or other non-nominative) case form of the noun, as in the Northern Hindko examples in Tables 34 and 35.

This feature, is, like the above-mentioned, widespread in South Asia and dominant as far as Indo-Aryan is concerned, although in this case it should be noted that the Indo-Aryan languages of HK are indeed spoken close to or immediately adjacent to Iranian languages with a dominant SOV/prepositional typology (Stilo 2009: 6).

6.3 Zero copula for predicate nominals

The possibility of expressing predicate nominals without an overt copula is investigated in WALS (W120A). There is an almost equal distribution between languages for which is it possible and for which it is impossible. A similar distribution is also found in the HK sample, although it was only possible to determine this for 18 out of 31 languages, and even then with a measure of uncertainty (Table 36).

The possibility of expressing predicate nominals with a zero copula has been noted for individual languages belonging to the Chitral, Shina and Kohistani groups. Often this is only an option in present tense, whereas with a past-time reference a copula is needed, as is shown with the Kalkoti examples in ‎(11). The findings do not allow for stipulating any particular geographical distribution.

This feature does not allow for any systematic comparison with other Indo-Aryan languages, but it should be noted that while overt copulas are typical in many of the main Indo-Aryan languages of the Subcontinent, a zero copula is instead the norm in eastern Indo-Aryan and in Sinhalese (Masica 1991: 336–337).

6.4 Position of polar question particles

The use and the position of polar question particles is covered in WALS (W92A). While such particles are frequent in the world’s languages, with a preference for the clause-final position, they are also entirely absent in a large group of languages, in which polar questions instead can be coded by intonation or a different word order. The present sample, although with less-than-ideal coverage, indicates a clear preference for final-position particles, such as the question particle -a in Brokskat cliticized to the finite verb: so ut-a? ‘Did he come?’ (Table 37)

Question particles are found in most Indo-Aryan languages outside of the HK region too. However, initial markers (e. g. in Urdu-Hindi, perhaps influenced by Persian) may in fact be as frequent as final markers (e. g. in Marathi, a feature shared with Dravidian). Polar questions formed simply by a distinct intonational pattern (e. g. in Gujarati) also occurs (Masica 1991: 388–389).

6.5 Clause chaining

A feature of a geographically similar scope to that of the basic word order SOV is the use of converb constructions (or conjunctive participle constructions, as they are sometimes referred to in the South Asian linguistic tradition) in complex sentences. An example from Gilgiti Shina ‎(1) illustrates how such clause-final converbs link successive and closely related events in a string, before being concluded with a finite verb. This is extremely common in the Indo-Aryan languages of the HK region. Strikingly similar chaining constructions, often corresponding to coordinate clauses in English and other European languages, are found in a wider macro-area, from Dravidian languages in southern India (Krishnamurti 2003: 440) to the Caucasus (Haspelmath 1993: 375–378).

As this feature is tendential rather than categorical, no attempt has been made to quantify or further specify the feature as it occurs in the region.

6.6 Complex constructions

Another tendential feature is the co-existence of left-branching and right-branching structures in the formation of complex constructions, but one which appears to show some interesting sub-areal differences in their respective distribution (Hook 1987). As shown in a smaller quantitative pilot study (Rönnqvist 2014), the further away from the dominance of strongly right-branching languages of wider communication, in this case Persian and Urdu-Hindi, a language is situated, the higher is the proportion of left-branching structures. The Palula sentence in ‎(13) is an example of a left-branching structure, in which an adverbial subordinate clause with a clause-final general subordination marker ta, precedes the main clause.

7.1 Numeral bases

In the categorization of numeral bases, the global sample shows a definite preference for decimal systems, whereas all other types are significantly less frequent. In this case, the present sample is particularly interesting in its overwhelming favouring of hybrid vigesimal-decimal systems, such as the one in Dameli, Table 38.

Only four languages show no traces of vigesimal organization, all four spoken in the southeastern corner of the region, thus bordering on the main Indo-Aryan distribution (Table 39).

Vigesimality is also reflected in many of the non-Indo-Aryan languages represented in the region, in Iranian, Nuristani and Tibeto-Burman as well as in Burushaski.

Elaborating further on the actual composition of numerals (hybrid vigesimal-decimal and decimal alike), reveals an interesting pattern with sub-areal aspects. Three different types were noted: one that consistently uses a 10+n and 20+n pattern, another that consistently uses the reverse pattern, n+10 and n+20, and a third that uses a mixed pattern, i. e. n+10 and 20+n (Table 40). The mixed pattern is used by most of the languages for which there is data, followed by the consistent n+base pattern, leaving two languages with a consistent base+n pattern (Map 6).

Examples of each of the three patterns are displayed in Table 41.

Map 6:

Numeral bases and composition in the Hindukush-Karakoram. (V=vigesimal(-decimal), D=decimal).

7.2 Contrasts in demonstratives

The WALS topic (W41A) is sorted under nominal categories. However, it could also be argued that such contrasts are lexical rather than grammatical, which is the reason for it to appear in this section. Globally, the most common system is one with a two-way distance contrast, as in English ‘this’ vs. ‘that’, followed by languages with a three-way contrast. More elaborate systems are very infrequent in the world-wide sample. The most common system, described or verifiable for at least half of the Indo-Aryan languages of the region, is one with three contrastive terms. In only one language, namely Grangali, do we have evidence, although far from firm or well-described, for four basic contrastive demonstrative terms (Table 42).

In a majority of the languages with what appears to be a three-way contrast, two primary features intersect, and only the first-mentioned is in a strict sense related to distance: a) proximal (within reach) vs. non-proximal (out of reach), and b) visible vs. non-visible. The three terms can thus be defined as +proximal+visible, -proximal+visible, -proximal -visible (Table 43).

In addition to that basic differentiation, another cross-cutting dimension often enters into it, namely emphasis or accessibility; the forms within parentheses are the emphatic elements in each of the terms displayed in Table 43. In many cases, this results in two contrasting forms of each basic demonstrative, one emphatic and one non-emphatic, the latter often used as third person pronouns.

Considerably complicating the picture is the availability of further specifications, especially applicable to the out of reach/visible category. Those specifications are frequently geomorphic, providing details as to e. g. the vertical position of a referent vis-à-vis the deictic centre.

7.3 Kinship systems

Although kinship systems are not treated in WALS, they are a highly promising domain for investigation, especially in the way they may reflect cross-community relationships in the region. As this is a vast field in itself, the present cross-linguistic comparison is very restricted and only one particular sub-domain has been selected for this study, namely kinship terms for one’s parents and their siblings. The values set up are the following: a) separate terms are used for all six types of relations, b) all three male relations have separate terms, but the term for mother’s and father’s sister is the same but different from mother, c) both father and father’s brother share terms, as do mother and mother’s sister; mother’s brother and father’s sister have separate terms, d) father and father’s brother are equal, but different from mother’s brother; all three female relations have separate terms, e) father and father’s brother are equal, but different from mother’s brother; the term for mother’s and father’s sister is the same but different from mother, f) father’s brother and mother’s brother share a term, while distinct from father; mother’s sister and father’s sister share a term, while distinct from mother. Interestingly, all six configurations have been verified in the sample languages (Table 44).

Although the present sample only gives an indication of distribution (Map 7), it appears that the pattern F=FB/M=MZ is an eastern or northeastern feature (Figure 2), also reflecting what has been posited as the ancestral terminology of Burushaski (Parkin 1987: 165) and the terminology used in Balti, the nearest Tibeto-Burman neighbour. If only looking at the F=FB pattern, it appears typical of the languages spoken in an uninterrupted central belt, stretching from the Kunar valley in the west to Gilgit-Baltistan in the east of the HK region.

Figure 2:

Palula kinship terminology (ɡaaḍ-/ɡeeḍ- ‘big’, lhook- ‘small’).

The maximum differentiating terminology, with six different terms is instead common in a southern belt, thus aligning itself with Punjabi kinship systems, while an “aunt” and “uncle” terminology is found in the northwest (consistently so in Khowar), with reflexes in adjacent non-Indo-Aryan communities in the Pamir and its surroundings (e. g. in Iranian Yidgha). The other configurations are in various ways asymmetrical and are found in the central parts of the region.

Map 7:

Kinship systems (uncles, aunts) in the Hindukush-Karakoram.