Падеж в каспийских языках: гилякский, мазандаранский, талышский, тати

Полный текст

Introduction

Gilaki and Mazandarani are two closely related Caspian languages that are part of the Central Iranian subgroup of Iranian languages (Figure). They are spoken, respectively, by approximately four million and three million people.

Origin of Caspian and Western Iranian languages according to A. Korn [1]
Source: сompiled by Vladimir B. Ivanov, Leyli R. Dodykhudoeva.

The third language examined in this study is Talysh. From a genetic and typological perspective, it differs in several respects from the two languages mentioned earlier, though it remains in close contact with them. Speakers of Gilaki and Mazandarani can easily understand one another, yet they do not comprehend Talysh, so Persian is used for inter-communication. Similar to Gilaki and Mazandarani, Talysh exhibits both lowland and highland varieties: the urban dialects belong to the first group, while the highland forms are commonly referred to as gāleši [2.  P. 65]. In Azerbaijan, most varieties belong to the Northern varieties. Yet, in the villages of Sarak and Digadi, located in the Astara district of Azerbaijan, a Charozhi variety is spoken; although its classification is open to question, it presumably belongs to the Central/Southern Talysh group [3.  P. 91]. Due to its unique disposition among distant dialect varieties of Talyshi, its speakers consider it a language. The Charozhis communicate with people from other settlements in Talysh. They are fluent in Talyshi, but the Talyshi find it difficult to understand the Charozhi variety.

In Iran, Talysh vernaculars employ an official orthography based on the Persian script [4]. According to W. Schulze, since 1989 in Lenkoran and Astara there has been increasing interest in restoring Talysh as a written language, previously using Cyrillic and later Roman- or Persian-based alphabets [5.  P. 7]. Today, however, the situation is different: our Talysh-speaking consultants tend to write in Cyrillic script.

As Geoffrey Haig points out, comparing Gilaki with Tati is complicated by terminological differences. For instance, what Rastorgueva et al. (2012) describe as the “Genitive” in Gilaki is termed the “Possessive” by Stilo. The morpheme in Tatic that we consider to share the same origin is labeled by Stilo (2018) as the “singular Oblique” or “Reverse Ezafe” [6.  P. 129].

Closer analysis has revealed specific irregularities in the oblique cases (genitive and accusative-dative) when compared with their counterparts in other well-known case systems. The subsequent discussion therefore addresses these features in Mazandarani, Gilaki, Talysh, and Tati.

Our conclusions can be summarized as follows: a) case distinctions are not essential for describing Gilaki and Mazandarani grammar; b) Talysh and Tati grammar do exhibit genuine cases; c) the ezāfe construction occurs in all four languages; d) sentence elements carrying new information (the rheme) may appear after the verb.

This article draws on fieldwork carried out between 2012 and 2014 in the northern Caspian provinces of Iran (Gilan, Mazandaran, Golestan, Semnan, and Tehran), as well as on interviews with a Talysh speaker in Moscow in 2019. Unless otherwise indicated, the data are either provided directly by the authors (for Persian) or obtained from consultants (for Mazandarani, Gilaki, and Talysh).

Ezāfe construction and the genitive case

By definition, in case construction only a dependent word can undergo case-sensitive labelling [7.  P. 162; 8.  P. 200; 9.  P. 1]. Here we provide a Persian example (1), where the ezāfe marker (ezf) ‑e is attached to the main word (daftar ‘copybook’). Such a structure is considered to be a classic example of an ezāfe construction.

(1) PERSIAN: daftar-e — sabz / copybook-ezf — green / ‘green copybook’

Thus, the Persian ezāfe marker -e/-ye in the attributive phrase (1) is not a case ending, since it is applied to the primary word daftar ‘copybook’, and not to the subsequent dependent word, the modifier sabz ‘green’. This order of the primary and dependent words will be called a direct ezāfe or head-driven phrase (i.e. head-initial construction).

Less frequently, we come across cases where the defining (dependent) word takes the first place in the Persian ezāfe construction [10.  P. 93]. Such an order of the dependent and primary words we shall call a reverse ezāfe:

(2) PERSIAN: tak-e — bute-hā / one-rezf — bush-pl / ‘solitary bushes’

In example (2) the word tak ‘one’ defines the word butehā ‘bushes’. Such a construction is less common in Persian. Comments on reverse ezāfe (head-final) constructions can be found in [11.  P. 19–20].

(3) PERSIAN: sāyer-e — rudxāne-hā / rest-rezf — river-pl / ‘other rivers’

In example (3), the ezāfe construction represents the primary form in Gilaki and follows the same word order as in Persian. The sequence sāyer-e [rest-REZF] could theoretically be interpreted as a genitive form, since it marks dependency. However, such an analysis is not adopted for Persian. As noted in [12.  P. 63], “Considering the Ezafe vowel to be case-related leads to undesirable stipulations”. Put differently, regardless of the arrangement of elements — whether the ezāfe structure is direct or inverted — Western Iranian languages do not associate it with grammatical case.

For Iranian languages more broadly, ezāfe can be defined as a construction in which two words are linked through the marker -ə/-e/-i. This is irrespective of whether the dependent element precedes or follows the head. Phonologically, the ezāfe marker attaches to the first element and forms one phonetic unit with it, but syntactically it governs the relation between both elements equally [10.  P. 20]. Comparable features have been observed in clitics and so-called phrasal affixes in other languages [13.  P. 10].

The case system in Talysh dialects diverges considerably from that of Gilaki and Mazandarani. As indicated in [14.  P. 87], the Asālemi dialect is central both geographically and linguistically among the Iranian varieties. Asālemi and Masali employ a two-case system for both singular and plural. By contrast, in Anbarani and Lankarāni, case distinctions appear only in the singular, with no plural case marking (Table 1).

The information on Asālemi was taken from [14.  P. 87; 15.  P. 66]. The two sources differ only in the oblique plural marker: E. Yarshater records -un, while D. Paul reports —(m)un. The paradigms for Masali and Anbarani are presented in [15.  P. 66]. Information on Lankarāni is based on [3.  P. 128] and has been confirmed through consultation with a native speaker. According to [3.  P. 128], the ending -ni occurs exclusively with nominals functioning as direct objects (see example 4).

(4) TALYSH (Lankarāni): kina-ni — vinde-da-Ø /girl-obl (oblique) — see-prs (present)-3sg / ‘He sees the girl’

Table 1
Case system in Talysh

Case	Asālemi		Masali		Anbarani		Lankarāni
	Singular	Plural	Singular	Plural	Singular	Plural	Singular	Plural
Direct	-Ø	-e	-Ø	-e	-Ø	-un/-ün/-ān	-Ø	-on
Oblique	-i	-(m)un	-i	-ān	-ə		-i, -ni

Source: сompiled by Vladimir B. Ivanov, Leyli R. Dodykhudoeva

According to D. Stilo [16.  P. 370], the nominal case system in Talysh, as well as in several other modern Central Iranian languages, consists of two cases. The oblique case, marked in Talysh with -í, originates from the merger of several Old Iranian oblique forms (genitive, dative, instrumental, ablative, and locative) into a single category [17]. The direct case, marked by zero in Talysh, developed from the fusion of the old nominative, accusative, and vocative. D. Paul further emphasizes that “Number is intertwined with a morphological case-marking system which, in common with many Iranian languages, manifests a distinction between direct and oblique case” [15.  P. 66].

W. Schulze proposes an expanded system for Northern Talysh nouns, adding two singular cases: the ablative-locative, marked by -o, and the feminine vocative, marked by -a [5. P. 17]. For personal pronouns, he suggests a four-case paradigm (similar to the one proposed by M. Purhadi for the Gilaki language), which appears to reflect an older stratum than the nominal system: NOM/ABS (absolutive), ERG (ergative), ACC (accusative), and POSS/OBL (possessive/oblique) [5. P. 35].

A shared phonological trait of both Talysh and Mazandarani is that, when a word ends in a vowel, neither the case suffix nor the ezāfe marker is attached to the stem, since vowel sequences are not permitted in the phonological structure. For instance:

ka-kā [room.obl-in] (Talysh Asālemi) ‘in the room’ [14.  P. 87] instead of *ka-i-kā;
zenā sere [woman.rezf house] (Mazandarani) ‘woman’s house’ [18.  P. 74] instead of *zenā‑e sere.

The unusual word order found in Persian (see the reverse ezāfe in example (2)) occurs regularly in Gilaki, Mazandarani, and Talysh. Below (5) is an illustration from the Lankarāni dialect of Talysh:

(5) TALYSH (Lankarāni): čāy-a — kul / tea-rezf — bush.dir (direct) / ‘tea bush’

Several features set Talysh apart from related languages:

1. The ezāfe marker is realized as ˗e in Persian and Mazandarani, and as ˗ǝ in Gilaki. In Talysh, however, its pronunciation varies across dialects: in Lankarāni it is ˗a, in Masali it alternates between ˗e and ˗ǝ, and in Anbarani it is ˗ǝ (data from [15.  P. 76]).
2. Stress placement differs as well. In Persian, Gilaki, and Mazandarani, stress consistently falls on the stem. In Talysh, by contrast, stress may appear either on the stem or on the ezāfe marker.
3. Another distinction lies in the scope of the ezāfe particle. In Persian, Gilaki, and Mazandarani, it marks all nominals. In Talysh, it regularly attaches to adjectives, whereas nouns are most often marked with ˗i/˗ǝ (the oblique case ending) and occasionally with ˗a/˗e, depending on the dialect and context.

Our survey of the literature on Talysh shows that the ezāfe element ˗a has been described under different labels. L. Pireyko refers to it as an “attributive construction marker” [3.  P. 131]; S. Vinogradova terms it a “modifier marker” [19.  P. 92]; D. Paul uses the designations “linking ˗a” [15.  P. 175] or “Persian ezafe” [15.  P. 76]. E. Yarshater [14.  P. 90] alternatively calls it a “modifier marker” and “suffix ˗a”. M. Naghzguy-Kohan [20.  P. 125] applies the labels “nešāne-ye ˗a” (marking the possessor) and “pasvand-e ˗a” (attribute marker). Since pasvand in Persian literally means “suffix,” this terminology requires clarification. It is important to stress that in Persian, Mazandarani, Gilaki, Tati, and Talysh, the ezāfe element does not form part of the word and therefore should not be analyzed as a suffix. Principles for defining word structure and word boundaries in Iranian languages have been elaborated in [21].

In the Lankarāni Talysh example (6), the ezāfe fuses with the final -a of the adjective toža ‘fresh’:

(6) TALYSH (Lankarāni): toža — meyv-on / fresh.rezf — fruit-pl / ‘fresh fruit’

A comparison of selected examples from Asālemi and Lankarāni is presented below (Table 2).

Table 2
The nominal phrase in Talysh

Asālemi	Lankarāni	English
vǝl-a bǝna [flower-rezf group]	vi-l-a dāsa [flower-rezf group]	flower bush
ju-a kanār-un-kā [stream-rezf bank-pl.obl-on]	ru-a kano [stream-rezf bank]	along the stream’s banks
šā cāk-a zua [shah.rezf good-rezf son]	podšo čok-a zoa [shah.rezf good-rezf son]	the king’s good son
xas-a baxt-a ruj-un [good-rezf fortune-ezf day-pl.obl]	šošotm-a ruž-on [fortunate-rezf day-pl]	fortunate days

 Source: сompiled by Vladimir B. Ivanov, Leyli R. Dodykhudoeva

Table 2 illustrates how elements within nominal phrases are linked by the ezāfe particle ˗a. In the Asālemi phrase xas-a baxt-a rujun ‘fortunate days’ (literally ‘good fortune days’), the presence of two ˗a markers reflects a recursive structure, connecting three words within a single nominal phrase. This resembles the Persian ezāfe chain, though the word order differs. According to W. Schulze, “the element -a is not case sensitive. Together with the oblique marker -i it covers the functional domain of the Persian ezāfe (cf. xaše-ye xošk [branch-ezf dry] ‘dry branch’ as opposed to Talysh xǝšk-a xol [dry-ezf branch]” [5]. Nevertheless, Schulze stops short of explicitly identifying -a as an ezāfe marker.

The examples below from Southern and Central Iranian Talysh are cited from M. Naghzguy-Kohan [20.  P. 125]:

(7) TALYSH: varg-a — ria /wolf-poss.sg — footprint / ‘wolf’s footprint’

(8) TALYSH: pas-a — šət / sheep-poss.sg — milk / ‘sheep’s milk’

(9) TALYSH: zarzang-a — gâz / bee-obl.poss.sg — sting / ‘bee’s sting’

(10) TALYSH: bəland-a — diyâr / high-lnk (link) — wall / ‘high wall’

(11) TALYSH: isbi-a — mazgat / white-lnk — mosque / ‘white mosque’

(12) TALYSH: tel’-a — gaf / bitter-lnk — word ‘bitter word’

M. Naghzguy-Kohan notes that, under the influence of Persian, the use of the linking ˗a in attributive constructions of the type “adjective + noun” (examples (11) and (12)) has expanded to also mark possession in “noun + noun” structures. In both cases, the pattern shows strong similarity to the Persian ezāfe [20]. His view aligns with ours, though he does not go as far as to classify this phenomenon explicitly as ezāfe. We propose that in both instances it should indeed be recognized as an ezāfe marker.

In our assessment, however, this development in Talysh dialects is not the result of Persian influence. This is supported by the fact that the same usage occurs in Lankarāni, a dialect that has been out of direct contact with Persian in Azerbaijan for more than a century:

(13) TALYSH (Lankarāni): vɨl-a — dāsa / flower.dir-rezf — group.dir / ‘bunch of flowers’

This pattern can thus be regarded as an inherent characteristic of Talysh in general. In a different example, however, our consultant employed the oblique ending ˗i in place of the ezāfe marker:

(14) TALYSH (Lankarāni): vɨl-i — kost / flower-obl — bush.dir / ‘flower bush’

We therefore propose that the ezāfe marker -e/-ə should not be regarded as a genitive case indicator, but rather as a general marker of determination in nominal phrases of Western and Central Iranian languages, functioning as a phrasal affix. It serves to link a wide range of nominals, including nouns, adjectives, pronouns, numerals, ezāfe prepositions or postpositions, and even larger nominal groups. The main distinction between Western and Central Iranian ezāfe constructions lies in the ordering of the head and its dependent elements. Our conclusions follow largely from Peysikov’s hypothesis that, although the ezāfe marker attaches phonetically only to the first constituent of the phrase, syntactically it applies to the entire construction.

The precise nature of the ezāfe marker in Western Iranian languages has recently been debated [11]. Comparative analysis of the arguments and hypotheses has led to the persuasive view that the Persian ezāfe is best understood as a phrasal affix: it is neither an inflectional affix within a word nor a separate linking element. This interpretation extends equally to the ezāfe markers of Mazandarani, Talysh, Tati, and Gilaki. The characteristics of a phrasal affix can be demonstrated in the following example:

(15) PERSIAN: behtarin — šahr-e — jahān / good.sprl (superlative) — city-ezf — world / ‘the best city of the world’

(16) TALYSH (Lankarāni): an — čok-a — šahar / most — good-rezf — city.dir / ‘the best city’

In the Persian example (15), the ezāfe marker is attached to the nominal phrase behtarin šahr ‘best city’. By contrast, in the Lankarāni example (16), an čok-a šahar ‘the best city’, it attaches to the analytic superlative form an čok ‘most good’.

Such functional parallels are well known to specialists studying minority Iranian languages such as Mazandarani, Gilaki, and Talysh. These researchers often rely on the theory of ezāfe developed for Persian, Tajik, and Dari, while at the same time maintaining the traditional notion of a genitive case in Central Iranian languages.

It should be emphasized, however, that attributive constructions in Talysh do not always align with ezāfe structures. A few examples from Mazandarani, Talysh, and Persian illustrate this point:

(17) MAZANDARANI: Hosayn-e — ketāb / Hussein-gen — book.nom (case-oriented interpretation) / Hussein-rezf — book (caseless interpretation) [18.  P. 133]

(18) TALYSH (Charozhi, Lankarāni): Hüseyn-i — kitob / Hussein-obl book.dir

(19) PERSIAN: ketāb-e — Hoseyn / book-ezf — Hussein / ‘Hussein’s book’

In example (18), the attributive construction in Talysh (Charozhi, Lankarāni) is formed by attaching the oblique case ending ˗i to the modifier. As noted in [22], certain Talysh dialects employ an ezāfe marker that closely resembles ˗i:

(20) TALYSH (Punali): xǝm-i — sar / jar-rezf — head / ‘top of the jar’ [22.  P. 187]

(21) TALYSH (Tulārudi): ǝšan — duš-i — sa / his — shoulder-ezf — on / ‘on his shoulder’ [22.  P. 570]

(22) TALYSH (Kišexālei): duš-i — sar-i — kā / shoulder-rezf head-ezf — on / ‘on his shoulder’ [22.  P. 723]

It remains unclear whether the ˗i described by Kalbāsi is the same as the one found in the Charozhi, Lankarāni example (18), Hüseyn-i kitob ‘Hussein’s book’. Her data do not include an attributive phrase of the type “noun + adjective.” In both Asālemi and Lankarāni, however, this relation may also be expressed with ˗a. Furthermore, Kalbāsi differentiates between the direct and oblique cases (hālat-e fā’eli and qeyr-e fā’eli, respectively). In the Tulārudi dialect, the marker may even be absent, with a zero ending (-Ø) used instead.

(23) TALYSH (Tulārudi): dǝ gerun tǝ / two —qeran-Ø / you ‘your two qerans’

In the Kešli vernacular, the ezāfe marker appears in a reduced form as a schwa (˗ǝ), resembling its pronunciation in Gilaki.

(24) TALYSH (Kešli): bǝzak-ǝ sa / jar-rezf head / ‘top of the jar’ [22.  P. 673]

Talysh: the object and the oblique case

According to L. Pireyko [3.  P. 168], the Talysh postposition ˗o(ro) may express the point of origin, the cause, or the goal of an action. Despite her claim that “˗o does not function as a direct object marker in any Talysh dialect,” our Lankarāni consultant provided the following example:

(25) TALYSH (Lankarāni): neč-i — hardɨše — zandaγ-o / theme — rheme / wolf-obl.s —eat.pst. 3sg — cow.dir.sg-spec.dir.obj / ‘A wolf ate the cow’

In example (25), the postposition ˗o is attached to the direct object within a pseudo-ergative construction, where the agent appears in the oblique case. Once again, in Mazandarani the direct object carrying “new” information is positioned after the verb; similar patterns are observed in the Talysh examples (27), (33), and (34).

W. Schulze notes: “Speakers of Northern 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)” [5. P. 8]. Yet in his samples, the locative actually precedes the verb. This suggests that placing an element after the verb is not determined by whether or not it is locative but rather by other factors. We argue that the decisive factor is informational status: under the rules of Differential Object Marking (DOM), thematic material tends to precede the verb, while rhematic, or new, information follows it. Locatives often fall into the latter category, since they typically introduce new information.

The way the direct object is expressed is closely tied to the overall syntactic alignment of the language, whether nominal or ergative. While most scholars agree that Talysh historically possessed an ergative construction, in present usage such structures are rare or nearly obsolete. As L. Pireyko observes, “the case system for subject and object is shifting toward the nominative type” [3.  P. 155].

(26) TALYSH (Asālemi): šā — dǝzd-i —xalat — dua / shah.dir.sg — thief-obl.sg — robe.dir.sg — give.pst.3sg / ‘The king gave the thief [a] robe of honor’

(27) TALYSH (Lankarāni): podšo — doše — ba — dɨz-i — xalat / theme — rheme / shah.dir.sg — give.pst.3sg — to — thief-obl.sg — robe.dir.sg / ‘The king gave the thief a robe of honor’

E. Yarshater [14. P. 89] notes that in ergative constructions, one object appears in the direct case while the other takes the oblique case, as illustrated in (26). In our Lankarāni illustration (27), however, the first object is marked instead by the preposition ba ‘to’. It seems that in both sentences (26–27) there is in fact only a single true direct object, xalat ‘robe’, whereas the second object (dǝzd-i/dɨz-i ‘thief’) functions as an indirect object. In Lankarāni (27), this indirect object is marked both with a preposition and an oblique suffix, while in Asālemi (26) it carries only the oblique case marker.

As argued in [23.  P. 143], in a proper ergative structure the transitive subject and direct object should be marked inversely. Yet in examples (26–27) both appear in the direct case, with the verb showing agreement with the subject.

A further departure from an ergative alignment is seen in Asālemi, where “past transitive verbs built on stem II have in each tense but two forms: 3^rd singular and plural, depending on whether the logical object is singular or plural” [14.  P. 105]. In Lankarāni and Charozhi varieties, the erosion of ergativity goes even further (examples 28–31): only the verb form (3SG) is retained, regardless of the person of the subject/object.

(28) TALYSH (Lankarāni): Āz — av-ɨm — vinde / I — he-1sg — see.pst.3sg / ‘I saw him’

(29) TALYSH (Charozhi): Tɨ — mɨni — vindǝ / you — me.1sg — see.pst.3sg / ‘You saw me’

(30) TALYSH (Lankarāni): Ama — tɨn-mon — vinde / we — you.2sg-us — see.pst.3sg / ‘We saw you’

(31) TALYSH (Charozhi): Tɨ — əmə — vindǝ / you — us.1pl — see.pst.3sg / ‘You saw us’

In Talysh example (28), the verb shows agreement with the subject, which appears in the direct case, while the direct object is marked by the oblique case. This pattern reflects a nominative-accusative alignment. Phrase (32) provides an additional illustration of the erosion of the ergative construction. This phenomenon is restricted to the past tense and does not occur in the present tense.

(32) TALYSH (Lankarāni): Āz — gɨl-I — čɨnedam / i — flower-obl — pick.prs-1sg / ‘(I) pick flowers’

In Talysh examples (32–33), similar to (28), the verb aligns with the subject in the direct case rather than with the direct object, which is marked by the oblique case.

(33) TALYSH (Lankarāni): Āz — ɨn — kitob-i — doda-m — ba — Maryam-i / theme — rheme / I — this — book-obl give-prs.1sg — to — Maryam-obl / ‘I give this book to Maryam’

(34) TALYSH (Lankarāni): Āz — ɨn — ahvolat-i — votɨme — ba — Hasan-i / theme — rheme — I — this — story-obl — tell.pst.3sg — to — Hassan-obl / ‘I told this story to Hassan’

In conclusion, although the ergative system in Talysh shows signs of simplification—where both subject and object may appear in the direct case—case distinctions are still preserved. A consistent pattern emerges in which the subject tends to occur in the direct case and the object in the oblique, reflecting a nominative-accusative alignment. In addition, in Lankarāni the direct object may also be marked with the postposition ˗o < ˗rā.

The noun phrase in Tati Language background

According to the latest hypothesis about the classification of Iranian languages [1], the Tati dialect group (hereinafter simply Tati) belongs to the subgroup of the Caspian languages of the Central Iranian branch.

In the present study, in contrast to the previous traditional classification, languages are divided not into Western and Eastern, but into Western, Central and Eastern groups. In the earlier classification, moreover, the proto-Western group was opposed to the proto-Central one in terms of isoglosses, but the proto-Eastern group had no such opposition. Thus, practically, all-Iranian, proto-Central and proto-Eastern linguistic groups were the same [1.  P. 249]. Nevertheless, Agnes Korn retains the Eastern branch in her diagrams [1.  P. 269, 272]. Therefore, in Figure 1, the Eastern branch is represented by a dotted line. Consequently, the division of Western and Eastern Iranian languages into Northern and Southern ones disappears.

Tati is spread among the communities living in northwestern Iran: in Iranian Azerbaijan, the provinces of Zanjan, Ramand, and in the vicinity of the city of Qazvin. It is divided into five smaller subgroups [24.  P. 107]: 1) Southern (Ramandi), 2) Khoini, Kho’ini^[1], 3) Northern (Khalkhali, Kajali, Shahroudi), 4) Harzan(d) i, 5) a subgroup located to the north and northeast of Qazvin in the villages of Kuhpaye, Rudbar, Alamut.

In turn, these subgroups are divided into smaller groupings based on their territorial specificity. So, for example, Pireyko identified that the first (Southern) subgroup includes the following dialects: Chali, Danesfani, Khiaradzhi, Khoznini, Esfarvarini, Takestani, Sagzabadi, Ebrahimbadi, Eshtehardi [24.  P. 107]. According to unofficial estimates, there are several hundred thousand Tati speakers.

In terms of segmental phonetics, Tati is quite close to the Persian and Talysh languages. In morphology, compared to other Iranian languages, it is distinguished by a relatively high degree of inflection; according to the traditional approach, in nominal inflection Tati has the categories of case, number and gender. Generally, the characteristic features of Tati noun phrase syntax are firstly, that the modifier precedes the head of the phrase, i.e. attributives are premodifiers^[2], and secondly, that the ergative construction is present. It is noteworthy that in attributive use, simple adjectives and adverbs are used, as well as adjectival or adverbial clauses, and noun adjuncts (=nouns modifying other nouns).

We have demonstrated that in the Gilaki and Mazandarani languages, the presence of a case category needs to be verified [25]. Our data demonstrated that, contrary to the traditional approach, there are no cases in these two Caspian languages. What was previously described as case is, in one instance, a reverse ezāfe, and in the other a postposition. However, similar scrutiny of Talysh, closely related to Tati, confirmed the presence of a case system in the Talysh language. Here we intend to examine the Tati language and to verify to what extent we can base its description on the category of case.

Syntactic functions within the noun phrase in Tati

In addition to synthetic word forms, we find analytical constructions in the morphology of Tati. Traditionally, the noun in Tati is assigned binary opposition in the following categories [24.  P. 109]: 1) gender (male, female), 2) number (singular, plural), 3) case (direct, oblique), 4) determination (definiteness/indefiniteness).

In plural number there are no formal markers for the feminine gender, i.e., in plural feminine nouns, the direct case is not opposed to the oblique case. In both cases, the zero marker (-Ø) is used. A similar phenomenon in the direct case is also typical for singular masculine nouns.

In the category of determination, indefiniteness is expressed only analytically by the numeral one.^[3] Definiteness can be articulated both analytically with the help of demonstrative pronouns, and synthetically by placing the nominal direct object in the oblique case. The rest of the grammatical meanings indicated above are expressed by inflections, including zero ones (-Ø).

According to Pireyko [24.  P. 109], adjectives in the attributive function are associated with a noun using the attributive marker -a, which, when examined, turns out to be the same ezāfe marker as the Talysh -a in the following expression:

(35) TALYSH: čāy-a — kul / tea-rezf — bush / ‘tea bush’

(36) TATI (Takestani): kamb-a — kār / small-rezf — work / ‘small matter’

(37) TATI (Takestani): paxt-a — bona / wide-rezf — roof (Takestani) / ‘wide roof’

Here the adjective does not carry any inflection: neither a marker that distinguishes adjectives from nouns, nor a case or number marker. The constructions (36, 37) are reverse ezāfe, the same as in Mazandarani serx-e jeme ‘red shirt’ and Gilaki rāst-ə gəb ‘truthful word’. Here is another example of reverse ezāfe in Tati:

(38) Naku-a — zania — kam-tar-e— tā — naku-a — mardak / good-rezf — woman — little-com-be.prs.3sg — than — good-rezf — man / ‘There are fewer good women than good men’

In addition to the ezāfe link between the head noun and the modifier, extending the attributive in noun phrases, a connection is possible using inflection with the oblique case:

(39) Hasan-e — čupēna-r — das / Hassan-obl — shepherd-obl — hand (Ebrahimbadi) / ‘hand of the shepherd Hassan’

(40) deraxt-e — lāqa / tree-obl — branch (Chali) / ‘tree branch’

“In Tatic, just as in Balochi and Gilaki, we find prenominal possessors marked by a suffix variously transcribed with -i, -í, and -e in Stilo [17.  P. 698–699], e.g. hæsæn-i kætæb ‘Hasan’s book’ (Harzani dialect). In one sub-group of Tatic languages (including e.g. Vafsi), the form of this case suffix varies in the singular according to the gender of the noun [17.  P. 694], while in other Tatic languages (including the Talyshi group) gender is lost, and this suffix has an invariable form in the singular. In some dialects of Tatic, there is a suffix that marks pre-nominal possessors, which may occur after the plural suffix, and with pronouns. This is illustrated in (12a—12c)” [6.  P. 129].

In our article, this is example (41). Here we see another instance of pure reverse ezāfe:

(41) Tāleqāni: Orāzāni dialect of Tatic: a. pa-i — bon / foot-rezf — bottom / ‘sole of the foot’
b.  boz-an-i — šir / goat-pl-rezf — milk / ‘the goats’ milk’
c. mǝn-i — xanæ / 1sg-rezf — house / ‘my house’ (according to: [17. P. 700]; glosses modified as in: [6. P. 130])

Stilo [17.  P. 700] considers this element as a “Reverse Ezafe”, which he identifies as distinct from the “Oblique suffix” of Tati. Haig’s view is that it is “the reflex of the old oblique, which has lost gender differentiation”. The element has “debonded”, “yielding exactly the same kinds of morpheme sequences that are attested in Gilaki and Balochi. As part of this general re-structuring of nominal inflection, the plural suffix becomes a general marker of plural number (rather than a composite morpheme expressing Oblique and plural), and then permits combinations with the debonded Oblique marker” [6.  P. 130].

Cardinal numerals are associated with the defined noun by adjacency. In some dialects, numerals are used in the singular [3.  P. 133]:

(42) sö — berā / three — brothers (Eshtehardi) / ‘three brothers’

(43) penj — göw / five — bull (Eshtehardi) / ‘five bulls’

In others, the noun is used in plural:

(44) bā — šaš — šuar-o / with — six— shepherd-PL.OBL (Chali) / ‘with six shepherds’

(45) dö — zaneg-e / two — woman-PL.DIR (Kajali) / ‘two women’

Ordinal numerals are associated with the noun they define through adjacency:

(46) duyyum — xodā / second — god / ‘second god’

Through adjacency, possessive pronouns are associated with the word they define:

(47) ešte — dǝta-r / your — daughter-SG.OBL (Shahrudi) / ‘your daughter’

(48) čame — ka / our — house (Kajali) / ‘our house’

(49) jay — berā / his — brother (Chali) / ‘his brother’

Object agreement connects the direct object and the predicate in an ergative construction [27.  P. 51]. The verb form agrees with the animate object in number, gender and person:

(50) Mardak-i— žaneg-i-a — behind / male-obl — female.dir-ind-f — see.pst.3sg.f (Kajali) / ‘A man saw a woman’

(51) Mardak-i — Mammad — behind(e) / man-obl — Muhammad.dir — see.pst.3sg.m (Kajali) / ‘The man saw Muhammad’

(52) Čemen — te — bebardeš / I.obl — you.dir — lead.pst.2sg (Eshtehardi) /‘I took you away’

In [28.  P. 10–11], in addition to the direct and oblique cases, the absolutive (absolute case) and ergative (ergative case) are introduced into Tati; however, the absolutive and ergative do not differ in form from, respectively, the direct and oblique cases. The peculiarity of Pireyko’s work is that, in glossing, the direct and oblique cases are denoted by the nominative (NOM) and the accusative (ACC). In the present article, the marker of the ergative, as well as the oblique case, is -ǝ, while for Pireyko it is -e [3.  P. 128; 24.  P. 109]:

(53) az — gav-ǝ — vin-ǝm / I.nom — bull-acc — see.prs-1sg (Khalkhali) / ‘I see a bull’

(54) Ahad-ǝ — a-š — bind / Ahad-erg — he.abs-3sg — see.pst.3sg (halkhali) / ‘Ahad saw him’

In the Orazani dialect of the Tati language, a classical reverse ezāfe is found [25.  P. 83]; this does not coincide with the inflection of the oblique case [6.  P. 130; 17.  P. 700; glosses adapted):

(55) pa-i — bon / foot-rezf — bottom / ‘sole of the foot’

(56) boz-an-i — šir / goat-pl-rezf / milk / ‘the goat’s milk’

(57) mǝn-I — xanœ / I-rezf — house / ‘my house’

The valuable information here is that this phenomenon is called reverse ezāfe by the Iranian authors writing in the English [29.  P. 203]. Perhaps it emerged in this dialect under the influence of Gilaki and Mazandarani [25.  P. 83–84]. Unfortunately, the Russian-language article [25] is unknown to Geoffrey Haig, therefore, he still designates instances of reverse ezafe as the genitive case, although in our article we have already shown the incorrectness of this approach.

Geoffrey Haig gives an example of debonding, i.e., the disconnection of case inflection of the masculine -i in the Dikin Maragei dialect of the Tati in the Alamut region:

(58) Sœr=t-i — me-jœn-én / head=2sg.poss-obl.m asp-hit-1sg / ‘I’ll hit you (man) on the head’

In this example, the oblique ending of the noun was separated from it and attached instead to the pronominal enclitic. In the feminine gender, such a disconnection does not yet occur [6.  P. 132]:

(59) Œz — das-œr=et — mi-n-m / I — hand-obl.f=2sg.poss — tam-see.prs-1sg / ‘I see your (woman’s) hand’

This confirms the existence of case in Tati along with the ezāfe construction. Despite the variety of terminology proposed by various authors, to correctly describe the Tati case system, only two terms are needed: the direct case and the indirect, i.e. oblique, case.

In the nominative-accusative structure, characteristic of the present tense of the verb, the functions of the direct and oblique cases are similar to the functions of the nominative and accusative cases, respectively (example 53). And in the ergative-absolutive structure, characteristic of the past tense, the functions of the direct and oblique cases coincide with the absolutive and ergative cases, respectively (examples 50–52, 54).

Results and discussion

A comparison between the traditional three-case descriptions of Gilaki and Mazandarani and the proposal in the present article that these languages lack a case system leads us to several conclusions. While, in theory, the ezāfe vowel and the postposition -re/-a in these languages could be interpreted as analytical case markers, the following points should be noted:

1. At present, there are no consistent or logically coherent definitions of such a case system. For example, the notions of adjectives marked with a genitive case ending or arguments supporting an accusative-dative marker after noun phrases remain unsubstantiated.
2. Treating the grammar in terms of case inflection introduces unnecessary complexity.
3. A case-free analysis provides a simpler, more accurate, and more efficient description that aligns with DOM principles. In this framework, the ezāfe construction replaces what is sometimes labeled the genitive case. The order of head and dependent elements can vary, and both may be regarded as equal in status. The ezāfe marker attaches phonetically to the first element, but syntactically it links the entire construction.
4. This caseless model, together with certain syntactic linking strategies in Gilaki and Mazandarani, corresponds to parallel phenomena in Western Iranian languages. It also allows for a unified description of Central and Western Iranian varieties.

This same grammatical model can be extended to Gilaki and Mazandarani dialects such as Shamerzadi and Velatru.

By contrast, Talysh and Tati do maintain case distinctions. Talysh dialects in particular show significant variation: Asālemi and Masali exhibit two cases in both singular and plural, while Lankarāni and Anbarani restrict case marking to the singular. In Asālemi and Lankarāni, ergative constructions are greatly reduced, effectively taking the form of impersonal structures. Asālemi employs only two verb forms (3rd person singular and plural), whereas Lankarāni uses just one (3rd person singular). Instead of an ergative pattern, Lankarāni has developed a nominative-accusative alignment, where the direct object is marked by the postposition -o < -rā, similarly to Gilaki, Mazandarani, and other Western Iranian languages such as Persian, Dari, and Tajik.

The dialects of the Tati group show further diversity. Their noun phrases display several strategies for linking elements, including the use of the oblique case, which suggests that nominal case marking plays a limited role in Tati overall.

Although all of the Iranian languages under discussion (Central and Western) are typologically SOV, speakers sometimes place rhematic, or new information, elements after the verb.

List of abbreviations

Ø (∅ Unicode character U+2205) zero ending
abs absolutive
acc accusative
asp aspect
com comparative
dat dative
dir direct
dom differential object marking
erg ergative
ezf ezāfe
f feminine
gen genitive
imp imperative
ind indefinite
lnk linker
m masculine
neg negative
nom nominative
obl oblique
pl plural
poss possessive
prs present
pst past tense
rezf reverse ezāfe
spec.dir.obj specified direct object
sg singular
sov subject-object-verb order
tam tense–aspect–mood element.

 

 

^{1 And also Di.}

^{2 However, cf. the Khoini dialect, where the adjective is not postpositive, but is used as a postmodifier URL: https://en.wikipedia.org/wiki/Kho%27ini_dialect (accessed: 12.07.2024).}

^{3 See similar marking of the indefinite singular in Ziyarati Mazandarani with ye/yek/yak/yag ‘a, an’, originally ‘one’ [26.  P. 21].}

Об авторах

Владимир Борисович Иванов

Институт стран Азии и Африки МГУ

Автор, ответственный за переписку.
Email: iranorus@mail.ru
ORCID iD: 0000-0002-2572-9581
SPIN-код: 5217-0695
Scopus Author ID: 56274320300
ResearcherId: D-3529-2013

доктор филологических наук, профессор, заведующий кафедрой иранской филологии

125009, Российская Федерация, г. Москва, ул. Моховая, д. 11

Лейли Рахимовна Додыхудоева

Институт языкознания РАН

Email: leiladod@yahoo.com
ORCID iD: 0000-0002-4567-9454
SPIN-код: 9490-5555
Scopus Author ID: 6505998146

кандидат филологических наук, старший научный сотрудник сектора иранских языков

125009, Российская Федерация, г. Москва, Б. Кисловский пер., д. 1/1

Список литературы

Дополнительные файлы