Lexico-Semantic Areality in the Greater Hindu Kush
An Areal-Typological Study on Numerals and Kinship Terms Jacqueline Venetz
Department of Linguistics
Independent Project for the Degree of Master (Two Year) 30 hp Typology and Linguistic Diversity
Spring Term 2019
Supervisor: Henrik Liljegren Examiner: Ljuba Veselinova
Project affiliation: ”Language contact and relatedness in the Hindukush region”, VR 421-2014-631
Lexico-Semantic Areality in the Greater Hindu Kush
An Areal-Typological Study on Numerals and Kinship Terms
Abstract
The Greater Hindu Kush designates a mountainous area extending from Afghanistan over Pakistan, Tajikistan and India to the westernmost parts of China. It is home to over 50 lan- guages from six different phyla; Indo-Aryan, Iranian, Nuristani, Turkic, Tibeto-Burman and the language isolate Burushaski. Due to its unique geographical setting, it is characterised by language contact and isolation, which lays the perfect ground for research on linguistic diversity, language convergence and genealogical relations.
The present study relies on data from the entire region and attempts to identify structural similarities based on lexical items from core vocabulary, numerals and kinship terms. The study reexamines the genealogical affiliation through lexical similarity and investigates areal patterns of vergence, i.e. the branching out or mergence of these patterns. Results reconfirm the established classification of the languages and indicate a certain level of structural simi- larity across language families for some features such as numeral bases, numeral composition and the terms for ‘parents’ and ‘parents-in-law’, yet it also shows great diversity for other features such as ‘grandchildren’ and one’s siblings’ partner.
Keywords
Greater Hindu Kush, areal typology, lexical areality, numerals, kinship terms, Indo-Aryan,
Iranian, Nuristani, Turkic, Tibeto-Burman, Burushaski
Lexikalisk Arealitet i Hindukushområdet
En areal-typologisk undersökning om numeraler och släktskapstermer
Sammanfattning
Hindu Kush är ett bergsområde som sträcker sig från Afghanistan över Pakistan, Tadzjikistan och Indien till västra Kina. Där finns över 50 språk från sex olika språkfamiljer: indoariska, iranska, nuristanska, turkspråk och tibeto-burmanska språk samt isolatspråket burushaski. På grund av områdets unika läge karaktäriseras språken av språkkontakt och isolation. Området lämpar sig således väl för forskning inom språklig mångfald, språkkonvergens och genealo- giska relationer.
Denna studie bygger på data från hela regionen och försöker identifiera strukturella likheter baserat på lexikala enheter från kärnordförrådet, numeraler och släktskapstermer. Studien undersöker den genealogiska tillhörigheten genom lexikala likheter och undersöker areala mönster av konvergens och divergens. Resultaten bekräftar den etablerade genealogiska till- hörigheten och indikerar en nivå av likhet i struktur, numeraler samt orden för ‘föräldrar’
och ‘svärföräldrar’. Resultaten påvisar också stor mångfald när det gäller andra termer såsom
‘barnbarn’ och ens syskons partner.
Nyckelord
Hindukushområdet, arealtypologi, lexikalisk arealitet, numeraler, släktskapstermer, indo-ariska,
iranska, nuristanska, turkspråk, tibeto-burmanska, burushaski
List of Figures
1 Geographical distribution of languages and linguistic genera in the GHK, rep-
resenting the traditional classification . . . . 3
2 The language sample of the Greater Hindu Kush . . . . 8
3 Visual representation of the cognate analysis . . . . 13
4 Phylogram based on cognate analysis . . . . 35
5 Simplified phylogenetic representation of the sample languages, based on Glot- tolog 3.4 (Hammarström et al. 2019) . . . . 36
6 Distribution of numeral bases (Map data © 2019 Google) . . . . 37
7 Distribution of numeral composition (Map data © 2019 Google) . . . . 38
8 Distribution of patterns for ‘parents’ (Map data © 2019 Google) . . . . 39
9 Distribution of patterns for ‘grandparents’ (Map data © 2019 Google) . . . . . 40
10 Distribution of patterns for ‘grandchildren’ (Map data © 2019 Google) . . . . . 40
11 Distribution of patterns for parents and their siblings (Map data © 2019 Google) 41 12 Distribution of patterns for ‘siblings’ children’ (Map data © 2019 Google) . . . 42
13 Distribution of patterns for ‘parents-in-law’ (Map data © 2019 Google) . . . . 43
14 Distribution of patterns for ‘partner’s siblings’ (Map data © 2019 Google) . . . 44
15 Distribution of patterns for ‘siblings’ partner’ (Map data © 2019 Google) . . . 44
16 Visual representation of the feature analysis . . . . 45
List of Tables
1 Kinship Abbreviations . . . . 5
2 Language Sample (classification based on Glottolog 3.4 (Hammarström et al. 2019)) . . . . 7
3 Cognate analysis example . . . . 8
4 Patterns for numeral composition . . . . 9
5 Features and values for kinship analysis . . . . 10
6 Examples for words more likely to be cognates . . . . 12
7 Examples for words less likely to be cognates . . . . 13
8 Numeral Bases . . . . 14
9 Ishkashimi multiples of ten . . . . 14
10 Dameli multiples of ten . . . . 15
11 Gojri (Afgh) multiples of ten . . . . 15
12 Munji numerals . . . . 15
13 Numeral Composition . . . . 16
14 The numerals ‘5’, ‘15’ and ‘25’ in Sawi, Palula and Khowar . . . . 16
15 Expressions of ‘parents’ . . . . 16
16 Expressions of ‘grandparents’ . . . . 17
17 Extended list of values for ‘grandparents’ . . . . 20
18 Expressions of ‘grandchildren’ . . . . 20
19 Expressions of one’s parents and their siblings . . . . 22
20 Expressions of ‘siblings’ children’ . . . . 24
21 Expressions of ‘parents-in-law’ . . . . 27
22 Expressions of one’s siblings’ partners with male/female ego . . . . 28
23 Expressions of one’s partner’s siblings . . . . 30
24 Summary of results: Numerals . . . . 32
25 Summary of results: Kinship Terms . . . . 33
Abbreviations
GHK Greater Hindu Kush
HKIA Hindu Kush Indo-Aryan
B brother
BD brother’s daughter
BS brother’s son
D daughter
DD daughter’s daughter
DS daughter’s son
f feminine
F father
FB father’s brother
FF father’s father
FM father’s mother
FZ father’s sister
m masculine
M mother
MB mother’s brother
MF mother’s father
MM mother’s mother
MZ mother’s sister
S son
SD son’s daughter
SS son’s son
Z sister
ZD sister’s daughter
ZS sister’s son
Contents
List of Figures . . . . i
List of Tables . . . . ii
Abbreviations . . . . iii
1 Introduction . . . . 1
2 Background . . . . 2
2.1 The Greater Hindu Kush . . . . 2
2.1.1 The Geographical Area . . . . 2
2.1.2 The Linguistic Setting . . . . 2
2.2 Previous Research . . . . 3
2.3 Areal Typology . . . . 4
2.3.1 Lexico-Semantic Areality . . . . 4
3 Method and Data . . . . 6
3.1 Data . . . . 6
3.2 Language Sample . . . . 6
3.3 Analysis . . . . 7
3.3.1 Cognate Analysis . . . . 7
3.3.2 Numerals . . . . 9
3.3.3 Kinship Systems . . . . 9
3.4 Comparison . . . . 11
4 Results . . . . 12
4.1 Cognate Analysis . . . . 12
4.2 Numerals . . . . 13
4.2.1 Numeral Bases . . . . 13
4.2.2 Numeral Composition . . . . 14
4.3 Kinship Terms . . . . 16
4.3.1 Parents . . . . 16
4.3.2 Grandparents . . . . 17
4.3.3 Grandchildren . . . . 19
4.3.4 Parents and their Siblings . . . . 21
4.3.5 Siblings’ children . . . . 24
4.3.6 In-Law Relationships . . . . 26
4.4 Summary . . . . 31
5 Discussion . . . . 34
5.1 Cognate Analysis . . . . 34
5.2 Numerals . . . . 37
5.2.1 Numeral Bases . . . . 37
5.2.2 Numeral Composition . . . . 37
5.3 Kinship Systems . . . . 38
5.3.1 Parents . . . . 38
5.3.2 Grandparents . . . . 38
5.3.3 Grandchildren . . . . 39
5.3.4 Parents’ Siblings . . . . 41
5.3.5 Siblings’ Children . . . . 41
5.3.6 In-Law Relationships . . . . 42
5.4 Areal Clusters vs. Genealogical Affiliation . . . . 43
5.5 Method Discussion . . . . 45
6 Conclusion . . . . 47
6.1 Summary of the Study . . . . 47
6.2 Further Research . . . . 47
Bibliography . . . . 51
Appendix I . . . . 52
Appendix II . . . . 55
Appendix III . . . . 61
Appendix IV . . . . 71
Appendix V . . . . 79
1 Introduction
The mountain region of the Hindu Kush is a multiethnic and multilingual melting pot bringing together over fifty languages from six different genera. A majority of these languages are members of the Indo-Aryan branch of the Indo-Iranian languages, while the remaining idioms belong to the Iranian, Nuristani, Tibeto-Burman and Turkic language families, as well as the language isolate Burushaski (Liljegren 2014: 136). Despite its linguistic diversity, the area as a whole is relatively underdescribed, which is due to the remoteness and the politically unstable situation of certain parts of the region. So far, only a handful of the languages have been studied in various length and depth and even less work has been done on an areal-typological level.
Given this background, many aspects of the languages of the Greater Hindu Kush remain to be uncovered and the present thesis aims to close one of these gaps, by shedding further light on the complex linguistic situation from a macro-typological perspective. The purpose of this study is to contribute to the understanding of the languages of the Greater Hindu Kush, especially in regard to their similarity in lexical structure. It aims to answer the following research questions:
1. How do the languages of the Greater Hindu Kush resemble/differ from each other struc- turally, based on numerals and kinship terms?
2. Can significant areal patterns be identified in regard to numeral bases, numeral compo- sition and various kinship terms?
3. Do these patterns mirror the genealogical affiliation of the languages or are they purely areal?
The investigation is carried out in two steps: first, the proposed genealogical relationship between the languages of the Greater Hindu Kush is reexamined by means of a cognate anal- ysis. Second, a number of lexical features are analysed, in order to detect patterns of areality in the area. These two analyses are then synthesised in a last step, where the genealogical affiliation is compared to the areal patterns.
For this endeavour, the study relies on extensive data gathered from speakers of a total of 59 different languages and language varieties spoken in the region. In a first step, a cognate analysis based on core vocabulary is performed, in order to verify the genealogical related- ness previously established by other researchers. Secondly, the numerals and kinship terms are analysed in terms of a number of selected structural features. Lastly, significant patterns in regards to the structural properties of these lexical items are identified, and determined whether these patterns mirror the genealogical affiliation of the languages or represent geo- graphical clusters.
The thesis is outlined as follows. In the second section, an overview of the geographical
area and its languages in question is given, as well as an introduction to areal typology with
particular focus on lexical areality. In section three, the data, language sample and method-
ological approach used for this thesis are presented. The results of the analyses are displayed
in section four and discussed in section five. Section six concludes the thesis by summarising
the important findings as well as predicting possible future research on the present subject.
2 Background
In this section, key notions and elements of this thesis are introduced. The first part is dedi- cated to the relevant region, both in terms of its geographical setting and linguistic context.
The second part contains an introduction to the concept of areal typology, with a special focus on lexical-semantics in regard to kinship terms and numerals.
2.1 The Greater Hindu Kush 2.1.1 The Geographical Area
The region that encompasses the languages relevant for this study is embedded within the highlands of Central and South Asian, as seen in Figure 1. Including parts of Afghanistan, Pakistan, Tajikistan, China and India, the area displays a high degree of linguistic and cultural diversity. As the region not only covers the Hindu Kush mountain range itself, but also sur- rounding areas like the westernmost foothills of the Himalayan arc, the disputed territory of Jammu and Kashmir, and several peripheral Afghan provinces, the term Greater Hindu Kush (GHK) will be used in this thesis, as suggested by Liljegren (2014: 134). Given its ethnic and re- ligious richness, the Greater Hindu Kush is not only linguistically speaking of interest, but also regarding its socio-cultural and political situation (Kreutzmann 1995). Especially Afghanistan has experienced decades of major political instability resulting in civil wars and international interventions. Pakistan has seen the rise of militant activities post 9/11, and has been in an ongoing dispute with India over the borderlands of Kashmir. These preconditions paired with geographically difficult terrain make the GHK quite hard to access, which in turn explains the lack of linguistic research done in the area, despite it being a goldmine of language diversity.
2.1.2 The Linguistic Setting
The Greater Hindu Kush is a major junction of four well-known phyla; it hosts the north- western outliers of the Indo-Aryan languages, the easternmost representatives of the Iranian languages, the westernmost extension of the Sino-Tibetan family and touches on the southern border of the Turkic branch (Liljegren 2014: 134-135).
The majority of the languages spoken in the GHK belong to the northernmost Indo-Aryan languages, which historically have been grouped under the term ”Dardic”. Since any striking similarities between these Dardic languages is most likely due to their longstanding contact in isolation from other Indo-Aryan languages, the languages are now considered to be an areal group instead of a genealogical one (Strand 2001: 251). Hence the group has increasingly been referred to in the purely geographical term ”Hindu Kush Indo-Aryan” (HKIA), which will also be employed in this thesis. The HKIA languages can be further divided into six subgroups, namely Chitral, Kunar, Pashai, Kohistani, Kashmiri and Shina (Bashir 2003: 824- 825). Not all Indo-Aryan languages within the confines of the Greater Hindu Kush are part of the geographical HKIA subgroup; a few of them belong to other sub-groupings, such as Domaki, Gojri or Pahari-Pothwari (Liljegren 2014: 137).
The linguistic neighbours of the HKIA languages, namely Iranian and Nuristani, are both
spoken around and partly within the HKIA sphere of influence. The Nuristani languages, con-
stituting the third main branch of the Indo-Iranian languages, are mainly based in the Afghan
province of Nuristan, and to a lesser extent on Pakistani grounds. The Iranian languages are
spread around the western part of the HKIA continuum, most predominantly represented by
Figure 1: Geographical distribution of languages and linguistic genera in the GHK, represent- ing the traditional classification
Pashto in the south and south-west, and by the Dari/Tajiki dialect continuum in the north (Liljegren 2014: 135).
The West Tibetan varieties found within the GHK realm, Balti, Ladakhi and Purik, are spoken in the Pakistan-administered Gilgit-Baltistan and the Indian-held state of Jammu and Kashmir respectively (Zemp 2018: 1). Kyrgyz and Uzbek, exponents of the Turkic branch, are situated at the periphery of the GHK in the north of Afghanistan. The language isolate Burushaski is mainly spoken in northern Gilgit-Baltistan, but a few hundred speakers are also located in the Indian state of Jammu and Kashmir (Yoshioka 2012: 2-3).
2.2 Previous Research
The main body of previous research consists of studies on individual languages or language families in varying length and depth, but still remain rather limited in their number. Especially Norwegian linguist Georg Morgenstierne has substantially contributed to the research of the Indo-Iranian languages in the first half of the last century. His work includes studies and tenta- tive grammars on a number of Indo-Iranian languages, such as Waigali, Pashto, Wakhi, Dameli, Pashai and Torwali, among others (Morgenstierne 1936; 1938; 1942; 1967). Early accounts of the Nuristani languages have been provided by the research of Strand (1973), Grjunberg (1980) and Degener (2002). The Hindu Kush Indo-Aryan languages have been extensively discussed in Masica (1991) and Bashir (2003).
Earlier areal-typological studies on the languages of the GHK are confined to a small
amount of features, but they already propose compelling evidence for areality. Such features
include the tripartite affricate/fricative distinction (Tikkanen 2008: 254-255), lexically con-
trastive tone (Baart 2014), alignment with curious ergative-accusative splits (Liljegren 2014), the prevalence of left-branching complex structures (Bashir 1988: 401-403), vigesimal numeral systems (Tikkanen 1988: 309), multi-valued deictic systems (Bashir 2003: 823) and grammati- cal evidentiality (Bashir 2006).
2.3 Areal Typology
As a relatively new field within linguistics, areal typology is situated at the junction of lan- guage typology and areal linguistics. The discipline has been defined as ”the study of patterns in the areal distribution of typologically relevant features of languages”. It not only describes the patterns themselves, but also explains the processes generating them, thus adopting both a synchronic and diachronic viewpoint at the same time (Dahl 2001: 1456). While the main inter- est of areal linguistics and areal typology coincide, their perspectives on the matter fundamen- tally differ; areal linguistics’ primary focus lies on the identification of linguistic areas, whereas language typology, and areal typology in particular, are more concentrated on linguistic fea- tures and their geographical distribution (Dahl 2001: 1456). From an areal-typological angle, diversity is just as significant as similarity and areal patterns are taken into account regardless of whether they correspond to the traditional notion of ”linguistic area” or not (Nichols 1992).
Dahl (2001) lists several crucial factors that are involved in the processes of areal diffu- sion such as migration routes, historical relations, bilingualism and the geography of the area encompassing a certain set of languages. In addition to these features, Aikhenvald & Dixon (2001: 1-3) mention four further explanations of linguistic similarity, namely universal prop- erties or tendencies, chance, genetic retention and parallel development, all of which need to be taken into account when investigating areal diffusion.
Koptjevskaja-Tamm (2010: 582-584) introduces a framework for the construction of areal- typological studies which combines both a micro- and a macro-perspective. The micro-per- spective adopts an approach similar to dialectology and traditional areal linguistics in that it provides a systematic and detailed description of linguistic domains across languages va- rieties. The macro-perspective complements the micro-typology by interpreting the findings against a general typological background (Koptjevskaja-Tamm 2010: 584). The combination of both perspectives thus permits a more fine-grained approach with a larger number of sam- pling point than what is usually common for a large-scale typological study, and allows for the identification of usual as well as unusual traits shared in the area (Koptjevskaja-Tamm 2010:
588).
2.3.1 Lexico-Semantic Areality
Analogous to areal typology, lexico-semantic areality is concerned with the diffusion of se- mantic features cross-linguistically in a geographical area (Koptjevskaja-Tamm & Liljegren 2017: 204). Koptjevskaja-Tamm & Liljegren (2017: 205) have identified a number of lexico- semantic phenomena that may serve as indicators of areal clustering. These phenomena in- clude lexico-semantic parallels, shared formulaic expressions, areal-specific lexicalisations and internal organisation of semantic domains.
Kinship Terms One of the key features researched in this study are kinship terms and ter-
minological systems. Systems of family relations are a remarkably promising domain for in-
vestigation, as they potentially mirror cross-community relationships in the region (Liljegren
2017: 146). Research of kinship, initiated by anthropologist Lewis H. Morgan in the nineteenth
century, has mainly been confined within the realms of anthropology and anthropological lin- guistics (cf. Carsten 2000, Schweitzer 2000, Holy 1996 and Dousset 2011). Linguistic research of kinship terminologies is primarily limited to the field of lexical typology (Koptjevskaja- Tamm, Rakhilina, et al. 2015: 437). A major contribution to the research of kinship terminol- ogy within linguistics has been done by Kroeber (1909). Kroeber argues against the classic distinction between classifying and descriptive kinship systems and instead details eight fea- tures he considers to be universal for all kinship terms. These features comprise generation, direct versus collateral, age difference in one generation, sex of the relative, sex of the connect- ing relative, sex of the speaker, consanguineal versus affinal, and the status of the connecting relative (Kroeber 1909: 78-79).
In her work on ”Typology of Kinship Terms”, Nikolayeva (2014) discusses the typological approach to study kinship terms and presents methods, and problems, of describing kinship systems. She points out that in order to objectively compare lexico-semantic systems, core symbols of the six simplest kinship terms should be used instead of taking another living language as the basis (Nikolayeva 2014: 30). More complex term are then expressed by com- bination of the core items. In the present thesis, the following abbreviations will be used:
Term Abbreviation
father F
mother M
brother B
sister Z
son S
daughter D
Table 1: Kinship Abbreviations
Numerals Within linguistics, numerals and numeral systems have long been interesting from a typological and historical point of view (Schapper & Klamer 2014: 285). A number of studies on the typological variety and distribution of numeral systems with respect to their ba- sic mathematical structure have been published beginning in mid-nineteenth century. Impor- tant contributions have been, among others, Greenberg (1979), Hurford (1975, 1987), Comrie (2013) and Hammarström (2007, 2010). While the 10-based (decimal) and 20-based (vigesimal) systems have been identified as the cross-linguistically most common ones, other arrange- ments such as body-part tally systems and 5- (quinary) or even 7-based (senary) systems have also been found among the languages of the world (Schapper & Klamer 2014: 249-250).
As the preliminary studies by Edelman (1999: 221) and Liljegren (2017: 143) have shown, the numeral systems of the languages of the GHK are peculiar, in that they overwhelmingly favour a vigesimal or hybrid vigesimal-decimal system. Different hypotheses have been put forward in order to explain this curiosity: Edelman (1999: 221-223) argues that the vigesimal system might be an innovation of these languages, as the original number system seems to have been decimal, which is still found in other relatives outside the GHK. Tikkanen (1988:
309) on the other hand sees potential Burushaski substratum at the origin of the vigesimal
system.
3 Method and Data
This chapter is dedicated to the data and methodology used for this thesis. The first section details the data collection process and the languages that have been sampled. The second section outlines the methodological approach and analytical procedure of the data evaluation.
In terms of its theoretical framework, the present study adopts a functional-typological perspective as defined in Croft (2003: 2). This analysis is exploratory in its nature and wholly data-driven.
3.1 Data
The data used for this study has been gathered in collaborative workshops with native speakers between 2015 and 2018. These multilingual workshops were held in Islamabad (Pakistan), Gilgit (Pakistan), Kabul (Afghanistan), Faizabad (India) and Srinagar (India) in addition to a number of individual sessions. The data collection was carried out by Henrik Liljegren within the frame of the research project ”Language contact and relatedness in the Hindu Kush region”, funded by the Swedish Research Council (2015-2019). The goal of the project is the creation of a typological profile of the region on the basis of data from 60 varieties spoken in the Greater Hindu Kush area.
The collected material includes, among others, the following data sets:
– Basic vocabulary word list (40 items), based on the Automated Similarity Judgement Program
1– Numerals (59 items), based on Eugene Chan’s questionnaire for the cross-linguistic project ”Numeral systems of the world’s languages”
2– Kinship terms (95 items), compiled by Henrik Liljegren
For the elicitation, the participants filled in written questionnaires that were provided in both English and either Urdu, Pashto or Dari, depending on the contact language of the consultant.
In addition to the written questionnaires, audio recordings were produced of the participants reading aloud their answers.
In order to avoid ambiguity, the terms of the kinship questionnaire were mostly of descrip- tive nature and not representing the actual lexical term used in English, Urdu, Pashto or Dari.
This is the reason why, for example, the terms corresponding to English cousin, are given as one’s parent’s sibling’s child. The consultants were thoroughly instructed on how to fill in the questionnaire and that no literal translations of the descriptive kinship terms were demanded.
However, this does not exclusively rule out the possibility, that certain translations might still be literal and not lexical.
3.2 Language Sample
The sample used for this thesis contains 59 varieties, with at least one representative of each phylum and, when relevant, of each subgroup. Table 2 gives an overview of all the languages as well as the percentage each of the subgroups makes out of the whole sample. Figure 2 shows their geographical distribution within the Greater Hindu Kush. The complete sample
1https://asjp.clld.org
2https://mpi-lingweb.shh.mpg.de/numeral/
of languages including their language code, exact location of the speaker, and time and place of the recording, can be found in Appendix I.
Family Subfamily Count Sampled Languages
Indo- European
Indo- Aryan
29 (49%) Kalasha, Khowar, Kashmiri (Ind), Kashmiri (Pak), Ba- teri, Gawri, Indus Kohistani, Torwali, Dameli, Gawar- bati (Afgh), Gawarbati (Pak), Alasai, Alingar, Alis- hang, Amla, Aret, Chalas, Korangal, Sanjan, She- mal, Brokskat, Kalkoti, Kohistani Shina, Kundal Shahi, Palula, Sawi, Shina (Ind), Shina (Pak), Ushojo, Gojri (Afgh), Gojri (Pak), Hindko, Pahari-Pothwari
Iranian 13 (22%) Dari, Ishkashimi, Munji, Parachi, Pashto (Afgh), Pashto (Ind), Pashto (Pak), Roshani, Sanglechi, Shughni, Wakhi (Afgh), Wakhi (Pak), Yidgha
Nuristani 6 (10%) Ashkun, Kamviri, Kati Eastern, Kati Western, Prasun, Waigali
Turkic Common
Turkic
2 (3%) Kyrgyz, Uzbek Sino-
Tibetan
Tibeto- Burman
3 (5%) Balti, Ladakhi, Purik Isolate Burushaski 2 (3%) Hunza, Nagar
Table 2: Language Sample (classification based on Glottolog 3.4 (Hammarström et al. 2019)) The sample is very tight and covers most languages spoken in the area, except for languages that are extinct or moribund. The broad range of the sample allows to exclude chance as an un- derlying factor of potential areal correlations. Further, the prevalence of the HKIA languages is also noticeable in the sample, making up nearly half of all the languages included. Follow- ing the research strategy suggested by Koptjevskaja-Tamm (2010: 582-589) for investigating geographical regions in areal-typological terms, the language sample contains a number of varieties from the same language, spoken in different countries or neighbouring areas. As mentioned in the previous chapter, this variation in languages will allow for the more detailed micro-perspective.
3.3 Analysis
Concerning the processing of the data, the audio files have been phonetically annotated in IPA by Noa Lange, reassembled in an Excel file and colour-coded according to language affiliation by Nina Knobloch. All of the following steps of the analysis have been carried out by the author of this thesis. As initially stated, only the three lists on basic vocabulary, kinship terms and numerals have been taken into account for the analysis. A number of features based on numeral terms and kinship systems have been selected. The exact procedure of analysis is detailed in the following sections.
3.3.1 Cognate Analysis
The first step was to reassess the genetic relationship of the different languages by identifying
cognates. In order to do so, a list of 83 items was used, which included the 40-item word list
as well as basic numerals and selected kinship terms. Words were considered to be cognates
Figure 2: The language sample of the Greater Hindu Kush
if they were deemed to be phonologically similar enough, to have developed from the same original word. All languages sharing the same cognate for one word would be assigned the same symbol as exemplified by the words ‘sun’ and ‘tooth’ in Table 3. The values of the symbols are of no particular importance; they merely serve as indicators of the different sets of cognates for each word.
language ‘sun’ symbol ‘tooth’ symbol
Ashkun su 2 dont 3
Balti ŋimah 3 soː 2
Bateri suːr 2 daːn 3
Brokskat suri 2 dæni 3
Burushaski, Hunza sa 1 ameː 1
Burushaski, Nagar sah 1 ameː 1
Dameli sir 2 dan 3
Dari, Darwoz aftoːw 7 danduː 3
Domaki toː ? don 3
Table 3: Cognate analysis example
Once all lexical items were examined, the assigned values were then fed into the software SplitsTree
3, which visualised the lexical similarity among the languages. SplitsTree uses an algorithm called NeighborNet, which allows for the construction of phylogenetic networks by agglomerating clusters, all without forming a hierarchy, as opposed to the traditional tree model (Bryant & Moulton 2004: 255). The thus obtained NeighborNet visualisation (see Figures
3The software is available at www.splitstree.org
3 and 16) shows the level of similarity between each language; the further away they are from each other, the less lexically similar they are.
3.3.2 Numerals
For the analysis of the numerals, two features have been selected, namely their numeral bases and their composition.
The base of a numeral system is defined by the value n, in a numeral expression constructed after the pattern xn + y, as for example in Swedish (1), which uses a 10-based system:
(1) Swedish fyr-tio-fyra four-ten-four
‘fourty-four’
WALS identifies six different values for the numeral bases, namely decimal, hybrid decimal- vigesimal, pure vigesimal, other base, extended body-part system and restricted. The hybrid decimal-vigesimal system differs from the pure vigesimal system in that it only expresses num- bers up to 99 vigesimally, to then switch to a decimal format (Comrie 2013). For this analysis, however, any language expressing at least one multiple of ten by any combination including the lexical numeral 20 has been regarded as vigesimal.
The numerals have further been analysed in regard to their composition of the base and the single digit. The different possible patterns are displayed in Table 4; either the base is always followed by the single digit as in French, or the single digit always precedes the base, as in German, or the composition changes, as in English:
language pattern ‘17’ ‘27’
French base + x dix-sept vingt-sept
German x + base siebzehn siebenundzwanzig English x + base, base + x seventeen twenty-seven
Table 4: Patterns for numeral composition
3.3.3 Kinship Systems
The kinship terms have been examined on the basis of their semantic reach and their structural composition, i.e. whether a kinship term is lexically expressed or semantically constructed, and if this construct is recurring in other languages. The sets of kinship terms that have been selected for the analysis, together with their assigned values, are shown in Table 5. The values have been set up in the course of the analysis and were not predefined.
The parents have been studied in terms of their composition, i.e. whether they are ex- pressed through a ‘mother-father’ compound or their own lexical expression. The identified values are (1) ‘father-mother’ compound (2) ‘mother-father’ compound and (3) ‘other’.
The grandparents have been analysed according to how descriptive the systems are; if a language differentiates between all four relations, it has been classified as a ‘four-way split’
type, if the difference only arises between the sex of the relative, the language was classified
as a ‘two-way split’ type.
The grandchildren have been analysed in a similar fashion, i.e. the languages were classi- fied according to whether they are differentiating between the sex of the grandchild, the sex of the connecting relative or both. Four values have been set up for classifying the possible patterns of expressing ‘grandchild’.
The parents’ siblings were classified in comparison to the terms for ‘father’ and ‘mother’, i.e. whether they use the same term for ‘father’ and paternal brother, respectively ‘mother’
and maternal sister, or entirely different lexical expression. This way of analysing yielded six different values that were possible for expressing one’s parents and their siblings.
One’s siblings’ children have been categorised analogous to ‘grandchildren’, which also resulted in four possible values, where a language either has four distinct lexical terms, only one lexical term, or varying combinations of grouping certain relations together under one term.
Regarding the in-law relationships, they all have been categorised similar to the grandpar- ents according to the amount of differentiation within each system. The siblings-in-law have been split up into one’s siblings’ partner and one’s partner’s siblings.
As mentioned in section 2.3.1, abbreviations of the six most basic kinship terms have been used and any further relations have been expressed with a combination of the initial six terms.
This analysis is by no means exhaustive and only serves as an entry point to the vast field of kinship terminologies in the languages of the GHK.
Feature Values Feature Values
parents
father-mother compound
siblings’ children
BS≠BD≠ZS≠ZD
mother-father compound BS=ZS/BD=ZD
other BS≠BD/ZS=ZD
grandparents
two-way split BS=BD=ZS=ZD
four-way split other
other
parents-in-law
two-way split
grandchildren
SS=DS/SD=DD three-way split
SS≠SD≠DS≠DD four-way split
SS=SD=DS=DD
siblings’ partners
two-way split
SS=SD/DS≠DD three-way split
other four-way split
parents and their siblings
F≠FB≠MB/M≠MZ≠FZ other
F≠FB≠MB/M≠MZ=FZ
partner’s siblings
two-way split
F≠FB=MB/M≠MZ=FZ three-way split
F=FB≠MB/M=MZ≠FZ four-way split
F=FB≠MB/M≠MZ=FZ other
F≠FB≠MB/M=MZ=FZ other
Table 5: Features and values for kinship analysis
3.4 Comparison
Once all the features have been analysed, they have been plotted on feature maps in order to trace potential patterns of areality. For this, Google’s MyMaps app
4has been used, which provided rudimentary maps. Unfortunately these maps are not equipped with scales and a subsequent insertion of the scales was not possible. For a more accurate representation and interpretation of the clusters, a scale would have been crucial. Clusters have been identified purely on the basis of visual estimate, and no specific geospatial analysis was conducted.
For further comparison, a similar approach has been adopted as for the cognate analy- sis; languages sharing the same pattern for a certain feature were assigned the same symbol, i.e. languages representing ‘parents’ as a ‘father-mother’ compound were assigned a differ- ent value than languages that express ‘parents’ with their own lexical term. The SplitsTree software was then used again to generate a graphic approximation of the similarity between the languages which revealed potential patterns of vergence, that is to say patterns of either divergence or convergence. These findings have subsequently been compared to the findings of the cognate analysis to see whether these structural patterns reflect the genealogical rela- tionship or if they indicate areal clustering due to factors such as language contact or parallel development.
4available at https://www.google.com/maps/d/
4 Results
The present chapter details the results from the analysis. The first sections specifies the find- ings from the cognate analysis based on basic vocabulary. The sections thereafter treat the results from the analysis of the numerals as well as the kinship terms.
The genealogical implications and geographical distributions of the following results will be addressed in section 5.
4.1 Cognate Analysis
The word list, together with additional items from the numeral data set and the kinship ques- tionnaire, have been analysed in terms of cognacy.
The analysis of the cognates has yielded some interesting results; it not only confirmed the previously established genealogical relationships between the languages of the GHK, it also uncovered certain tendencies within the sample. Several lexical items have shown to be more or less uniform throughout the entire sample and across language families, while other lexical items appear to be unique, sometimes even within language families.
Generally, words like pronouns, body-parts, numbers, core kinship terms and terms re- lated to nature tend to be cognates more often. Table 6 presents a few examples, where most languages share cognates for each of the words. Obvious outliers are the non-Indo-European languages, such as Burushaski, Kyrgyz and Purik.
On the other hand, words like complex kinship terms, verbs and individual words from certain categories displayed more variation, even within language families. This can be ob- served in Table 7, where even related languages such as Sawi, Gawri, Kalasha and Pashai have very different lexical entries for the same word.
‘I’ ‘ear’ ‘leaf’ ‘name’ ‘mother’
Ashkun (Nuristani) aj kamaʈ͡ʂə paːr nam aɽaj Purik (Tibeto-Burman) ŋa hr̥naː loːma miŋ ama
Gawri (Kohistani) jɑ kɑn pɑɬ nɑːm jaj
Sawi (Shina) ma kaːɳ paːɬu naːm jeːj
Hunza (Burushaski) d͡ʑe altumal tap ajik ami Gawarbati, Pak (Kunar) ãː kʰamʈa faʈa naːm d͡ʒaːj
Parachi (Iranian) ɑːn guːʃ puːn nɑːm mɑː
Gojri, Afgh (Indo-Aryan) hũː kaːn baːmfru naː maː Kashmiri, Ind (Kashmiri) bɨ kan panɨ naːw moːd͡ʒ
Kalasha (Chitral) a kõ põ nom aːja
Kyrgyz (Turkic) men qulaq berk ɑt ene
Chalas (Pashai) a χoɽ paʈek nom aːi
Table 6: Examples for words more likely to be cognates
The lexical similarity of the languages has also been rendered visually by using the SplitsTree software, which generated a NeighborNet seen in Figure 3. The lines between languages indi- cate their lexical similarity; the closer and shorter a line is, the more similar the languages are to each other. A distance matrix
5has also been calculated and it follows the same principle;
5Due to its size, the distance matrix could not be included in this document, but it is accessible viathis link.
‘liver’ ‘hear’ ‘mountain’ ‘see’ ‘wife’s brother’
Ashkun (Nuristani) vijon purud͡zu kanɖrə vendə aːj mi Purik (Tibeto-Burman) t͡ʃʰinma ɲanma riː juŋma nunu Gawri (Kohistani) jan bud͡ʒuːg kʰɑn t͡ʃʰaʈɑːg ʃɑːʃur
Sawi (Shina) jiːnuː bud͡ʒiloː ɖaɖu paʃiloː awχaj
Hunza (Burushaski) akin dojalas t͡ɕʰar barenas arik Gawarbati, Pak (Kunar) andeːt͡s ʂuɳɖaːwa daɽa - gala baːlawa ʐami Parachi (Iranian) d͡ʒigar hatu dahɑːr duɽuː χusur bura Gojri, Afgh (Indo-Aryan) pʰapru suɳun ɖʰaːkuː deːkʰaɳ saːɽoː Kashmiri, Ind (Kashmiri) krehən maːz bɔːzun baːl wut͡ʃʰun həhər Kalasha (Chitral) ʒaŋgu kõːkarik denta d͡ʒagek weːwaj
Kyrgyz (Turkic) bor uqan qɯr kɯrgi quda
Chalas (Pashai) zəɽ aːrak izor paɽajk ʃiwuɽuk
Table 7: Examples for words less likely to be cognates the lower the number, the more similar two languages are to each other.
Figure 3: Visual representation of the cognate analysis
4.2 Numerals
4.2.1 Numeral Bases
The analysis fo the numeral bases has been based on a more generalising definition of viges-
imality, as discussed in section 3.3 and section 5.5. Adopting said broad definition has led to
the present results, where a majority of the GHK languages display a vigesimal-based system, as has already been indicated in previous studies (Liljegren 2017: 144). This does not reflect the global tendency, which clearly favours the decimal system in 64% of the world’s languages (Comrie 2013). Only a third of the GHK languages adhere to the decimal type, as seen in Table 8.
System Distribution Vigesimal 37 (64%) Decimal 21 (36%)
Total 58
Table 8: Numeral Bases
The languages displaying a decimal system are quite uniform and no unexpected patterns have been identified (9).
Number Translation
ten daː
twenty biːst thirty siː forty t͡ʃəl fifty pind͡ʒɑː sixty ʃaːst seventy aftɑːd eighty aʃtɑːd ninety nawad hundred sad
Table 9: Ishkashimi multiples of ten
As discussed in the previous chapter, any language using the numeral 20 at least once to form any other multiple, has been considered vigesimal. While most of these languages consistently use the lexical numeral 20 as a means of expressing multiples of ten, as seen in example (10), in some cases the system deviates slightly from this constant vigesimal system. Afghan Gojri, for example, has its own numeral for ‘forty’ t͡ʃɽiː but switches to a vigesimal system for the remaining multiples of ten, as seen in (11).
One very curious case is Munji. While superficially Munji appears to have a normal decimal- based system, it becomes clear after close observation that Munji actually does not have any ten-based numerals at all. Instead, Munji only has numerals for 0 to 9, and any number higher than nine is a mere combination of these basic numerals, as for example juː o sifɛr – ‘ten’, which literally means ‘one and zero’. All single digits and the multiples of ten for Munji are given in Table 12.
4.2.2 Numeral Composition
The actual composition of the numerals reveals three different patterns: an n + base type, a
base + n type, as well as a mixed pattern, which uses n + 10 but switches to base + n for any
Number Translation Composition
ten daʃ 10
twenty biʃiː 20
thirty biʃi oː daʃ 20 + 10 forty du biʃiː 2 + 20 fifty du biʃi oː daʃ 2 + (20 + 10) sixty traː biʃiː 3 + 20 seventy traː biʃi oː daʃ 3 + (20 + 10) eighty t͡ʃoːr biʃiː 4 + 20 ninety t͡ʃoːr biʃi oː daʃ 4 + (20 + 10) hundred pãːt͡ʃ biʃiː 5 + 20
Table 10: Dameli multiples of ten
Number Translation Composition
ten da 10
twenty biː 20
thirty triː 30
forty t͡ʃɽiː 40
fifty daːte t͡ʃɽiː 10 + 40 sixty trɛː biː 3 + 20 seventy daːte trɛː biː 10 + (3 + 20) eighty t͡ʃaːr biː 4 + 20 ninety daːte t͡ʃaːr biː 10 + (4 + 20)
hundred suː 100
Table 11: Gojri (Afgh) multiples of ten
Number Translation Number Translation
zero sifɛr ten juː o sifɛr
one juː twenty lə o sifɛr
two lə thirty çiraj o sifɛr
three çiraj forty t͡ʃəfiːr o sifɛr four t͡ʃəfiːr fifty pɑːnd͡ʒ o sifɛr five pɑːnd͡ʒ sixty ɑːχʃe o sifɛr
six ɑːχʃɛ seventy ɑːvde o sifɛr
seven ɑːvdɛ eighty ɑːʃce o sifɛr eight ɑːʃcɛ ninety new o sifer nine nɛw hundred juː o lə siferiː
Table 12: Munji numerals
number above 19. Table 13
6summarises the distribution of the different patterns. The sample displays a tendency towards the mixed pattern, with almost half of the languages adhering to that pattern. The patterns n + base and base + n are both equally spread among the other half of the languages.
Composition Distribution n + 10, 20 + n 26 (46%) n + 10, n + 20 17 (30%) 10 + n, 20 + n 14 (24%)
Total 57
Table 13: Numeral Composition
Table (14) gives an overview of the three types, with examples illustrating each of them. In- terestingly, all of the languages in this example are Indo-Aryan and while their lexical forms are similar, the structural make up of the numbers is clearly different, which points towards an areal feature rather than a genealogical one.
‘5’ ‘15’ ‘25’
Sawi n + 10, n + 20 paːnd͡ʒ pand͡ʒiʃ paːnd͡ʒɑːn biʃ Palula n + 10, 20 + n paːnd͡ʒ pand͡ʒiːʃ bʰiʃeː paːnd͡ʒ Khowar 10 + n, 20 + n põːt͡ʃ d͡ʒoʃ põːt͡ʃ biʃir põːt͡ʃ Table 14: The numerals ‘5’, ‘15’ and ‘25’ in Sawi, Palula and Khowar
4.3 Kinship Terms 4.3.1 Parents
The expression of ‘parents’ is predominantly a compound formed by the juxtaposition of the terms for ‘mother’ and ‘father’. Out of these languages, a majority of 34 languages put ‘mother’
first and ‘father’ second, while the remaining 15 form the compound the other way round. Ten languages express ‘parents’ with their own lexical item. Table 15 summarises these numbers.
Parents Distribution
mother-father compound 34 (58%) father-mother compound 15 (25%) special lexical item 10 (17%)
Total 59
Table 15: Expressions of ‘parents’
Kundal Shai, see example (2), is one of these languages, that puts ‘mother’ first and ‘father’
second. Uzbek is an example of a language that places ‘father’ first, followed by ‘mother’, as seen in (3).
6This table excludes Pahari-Pothwari as well as Kundal Shahi, as the numeral terms for these languages are missing in the data.
(2) Kundal Shahi [shd]
meːl mother
maːl father
‘parents’
(3) Uzbek [uzs]
ɑːta father
ɑːna mother
‘parents’
The remaining ten languages have a unique lexical term for ‘parents’ which is not derived from ‘mother’ and/or ‘father’ as in example (4):
(4) Dari (Darwoz) [prs (d)]
a. bɑːba – ‘father’
uːma – ‘mother’
b. χuːnawɑːda – ‘parents’
4.3.2 Grandparents
The ways of expressing ‘grandparents’ among the GHK languages can be divided up into three different patterns. The most common way to designate one’s grandparents is to simply dis- tinguish between ‘grandfather’ and ‘grandmother’, regardless of their paternal or maternal affiliation. A good third of all languages in the sample differentiate between all four possible types of relations. Two languages adhere to their own patterns, which do not fit into either the two-way nor the four-way split type. The different patterns and their distribution are given in Table 16
7.
Value Sample representation
two-way split 37 (64%)
four-way split 19 (33%)
other 2 (3%)
Total 58
Table 16: Expressions of ‘grandparents’
Most languages adhering to the first type either have distinct lexical items for ‘grandfather’
and ‘grandmother’, as in 5, or items based on the term of ‘father’ and ‘mother’ with additional markers indicating the older generation.
7Pakistani Kashmiri has been disregarded in this analysis as the term for ‘mother’s father’ is missing in the data.
(5) Brokskat [bkk]
a. duːdo – ‘grandfather’
b. dædeː – ‘grandmother’
The additional marking of the senior generation is either done by reiteration of the original parental term (6) or by adding an adjective meaning ‘big’ or ‘old’ (7).
(6) Sanglechi [sgy]
a. taːt taːt father.father
‘grandfather’
b. naːn naːn mother.mother
‘grandmother’
(7) Sawi [sdg]
a. gaːna old.m
baːbu father
‘grandfather’
b. gaːni old.f
jeːj mother
‘grandmother’
The four-way split type can further be divided into languages subscribing to a FF≠MF/FM≠MM type or a FF≠MF≠FM≠MM type. The FF≠MF/FM≠MM type is characterised by phonologically similar words for both ‘grandfather’ with the distinction between paternal and maternal affil- iation is made apparent by means of different consonants. The paternal side is usually defined by the use of bilabial consonants (8a) while the maternal side is marked by nasal consonants, as it is the case in Palula (8b).
(8) Palula [phl]
a. dôːdo – ‘father’s father’
dêːdi – ‘father’s mother’
b. môːmo – ‘mother’s father’
mêːmi – ‘mother’s mother’
This subtype also includes languages that add a specific maternal marker to the otherwise identical lexical item for both ‘grandmother’ and ‘grandfather’. This maternal marker has been found in four languages of the sample, and it is usually a term derived from ‘mother’ as seen in Kashmiri (9):
(9) Kashmiri (Ind) [kas (i)]
a. maːd͡ʒi mother
bɨɖi bab grandfather
‘maternal grandfather’
b. maːd͡ʒi mother
naːnʲ
grandmother
‘maternal grandmother’
The FF≠MF≠FM≠MM pattern expresses all four types of relations lexically different. In this regard, Gawarbati (Afgh) strikes as particularly special, as it appears to have lexical items for
‘father’s father’ and ‘father’s mother’ but not for the maternal grandparents. In fact, ‘mother’s
father’ seems to be a combination of ‘father’ and the lexical item for ‘paternal grandfather’, as
seen in (10):
(10) Gawarbati (Afgh) [gwt (a)]
baːp father
buɖa
(paternal).grandfather
‘maternal grandfather’
The term for ‘maternal grandmother’ d͡ʒaːjani d͡ʒaːj looks like a literal translation of ‘mother’s mother’.
Equally curious is the case of Kalkoti, displayed in example (11), where no clear pattern can be made out.
(11) Kalkoti [xka]
a. leːg baːn – ‘father’s father’
daːd – ‘father’s mother’
b. deːd – ‘mother’s father’
maːm – ‘mother’s mother’
The two languages classified as ‘other’ – Pashai (Aret) and Uzbek – do in fact display a three- way split. The Aret variety of the Pashai languages uses a FF≠MF/FM=MM type, where both maternal and paternal grandmother are expressed with the same lexical item but a difference is made between FF and MF, as exemplified in (12):
(12) Pashai (Aret) [aee (at)]
a. bəːwəs bɑːbeː – ‘father’s father’
gaɽeːam – ‘father’s mother’
b. bɑːbeːam – ‘mother’s father’
gaɽeːam – ‘mother’s mother’
Uzbek follows a very curious FF=MF/FM≠MM pattern, where both grandfathers share the same term, but not the grandmothers. The grandfathers are called ‘big fathers’ – ɑːta ‘father’, buːwa ‘big’, while the grandmothers are expressed more descriptive. The Uzbek example is shown in (13):
(13) Uzbek [uzs]
ɑːta mɑːmɑː – ‘father’s mother’
ɑːna mɑːmɑː – ‘mother’s mother’
Table 17 is taking these different sub-types into account and gives a more detailed overview of the expression of ‘grandparents’.
4.3.3 Grandchildren
Grandchildren are expressed in several different ways. Half of all languages simply have two
separate terms to refer to either ‘grandson’ or ‘granddaughter’. The most descriptive type,
Value Sample representation
FF=MF/FM=MM 37 (64%)
FF≠MF/FM≠MM 15 (26%)
FF≠MF≠FM≠MM 4 (7%)
FF=MF/FM≠MM 1 (1%)
FF≠MF/FM=MM 1 (1%)
Total 58
Table 17: Extended list of values for ‘grandparents’
where each relation has its own lexical term, makes up 22% of the sample. The simplest de- scriptive type, namely only one term for all relations, has been found in eight languages. The remaining eight languages of the sample display their own patterns that do not fall into any of the above mentioned categories. The different patterns and their distribution are shown in Table 18.
Value Sample representation
SS=DS/SD=DD 30 (51%)
SS≠SD≠DS≠DD 13 (22%)
SS=SD=DS=DD 8 (14%)
SS=SD/DS≠DD 3 (5%)
other 5 (8%)
Total 59
Table 18: Expressions of ‘grandchildren’
The predominant type found in the sample simply distinguishes between the sex of the grand- children, disregarding the sex of the connecting relative, as exemplified in the Pakistani variety of Pashto (14):
(14) Pashto (Pak) [pbu (p)]
a. nwaseː – ‘son’s son’
nwasej – ‘son’s daugther’
b. nwaseː – ‘daughter’s son’
nwasej – ‘daughter’s daughter’
This pattern is followed by the most descriptive type, where all four relations are expressed differently, as in Hindko (15):
(15) Hindko [hno]
a. poːtraː – ‘son’s son’
poːtriː – ‘son’s daughter’
b. dʰeːtraː – ‘daughter’s son’
dʰeːtriː – ‘daughter’s daughter’
Parachi (16) is one of the languages that adheres to the least descriptive pattern, with only one lexical item for ‘grandchild’:
(16) Parachi [prc]
a. nawɑː – ‘son’s son’
nawɑː – ‘son’s daughter’
b. nawɑː – ‘daughter’s son’
nawɑː – ‘daughter’s daughter’
Three languages in the sample display a SS=SD/DS≠DD pattern, where the son’s descendants share the same term while a distinction is made between the daughter’s male and female offspring, as seen in Kyrgyz (17):
(17) Kyrgyz [kir]
a. nibire – ‘son’s son’
nibire – ‘son’s daughter’
b. d͡ʒen – ‘daughter’s son’
t͡ʃibire – ‘daughter’s daughter’
A few special cases have been identified that do not resemble any of the previously defined patterns. One such case is Uzbek, where both granddaughters share the same term qiz niːwara, which is clearly a compound of ‘daughter’ qiz and ‘grandchild’ niːwara. The son’s son follows a similar pattern with ‘son’ wuʁul and niːwara. Curiously enough, however, the term for
‘daughter’s son’ is a compound of bat͡ʃa and niːwara, where bat͡ʃa is most likely a Persian loan for ‘son’.
Equally curious is the case of the in Darwoz spoken Dari variety, which has the Persian word for ‘son’s son’ nuwasa, aand a derived form of ’sister’ χɑːr as both ‘son’s daughter’ and
‘daughter’s son’ χɑːrzɑː. ‘Daughter’s daughter’, however, is the compound daχtari χɑːr, which literally translates to ‘daughter’s sister’.
Gawri falls in the same category, as it has one term covering two kinship relations, but as opposed to Uzbek and Darwoz Dari, Gawri summarises both ‘daughter’s son’ and ‘daughter’s daughter’ under the term nosaj, while ‘son’s son’ is expressed through poːɬ and ‘son’s daughter’
through peːɬ.
Both Waigali and the Afghan variety of Wakhi have the same term for all relations but one;
in Waigail’s case, the ‘son’s son’ is expressed as nawaː, while all others fall together under the term nut.
Afghan Wakhi covers all relations under the term nəpɨs, except for the daughter’s daughter which is a compound formed of ‘daughter’ ðəɣd and ’child’ zəman. Interestingly enough, the term for ‘child’ in Afghan Wakhi is wəxtək, but still zəman in the Pakistani variety of Wakhi.
4.3.4 Parents and their Siblings
The parents’ siblings have been compared not only to each other, but also to the respective
terms for ‘father’ and ‘mother’. Six different patterns have been identified in the data: (1)
separate terms for all six types of relations, (2) the male relations have different terms; the
term for mother’s and father’s sister is the same but different from mother, (3) father’s and
mother’s brother are the same but differ from father; father’s and mother’s sister are the same
but differ from mother, (4) father and father’s brother are the same but differ from mother’s brother; mother and mother’s sister are the same but different from father’s sister, (5) father and father’s brother are the same but different from mother’s brother; mother’s and father’s sister are the same but different from mother, and (6) father, father’s brother and mother’s brother are all distinct; mother, mother’s sister and father’s sister are the same.
The most predominant pattern is the one differentiating between all possible relations, making up two thirds of the entire sample. All other types are considerably less common and each one is only found in a handful of languages. Two languages do not adhere to any of the defined values and are thus classified as ‘other’. Table 19
8summarises the patterns and their distributions.
Value Sample representation
F≠FB≠MB/M≠MZ≠FZ 37 (64%)
F≠FB≠MB/M≠MZ=FZ 8 (14%)
F≠FB=MB/M≠MZ=FZ 6 (10%)
F=FB≠MB/M=MZ≠FZ 4 (7%)
F=FB≠MB/M≠MZ=FZ 2 (3%)
F≠FB≠MB/M=MZ=FZ 1 (2%)
Total 58
Table 19: Expressions of one’s parents and their siblings
The most descriptive type has six different lexical items for each of the relations (18):
(18) Gojri (Pak) [gju (p)]
a. baːp – ‘father’
t͡ɕaːt͡ɕu – ‘father’s brother’
maːmoː – ‘mother’s brother’
b. mãː – ‘mother’
moːsiː – ‘mother’s sister’
pʰupʰi – ‘father’s sister’
Type (2) has five different lexical items, the only terms overlapping are the ones for mother’s sister and father’s sister. The Pashai variety of Alingar is one example of a type (2) language (19):
(19) Pashai (Alingar) [psi (ar)]
a. bwɑː – ‘father’
kɑːkɑː – ‘father’s brother’
mɑːmɑː – ‘mother’s brother’
b. ɑːi – ‘mother’
mɑːmoː – ‘mother’s sister’
8Balti is missing in this analysis, as the term for ‘father’s brother’ has not been recorded.
