Northern Talysh 3895866814

Citation preview

Northern Talysh 1

Wol~gang Schulze

Languages of the World/Materials 380 LINCOM EUROPA

I

LW/M380

3

NORTHERN TALYSH

Contents

LINCOM EUROPA Freibadstr. 3 D-81543 Muenchen [email protected] http://home.t-online.de/home/LINCOM.EUROPA

Ali rights reserved, including the rights of translation into any foreign language. No part of this book may be reproduced in any way without the permission of the publisher.

Printed in E.C. Printed on chiarine-free paper

Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data A catalogue record for this publication is available from Die Deutsche Bibliothek (http://www.ddb.de)

ISBN 3 89586 681 4

1. lntroduction 2. The language 3. Palangi ahvolot 4. Phonemes and Functional Load 4.1 A synchronic survey 4.2 The diachrony of the Northem Talysh sound system 5. The morphological inventory of Northem Talysh - an overview 5.1 Nouns 5.2 Adjectives 5.3 Pronouns 5.4 Verb 5.5 Numerals 5.6 Conjunctions 5.7 Particles 5.8 The bound morphemes of Northern Talysh 6. Actancy in Northern Talysh 6.1 Case marking of nouns 6.2 Pronouns, critics, and relational behavior 6.2.1 Pronouns 6.2.2 Agreement: Clitics and suffixes 6.3 Tense marking, critics, and relational behavior 7. Beyond the default: The periphery of relational behavior in Northern Talysh 8. Conclusions

5 6 8 9 9

15 17 17 18 19

22 25 26 26 27 29 29 34 34 42 45 58 66

Appendix I: Palangi ahvolot: The text Appendix Il: Palangi ahvolot: An etymological index Appendix lii: Abbreviations

89

References

91

lndex

93

68 78

LW/M380

5

NORTHERN TALYSH

1. lntroduction 1 The present booklet is a report on the morphosyntax ofNorthem Talysh as documented in an oral account that I recorded in Baku (Azerbajdzhan) in 1986. The dialect of this account which has been told to me by Novruz Mamedov is that of Shuvi. In this booklet, I will concentrate on the salient features of Northem Talysh morphosyntax in terms of an 'actancy typology' (coding of relational primitives, agreement, TAM system, word order etc.). In order to cover the high degree of co-paradigmatization, the analysis is formulated in terms of an interpretative 'text grammar' which is supplemented by data from other sources as weil as from own field notes. The morphosyntactic features ofNorthem Talysh are interpreted in the tradition of Functional Typology (making specific reference to the 'Grammar of Scenes and Scenarios' as developed in Schulze 1998). This includes references to both the formal and fimctional diachrony of Talysh as weil as to parallels from other languages in the area. The appendix gives the text itself, provided with interlinear glosses and translation. The interpretative part (section 6) is preceded by a comprehensive treatment ofNorthem Talysh phonology from a synchronic point to which is added a survey of the morphological inventory of Northem Talysh as it shows up both in written sources and in my own field notes. In this section (as weil as in section 6) emphasis has been laid on the question to which extent the Northem Talysh morphology can be described from an etymological perspective. lt should be noted that the description ofthe diachronic background ofNorthem Talysh is limited by the fact that there are hardly any reliable data on earlier stages of the language. Though Talysh clearly belongs to the Iranian language group, we cannot claim that either Old Persian or Avesta- being the earliest attested Iranian languages - represent immediate forerunners of the language. In order to cover this aspect, I will often use the symbol •.,,,• tc- ;ndicate a relationship that is not based on immediate genetic 'vicinity'. The same holds for the lexical index given in Appendix II.

1 I would like to thank Gilbert Lazard who has extensively commented upon an earlier version ofthis paper. Admittedly, I could not always subscribe to his comments. Thus, in case parts of my analysis turn out to be erroneous, the reader should in no case blame the reviewer. All faults and other shortcomings naturally remain in my own responsibility.

6

LW/M380

NORTHERNTALYSH

2.Thelanguage

Northern Ta/ysh

The Northern Talysh Area

Northern Talysh is a somewhat artificial term for those variants of Talysh (tolas) that are spoken by some 80.000 people in Southern Azerbajdzhan. The northern border ofthe Talysh area can be roughly characterized by the river Viläzh-Chaj (Mugani Steppe) though Azeri speaking communities are to be found especially at the right side of the lower course of the ViläzhChay, down to the swamps ofKumbashi (cf. Miller 1953:9). In the West and Southwest, the area is confined by the so-called Talysh Massif whereas the Southern border is indicated by the Astara-Chaj. Both the Talysh Massif and the Astara Chaj also form the border line between Azerbajdzhan and Iran. In fact, the term 'Northern Talysh' is motivated by administrative parameters rather than by linguistic features. Linguistically speaking, the region in question represents the Northem part of a dialect continuum in which the variant of Astara plays the role of a transitional dialect linking the Northern variants with the Talysh dialects of Iran (these have about the same number of speakers as the Northem variants; see Lazard 1978, 1979a, 1979b for a description of the Mäsulei dialect of Southem Talysh and Yarshater 1969 for a general description of Southem Täti). Southem Talysh covers the variants of Tularud, Pare Sar, Mäsäl, Mäsule, and Zide. Azerbajdzhan cities with more than 20% speakers ofTalysh are Ashagy-Nüvedi, Astara, Avrora, Boradygakh, Gaftoni, Jardykly, Kiroevsk, Kizhab_a, Lenkoran or Länkäran (Talysh länkon, lit. 'landing place, moorings'), Lerik (Talysh lik, ht. 'ground'), Massally, Narimanabad, and Port-Il'ich. Talysh is generally classified as a Northwest (NW) Iranian language (see Lecoq 1989, Windfuhr 1989 Paul 1998 for a classification ofNW lranian). lt can be regarded as one ofthe descendants of ~ Northwestem Old Iranian dialect that is perhaps related to Median and (Middle-)Parthian (cf. Schmitt 1989 for a survey of Middle lranian languages). Its immediate predecessor has sometimes been identified as 'Middle-Median', the hypothetical_ language of the ancient province Media Atropatene (Azerbajdzhan), cf. Schmitt 1989:101. Other Northwest Iranian languages are (among others) Baluchi, Gazi (Esfahan), Göräni (Kurdistan), Kurdish, Parächi (Afghanistan), Sangesari (Central Iran), Semnäni (Central Iran), Sivandi (near

LW/M380

7

NORTHERN TALYSH

Shiraz), Southem Täti (ÄzarI and Eshtehärdi), and Zäzäki (Kurdistan). The closest relatives of Talysh perhaps are the Shahrudi and Takistani variants of Southem Täti (cf. Henning 1954; Yarshater 1959 for a description of Shährudi Täti). Till recent, Talysh has been an unwritten language in Azerbajdzhan as weil as in Iran. From 1931 to 1935, some efforts had been undertaken to develop a written Northem Talysh tradition (based on the Latin script). This enterprise led to the publication of a number of Talysh texts, including school books and a newspaper (Sa Tolas 'Red Talysh', in fact a bilingual publication (Talysh/Azeri)). Political reasons conditioned that this tradition had been abandoned in 1935. Today, a written version ofTalysh is occasionally used in smaller villages located in the Talysh mountain range (e.g. in Archevan, Artupa, Pensa, and Tengerüd). Additionally, a growing interest to re-introduce Talysh as a written language (earlier Russian based, now Latin or Persian based) has developed both in Lenkoran and Astara since 1989. Typologically speaking, the architecture of Northem Talysh grammar is characterized by modest inflection that becomes more and more supported by features of agglutination (probably due to both inherent structural processes and language convergence towards the Azeri variants of Southem Azerbajdzhan and Northwestern Iran). Talysh occupies a somewhat fuzzy position with respect to the continuum of head and dependent marking: Though the marking of relational functions is documented with nouns in the singular as well as with pronouns, head marking is more prominent. However, typical head marking features such as personal inflection ofverbs become rather vague because ofthe heavy tendencies in Talysh to use a system of floating clitics (see below). The bulk of morphological elements is related to the verbal paradigm (tense, aspect, mood (TAM), agreement, negation), whereas nominal paradigms are often restricted to {wo or three variants. Yet, the picture is on the way to change because a significant number of former postpositions tend to be agglutinated on the noun creating a new paradigm of locative cases. Else, locatives and related categories are expressed by prepositions, postpositions, and circumpositions, both simple and complex. There are both synthetic and analytic TAM forms. Analytic structures dominate the verbal paradigm, but it should be noted that they often become 'synthetisized'. This fusional process is characteristic especially for present-day Northem Talysh. Talysh does not coherently encode the relational primitives S (subjective), A (agentive), O (objective). The language knows both an accusative and an ergative behavior ofthese primitives (see Schulze (in press) for the notion of 'relational behavior'). This split is TAM dominated just as in some other Northwest Iranian languages and distinguishes an accusative behavior of present tense based structures from an ergative behavior of certain past tense based structures. Except for rather artificial purposes, Talysh does not use grounding strategies (passives or antipassives) - however, impersonal passive-like structures are sometimes observed. The morphosyntactic coding of pragm,i.tic dimensions is rather significant. Both explicit focus particles and the exploitation of morphological means to secondarily signal pragmatic aspects apply.

LW/M380

8

NORTHERN TALYSH

Speakers of Northem Talysh have a strong preference for verb final structures. Else, only locatives (especially allatives) are allowed to occupy the sentence final position (preceded by the verb). In texts translated from Russian, other constituents may appear after the verb (the same holds for Russian trained younger speakers especially in Baku). Relational primitives are aligned according to an accusative pattem (SV and AOV). Within noun phrases, elernents show a centripetal ordering (noun final) which includes AttrN, GN etc.; only certain postpositions are allowed in NP final position. Northem Talysh has no gender distinction except for a vocative that is zero with human males, but -a with females. Clause linkage is rather asyndetic in Talysh: Though both strategies of subordination and coordination are available, they are rarely used. Subordination - if ever present - is controlled by the serialization of TAM forrns to which case-like elements or postpositions are incidentally added. A remarkable feature of subordination is the possible backgrounding of S and A (> possessive/obliquus) in (temporal) embedded clauses.

3. Palangi ahvolot Northem Talysh has not yet experienced a broad textual documentation. The main source so far exploited by linguists is Miller 1930. To my knowledge, the text presented in this booklet represents the only !arger Northem Talysh text document that has been published since Miller 1930. The text (20 min. of recording) presents a rather trivial story about a man who goes hunting. He is accompanied by (at least) two dogs and suddenly realizes the trace of a 'tiger' or 'leopard' (Talysh palang translates both 'tiger' and 'leopard' - if we relate the story to a local event, we should interpret palang as 'leopard'). After numerous incidents the man manages to kill the leopard and carries it home with the help of his father. The story is told by EGO and, in consequence, is characterized by a rather egocentric perspective. lt can be subdivided into 77 inforrnation chunks that rang from very short structures (three words, cf. chunk 4) to rather !arge and complex structures (e.g., 32 words in chunk 46). Statistically speaking, the text contains 789 tokens that represent 395 types. If we neglect word forrnation and any diachronic analysis these types are derived from 234 lexical entries. 96 of these have more than one representation, whereas 147 are hapax legomena (regarding the text). Tue story does not start with a typical intrada. lt also lacks a specific local setting: The only thing we leam is that EGO's occupation is related to woods (a hunter?) and that EGO Jives in a small rural village (Talysh dl) not too far away from some woods. Tue story seems to be based on a true incident which happened in Soviet times. This comes c!ear from the last two chunks (76-77): EGO teils that the incident became known in the rayoni markaz which translates the Russian term rajcentr 'regional center' (am hadisa sada rasada ba rayoni markaz 'these news reached the regional center') and that he had received an award (manson makofot ( ...) kärde 'they gave me an award'). According to my inforrnant, this type of 'auctorial' narration is rather usual in Southem Azerbajdzhan.

LW/M380

9

NORTHERN TALYSH

4. Phonemes and functional load 4.1 Synchronie survey All twenty-nine phonemes that can be described for Northem Talysh also appear in the text . ., discussed in this booklet. (1) lists the vowel phonemes:

n ,, (1)

a~ä - - ä Note that ü (IY/) has a free variant /u/, whereas /a/ often is heavily palatalized (Ire/). Secondary correlations are not distinctive in Northem Talysh, yet ti (/o/) is norrnally long, whereas vowels followed by /_nC/ or /_C#) tend tobe nasalized, cf. bavindom /'btv1dom/ 'I should see'. The twenty-two consonants (including sonants) are listed in (2):

(2)

Labials: b /bl,p /p/,f/f/, v /v/ Dentals: d /d/, t lt/, s /sl, z /z/ Palatals: J /d?,f, /tJ/, l?,I, IJ/, Velars: g lg/, k /k/, g /y/, x /xi Pharyngeals: h /h/ Sonants: m Im/, n In/, l III, r Ir/

c

z

s y /j/

Tue 789 lexical tokens ofthe text correspond to 3550 phonemic tokens which gives us a moderate ratio of roughly 4.5 phonemes per token. Tue basis to calculate the functional load ofthe 29 Talysh phonemes is the frequency !ist given in (3): (3)

All

Initial

3550

%

782

%

Medial 1986

o/o

Final 782

o/o

a

645

18.16

20

02.55

404

20.34

221

28.26

229

06.45

2

00.25

109

05.48

118

15.08

a

220

06.19

19

02.42

165

08.30

36

04.60

n

219

06.16

25

03.19

149

07.35

45

05.15

m

173

04.87

34

04.34

66

03.32

73

09.33

102

13.04

41

05.24

b

167

04.70

147

18.79

20

01.00

e

165

04.64

5

00.63

58

02.92

d

162

04.56

62

07.92

100

05.03

0

144

04.05

15

01.91

88

04.43

NORTHERN TALYSH

10

LW/M380

11

LW/M380

NORTHERN TALYSH

!11

03.71

12

01.53

116

05.84

4

00.51

123

03.64

42

05.37

67

03.37

14

01.79

122

03.43

II

01.40

103

05.18

8

01.02

115

03.23

45

05.75

60

03.02

10

01.27

109

03.07

18

02.30

67

03.37

24

03.06

132

r g ~

II

106

02.98

29

03.70

67

03.37

10

01.27

k

103

02.90

50

06.39

49

02.46

4

00.51

p

83

02.33

55

07.03

27

01.35

1

00.12

ü

77

02.16

1

00.12

48

02.41

28

03.58 00.51

s

75

02.11

29

03.70

42

02.11

4

~

73

02.05

63

08.05

8

00.40

2

00.25

6

00.76

V

64

01.80

3

00.38

55

02.76

61

01.71

30

03.83

23

01.15

8

01.02

4

00.51

14

01.79

X

39

01.09

6

00.76

29

01.46

z

39

01.09

7

00.89

18

00.90

02.94

9

00.45

03.58

II

00.55

h

g

z 3

32

00.90

23 17

28

00.78

22

00.61

00.12

19

00.95

00.33

8

01.02

3

00.15

3

00.38

6

00.30

12

II

00.30

(4)

n

m Core ~

0

d

r 1 b

00.25 00.12

2

00.25

If we try to interpret this !ist we have to bear in mind that the frequency of some of these phonemes is determined by idiosyncratic terms that characterize the text: For instance, initial p has 55 records ofwhich 36 are based on the word for 'leopard/tiger' palang. Initial/, on the contrary, has a rather broad documentation in the Talysh lexicon (Pirejko 1976:221-224 lists some 60 lexemes), but is restricted to the hapax.fikr 'thought' in the text. Furthermore, some phonemes have a significantly different frequency with respect to their position in the word (cf. b which has the highest rate when initial (18.79 %), but which is weak in medial position (1.00 %) and which is not allowed in final position). In order to take this aspect into consideration, the following calculation has been done: All phonemes have been ranked from 1 (highest frequency) to 29 (lowest frequency) for all three positional variants. For instance, e occupies rank 24 (initial), rank 14 (medial), and rank 3 (final). These ranks are added up (41) and again rated according to the cumulative rank. Hence, e would be associated with the relative rank 12. Finally, this relative ranking is added up with the absolute ranking of the phonemes (column 2 in (3)) and the result is divided by 2. (4) lists the cumulated rating of all phonemes:

Relative Ranking

Frequency Ranking

IMF

Close periphery 2

Final Ranking (R+F-Rank ./. 2)

Distant

Periphery

k

1 3 3,5 4 5 7 8 8 9,5 9,5 9,5 11,5 12 12 13 17 17,5 18 18,5 18,5 19,5 21,5 23 23,5 25,5 25,5 28 28 29

1 2 4 3 8 10

4 3 5 2 14

5

II

7 6

9 13 8 7 15 12 10 6 16 17 19 22 21 18 20 24 23 26 25 28 27 29

II 12 8 12 14 20 18 18 17 15 16 21 23 22 24 25 26 27 28 29

In (4) the inventory of Talysh phonemes is interpreted as a prototypically organized system: The core domain is characterized by phonemes that have a final ranking from 1 to 11 (a, n, "• m, i, s, t, o, g, d, e). The immediate periphery is represented by ranks 12 through 20 (ä, r, [, b, k, P, s, v, y, ü), whereas the distant periphery is marked by ranks 21 through 29 (c, z, x, g, h, i, f, 3). From this ranking we can conclude that the phonemes that belang to the core domain are rather unmarked and 'expectable' or predictable. Moreover, they have a much higher functional load than those of the two peripheries. This assumption corresponds to the fact that 84.08 % (in numbers 544) phonemes of the totality of 647 phonemes that appear with bound grarnmatical morphemes belong to the core domain (against 102 phonemes located in the closer periphery (periphery I) and one single phoneme associated with the distant periphery (periphery 11)), cf. (5):

12

LW/M380

(5)

Rank

a

Core

Peripery 1

1

Phonemes in mor-

Phonemes in mor-

phemes: absolute

phemes: rank

130

1

%

% of phonemes in morphemes

20.15

LW/M380

ü

13

35.08

12.67

31.16

03.70

20.09

:,

25.45

08.65

k

23.30

03.70

d

3

NORTHERN TALYSH

15.43

00.36

12.50

01.23

10.66

01.23

09.09

00.15

n

2

45

6

20.54

06.95

:,

3,5

56

5

25.45

08.65

b

21.55

05.56

p

02.40

00.30

21.10

03.55

a

01.88

00.30

20.54

06.95

00.75

00.15

m

4

68

4

39.30

10.51

§

i

5

82

3

35.08

12.67

n

§

7

23

12

21.10

03.55

0

8

24

9

16.66

03.70

t

8

---

---

---

---

d

9,5

25

8

15.43

03.86

g

9,5

---

--

--

---

a

11,5

2

15

01.88

00.30

e

12

89

2

53.93

13.75

1

12

1

17

00.75

00.15

-m

r

12

---

---

---

--

-i

b

13

45

7

21.55

05.56

k

17

24

10

23.30

03.70

p

17,5

2

16

02.40

00.30 01.23

s

18

8

13

10.66

j

18,5

8

14

12.50

01.23

V

18,5

--

---

---

---

ü

19,5

24

II

c

21,5

-----

-----

z

23

X

23,5

--

g

25,5

---

phery II h

25,5

---

Peri-

NORTHERN TALYSH

31.16

(8)

m

---

---

---

-------

---

f

28

--

-----

--

1

18

09.09

00.15

infinitive (when stressed), auxiliary past 3so clitic/suffix !so obliquus singular

As expected, the functional load of the formal categories mentioned in (7) is considerably high. This aspect will be discussed in the next section of this booklet. The high grammatical value of -e, -m, and -i becomes obvious ifwe relate their frequency to the final position, cf.:

---

28 29

-e

03.70

z

3

(7)

e

--------

--

Obviously, e, m, and i play a central role in the phonemic architecture ofTalysh morphology, at least with reference to the kind of text represented by the account underlying this analysis. These three elements represent 36.93 % of all 'morphological' phonemes in tht text. Contrary, for instance, to b, k, etc. they encode morphemes themselves, cf.

Final 102 73 118

Morphological 89 68 82

% 87.25 93.15 69.49

This rate becomes even higher if we take into consideration that both -e and -m may appear in medial position when followed by other morphemes in a given agglutination chain, cf. (9)

zan-a-m-e [PA 64] know:PAST-PERF-ISG:A-AUX:3SG 'Iknew'

If we have a closer look at the frequency of phonemes in bound morphemes compared to their general frequency, we arrive at an interesting picture (cf. column 2 in (6)):

kärd-e-nJa [PA 60] make-AOR-TEMP

(6)

%

% of phonemes

%

'when making .. .'

in morphemes

in morphemes

1:

% ofphonemes

53.93

13.75

a

20.13

05.40

39.30

10.51

0

16.66

00.37

Those 'morphological' phonemes that are located in the periphery ofthe Talysh phoneme system are more marked than the phonemes discussed so far. For instance, ü is restricted to the locative morpheme -kü, and J appears only with the rather infrequent temporal converb -nja.

14

LW/M380

NORTHERN TALYSH

One exception seems tobe initial b-: Contrary to its general peripheral status, it is rather frequent in this position (167 occurrences). This fact is matched by the high rate for b- in bound morphological units (36 occurrences). But the functional load of b- is rather low which conforms to its peripheral status: lt basically encodes the modalization of verbal forms (future, conjunctive, optative). A second function is related to locatization, cf. (10)

ha-

b:}be-

(8) (15) (13)

Future Modal Preverb

A parallel degree of markedness is given for b- in general: lf we deduct the records for 'morphological' b- (36), we arrive at 133 records for b- in 'free' elements. 60 ofthem are related to the preposition ba- 'in, into'. Only 19 occurrences of b- can be related to pure lexical forms (only base forms are given): (11)

bärde bäslamis badan barbar b:}Jang b:}n b:}r beg bitov bi3 büd:}n

(1) (1) (1) (1) (1) (1) (9) (!) (!) (!) (l)

ba babesta b:} be

(60) (2) (5) (25)

NORTHERN TALYSH

with the palang 'leopard/tiger' the default interpretation of which are transitive structures. Still, the statistics would slightly lead to another result if we include the phonemic structure of 'free' functional morphemes such as personal pronouns, deictic elements etc. For instance, z has an absolute rate of0l.09 % (39 tokens). But in 15 instances, z is conditioned by the occurrence of the first person singular pronoun äz 'I' to which the focus marker -an may be added (az-an). Yet, I refrained from including these structures in the calculation of 'morphological' phonemes because this would presuppose a decision about the lexical or grammatical status of the structure in question. For sake of simplicity, I have treated 'free' morphemes as pseudolexical or indexal structures. The conclusion we can draw from these statistics is that the prototypical organization ofthe Talysh phoneme system as represented in the text meets the prerogatives ofmarkedness and functional Ioad. The grarnmatical (morphological) inventory is phonemically adjusted to this prototypical structure which leads to a harmonic coupling of both domains.

4.2 The diachrony ofthe Northern Talysh sound system 'carry' 'turn' 'body' 'noise' 'arm' 'ground' 'bush' 'moustache' 'whole' 'smart' 'side'

Else, initial b- has the following distribution Gust as in (11), only the base forms are given): (12)

15

LW/M380

Preposition ('in, into') Postposition ('after') Deixis (prox) Light verb ('become')

Again note that the statistics discussed so far depend on the textual character ofthe underlying source: Tue high frequency of -e, -m as well as of b- (in ba) is due to the basic setting ofthe account: A first person (-m) talks about an event in which he and a third person (-e) is involved. This event is characterized by a chain of movements which accounts for the high frequency of the directional preposition ba. The multiple occurrences of the multifunctional oblique marker -i has to be related to the interactional event structure: EGO mainly interacts

In order to describe the diachronic background of the contemporary sound system, I refer to Pirejko (1991:95-124). lt should be noted that I do not intend to discuss the single developments in detail here. The following overview simply aims at a rudimentary orientation that helps the reader to follow the subsequent discussion ofthe diachrony ofNorthern Talysh morphology. a) Vowels Iranian *a *ä *i *i' *u *ü *ai *(a)ya *au *(a)wa

Northem Talysh a,ä,a,e

o, rarely a ;J

ü, a, rarely e ü, rarely e i, rarely e ü 0

Accordingly, Northern Talysh vowels reflect the following Iranian vowels:

ä

<
härd-ovnie 'feed' pare ,fly' > por-nie ,Jet fly' [vowel harmony applies]

Allomorphs are uni, -ni. 5.4.4 Non-finite verbal forms 5.4.4.1 Infinitive < *-en < Iranian *-anai -e -ie .,,. Persian -idan Tue distribution of these two infinites is based on the lexicon. ie-infinitives tend to be replaced by e-infinitives, e.g. ganie :::z gane 'fall'.

23

LW/M380

5.4.4.2 Participles Perfect(ive) General Present Future

STEM:PAST-a ba-STEM:PAST a-STEM:PRES STEM:PAST-anin

< *-aka

NORTHERN TALYSH

kärd-a 'done' ba-kärd 'doing/done' a-ka 'doing' kärd-anin 'going to do'

5.4.4.3 Gerunds / Converbs Gerunds are based on the infinitive to which a postposition is added (the pure infintive also encodes a modal converb). Tue structure INF-ada has developed towards a general converb (often eo-temporal). 5.4.5 Finite verbal forms 5.4.5.1 Personal inflection Verbsare inflected for person and number. There is neither an inclusive/exclusive distinction nor is gender marked for persons. Northem Talysh has two types ofpersonal agreement markers: a) suffixes, b) floating clitics, see section 6.3 Suffixes (Type!) encode the Subjective and the Agentive in 'accusative-based' tense forms, floating clitics (Type2) encode the Agentive in 'ergative-based' tense forrns. 5.4.5.l.l Suffixes (Type!)- Sand A in (S=A;O), S in (S=O,A) lsg -m lpl -mon -s 2pl -on 2sg 3sg -0, -e 3pl -n 5.4.5.1.2 Floating clitics (Type2)-A in (S=O;A) lsg -(a)m lpl -(a)mon 2s g -a 2pl -(a )on 3sg -(a)s / -(a)z 3pl -(a)son 5.4.5.2 Tense forms 5.4.5.2.l Type! Present STEM:INF-da-Typel Future ba-STEM:PAST-INF-Typel AUGM-STEM:PRES-PAST-Typel Past (imp.) [strong modal connotations] Past II (dur.) PRES + AUX:PAST-Typel Past Condit. PART:PERF + AUX:COND

kärd-e-da-m ba-kärd-e-m a-ka-i-m

,Ido' 'I will do' ' 'I was doing'

kärd-ed-a b-i-m kärd-a b-ai-m

'I was doing' 'ifl did'

24

LW/M380

5.4.2.2.2 Typel + Type2 Past(perf.) STEM:PAST-i-Typel (intrans.) STEM:PAST-Type2-e (trans.) Perfect STEM:PAST-a-Typel (intrans.) STEM:PAST-a-Type2-e (trans.) Pluperfect PART:PERF + AUX:PAST (intrans.) PART:PERF-Type2 + be (trans.)

NORTHERN TALYSH

§-i-m kärd-arn-e om-a-m kärd-a-m-e om-a b-i-m kärda-m be

5.4.5.3 Mode Conjunctive ba-STEM:PRES-a(sg, 3pl) } ba-STEM:PRES-a(l pi) -Type 1 ba-STEM:PRES-i(2pl) Conj. Past: Optative Necessitive

PART:PERF + AUX:C0NJ bi-STEM:PRES-o-Type 1 PART:FUT-PAST(-i-)-Typel

'I went' 'I did' 'I came' 'I have done' 'I had come' 'I haddone'

'ifl do'

kärd-a b-o-m bi-k-o-m kärd-anin-i-m

'Iwoulddo' 'wouldldo' 'I must do'

5.4.5.4 Imperative Imperaties are formed with the help of the present stem, to which the modal prefix ba- is added: ba-ka 'do!' The prefix is missing with preverbs: o-ka (back-make) 'open!'. 5.4.5.4 Evidential Evidentials ('by hearsay', inferential) are construed with the help of ban < *ba-n (perfect of be- 'tobe' plus focal particle or 3PL:PERF) which can be added to any indicative tense form:

(G14) zoa-m palang-as vind-a ban son-lSG:POSS Ieopard-3sG:A see:PAST-PERF:AUX:3SG INFER 'My son has seen a leopard, they say.' 5.4.5.5 Negation Negation is indicated either by special inflectional forms or by the negative particle ni 'not: Conj. Past Conj. Future Necess. Past (dur.) Past (imperf.) Past (perf.) Perfect

kärd-a na-bay-m na-k-a-m a-k-a-ni-m kärd-anin-ni-m kärd-e-da na-b-i-m n-a-k-i-m na-kärd-am-e na-kärd-am-e

'ifl did not do' 'ifl do not do ' 'I will not do' 'I must not do' 'I was not doing' 'I was not doing' 'I did not' 'I have not done'

25

LW/M380

kärda-m na-be kärde-da-ni-m ma-ka

Pluperfect Present Prohibitive

NORTHERN TALYSH

'I had not done' 'I don't do' 'don't do'

5.4.5.6 Diathesis Northern Talysh uses the passive diathesis in order to completely mask the agentive function in transitive structures. An agentive cannot appear in the periphery of a passive structure. Such 'impersonal' passives are marked by the perfect(ive) participle to which the auxiliary be is added:

(G 15) acavon so-ada palo pät-a be-da 3PL:POSS yard-INESS pilav cook-PERF AUX:INF-PRES 'Their pilav is cooking in the yard' (*the pilav is cooked by them in the yard') Note: Such constructions can also be interpreted in terms of Iong distance possession ('they have a pilaw cooking in the yard') from which a new type of 'ergative structure' may emerge. Tue 'active' (accusative) variant would be: (Gl6) avon pilo päte-da-n so-ada they pilaw cook:INF-PRES-3PL:Ayard-INEss 'They cook a/the pilav'.

5.5 Numerals 1 da 2 3 se 4 eo 5 pen3

sa§

6 7 8 9 10

häft hast nav da

20 30 40 50

vist si ca! pen30

60 70 80 90

sest häfto hasto nava

100 200 300 400 500

sa da-sa se-sa co-sa pen3-sa

600 700 800 900 1000

sas-(s)a häft-sa hast-sa nav-sa hazo

26

LW/M380

First decade : a) Synthetic: yonza 11 12 donza senza 13 14 corda ponza 15

16 17 18 19

NORTHERN TALYSH

LW/M380

go guyä hä hatto ki mavot ne xay

sonza häfda hasta nonza

27

NORTHERN TALYSH

emphasis 'as if 'yes' 'even' emphasis 'it seems' 'no' 'no'

b) analytic: 10 va X: davai (11), davada (12), davase (13) etc. 5.8 The bound morhemes ofNorthern Talysh

Higher numbers are construed as: (X) 1000 - (X) 100 (da 'with') (X) 10 - X 2419: da hazo cosa (da) nonza 3286: se hazo dasa (da) hasto sas Ordinal numbers are based on the cardinals to which -um, -om, -am (Old Persian -ama the use ofwhich has been extended to 'two' and 'three') is added (often followed by the Azeri suffix Ji): dav-am(-Ji) 'second', sej-im(-Ji) 'third' etc. 'First' differs from this paradigm (ijn < *i-

in). 5.6 Conjunctions 5.6.1 Coordination (selection) 'and' (Arabic wa) va, v;, ammä, arno 'but' (Arabic 'ammä) 'or' yä gä ... gä 'and ... and'

5.6.2 Subordination (selection) ki general subordination, often preceded by a nooun (vaxta ki 'when' etc.) gora 'in order to' < Azeri körä < gora-s (gora + 3sg:clitic) 'because' goros agar,agarn 'if ""'Avesta hakar ABS/OBL, pronoun > ACC) has undergone a further split in the pronominal system (cf. Windfuhr 1998:260, Lecoq 1989:302): Here, 1s0:0 as weil as 3so:o and 3PL:o are split according to the question whether O is associated with an accusative (present tense) or ergative (past tense) behavioral pattem: Based on an innovated 'nota accusativi' a-, the first singular uses adem for O in S=A;O, but a-cem for O in the tripartite (basically ergative) pattern S;O;A. Tue third person has ajo vs. o (or oa) in the singular, and aj(u)an vs. oe in the plural. The fact that shortened forms of the accusative pronouns (man instead of mani etc.) are sometimes observed in Northem Talysh may represent the starting point of a parallel paradigm in Northem Talysh (cf. (70) below). Yet, it is important to note that in Hazärrüdi Täti the split is based on three different stem forms at least in the first person singular:

37

LW/M380

ISG 2SG 3SG IPL 2PL

Talysh ma ta ay ama sama

Young Avesta mana, encl. me tauua, encl. te auurujhe ? Encl. -nä ahrnakam,encl. -nö lJsmakam, encl. -v5 yosmakam, encl. -vo OldAvesta m:>.na, encl. möi tauua, encl. töi, te

Old Persian mana -tajy avahi,iya amäJJam

?

lt can be assumed that the oblique pronouns had gradually replaced the ab~ulutives (casus rectus) in the personal paradigm except for the first person singular. This is especially true for the plural pronouns, cf. IPL Avesta vaem vs. Talysh ama, 2PL yüs (Gathas), eise yuiäm or yuiam vs. Talysh sama; the situation with respect to the 2s0 remains doubtful: Talysh -a may stem from both *-u (and *ü) and *-a (Pirejko 1991:103-4). Hence it can be compared to Avesta tu, Old Avesta tuuäm, Young Avesta tüm as weil as to Old Avesta tauuä (genitive) (Hoffmann & Forssman 1996:161). Ifwe claim that Talysh ma (!so) stems from *mana (genitive) (Old Avesta mä.na, Young Avesta mana) we should assume that the same is true for the second person singular (ta < *tava). But note that the use of second person agents in past tense assertive (indicative) structures where we should expect tava is rather infrequent (see Schulze 1998:532 on the structural coupling of speech act participants (SAP), agentivity, and modality). Ifwe assume that second singular SAPs norrnally occur in interrogative structures, it may weil be that - in these instances - A is regularly demoted to S (see Gerstner-Link (ms.)). Hence, ta should be better interpretd as nominative (or absolutive) in contexts like (28). Note that again the floating clitic -a (2sG:A) remains ergative. (28)

[ta] ma-ni konjo bagam-a kard-e [Miller 1953:180) you:SG:N0M:A(>s) I:OBL·ACC where Iove-2SG:A make:PAST·AUX:3SG 'Where did you fall in love with me?'

As has been said above, the first person singular has preserved the distinction documented in Old Avesta azäm (NoM) vs. mä.na (GEN) (Talysh äz vs. ma). This fact corresponds to the general observation that utterances are structured according to a strong egocentric perspective in the South East Caucasian linguistic area in which Talysh has participated at least in earlier times (cf. Schulze 1999, Schulze (forthcoming b)). The use of former genitively marking pronouns to encode A in a specific set of past tense utterances is based on structures Iike Old Persian

LW/M380

(29)

38

NORTHERN TALYSH

ima tya manä kar-t-am ParBavaiy [Kent 1953:125 = Darius, Bagistan III, 10]

(31)

PROX:NOM:SG:N what:NOM:SG:N l:GEN make-PART:PAST-SG:N Parthia:LOC 'This [is] what I have done in Parthia' (- 'what is done by mein Parthia') which illustrates the Old Persian prototype manä krtam astiy 'I have done', cf. section 6.3 and Benveniste 1952:56 (Talysh > ma kärdame, see below for a description of ergative behavior in Talysh; note that Bossong 1985 denies a genitive background of the oblique case in Iranian. Instead he proposes the dative to represent the starting point of oblique techniques). Interestingly enough, Northem Talysh seems to copy this 'possessive' interpretation of certain transitive structures with in subordination, cf. section 6.1 (18) and

semi=precious=stone:NOM:M REL:NOM:M turquoise:NOM:M PROX:N0M:M from Khorasm-ABL AUGM-bring-PASS-3SG REL:NOM:M here make-PART:PASS:PAST:3SG 'The turquoise stone has been brought from Khorasm [and] has been treated here.' Tue preposition *haca and the oblique forms ofthe personal pronoun have fused, resulting in the following output: (32)

lsg 3sg

(30a) cayi püst-amon pe-gat-e [PA 74] he:POSS skin-lPL:A up-take:PAST-AUX:3SG 'We took his skin... ' cama gülla gänd-e [PA 65]

l:ross:A bullet throw:rAST-AUX:3sG '[When] I shot' (lit. 'I threw a/the bullet') (c)

cayi ca do-kü ba ma-sa jo be [PA 40] he:POSS:A he:POSS tree-LOC[ABL] to l:OBL-SUPERseperate LV:PAST:3SG '[When] he left his tree [and moved] towards me .. .'

(d)

( ...) palang-i ba bar da-s-a ( ...) [PA 19] ( ...) leopard-OBL to bush in-go:PAST-PERF '(...) [when] the leopard went into the bush ( ...)'

(30a) shows a standard possessive construction. (30b) and (30c) demonstrate the use ofthese possessives to encode A in a quasi-subordinative structure: (30b) further illustrates that the verb is less finite than standard verbal forms: it lacks personal clitics. (30d) illustrates that the use of the oblique case in such possessive structures also applies for intransitive structures with overt S-referents. Obviously, A (and S) marking based on the possessive paradigm establishes some kind of (mostly temporal) subordination which in spoken Northem Talysh normally is asyndetic and lacks explicit subjunctions. Tue pronominal possessives themselves are derived from an old preposition that appears in Old Persian as hacä, in Avesta as haca or hacä (> *ca-). This preposition that is related to the IE stem *sekf!- 'to follow' has a strong ablative function in Old Iranian cf.

NORTHERN TALYSH

käsaka haya al.Jsaina hauv hacä uvärazmi-yä a-bar-iy-a haya idä kar-ta [Brandestein & Mayrhofer 1964:87 = Darius, Susa f, 39-40]

2sg

(b)

39

LW/M380

!pi 2pl

*(ha)ca-ma *(ha)ca-t:i *(ha)ca-ay *(ha)ca-ama *(ha)ca-~(:i)ma

> > > > >

cam:1,a~m:1 *c(a)t:i

>

•act:i

cay cama, acama *c(a)~:ima

>

•c~:ima >

>

:i~ta

~:ima

Variants starting with a- (acama, acama, acavon etc.) still reflect the initial segment *ha-. Again the third person plural ((a)cavon < *(ha)ca-av-on) is exempted because avon, originally the plural marked rectus av, is a much younger pronoun that did not participate in the formation of oblique case forms. Tue metaphorization of an ablative function towards a possessive function is a typological standard, cf. the case syncretism genitiveablative in Old Greek, the German and English genitive based on von resp. o/etc. The two enclitic paradigms mentioned in (22) are distributed with reference to the underlying type of relational behavior. In accusative structures, the first row is used to crossreference A and S, cf.: (33a) äz-an (. ..) ba cayi düs tala-da b-i-m [PA 29] l:NOM-FOC ( ...) to HE:POSS behind run:PAST-PRESLV-PAST-ISG:S 'lt was me who was running behind him.' (b)

äz ogärd-a-m ba märd-a palang-i-sa [PA 68]

l:ABs tum:PAST-PERF-lSG:s PROX dead-REL leopard-0sL-SUPER 'I tumed to this dead leopard. • (c)

äz hamisa ba-yi tämso a-ka-m [PA 75]

l:ABS always to-HE:OBL look AUGM-make:rREs-lsG:A 'I always looked at it.' In (33a) and (33b), the suffix -m encodes S, whereas it is triggered by A in (33c). Tue singular set of affixes used to encode S=A should best be related to the corresponding secondary endings in Old Iranian (Avesta -m, -s, Old Persian -m, *-a) . The primary ending *-si > Avesta -

LW/M380

40

NORTHERN TALYSH

hi, or -hi, Old Persian -hy would produce -i in Talysh (cf. Modem Persian mikon-i 'you (singular) are saying'). The third person clitic -0, however, presupposes the primary ending *ti which yields *-y > -0 with thematic stems. Tue plural morpheme -mon (IPL) seems to represent the pluralized clitic -m (!so) though we should expect *-m-an rather than -mon. lt replaces Old Iranian *-mahi (Avesta -mahl) (primary) or *-ma (Avesta -mä, -ma) (secondary). Tue second person plural is simply -on which proposes the pluralization of a zero-morpheme used for the second person singular (*-0-on instead of *-s-on). If we take into consideration the Old Iranian data, we should start with Avesta -0a, -0ä (no data for Old Persian). *-0- normally becomes zero in Talysh (cf. Talysh ro 'way' < *ra0a, pän 'broad' < *pa0ana-, see Pirejko 1991:114). The vowel following *-0- is normally dropped, too. From this we may conclude that the zero element in -0-on does not represent a variant of the second person singular, but the regular reflex of *-0a (2pl, secondary ending). The third person plural morpheme -n is regularly derived from the secondary ending *-nt (Avesta-ni, -aJ). As I will show in section 6.2.2, the tendency to use an additional set of personal clitics has its sources already in Old Iranian, cf.

LW/M380

(d)

bega-m buma [Yar-Shater 1969:154] brother-1 so:ross come:PAST:PRET 'My brother came.'

(e)

meraxxas-i mi-ar-em [Yar-Shater 1969:155) free-2s0:o PRES-LV:TRANS:PRES-1 SG:A 'I set you free.'

kära-sim hacä darsam a-tarsa [Brandenstein & Mayrhofer 1964:84 = Darius, Bagistan I, 50-51) people:NOM:M-3SG:IO from much AUGM-fear:IMPERF:3SG 'The people were much afraid of him.'

Iranian personal clitics generally refer to an oblique function (genitive, dative, accusative, etc.) which does not allow us to derive the clitics from old personal endings that normally follow an S=A pattem encoding a nominative case (but see (41 )). In modern Iranian languages there is a strong tendency to structurally and formally couple possessive clitics and clitics to mark the {0-10} domain, cf. Modem Persian (33,a-c) and Southem Täti (Takestani) (33,d-e): (35a) man madar-e xod-am-ra dust dar-am [Alavi & Lorenz 1988: 105) l:NOM mother-REL REFL-ISG:POSS-ACC:DEF love have:PRES-ISG:A 'I love my mother.' (b)

ta-t na-pors-and harf na-zan [Alavi & Lorenz 1988:104] till-2so:o NEG-ask:SUBJ:PRES-3PL:A word NEG-hit:PRES:2SG:IMP 'Don't speak before someone has asked you.'

(c)

bera-yas xand-am [beras xundam] [field notes] for-3s0:10 read:PAST-ISG:A 'I read [it] to him.'

NORTHERN TALYSH

In some Northwestem Iranian languages such as Kurmanci Kurdish, the clitics preserve their 0-orientation even in 'ergative' structures, cf. (36a) ez hesp-f di-bin-im [Bedir Khan & Lescot 1986:86) l:NOM horse-ACC:DEF:M DUR-see:PRES-ISG 'I see the horse.' (b)

(34)

41

~ven ez dft-im [Bedir Khan & Lescot 1986:153] shepherd:OBL:M l:ABs see:PAST-lso:o 'Tue shepherd saw me.'

In Northem Talysh, personal clitics show a significant shift regarding their relational behavior. They generally refer to A in past tense transitive structures (contrary to O as illustrated for instance by the Kurdish example (36b)), cf.: (37a) ma i tifang-a posna-m i-a ba pa/ang-i kalla [PA 52] I:oBL a rifle-REL butt-1s0:A hit:PAST-PERF to leopard-oBL head 'I have hit the rifle butt on the head ofthe leopard.' (b)

[tifang] pe-gat-a-m-e [PA 1) [rifle] up-take:PAST-PERF-1 SG:A-AUX:3SG 'I took up [the rifle].'

The grammaticalization of personal clitics as A markers in Talysh has obviously started with the possessive domain that was part of the functional scope of clitization in earlier (Western) Iranian, cf.:

LW/M380

42

NORTHERN TALYSH

(38)

The inner circle of this scheme symbolizes the functional scope of oblique pronominal clitics in Early lranian (genitive, dative, accusative). This functional domain can be subjected to metaphorical processes that are based either on the possessive (genitive) sub-domain or on the O-oriented (dative/accusative) sub-domain. Talysh gives an example for the first option, whereas Persian, Kurmanci Kurdish and other Iranian Ianguages illustrate a strong Oorientation of clitization. These processes are based on the neutralization of the diathetic properties of the underlying past participle (*-ta-). lt is important to note that the functional specification of clitization is not conditioned by an appropriate relational behavior: Both TaIysh (A-orientation) and Kurmanci Kurdish show an ergative behavior of certain past tense transitive structures that is characterized by the correlation OBL ➔ A and ABS ➔ 0 (resp. IO), cf. section 6.3. But whereas the A-orientation of oblique clitics in Talysh matches the oblique case marking of their (pro)nominal heads, the O-orientation provokes a remarkable conflict: Ifpresent, the overt O-function is marked by the absolutive which corresponds to the casus rectus whereas the clitics retain their oblique background (cf. ez (Ass) vs. -im (historically speaking: OBL) in the Kurdish example in 36b).

6.2.2 Agreement: Clitics and Suffixes

The provenience ofTalysh clitics from shortened oblique pronouns cannot be questioned with respect to the first person singular (-(a)m < *-ma < *maiy (genitive), cf. Old Persian -maiy, Old Avesta -möi, Young Avesta -me). (39) illustrates the use of clitics to encode a (backgrounded) A in Old Persian: avaBä-säm hamaranam kar-tam [Kent 1953:121 = Darius, Bagistan II, 27] thus-3PL:POSS(>A) battle:NOM:SG:Nmake-PART:PAST-NOM:SG:N 'Thus they fought the battle' (~ 'thus the battle was fought by them')

As has been said above, the third person clitic -(a)s reflects the Old Iranian clitic *-sai(y) which is a genitive(-dative) of the pronominal stem *sa-. Again, the second person singular causes problems: If we assume ta to be the form that underlies the clitic -a we are confronted with the probable 'absolutive' (or nominative) function ofthis pronoun (diachronically speak-

NORTHERN TALYSH

ing). However, if we have a closer Iook at other Iranian Ianguages that behave like Talysh we are forced to assume a possessive (oblique) marker *taiy (Old Avesta -töi, -te, Young Avesta te) to represent the underlying form. Obviously, the reflex of *taiy in Early Talysh differed from that of *tü etc. Quite parallel to the past participle marker *-ta which is generally lost in Talysh, posttonic *-taiy was likely to Ioose its *-t- in intervocalic position (contrary e.g. to Täti: Here, *-t- is subjected to rhotacism). Contrary to historical *-a that was generally dropped, the vocalic reflex ofthe group *-aiy (> *-e) yielded -a (cf. Pirejko 1991:103). The plural represents a more recent pluralization of the corresponding singular clitics - in fact an innovation within Early Northwestem lranian. (40) summarizes the processes related to the Talysh personal clitics: (40) ISG 2SG 3SG IPL 2PL 3PL

Old Iranian *-maiy *-taiy *-saiy [Avesta -n;i] [Avesta -b;i / -vö] [Old Persian -sam]

Early Talysh *-mai >*-m:i *-tai > *-(t):i *-sai > *-s:i *-m(:i)-rui *-(t):i-rui *-s:i-rui

Northem Talysh -m -:i -s -m-on -(:i)-on -s-on

Interestingly enough, Northem Talysh does not show any clear evidence for the preservation ofüld Iranian personal clitics in the accusative (cf. Avesta lsa -mä, 2sa -Bßä, IPL-nä, -nö, 2PL vä, vö). Obviously, the tendency towards a genitive-accusative syncretism became notorious already in an early version ofTalysh (or in Late Western Iranian?). Yet, contrary to the nominal oblique marker -i that is derived from the old genitive marker -hyä, the former genitive clitics were not used to encode O in accusative structures, cf. Modem Persian (41) and Northem Talysh (42): (41a) dast-e kui!ek-e sejid-at-rti did-am [field notes] hand-REL little-REL white-2sa:ross-0:DEF see:PAST-1 sa:A 'I saw your little white hand' (b)

(39)

43

LW/M380

dar xane-i did-am-at ke certig-as rousan bud [field notes] in house-RELREL see:PAST-ISG:A-2SG:ORELiight-3SG:POSS burning be:PAST:3sG:S 'I saw you in a house in which a light was burning.'

(42a) balet-a bari-a be [Miller 1953:123] fence-2sG:POSS break:PAST-PERF LV:3SG 'Your (sg.) fence had been broken.'

LW/M380

(b)

44

NORTHERN TALYSH

ta-ni vogänd-amon-e [Miller 1953: 117] you-ACC send:PAST-IPL:A-AUX:3SG 'We sent you (sg.)'

(46)

(43)

**ama-a vogänd-amon-e we:ABS/ERG-2SG:0 send:PAST- 1PL:A-3SG [cf. Modem Persian ma-tferestad-im]

Contrary to some Southwest Iranian varieties, such as most Northern Täti dialects, the casus rectus of personal pronouns cannot be subjected to clitization, cf. (44) which illustrates this procedure with the help of an example from Central Northem Täti (dialect of Dagkushchu and Arisküsh): (44)

in tü-rä cül jüft sätäl-tü [Grjunberg 1963:39] PR0X you:SG-OBL forty pair sock-2S0 'Here [are] your (sg.) forty pairs of socks.'

In (44), the element -tü (2s0) represents a clitic that is co-referential with the oblique pronoun tü-rä (you-OBL) indicating possession. lt is based on the corresponding personal pronoun (tü) that marks the casus rectus. Quite often, these nominative-based clitics indicate possession even without a preceding oblique pronoun, cf. Central Northem Täti: (45)

may-män män-ä kuft [Grjunberg 1963 :39] mother-tso 1-oBL beat=up:PASr:3sG:A 'My mother has beaten me up.'

Tue fact that the casus rectus is exempted from clitization in Northern Talysh coincides with the Jack of an 'accusative' function of the oblique clitics. lt can be assumed that the realignment of actance encoding in sentences marked by a perfective past tense, that is the 'ergativization' of parts of the Talysh morphosyntax, strengthened those functional domains of the paradigm ofpersonal pronouns that were involved in this process, cf. (46) which Jists the case functions of clitics in Old Iranian, Northem Täti, and Northem Talysh:

Old Iranian

N. Täti

ACC GEN

OBL NOM

NORTHERN TALYSH

N. Talysh

s OBL

A 0 Poss

(41a) and (42a) illustrate the use of clitics to encode possession. (41b) shows the corresponding clitic -t in O-function, whereas the Talysh example in (42b) has an overt ('accusative' marked) pronoun in the same function (tani). lt is impossible to construe sentences with 0clitics, such as:

45

LW/M380

OBL

6.3 Tense marking, clitics, and relational behavior

The tense, aspect, mood system (TAM) ofNorthem Talysh is historically based on the standard Iranian distinction between a present tense (or durative) stem and a past tense (perfective) stem. Diachronically speaking, most ofthe past tense stem forms had been derived from the past participle *-ta. This paradigmatic structure that is preserved, for instance, in most of the Southem Tati dialects had conditioned the realignment of parts of the Talysh morphosyntax ('ergativization') just as it can be described for a considerable number of other New Iranian languages (among others Pashto, Jaghnobi, Sernnäni, Göräni, Zäzäki, Kurdish, Eshtehärdi Täti, Baluchi; cf. Pirejko 1979, Bossong 1985 for a general overview). However, with most verbal stems, this diptotic system was harmonized in Northern Talysh in favor of the present tense stem form. Today, this process still has not come to an end: Verbs that have their infinitive marked by -ie < Old lranian *-itan (~ Modem Persian infinitive -idan) (instead of -e < Old Iranian *-anaiy) normally have retained the opposition present/past stem, cf. (47)

Infinitive vask-ie toz-ie /toz-ie gan-ie can-ie

Present stem vasktoz- / tozganc:in-

Past stem v:isk-itoz-i- / toz-igan-ic:in-i-

'to lie' 'to drive' 'to fall' 'to gather'

Yet, in modern Northern Talysh, the use of past stems marked by -i becomes more and more obsolete, cf.: (48)

cayi ba zamin e-gan-e [PA 42] he:ross to earth down-fall(:PAST)-AUX:3so '[When] he has fallen down .. .'

Here, egane replaces the expected form e-gan-i-e. The tendency of reducing the old diptotic paradigm should be related to the general trend in Talysh towards agglutination: Tue inflectional diversity of verbal stems is subjected to a process of 'stem stabilization', by which is meant that the Iexical domain becomes structurally (and formally) separated from the inflectional domain. In other words: The lexical component is freed from inflectional 'burden'. This

46

LW/M380

NORTHERN TALYSH

process probably results from language contact with Azeri. As to be expected, some verbs of high frequency have partly retained the present/past dichotomy though they do not belang to the ie-class, cf.: (49)

Infinitive bärd-e kärd-e värd-e härd-e do-e no-e z-e s-e om-e

Present stem bäkävähädana-

zansan(v)o-

Past stem bärd- / bäkärd- / kävärd- /vähärd- / hädonozsom-

'to carry away' 'tomake' 'to bring' 'to eat' 'to give' 'to lay' 'to hit' 'to carry' 'to come'

In a formal perspective, the TAM system of Northem Talysh is characterized by strong fusional tendencies within a basically analytic paradigm. Only in small portions, the Old lranian inflectional system has been retained. Major tools to indicate tense forms are tense-specific agreement patterns as weil as certain tense-like morphemes. Both patterns are formally coparadimatized and functionally coupled. Two series of personal markers apply: 1) a standard set of morphemes that is derived from the Old lranian system of personal agreement. 2) the paradigm of (floating) personal clitics described in section 6.2, cf. (50): (50) !SG 2SG 3SG IPL 2PL 3PL

Type 1 -m

-s -0 -mon -on -n

Type2 -(:l)m -:l -(:,)s -(:,)mon -(:l)on -(:,)son

Note that Type l represents bound morphemes or suffixes that are not allowed to 'float' within the sentence. However, from a historical point of view, we have to assume that Type 1 elements are derived from a structure 'auxiliary + AGR' (or 'copula') that was cliticized to specific verbal forms. This comes clear for instance from the present tense, cf. (51)

s-e-da-m carry-INF-PRES- JSG 'I am carrying; I carry'

47

LW/M380

NORTHERN TALYSH

Tue present tense is synchronically construed on the basis ofthe infinitive -e (-ie) to which an element -da- is added. Miller (1953:146) has convincingly argued that the Talysh present tense copies the Azeri vemacular type: (52)

al-mak-ta-ytm carry-INF-LOC-ISG 'I am carrying'

In Azeri, this construction is based on the locative marking ofthe verbal noun almak 'to carry' which produces the typologically and cognitively expected durative or imperfective reading (cf. German ich bin am tragen '1 am carrying'). Northern Talysh has obviously borrowed the locative marker -da- (in Azeri -da is assimilated to -k ofthe infinitive marker producing -ta). lt also serves as a general locative (inessive) morpheme, cf.: (53)

cama nima ro-a-da ba-yi gülla gänd-e [PA 40] l:ross(>A) half:REL road-REL-LOC to-3so:OBL bullet throw:PAST-AUX:3so 'When I shot at him half-way to me.'

In Talysh, the personalization of the locative infinitive is probably based on a now lost intransitive auxiliary that is represented e.g. by Old Persian ah-las- (Iso amiy etc.), cf. (54)

s-e-da-m < *s-e-da [ah]-m(iy)

which gives us a basic reading 'I am in carrying'. The structure "'ahmi(iy) etc. then became cliticized in analogy to residues ofthe older (lndo-Iranian) type ofpersonal inflection (cf. Old Persian akunavam, Sanskrit akrvavam 'I did'). The underlying auxiliary also forms the basis to encode secondary (analytic) TAM categories. In these instances, the *bhu- variant (Talysh b-) is used, cf. the analytic imperfect: (55)

Talysh: Azeri:

om-e-da b-i-m gel-mek-te i-di-m come-INF-LOCAUX-PAST-ISG 'I was coming'

The structure of the analytic imperfect is a strong argument against the assumption of Paul (1998:173) who relates Talysh-d- (sie!) to a Northwest-Iranian present stem formative •-nt < present participle *-nt (segmenting tiz sedam '1 am going' as az se-d-am (sie!)). Another formerly analytic tense form is represented by the future tense. Talysh employs the modal preverb ba- (cf. the preposition ba 'to') to which the infinitive is added (fol-

LW/M380

48

NORTHERN TALYSH

lowed by Type 1 suffixes), cf. ba-kärd-e-m 'I will/ shall make' (this corresponds to a reading 'I am to (the) making'). Residues ofthe older inflectional paradigm based on Type 1 suffixes are tobe found in a restricted number ofTAM forms. Tue most salient structure is represented by the imperfect (durative): lt employs the Indo-Iranian augment (a-) and a past tense morpheme -i- that is analyzed as the continuant ofthe Old Iranian optative *-äi- by Miller (1953:154), cf. (52) which demonstrates the modal connotation ofthe Talysh synthetic imperfect: (56)

a merd a-mand-i basa ka-i küm-i-sa i sor [Miller 1953:153) DIST man AUGM-stay-PAST:3SG [IMPERF] to up house-OBL roof-OBL-SUPER one year 'That man should stay on the roof of the house for one year.'

Tue modal function of the synthetic imperfect also explains why the negative future is based on this tense/mood form, cf. a-dä-n-i-m (Auo-give:PRES-NEG·PAST-lso) 'l won't give'. The slot taken by -i- in the imperfect has developed to a general T AM slot in verbal structures that

LW/M380

49

NORTHERN TALYSH

The symbol '>S' denotes 'motion to subjective'. Assuming that at this stage of the language the unmarked word order would have been SV - AOV and GN within the noun phrase, we should assume that (peripheral) A had been placed before the nominalized participle that played the role of a noun in a predicative structure. A synchronized Talysh version of (58) would be: (59)

**pa/ang merd-i kast-a e leopard:NOM man-OBL kill-PART:PAST[:PASS] AUX:3sG 'Tue man killed the leopard.'

Now compare (59) to what we finally get (as one variant) in Northem Talysh: (60a) merd-i palang kast-a-s-e [field notes] man-OBL leopard:ABS kill-PERF-3SG:A-AUX:3sG 'The man killed the leopard.'

have a Type 1 agreement paradigm, cf. (b) (57)

Present conditional: b::i-vot-::i-m b::i-vot-o-m Present optative: a-vot-i-m Imperfect:

SUBJ-say-C0ND-ISG SUBJ-say-oPT-ISG AUGM-say-PAST-lSG

'I would say' 'I should say' 'I was saying'

Structurally speaking, this slot is also taken by the morpheme -a- that encodes perfective (aorist) TAM forms, cf. gani-a-m (fall:PAST-PERF-lso) 'I have fallen'. Yet, -a- should be regarded as a participle marker rather than a TAM formative in a historical perspective: lt is assumed to have developed from the Old Iranian past participle (passive) *-(ta-)aka (cf. Miller 1953:189, Pirejko 1991 :150). As known from quite a number of other Iranian and Indo-Aryan languages, the strong S=O (passive) orientation underlying the *ta-participle has provoked a decisive split in the relational behavior ofthose actants that are used with *ta-derived TAM forms (see among many others Edel'man 1974, Pirejko 1979, and Payne 1979 for the Iranian data, Peterson 1998 for a recent discussion ofthe Indo-Aryan data). The *ta-participle obviously underwent strong nominalization that was structurally coupled with the association of a certain degree of referentiality especially in transitive structures. The participle then acquired a predicative reading that could be additionally marked by a possessor via clitization of the corresponding Type 2 elements. For Talysh, we have to assume that the *ta-participle gradually strengthened its S=O orientation. Tue A-function was introduced with the help of the (geni-tivebased) oblique case (cf. German er ist mein Gefangener 'he is my prisoner'). Prototypically, we would expect something like (58) in Early Talysh (or Pre-Talysh): (58)

NP:NOM:o(>s) [NP:oaL:A VERB:PART:PAST:PASS]NP AUX:AGRo>, :PRESIPAST

merd-i palang-on kast-a-s-e [field notes] man-OBL leopard-PL:ABs kill-PERF-3sG:A-AUX:3so 'The man killed the leopards.'

I have added an example that has a plural patient (pa/angon) in order to show that the auxiliary is no longer sensitive for number agreement with 0. Tue residue of a former categorial agreement with the patient can be found in parts in Southem Täti: e.g., in EshtehllrdI, the verb tends to agree with human or at least animate O as for number and gender in certain past tense forms, cf.: (61a) Hasan-e hasar-es darund bu ke zaminlarz be-bu [Yar-Shater 1969:237] Hasan-OBL yard-3sG:A sweep:PAST:M AUX:PAST:M when earthqake PAST(MOD)-be:PAST:M 'Hasan had swept the yard when the earthquake occurred.' (b)

Hasan-e vafra-s derund-a bia ke .... [Yar-Shater 1969:237) Hasan-OBL snow-3sG:A sweep:PAST·F AUX:PAST:F when .... 'Hasan had swept the snow when .... '

(60) differs from (59) with respect to at least to points: First, word order has changed from OAV to AOV, and second, the verbal structure includes a possessive clitic (-s). Remember that intransitive constructions do not use Type 2 clitics, but Type 1 suffixes, cf.:

LW/M380

50

NORTHERN TALYSH

(62a) sapa ve davax-a [PA 123] dog much bark:rAST-PERF:3so:s 'The dog has barked loudly.' (b) palang-i (...) narra-s ka [PA 20) leopard-oBL ( ... ) roar-3sG:A make:rAST:PERF 'The leopard ( ... ) has roared.'

51

LW/M380

NORTHERN TALYSH

In Northem Talysh, the verb never agrees with SAPs in O function (contrary to some Southem Talysh dialects that may agree in number, cf. 6.1): as has been said above (6.2), personal pronouns that have O function must not be marked by the absolutive. Instead, an accusative-like case form (-ni) is used (cf. ma-ni in (63)). This gives us a tripartite substructure in certain past tense forms, cf.: (67)

From this we should expect that O-agreement in transitive structures is based on Type 1 suffixes, too: Remember the strong S=O orientation of the *ta-participle which conditioned a parallel behavior of S and O with respect to agreement in Talysh. But does O really represent the trigger of agreement with the auxiliary in transitive constructions (as suggested by (59))? Consider (63): (63)

palang-i ma-na-s gat-a [PA 44) leopard-OBL I-Acc-3sG:A take-PERF 'The leopard has taken me.'

Prototypically, we would expect something like (64) which is illustrated by an example from Southem Täti (EshtehärdI) in (65): (64)

**palang-i äz-as gat-a-m leopard-osL l:ABs-3sG:A take-rERF-lsa:o

(65)

This kind ofO-split (O[nsAP] > ABS, O[sAP] > Ace) is directly associated with the Person Hierarchy and should be interpretd in the context of 'social deixis': SAPs in O-function cannot be processed in analogy with nSAP (whether human or not), but have tobe referred to with the help of a historically deictic (or locative-allative) element -ni that has become grammaticalized as an accusative (remember that -ni is also used with definite nouns ending in a vowel in O-function, cf. (16)). In consequence, the O-function looses its verbal binding conditions: The construction becomes less O-centered - or, less transitive (cf. Schulze 1998:457-491 for a discussion of weighting processes within transitive structures). Kurmanci demonstrates that the splitting ofO effects the agreement pattem, cf. (68) and again (66b):

az-as biJriJsta-ym [Yar-Shater 1969:155) l:ABS-3SG:A send:PAST-ISG:O 'He sent me.'

(68)

min bi van r;irok-an ne di-zan-i [Bedir Khan & Lescot 1986: 157] l:OBL with PR0X:PL girl-PLneg DUR-knOW:PAST-PAST 'I did not know these girls.'

Accordingly, we can assume that O-agreement in Talysh 'ergative' constructions is split just as O is split, cf.:

I have mentioned in section 6.2 that e.g. the Kunnanci dialect of Kurdish has focused on an S=O agreement in such instances, cf. again: (66a) ez ket-i-m erde [Bedir Khan & Lescot 1986:124) l:ABS fall-PAST-ISG:S earth:OBL 'I feil down to the ground.'

(b)

ker;ke ez dit-im [Bedir Khan & Lescot 1986:153] little=girl:oBL:F I:ABS see:PAST-ISG:0 'The litlle girl saw me.'

(69) SAP nSAP

Case ACC ABS

AGR

+

>

zero-function

If this analysis is correct, the element -e in (60) in fact represents the relict of a former Oagreement pattem (as is normally proposed for other NW-Iranian languages, too). lt can be related to the Old Persian copula astiy (Avesta astt) and also serves to construe predicative structures (cf. bo ma-ro haif-e (for I:oBL-for pity-Aux:3so) 'I feel sorry for him/her/it', lit. '(s)he/ it is a pity forme'). In order to fully account for the construction described in (60), we have yet to explain the change in word order as weil as the syntax ofthe A-clitics (Type 2). We have seen that the

LW/M380

52

NORTHERN TALYSH

expected word order would have been something like OAV resulting from O[NP:OBLA]V < O[>S]-[NP:POSSA Nv] AUX. 3 Obviously, this structure has been re-aligned according to the general preference in Talysh to place the A-function in the sentence initial topic position. Tue clitics (Type 2) seem to represent (anaphoric) traces of this process that is based on a topicalization procedure, cf. (70)

Tue actual way to encode verbal possession ('have') can be regarded as a relict ofthe underlying 'possessive' construction: Talysh lacks a verb 'have'. Instead the language refers to the strategy oflong distance possession, cf.: (71)

ay ba cama riz nison-as do-e [field notes] he:OBL to I:POSS trace sign:INC-3SG:A give:PAST-AOR:3SG 'He showed me the trace.'

(d)

ba katto-i (< *-s) sekayat kard-e [Miller 1953:168] to chairman=of=parish=council-3sG:A complaint make:PAST-AOR:3sG '(S)he complained to the chairman ofthe parish council.'

(e)

cama-s glai müaxol bari-e [Miller 1953: 168]

Note that the overt possessor is marked by the obliquus and not by the possessive form cayi. This indicates that (71) is treated as a transitive structure rather than an intransitive one. (71) has retained features of O( harda). But note that -a effects other final vowels, cf.: (75)

(b)

pa/ang-i vind-as-e ( ... ) [PA 20) leopard-osL see:PAST-3SG:A-AUX:3sG ( ... ) 'The leopard saw [that] ( ...)'

(c)

( ... ) epast-amon-e värd-amon-e ba ka [PA 72] ( ... ) wrap=up:PAST-IPL:A-AUX:3SG carry:PAST-IPL:A-AUX:3SG to house '( ... ) we wrapped [it] up [and] carried [it] home.'

ma-n-a (< *ma-ni-a) ba därd da no-a [Miller 1953:172) I:OBL-ACC-2SG:A to pain into lay-PERF 'You (sg.) have hurt me.' (lit. 'you have laid me into pain.')

An alternative to the assumption mana < *ma-ni-a seems possible if we can give further evidence that the 'shortened' accusative form of the pronouns in 0-function is restricted to the past tense constructions in question (cf. the discussion of the Hazarrudi Tati paradigm in

allows two observations. First, floating clitics may not appear to the right of a verb or before their head (that is in cataphoric function), cf.

(26)). Obviously, the missing of 2s0 clitics is directly related to the basic function of floating in Talysh, namely focusing: Tue (forrner) anaphoric function of clitics produced marked structures ofthe type 'constituent + clitic' that became grammaticalized as focused structures, cf.: (76)

(79)

palang-i merd-as kast-e [field notes] leopard-osL man-3sG:A kill:PAST-AUX:3sG 'The leopard killed the man.' < *leopard;'sroP manyFO OBL, 0 > ABS). But note again that even in such structures agreement has been realigned according to an accusative orientation (see Schulze 1999 for an evaluation of competing strategies within the Accusative Ergative Continuum (AEC)). If personal pronouns that encode Speech Act Participants are involved, the picture changes: On the one hand, SAP in 0-function are referred to in a more 'distant' way. They are normally encoded by the (directional) ni-accusative, cf.:

7. Beyond the default: The periphery of relational behavior in Northern Talysh

(89)

(87a) hamin ba vara-da ba-na ki mammad-as cayan värd-a [Miller 1953:214] just to place-INESS coNo-reach:PRES:3sG REL Mammad-3sG:A from=there bring:PAST-PERF 'He should go to just that place from where he has brought Mammad.'

(b)

a moy-as ki gat-a be [Miller 1953:175] DIST fish-3sG:A REL take:PAST-PERF AUX:PAST:3sG 'That fish which (s)he had caught...'

Ifwe summarize the relational behavior ofS, A, and O as it is illustrated by case marking and agreement, we arrive at the following picture: Prototypical core

Periphery

60

LW/M380

NORTHERN TALYSH

This metaphorical extension of O-marking techniques is obviously related to the inherent definiteness of personal pronouns. Still, this kind of O-split should be interpretd in pragmatic terms, too, because we have seen that nSAP referents in perfective structures do not participate in the technique of Differentiated Objective Marking (DOM), as is true for O-referents in irnperfective structures (see below). On the other hand, all SAP pronouns except the first person singular normally do not use a distinct case form to encode A-function in perfective structures. ta, ama, and sama obviously conform to the standards ofthe Person Hierarchy, cf.: (90)

S

II (91)

0

A

ta, ama, sema

s

Type l suffixe~

Case

tani, amani, samani 1 Agr

0 A

>

>

[*AUX]

1

Type 2 clitics

(90) is a scherne ofthe 'accusative' behavior of overt personal pronouns (except SAP(l)). This scherne in contrasted by the tripartite agreement system in (91). The fact that overt pronouns can be 'underdifferentiated' with respect to the marking of S and A in 'ergative' structures should be explained by referring to the lexical status of pronouns: Pronouns normally own much stronger autonornous sernantics than personal clitics. If we assurne that the Person Hierarchy is - arnong others - based on the conceptualization of Speech Act Participants, it is rather probable that personal pronouns - the lexical representation of SAP concepts - are arnong the first candidates to observe this hierarchy. The less a given elernent has a sernantic reading the more its functional domain becomes relevant: lt can be exploited to differentiate sub-strategies that are not based on the conceptualization of SAP in general, but on some specific properties such as agentivity only (for a rather parallel split in Tabasaran (Southeast Caucasian) see Schulze 1999). Interestingly enough, the first person singular is exempted from this split. The corresponding pronoun uses az for S=A and ma for A[ ;=S] in perfective structures, cf.

(92)11--:--:-a-----l1-la__>_ _m_a_n_i____

....J

(93)

s

Type l suffixe

A

Type 2 clitic

0

>

[*AUX]

1

NORTHERN TALYSH

Accordingly, the tripartite system of agreement is copied by a parallel behavior of the overt pronoun. This fact should be interpretd in terms of a 'left shift' on the Person Hierarchy that represents a kind of 'agentive overstreching' (cf. Schulze 1999 and Schulze (forthcoming a, chapter 2.1) for parallel data in East Caucasian): The inherent agentivity that is associated with the concept EGO underlying SAP(l) receives an additional (iconic) morphological signal, cf. (note that the term 'ergative' is used here to indicate the ergative component in the tripartite behavior): (94)

ta, ama, sema ..__ _ _ _ _ _ _ _ _ _ __

61

LW/M380

SAP(l) S;=A;O Ergative S;=A

SAP(2) S=A;O Accusative S;=A

SAP(l).PL S=A;O Accusative S;=A

SAP(2).PL S=A;O Accusative S~A

nSAP S=O;A Ergative S;=A

Naturally, this picture slightly changes if we allude to the diachrony of this co-paradigmatic structure. We have seen in section 6.2 that the overt pronouns for plural SAPs (ama and sama) as weil as for nSAP (av(on)) result from Old lranian oblique forms. There are no traces ofthe old nominatives that are documented e.g. by Avesta vaem 'we' (Old Persian vayam), yüiäm or yüi (Old Avesta) resp. yuiam (Young Avesta) 'you (plural)' and huuö (Old Avesta) resp. häu (Young Avesta) 'they' < distal (cf. Old Persian hauv (m.f.), but note Old Persian ava (n.)). The tendency towards an 'obliquifization' of the paradigms of personal pronouns represents a well-known feature in many younger Indoeuropean languages and should be interpretd in terms of pragmatic clefting strategies. However, they normally do not affect the basic relational behavior, although the process in question may temporarily result in a 'neutral' behavior (S=A=O). lt has been said above that the strategy in subordination to encode S and A by the oblique case in its possessive function levels the clear-cut distinction between perfective and imperfective clause types (note that I use the terms 'perfective'/'imperfective' in a prototypical sense). This strategy shows that variations in case marking still have a considerably high pragmatic value. lt has been claimed in section 6.1 (cf. (l 7)) that the basic function ofthe 'casus rectus' (absolutive) is to encode the unmarked core domain whereas the oblique case symbolizes any kind of peripherization. With O in accusative strategies, this peripherization expresses a pragmatic peak related to definite (or afore-mentioned) referents. With A in ergative strategies, it signals the existence of a causal periphery that has gradually usurped properties of the core domain (a 'metonymic' process in terms of the 'Grarnmar of Scenes and Scenarios', cf. Schulze 1998:557-575). We should expect that the dynarnic relation between core and periphery allows a certain degree of variation related to case marking and agreement. This potential becomes even more salient if we take a look at how aspects of word order based relational behavior are co-paradimatized: (95) lists the relational primitives of the first 14 clausal units of palangi ahvolot.

LW/M380

(95)

62

NORTHERN TALYSH

1. SVaOV aV 2. SVaOV 3. SV 4. SV 5. SVaV SV SV 6. ---

7. AOV 8.aOVaV 9. aVaV SV 10. aOV 11. aVO[SV] 12. aOV 13. SV aVO[aOV] 14. SV AOV a[O]V Tue symbol a indicates either a covert actor (marked by agreement only) or a deleted, coreferential actor. lt comes clear that referential tracking is established on a pure accusative basis in the passage in question. S and A both have pivot function and never follow a verb. There seem to exist two 'pragmatic domains' (or slots): a) an initial topic domain, and b) a clause final cataphoric domain that has focus properties. Yet, there is a clear distribution ofthe three relational primitives with respect to these domains: S and A occupy the topic domain whereas only O (as weil as 10) may or may not appear in clause final focus as long as this clause is followed by another clause in coordination or subordination. In other words: Word order in Northem Talysh is rather stable and accessible to pragmatic variation only to a very limited extent. Consequently, we can assume that pragmatic strategies will rely more on variations in case marking or agreement than on word order. In his treatment ofNorthem Talysh, B.V. Miller has devoted a special chapter to what he calls „kontaminirovannye formy ergativnyx postroenij" (Miller 1953:175-182) [contaminated forms of ergative structures]. He discusses a number of Talysh sentences that exhibit case marking strategies different from the prototypical structures discussed so far. As expected, it is basically the ergative behavior that experiences significant modifications. On the one hand, A marked by the oblique case tends tobe replaced by the 'casus rectus' or absolutive without affecting the agreement pattem. We have seen that this strategy is already standard with personal pronouns others than the first person singular. Inpalangi ahvolot, the absolutive marking ofthe first person singular in an ergative based A-function occurs twice, cf.: (96a) asafna äz-an bäslamis-am kärd-e [PA 15] now l:ABS-FOC tum-lsG:A make:PAST-AUX:3SG 'Now, I too tumed around.'

LW/M380 (b)

63

NORTHERN TALYSH

äz palang-i ba bar da-s-a vara-ka tifang-am pekarn-i [PA 19] l:ABS leopard-oaL to bush in-go:PAST-PERF place:OBL-LOC rifle-tsG:A lift=up:PAST-PAST 'When the Ieopard went into the bush, I lifted the rifle on the spot.'

The raising of expected ma (oaL) to az (ABs) obviously results from the specific pragmatic function that is associated with äz in the given context: In (96a), the pronoun is followed by the focus marker -an,4 in (96b) we have to deal with the clefting of äz that serves to mark focus, too (note that the oblique case palangi (instead of palang) is due to embedded A-split in subordination). Unfortunately, the text palangi ahvolot does not give clear evidence whether the focus marker -an may have the same effect with third person pronouns or nominal referents. In PA 43, for instance, -an is added to the obliquus marked agent, cf.: (97)

palang-i-an (. ..) av-as gat-e [PA 43]

leopard-oBL-FOC (... ) he:ABs-3sG:A take:PAST-Aux:3SG 'The leopard (... ) took him.' Yet, the raising of A:oBL to A:ABs is not uncommon, especially in switch reference; cf.: (98)

av ba do rost be tüla cayi düs gat-e [PA 30] he:ABs to tree hide LV:PAST:3sG hound he:ross tail take:PAST-AUX:3sG 'He hid in the tree [and] the hound grabbed its tail.'

Here tüla 'hound' replaces the expected form tiilay. However, we cannot teil for sure whether the form tüla results from a switch in case marking or whether we have to deal with a phonetic process (tülay > tüla). The examples given by Miller (1953: 176-181) in fact hint at the possibility to replace the oblique case by the absolutive, cf.: (99)

ien os e-kärd-as-e [Mill er 1953: 176]

woman:ABS soup out-make-3sG:A-AUX:3sG 'The woman poured out the soup.' (98) differs from (99) in that it does not show agreement anymore. 5 This fact relates (98) to the standard way imperfective sentences are construed (remember that the third person singular affix is zero here), cf.:

4

The suffix -an has focus function (often contrastive) that sometimes merges with topicalization. This claim has to be relativized: There is a rather arbitrruy distrribution of düm (postposition: 'behind', noun: 'tail') and düs (noun: 'tail') in the text. Though the existence of düs 'tail' had been confinned by my infonnant, it may weil be that düs in (98) (as weil as in PA 29) stems from*düm-as (tail3SG:A). In this case, (98) would include a floating agreement clitic. 5

LW/M380

64

NORTHERN TALYSH

(100) tüla cayi düm-i gat-eda [field notes] hound he:Poss tail-OBL take-PRES:3sG:A 'Tue hound takes its tail.' The tendency to drop the agreement clitics is rather strong especially with the pluperfect. This tense form represents the perhaps most 'accusativized' structure within the perfective paradigm, cf.: (101) basta däs värd-a b-i-m a xanjil-i [Miller 1953:179] to:REFL:POSS hand take:PAST-PERFLV:PAST-PAST-ISG:ADIST knife-OBL 'I took the knife [khindzhal] in my (own) hands.' Tue expected form would be: (101 ') basta däs a xanjil-am värd-a be to:REFL:POSS hand dist knife-lsG:A take:PAST-PERFLV:PAST:AUX:3SG which was accepted as 'normal' by my informants. Note that the structure in (101) violates the Talysh constraints on sentence final constituents probably due to interferences with Russian (the idea to interpret xanjil-i as some kind of afterthought would call for a cataphoric pronoun in the preceding segments which, however, does not occur). The example illustrates the second major tendency in the dynamics of case marking in Northem Talysh, namely the introduction of O-splitting techniques in the perfective paradigm (xanjili instead of xanji/). O-split is quite often met in subordination, a structure which - as we have already seen - shows a rather accusative behavior, cf.: (102a) cayi gac-i sipi kä [PA 54] he:ross teeth-oaL white make:PAST:PERF 'When he bared the teeth .. .' (b)

cama a palang-i yali i-a gora [PA 77] l:ross rnsr leopard-oaL alone hit-PERF because 'Because I killed the leopard on my own ... '

(c)

cama tifang-i gordon-e [PA 55] l:ross rifle-oaL tum=around:PAST-AOR 'When I tumed around the rifle .... .'

Tue tendency to extend the functional domain of the oblique case becomes apparent from another fact: In some instances, this case form is used in prepositional phrases, cf.:

LW/M380

65

NORTHERN TALYSH

(103) zan-a-m-e ki ba tüla gtlnd-a gülla bapalang-i ba-gan-e [PA 64] know:PAST-PERF-ISG:A-AUX:3sG that to hound throw:PART:PAST:REL bullet to leopard-OBL FUTfall-FUT:3SG 'I knew that the bullet shot at the hound would hit the leopard.' Here, ba palangi 'to the leopard' replaces the expected form ba palang. lt can be assumed that this development is related to the fact that ba calls for an obliquus (possessive) with pronouns (cf. (99) basta < *ba-asta to-REFL:ross, ba cama 'to me' etc.). If we summarize the periphery of relational behavior in Northem Talysh, we arrive at the following picture: On the one hand, there is a strong tendency to realign the case marking of nominal and pronominal constituents in transitive perfective constructions according to an accusative behavior. However, this development is not automatic. Rather, we observe a pragmatic motivation (A-focusing) as well as structural conditions (accusative O-split in subordination) that gradually reformulate the way how case marking in 'ergative' transitive sentences applies. On the other hand, the possessive component ofthe obliquus experiences a functional extension that is perhaps best documented by its use to encode S and A in subordination. (102) illustrates that the possessive construction may even penetrate the domain of main clauses, cf.: (104) cayi ca do-kü ba ma-sa jo be cama nima ro-ada ba-yi gülla gänd-e [PA 40] he:POSS(>A} he:POSS tree-LOC to l:OBL-SUPER separate LV:PAST:3SG l:POSS(>A) halfREL way-INESS to-he:OBL bul!et throw:PAST-AUX:3SG 'When he left his tree towards me, I shot at him half-way.' The main clause cama nima roada bayi gülla gände is construed just as the preceding embedded clause (instead of expected ma nima roada bayi gülla-m gttnde). 6 But contrary to the older type ofpossessive (or oblique) constructional type that underlies West Iranian 'ergativity' this new possessive type has a clear accusative orientation. This orientation becomes even more apparent ifwe take into consideration the loss ofagreement. A sentence type like (105a) cama ba di om-e [field notes] I:ross(>s} to town come:PAST-AOR 'When I came into the town ... .'

6 One could speculate whether the second clausal structure is again embedded into the following sentence (PA 41 ): nima vote gocaga gülla cayi pesti barasto (' it seemed that the little bullet broke his backbone'). However, my informant produced a lengthy pause between PA 40 and PA 41 which would argue against such an interpretation. Also, the assumption of a doubled embedded structure does not meet the textual semantics of this passage.

LW/M380

(b)

66

NORTHERN TALYSH

cama am kogaz-i rosn-i [field notes] I:ross(>A) PR0X letter-OBL receive:PAST-AOR 'When I received this letter.. .'

does not show any 'ergative' behavior anymore though it belongs to the perfective paradigm. The extension of functional properties related to the oblique case incidentally gives rise to hypertrophic techniques that are characterized by a seemingly unmotivated use of i-fom1s. A good example is

LW/M380

This sentence violates standard constraints on Northem Talysh morphosyntax at least with respect to three aspects: 1) the word order is OVA; 2) the oblique case is used to encode both A and O; 3) the sentence lacks agreement because the trigger (müz) occupies the sentence final position. We have seen in section 6.3 that the trigger of agreement must always precede its clitic. In (106), agreement fails because müz is not followed by a constituent that could serve as the host of the clitic (müzi cama nemek-as härda would be the standard version of (106)). Yet, the value of the example in (106) remains doubtful. Miller mentions the same phrase in Miller 1953:190 to demonstrate the technique ofsubordination and translate it as 'nasu so!' s"evsaja muxa' ('the fly eating our salt'). Unfortunately, I was not able to retrieve this example in Miller 1930. Yet, it depends on the context to decide whether the oblique fomi muzi encodes the agentive of härda 'eating/eaten' or whether it represents the agentive of a superordinate clause.

7. Conclusions In this booklet, I have tried to describe some salient features ofNorthem Talysh morphosyntax as they become transparent from the text source underlying this analysis. The idiosyncrasies ofthis text did not allow to draw a complete picture ofboth the core and periphery ofthe Talysh granimatical system. In order to obtain a more comprehensive data base, I had to add some data from Miller 1953 (which again are taken from Miller 1930). The analysis has shown that Northem Talysh operates on the basis of a rather limited morphological inventory that is based on the core domain of its phonological system. Northem Talysh rarely exploits phonological elements of the marked (peripheral) domain to encode marked grammatical categories or to augment its morphological devices. Instead, the bulk of grammatical categories is based on emergent structures: Morphemes acquire a specific function mainly through co-paradigmatization which corresponds to the structural coupling ofbase functions. Tue high

NORTHERN TALYSH

degree of co-paradigmatization establishes some kind of fomial morphological network that does not allow to treat morphological segments separately. In this sense, Northem Talysh grammar represents a rather interesting 'mixture' between agglutinating strategies on a rather low Jevel and inflection. The complex system of structural coupling leads to various types of splits in the behavior of relational primitives most of which I could touch upon in a preliminary way only. Many of the features in question still await a more comprehensive treatment that should also encompass aspects of recent language contact (mainly with Azeri) and typological parallels in other present-day NW-Iranian languages (mainly Southem Täti and Zäzäki).

(106) cama nemek-i härd-a müz-i [Miller 1953: 176] we:ross salt-oBL eat:PAST-PERF fly-oBL 'The fly has eaten our salt.'

67

LW/M380

68

NORTHERN TALYSH

Appendix 1: Palangi ahvolot Note: The text palangi ahvolot is given in a rather literary translation. The leopard and the dog are generally referred to by 'he'. See Appendix III for the glossing.

1. az sahat hast-ada be-s-i-m kay-kü, asta visa a-s-a sap-on xorak-on-am do, tifang-am tamiz ka, pe-gat-a-m-e s-i-m ba visa.

l:ABS hour eight-INESS out-go:PAST-AOR-ISG:S house-Loc REFL:POSS wood AUGM-gO:PART:REL dog-PL food-PL-ISG:A give:PAST:PERF, rifle-lsG:A clean make:PAST:PERF up-take:PAST-PERF-ISG:A-AUX:3SG go:PAST-AOR-)SG:S to wood. 'I left my house at eight; having fed my hunting dogs, having cleaned the rifle, [and] having taken [it] up I went to the wood.'

LW/M380

69

NORTHERN TALYSH

8. av-am hattä dast-a gayis-i dabäst-e s-i-m palang-i riz-i-sa "'' he:ABs-lsG:A even field-REL lead-oBL fasten:PAST-AUX:3so go:PAST-AOR-lso:s leopard-osL trace-OBLSUPER 'I even fastened him [to] a (field)lead [and] I followed the trace ofthe leopard.'

9. be-s-i-m käm s-i-m av da-s-e ba yol-a bar. out-go:PAST-AOR-ISG:S a=Iittle go:PAST-AOR-ISG:S he:ABS to-go:PAST-AUX:3SG:S to big-REL bush. 'I went [much], I went a little, he went to a big bush.'

10. ayo-m bar-i atrof-am riz kärd-e. then-lsG:A bush-osL surroundings-Iso:A trace make:PAST-Aux:3so 'Then I checked the surroundings of the bush.'

l:ABs three hour walk:rAST:PERF-LOC after l:OBL leopard-osL trace-lsG:A see:PAST-Aux:3sG 'After a walk of three hours, I saw the trace of a leopard.'

11. ba-vind-e-m ki dogardän-an palang ca bar-i-kü davärd-a-ni. Fur-see:PRES-INF-ISG:A that really-Foc leopard from bush-osL-LOC corne:PAST-PERF-NEG:Aux:3so:s 'I would realize that in fact the leopard had not come from the bush.'

3. hark külük vo-a be. strong rain fall:PAST-PERF LV:PAST:AUX:3SG •A heavy rain had fallen.'

12. ba mament-ada sapa-m va do. this moment-INESS dog-lSG:A free give:PAST:PERF:AUX:3so 'In this moment, I let go the dog.'

4. havo mülayim be.

13. sapa ve davax-a na-zn-a-s-e av ca bar-i-kü be-gand-e. dog much bark:PAST-PERF:3so:s NEo-can:PAST-PERF-3SG:A-AUX:3so he:ABs frorn bush-oBL-LOC outchase:PAST-AUX:3so 'The dog Joudly barked [but] he could not chase him out ofthe bush.'

2. az se sahät nav-a-kü babesta ma palang-i riz-am vind-e.

weather lenient LV:PAST:AUX:3so 'The weather was Ienient.' 5. äz be-s-i-m käm s-i-m palang-i riz-i-sa, cama sap-on-kü ila palang-i bü-kü tärs-a, ba düm mänd-e.

l:ABs out-go:PAST-AOR-ISG:s a=little go:PAST-AOR-ISG:S leopard-oBL trace-OBL-SUPER l:POSS dog-PL-LOC one leopard-osL smell-Loc become=afraid:PAST-PERF:3so:s to back stay:PAST-AUX:3so:s 'I went and went a little following the trace ofthe Ieopard; one ofmy dogs had become afraid ofthe smell ofthe Ieopard [and] stepped back.' 6. taxminan vist-si metra cama-kü diaro abi. about twenty-thirty meter l:ross-Loc away there. •About twenty to thirty meters away from me.' 7. anJax galay-m-i ofci-a sapa hark riz-i-sa se habas-kor be. there one-Iso:ross-one hunting-REL dog rnuch trace-osL-SUPER go:INF wish-rnaking LV:PAST:3so:s 'There, one ofrny hunting dogs wanted to follow the trace.'

·., '.t

14. ba mament-ada am hayvon bar-i-kü be-na-s-e gora, yani ki az ayi ba-vind-o-m, vind-e-m na-zn-a. . this moment-INESS prox anirnal bush-osL-LOC out-NEG-go:PAST-Aux:3so because that=1s CONJ l:ABS he:OBL SUBJ-see:PRES-OPT-ISG:Asee-INF-!SG:ANEG-can:PAST-PERF . ' 'In this moment, as this animal did not come out of the bush - so that I would see h1rn - I could not see him.' 15. asafna az-an baslamis-am kärd-e bar-i gardo gard-a-y. now l:ABS-FOC turn-lso:A rnake:PAST-AUX:3S0 bush-OBL around move:PAST-PERF-AUX:3SO 'Now I tumed around, too, [and] moved around the bush.'

LW/M380

70

NORTHERN TALYSH

16. vot-e-m ki bada palang ba-vind-o odam gtird-eda, be-s-i bar-i-kü, tlz bayi illa gülla bagänd-o-m.

say-INF-lso:ross coNJ come=on leopard SUBJ-see-oPT:3sG:A man move-PRES:3so:s out-go:PAST-AOR bush-oaL-LOC l:ABs to:he:OBL one shot SUBJ-throw:PRES-OPT-lso:A 'I said [lit. '[it was] my saying']; 'Come on!' so that he would see a man walking; coming out from the from the bush; I would shoot at him.' 17. ammä palang be-na-s-e.

but leopard out-NEo-go:PAST-AUX:3so 'But the leopard did not come out.' 18. palang-i bar-i-kü be-na-s-e ba-m-as ve tasir kärd-e gora cama badan astan-bastan lärz-a.

leopard-oaL bush-oaL-LOC out-NEo-go:PAST-AUX:3so at-1:0BL-Jso:A much effect make:PAST-AUX:3so because hoss body REFL:0BL-to:REFL:0BL shake:PAST-PERF. 'Because the leopard who did not come out of the bush impressed me very much, my body trembled.' 19. äz palang-i ba bar da-s-a vara-kü tifang-am pekarn-i ba po gada-gada damänd-i-m ba cayi /ona da-s-e.

1:ABs leopard-oaL in bush in-go:PAST-PERFplace:oaL-LOC rifle-lso:A lift=up:PAST-AOR with foot carefully begin:PAST-AOR-lso:s in he:ross cave in-go-INF 'When the leopard came into to bush, I lifted the rifle [and] carefully started to climb into his cave.'

LW/M380

71 .

NORTHERN TALYSH

24. ve sahat-am kam sahat-am ba cay düm anjax tüla-m cay düm-i-kü däs na-kas-a. much time-1s0:s little time-Iso:s to he:Poss behind but hound-Iso:Poss he:ross tail-oaL-LOC hand NEG-leave-PERF 'I was behind him a long time, a short time; but I did not set free my dog behind him.' 25. ve vara-kü ogtird-a tüla ba-gat-o anjax tüla ve bij be gora astan-as ba cayi gir na-do.

much place-LOC turn:PAST-PART:PERF hound sua1-take-0PT:3so:A but hound much smart LV:PAST:3so:s because REFL:0BL-3SG:A to he:POSS paw NEo-give:PAST:PERF 'From all around he turned to take [the dog], but because the dog was very smart he did not give himself into his paws.' 26. taxminan lona-kü pensa metra-y tal-a tlz-an ba cayi düm. about cave-wc five=hundred meter-oaLrun:PAST-PERF:3so:s 1-Foc to he:Poss behind 'He ran away some 500 meters from the cave, I too [ran] behind him.' 27. mavot ayi los-as hest-ada-n.

it=seems he:OBL prey-3s0:POSS be:PAST-PRES-3PL:S 'Obviously he had preys.' 28. av ayo damänd-e ba do pes-e.

he:ABs there start:PAST-Aux:Jsoto tree climb-1NF 'There he started to climb on a tree.' 29. tlz-an gara nafas-a-y ba cayi düs tala-da b-i-m.

20. palang-i vind-as-e ba cayi /ona odam da-s-eda, da kara-s narra-s kä.

leopard-oaL see:PAST-3SG:A-AUX:3so in he:ross cave man in-go-PREs two time-Jso:A roar-Jso:A make:PAST:PERF . 'Tue leopard saw [that] a man was entering his cave [and] roared two times.' 21. dia ma-ka cayi narra kard-e hany-an ba cama xos om-a. note PROH-make:PRES he:ross roar make:INF already-Foc to l:ross wish come:PAST-PERF:3so:s. 'No wonder, that his roaring began to please me.' 22. metra-yan s-i-m ba nä palang asta lona-kü be-gan-i. meter-Foc go:PAST-A0R-lso:sto forward leopard REFL:ross cave-Loc out-go:PAST-AOR:3so:s 'I moved a meter forward, [and] the leopard came out of his cave.' 23. palang-i be-gan-e tlz-an be-gan-i-m ba cay düm. leopard-oaL out-fall:PAST-AOR:Jso:s 1-FOC out-fall:PAST-AOR-lso:s to he:ross behind 'The leopard came (lit. fell) out [and] I too came (lit. fell) out behind him.'

l:ABS-FOC strong breath-PERF-AUX:Jso to he:OBL behind run:PAST-PRES LV-PAST-ISG:S 'I, too, ran behind him - heavily breathing.' 30. av ba do rost be tüla cayi düs gat-e. he:ABs in tree comfortable LV:PAST:3so:s dog(:OBL) he:ross tail take:PAST-3so:s 'He hid in the tree [when] the dog grabbed his tail.' 31. ogärd-e tül-as vist-si metra-s talvon-a. tum=around:PAST-Aux:Jso:s dog-Jso:A twenty-thirty meter-Jso:A chase=away:PAST-PERF 'He turned around [and] chased the dog away twenty-thirty meters.' 32. palang ogärd-e ba do haft-hast metra mänd-a be cayo-s-an astan so do ba do. leopard turn=around:rAsr-Aux:3so:s to tree seven-eight meter stay:PAST-PERF LV:Jso:s from=there3so:A-FOC REFL:0BL throw give:PAST:PERF to tree 'The leopard tumed around stopped some seven eight meters away from the tree [and] threw himself onto the tree.'

LW/M380

72

NORTHERN T ALYSH

33. gasam ba-bü ba xodo av coko ki ba do väst-e ca do-kü balang-i saräf-i hask-a xol arasy-a egan-i ba zamin iago a palang-i do lorzon-i.

oath SUBJ-LV:coNo:3so:s to god he:ABs how coNJ to tree jump:rAsr:3sG:s he:ross tree-LOC arm-oBL thickness-oBL dry-REL branch break:PAST·PERF:3so:s fall=dawn:PAST·PAST:3sG:s to earth so msr Ieopard-oBL tree shake:rAST-AOR 'Let the oath be with God! When he jumped towards the [lit. his] tree, a dry branch thick as an arm broke [and ] fell from the tree dawn on the earth - So (strongly] that Ieopard shook the tree.' 34. ma asta albahal roson-i ba ca do-yi ban. l:OBLREFL in=this=moment direct:PAST·AOR:3soto he:ross-tree-oBL ground 'In this moment, I reached [lit. directed myself] to the foot ofhis tree.'

35. ma ba-yi-m cokcoki-m dia kärd-e. l:oBL to-he:OBL-ISG:A much-IsG:A look make:PAST·AUX:3so 'I stared at him.' 36. cayi a harakat-on ma vind-e gora ma na-pi-a ba-yi gülla ba-gänd-o-m. he:ross DIST movement-rL I:oBL see:PAST-Aux:3SG because l:OBL NEG-want:PAST-PERF to-he:OBL bullet SUBJ-throw-orr-1 SG:A 'Because I saw him moving, I did not want to shoot at him.' 37. anJax tifang-am däs-ada hozo be. but rifle-IsG:ross hand-INESs ready LV:PAST:3sG 'But I had my rifle ready in [my] hand[s].' 38. ba cayi beg-on-om ve tämso kärd-e. to he:ross moustache-PL-ISG:A big Iook make:rAsr-Aux:3sG 'I stared at his moustache.'

39. cayi cas ba cama cas-i gan-e astan cayo-an dik cama kalla-sa ce do. he:ross eye to l:ross eye-osL fall:rASr-Aux:3sG REFL:OBL from=there-Foc directly I:ross head-suPER jump give:PAST:PERF

'When his eye[s] feil into my eye(s], he himselfmade ajump from there directly on my head.' 40. cayi ca do-kü ba ma-sa Jo be cama nima ro-ada ba-yi gülla gänd-e. he:OBL he:ross tree-Loc I:oBL-SUPER separate LV:PAST:3so:s l:ross half:REL way-INESS to-he:oBL bullet throw:PAST-AUX:3SG 'When he left his tree towards me [and] I shot at him half-way.'

LW/M380

73

NORTHERN TALYSH

41. nima vot-e gocag-a gülla cayi pesti ba-rast-o. half say-INF little-REL bullet he:ross backbone sus1-break-orr:3sG:A 'lt seemed that the little bullet broke his backbone.' 42. cayi ba zamin egan-e hamin sapa ki cayi-kü tärs-a-da be ras-a palang-as gat-e. he:OBL to earth fall=down:rAsr-Aux:3sG:s PROX:EMPH dog coNJ he:oBL-LOC be=afraid-PERF-PRES LV:PAST:3SG:S approach:PAST-PERF:3SG:S Ieopard-3sG:A take:PAST-AUX:3SG 'When he feil down to the ground, just this dog which had been afraid of him attacked the approaching leopard.' 43. palang-i-an gänd-e gac av-as gat-e. leopard-OBL-FOC throw:PAST-AUX:3SG teeth he:ABS-3SG:A take:PAST-AUX:3SG 'The leopard, too, threw [himself at him and] bit him.'

44 sapa vind-e palang-i mana-s gat-a damänd-e tal-a. dog see:PAST-AUX:3SG leopard-OBL l:ACC-3SG:A take:PAST-PERF begin:PAST-AUX:3SG run:PAST-PERF :3SG:S 'The dog saw that the leopard took him [*me] [and] began to run [away].' 45. düm-a-kü sapa büdan-as gat-e. tail-OBL-LOC dog:OBL side-3so:A take:PAST-AUX:3SG 'He took the dog' s side near the tail.'

46. ba mament-ada ma ve harakat kärd-e ba palang-i kalla-sa i gülla ba-ian-o-m lakin av da sapa albayaxa be gora äz tärs-a-m ki gülla ba-gan-o da sapa ba cayi kalla-m na-zn-a gülla gänd-e. PROX moment-1NESS l:OBL big movement make:PAST-AUX:3sG to leopard-OBL head-suPER one bullet sus1-hit:PAST-OPT-lsG:A but he:ABs with dog scuffie LV:PAST:3so:s because l:oBL fear:PAST-PERF-Isa:s coNJ bullet sue1-fall:past-orr:3sa:s with dog to he:ross head-IsG:A NEG-be=able:rAST-PERF bullet throw-INF 'In this moment, I moved [in order] to shoot into the leopard's head, but because he [still] fighting with the dog I feared that the bullet might hit the dog [and] I was not able to shoot at his head.' 47. hamin mament-ada palang i kara güca be o b-e sapa astan-basta[n] bar-bar kärd-e karde damänd-e se ba palang-i Java. PROX:EMPHmoment-INESS leopard one time yawn:PASTLV:PAST:3SG:sopen LV:PAST:3SG:S dog REFL:OBLto-REFL:OBL noise make:PAST-AUX:3SG make:PAST-AUX:3SG begin:PAST-AUX:3SG go:INF to leopard-oBL stomach 'Just at this moment, the leopard now yawned; when [he] opened [his mouth] the dog himself barked and barked [and] began to run into the leopard's stomach.'

LW/M380

74

NORTHERN TALYSH

48. gasam ba-bü ba xodo avada zoa-m a-p-i ki palang-i nezi se ne ba cayi tarafdia ba-ka. oath susJ-Lv:cOND:3sG to god hold guy-lsa:s AUGM-want:PRES-PAST CONJ leopard-oBL nearness go:JNF NEG:Aux:3sG to he:oBL side look sUBJ-make:PREs:coNn 'Let [my] oath be with god: I'd like to see a hold guy who would not direct himselftowards the leopard [and] look in his direction.'

49. ba cayi gora ki av sapa ebard-i-ada cayo cas-on hä gala nalbaki gadar ba-be. to he:ross because coNJ he:ABs dog swallow:PAST-PAST-PART:PAST from=there eye-PL each one dish like FUT-LV:PAST:3so 'As he swallowed the dog [his] eyes would become Iike dishes.' 50. cayi gav-an iogo o ba-be ki kämina vist-i penJ si santi o ba-be. he:ross mouth-Foc such open FUT-LV:PAST:3socoNJ at=least twenty-oBL five thirty centimeter open FUT·LV:PAST:3SG

LW/M380

75

NORTHERN TALYSH

55. asafna cama tifang-i gordon-e tifang-i lüla-m da-gänd-e be cayi güs-i da/a. now l:POSS rifle-oBL tum=around:PAST-AUX:3so rifle-OBL pipe-lSG:A in-throw:PAST-AUX:3SG to he:POSS ear-osL inside. 'Now, having turned around my rifle, I rarnmed the pipe of the rifle into his ear.' 56. gärd-a-gärd-a palang-i nafas be-ma gan-i iago be ki tel-a zahar taki albahal-om tifang ba po pekarn-i cayi güs-i dalada-m i gülla i-a. move:PAST-PERF-move:PAST-PERF leopard-OBL breath to-1:0BL fall:PAST-PAST so LV:PAST:3so:s CONJ acid-REL poison as in=this=moment-Iso:A rifle to foot pull=up:PAST-AOR:3so he:ross ear-OBL inside-Iso:A one bullet hit:PAST-PERF 'When the leopard's breath feil onto me while he was moving it was so [as if] [it were] an acid poison; just now I pulled the trigger [and] shot a bullet into his ear.'

'His mouth would be opened so much that it would be opened at least 25 to 30 centimeters.'

57. sapa äl xün-i dala-da be taraf-ada var-var kärd-e-da be. dog crimson blood-osL inside-INESS LV:PAST:3so:s side-INESS growl make:-PRES-PRES LV:Aux:3so 'The dog which was bloodstained started to growl.'

51. ma ve harakat kärd-e üian-an ba cayi kalla galla gülla ba-ian-o-m sapa cayi gav-ada be gora ma ba sapa-m hayji om-a vot-e-m ki gülla ba-gan-e sapa ba-kast-e. I:oBL big movement make:PAST-AVX:3sa again-Foc to he:OBL head one bullet sua1-hit-0PT-lso:Adog he:ross mouth-INEss LV:PAST:3so:s because I:oBL to dog-Iso:A pity Lv[come]:PAST-PERF say-INFIso:ross coNJ bullet FUT-fall-INF:3so:s dog Fur-kill-INF:3so:A

58. palang-i oxom-a nafs-ada i narra-s da. leopard-OBL Iast-REL breath-INEss one roar-3so:A give:PAST:PERF 'The leopard roared breathing his last.'

'Again I tried hard to shoot a bullet in his head [but] because a bullet might hit the dog in his mouth I had pity with the dog- saying that the bullet might fall [and] kill the dog.' 52. ma i tifang-a posna-m i-a ba palang-i kalla. I:oBL a rifle-REL butt-lso:A hit:PAST-PERF to leopard-osL head 'I hit [the] rifle butt on the leopard's head.' 53. ba vaxt palang-i vind-e odam ras-a ba-ma asta gac-as sipi kä. PROX time leopard-oBL see:PAST-AUX:3so man come=close:rAST-PERF:S to-1:0BL REFL:POSS teeth-3s0:A white make:PAST-PERF

'Now when the leopard saw that a man came close to him [lit. 'me'] he bared his teeth.' 54. cayi gac-i sipi kä ang-on-as las be sopa nepax-a cayi gav-i-kü.

he:POSS teeth-OBL white make:PAST:PERF jaW-PL-3S0:POSS weak LV:PAST:3so:s dog escape:PASTPERF:3so:s he:POSS mouth-OBL-LOC 'When he bared [his] teeth, hisjaws became weak [and] the dog escaped from his mouth.'

59. äz i kara-n ba-vind-e-m taxminan penJo-sest metra mo-a pa/ang naroldamis be-da. l:ABs one time-Foc FUT-see-INF-lsG:A roughly fifty-sixty meter young-REL leopard roaring LVPRES:3so. 'This time I would realize that some fifty to sixty meters [away] a young leopard is roaring.' 60. cama ba cayi taraf dia kärd-e-nJa sopa tal-a äz-an tal-a-m. I:ross to he:ross side look make:PAST-AUX:3so-TEMP dog run:PAST-PERF:3so:s 1:ABS·FOC run:PAST-PERFIso:s 'When I looked at him, the dog ran away [and] I run, too.'

61. a palang ogärd-e ba ces tüla se ca palang-i düm-os gat-e. rnsr leopard turn=around:PAST-AUX:3sG:s to down hound go:PAST:AUX:3sG:s from leopard-OBL tail3so:A take:PAST-AUX:3so 'That leopard turned down to the hound; he (the hound) went and took the tail of that leopard.' 62. av tal-a tüla-n ba cayi düm-a ängal be. he:ABs run-PERF:3so:s dog-Foc to he:ross tail-RELnode LV:PAST:AUX:3so:s 'He ran away [and] the hound, too, got entangled in his tail.'

LW/M380

76

NORTHERN TALYSH

LW/M380

77

NORTHERN TALYSH

63. ma hani ba palang-i ne tüla-m düz gat-e ba mizon gända-m-e.

71. ba-da 3amät bollica hie na-bü ba cayi märd-a tämso ba-ka-n.

l:oBL already to Ieopard-oBL NEG dog-lsG:A right take:PAST-AUX:3soto aim throw:PAST-ISG:A-Aux:3so 'I aimed at my hound, but not at the Ieopard.'

suai-give:2so:JMP people exceptional nothing NEo-be:COND:3so:s to he:ross die:PAST-PART:PERF Iook SUBJ-make:PRES-3PL 'Let the people - nothing should be [more] exceptional - see his dead [body].'

64. zan-a-m-e ki ba tüla gänd-a gülla ba palang-i ba-gan-e.

know:PAST-PERF-lsG:A-AUX:3so CONJ to hound throw:rAST-PART:PERF:REL bullet to Ieopard FUT-fallINF:3so:s 'I knew that the bullet shot at the hound would hit the Ieopard.' 65. cama gülla gänd-e palang-i i narra-s do.

I:ross bullet throw:PAST-AUX:3so Ieopard-oaL one roar-3so:A give:PAST:PERF 'When I shot, the leopard roared.' 66. astan-as tüla-kü etarn-i se.

REFL:OBL-3SG:A hound-LOC free:rAST-AOR go:PAST:3SG:S 'He freed himselffrom the hound [and] ran away.'

72. s-i-m asta dada-m pegat-e asta äsp-emon-an pegat-e hamin palang-amon no asp-i-sa epast-amon-e värd-amon-e ba ka. go:PAST-AOR-ISG:S REFL:POSS father-lso:POSS take-lNF REFL:POSS horse-lPL:A-FOC take:PAST-AUX:3SG PROX:EMPH Ieopard-lPL:A for horse-OBL-SUPER wrap=up:PAST-IPL:A-AUX:3S0 carry:PAST-IPL:A-AUX:3S0 in hause 'I went to call my father [and] we took our horse; we wrapped this Ieopard around the [back of the] horse [and] carried him harne.' 73. 3amät om-a bollica-son tämso kärd-e.

people come:PAST-PERF:3SG:S exceptional-3PL:A Iook make:PAST-AUX:3SG 'The peole came [and] Iooked at [this] exceptional [thing].'

67. s-i-m ca palang-i-kü ba-vind-e-m zago ki ba darozi mü ogänd-a ba zamin.

74. cayi püst-amon pegat-e hasg-amon kärd-e z-a ba divo.

go:PAST-AOR-lso:s from Ieopard-oBL-LOC rur-see-INF-lso:A such coNJ to length hair spread:rASTPERF:3so to earth 'I went to the Ieopard to see that he spread his hair on the earth.'

he:POSS skin-lPL:A take:PAST-AUX:3sodry-lPL:A make:PAST-AUX:3sohit:PAST-PERFtO wall 'We took his skin, dried [it] [and] hang [it] on the wall.'

68. äz ogärd-a-m ba mard-a palang-i-sa.

l:ABS turn:PAST-PERF-ISG:S PROX die:rAST-PART:PERF:REL leopard-OBL-SUPER 'I turned to the dead Ieopard.'

he:ross skin very beautiful LV:PAST:Aux:3so:s because l:ABS always to-he:OBL look AUGMmake:rREs-1 so:A 'Because his skin is very beautiful, I always Iooked at it.'

69. s-i-m tika-m däs dasü-a ba-yi.

76. ära-kü se eo rüz davard-e-kü bapesta am hadisa sa-da ras-a-da ba rayon-i markaz.

go:PAST-PAST-lso:s a=little-lsG:A hand rub:PAST-PERFto-he:oBL 'I went [and] rubbed my hands a little bit because ofthat.'

this-LOC three four day pasS-INF-LOC after PROX Story come:PAST:PERF-PRES reach:PAST-PERF-PRES to rayon-oBL center 'After three, four days had passed since that [event], this story had reached the regional center.'

75. cayi püst ve gasang be gora äz hamisa ba-yi tämso a-ka-m

70. ma pi-a cayi püst-i pegat-o-m asafna fikr-am kä ki an palang-i bitov ba di bärd-e basta-m sahrat zan-a.

l:OBL want:PAST-PERF he:ross skin-OBL take=off-OPT-ISG:A now thought-lSG:A make:PAST:PERF CONJ PROX Ieopard-oBL whole to village carry-INF to:REFL-ISG:Afame know:PAST-PERF 'I wanted to skin him; [but] now I thought that carrying him unskinned to the village I would experience fame.'

77. cama a palang-i ya/i i-a gora cama sakl-ason kas-a man-ason makofot lap do[ .. ]mis kard-

e. I:ross msr Ieopard-oaL alone hit:rAST-PERF because I:ross photo-3PL:A draw:rAsT-PERF l:ACC-3rL:A award much [?] make:rAsT-AUX:3so 'Because I killed the Ieopard alone, they took a picture fromme [and] gave me an award [...].'

78

LW/M380

NORTHERN TALYSH

Appendix II: Palangi ahvolot: An etymological index The following index contains all lexical tokens as documented in the text palangi ahvolot. Verbs are given in the infinite, nouns in the nominative/absolutive case. The reader should note that the etymological commentary does not aim at a comprehensive treatment of the history ofthe Talysh Iexemes. For those lexemes that clearly belong to the Iranian stratum, paralleis from Persian, Old Persian, Avesta (sometimes Sanskrit) and/or Pehlevi are given. Loans that ultimately stem from Arabic are indicated as 'Arabic' irrespective ofwhether the process of borrowing has taken place via Persian and/or Azeri. 20 lexical types (8.47%) remain obscure - I refrained from taking far-fetched parallels into consideration. The following table Iists the different sources together with their frequencies (as they show up in the text pa/angi

ahvolot): 150 33 26 20 2 2 2 234

64.10% 14.10% 11.11% 8.47% 0.85% 0.85% 0.85% 0.42% 100%

Iranian Azeri Arabic Unclear European Onomatopoetic Russian Babytalk Total

English

Source

Synchronie correspondences

Old Persian / Median/ Pehlevi

Avesta

A a abi

that there

Iranian Iranian

äl albahal

crimson in this moment

Azeri Azeri

N. Talysh

English

Source

Synchronie correspondences

albayaxa ammii an3ax ang ängal ära aresie äsp atrof av avada ayo äz

scujjle but there, but jaw node this break horse surroundings he hold there I

Azeri Arabic Azeri Azeri

älbäyaxa 'amma an3ag äng

Iranian Iranian Arabic lranian

asb, asp pu-af ü

Iranian lranian

*ay-o

be ba-pest-a 'to back' badan (P)

NORTHERN TALYSH

Old Persian / Median/ Pehlevi

Avesta

Med. aspahauv

häu

äwm

B

The high frequency of lexemes that can be related to Iranian clearly shows that Northem Talysh has preserved its Iranian characteristics in the lexicon, too. Yet, the considerable number of Azeri loans (most of them rather frequent in the text) demonstrates that the convergences of Northern Talysh grammar towards Azeri is supported by the lexicon of the speaker. N. Talysh

79

LW/M380

OP a- (rRox) OP abi 'towards' al älbähäl

aibi

ba babesta, bapesta badan bar-bar hark baslamis

to, in, with after

Iranian Iranian

body noise much turn (inc.)

Iranian Onomat. Azeri Azeri

beg bi3

moustache smart

Azeri Azeri

bitov bollica bü büd;m bärde ba balang ban bar

who/e exceptional smell side carry this arm ground bush

Azeri Iranian Azeri lranian Azeri Iranian Iranian Azeri?

bärk 'strong' baslamag 'begin' b1g bi3 'smart, tricky' bollu3a boj bud 'hip' barbordan bu bazü (P.)? bun (Phi.) bün bärä 'ambush'?

baoiöibar-

buna-

80

LW/M380

N. Talysh

English

Source

Synchronie correspondences

NORTHERN TALYSH

Old Persian / Median/ Pehlevi

Avesta

c cas

eye from=there jump down Jour much how from

Iranian Iranian Iranian

cesm ca-ay-o (ABL)

Iranian Azeri lranian Iranian

cähär cok cetor ?

da dabäste

with fasten

Iranian Iranian

bastan

dada dagände damände

Jather throw into start

Babytalk Iranian Iranian

< da-gände da-+mändan

cayo ce ces

eo cokcok coko

c3

OP casma

ca8wärö

haca

hacä/haca

hadä OP ba"d- (part. pasta-) + da-

hadä bafid-

man- 'stay, wait' dasta-

man- 'stay, wait' zasta-

D

däs dast dasüe davaxe davärde di

hand jield rub bark come village

Iranian Iranian Iranian Iranian Iranian Iranian