CASES, forms and uses, in Iranian languages and dialects.

General observations. The term case is used on at least three linguistics levels. It refers to:

1. the semantic role of a noun (phrase) relative to another noun (phrase) and/or to the verb, such as agent, patient, experiencer, and possessor;

2. the syntactic function of a noun (phrase) relative to another and/or to the verb, such as subject, direct object, and indirect object, and dependent noun (phrase);

3. the morphological means, “case”-endings, and/or pre/postpositions, often together with word order, by which semantic role and/or syntactic function of a noun (phrase) are distinguished, such as nominative, accusative, dative, and genitive.

The distinctions between these three levels may be illustrated by the following examples. Semantically, in both “He hit her” and “She was hit by him,” HE is the agent (the person who hits), and SHE the patient (the person being hit). Syntactically, however, “he” is the subject in the active sentence but “she” in the passive sentence, morphologically marked by the opposition of “he” and “she” vs. “her” and “(by) him.”

The interrelation of the three levels is reflected in traditional Western grammatical terminology. Nominative, accusative, dative, genitive, and, further, vocative, instrumental, ablative, locative, etc., are not exclusively morphological terms, but also imply the syntactic and semantic functions and roles which the endings of these cases most typically, but by no means only, mark.

For example, an “accusative,” with its case ending, may most typically mark the semantic patient and the syntactic direct object, but it may also mark, semantically, temporal and locational “goals” (“accusative of time/place”) and, syntactically, temporal and directional “objects.”

Such multi-functionality of morphological marking may be illustrated by an example from Persian. In Persian, the (specific) direct object is marked by the postposition -rā, called, or identified as, the “accusative” in traditional European grammars of Persian. The same postposition, however, may also, though not obligatorily, mark temporal and directional “object,” for instance, direct object: man ū-rā dīdam “I saw him/her,” emšab-rā īnjā bāš “Stay here tonight,” hama šahr-rā gaštam “I walked around the whole city.”

The terms of Near Eastern grammatical tradition similarly imply reference to all three levels. For example, the term fāʿel “doer” refers to the semantic “agent” as well as the syntactic “subject” of a sentence, and the morphological marker -u (in Arabic); and the term mafʿūl “(what is) done” refers to both the semantic “patient” and the syntactic “object” of a sentence, as well as to the morphological marking of it by -a in the case of a mafʿūl-e bī-wāseṭa “direct object,” as opposed to mafʿūl-e bī-wāseṭa “indirect object.”

In the following survey of cases in Iranian languages the focus will be largely on the morphological distinction of cases, which is the most tangible, rather than on semantic roles and syntactic functions. Typically, in the history of languages, morphological differentiation is cyclical: a stage with no or minimal differentiation is followed by a stage with increased differentiation, which in turn is followed by a stage of decay of the case system into a minimal binary opposition, direct case vs. oblique case, or total loss of formal differentiation other than by pre- or postpositions and/or word order. These rudimentary stages are then followed by newly differentiated case marking, usually by the development of pre- or postpositions into ending-like, or prefix-like case markers. Such cyclical developments are well represented and documented in the history of the Iranian languages, as well, and each of the many Iranian languages of today represents various stages in that cycle.

Origins. Iranian inherited its nominal inflection from Indo-European via a common Indo-Iranian stage. The typical Indo-European noun consisted simply of a root or a root followed by a “stem” formant. While the root gave the basic meaning, the stem formant represented “grammaticalized” meanings; in some instances it had been reduced to the simply morphological functions of identifying the morphological “class.” For example, in Old Iranian, the stem formant -tar inherited from Indo-European (cf. Lat. -tor- in fac-tor-, etc.) marks agent nouns, e.g., from the base dā- “to give” (OInd. dā-) or “to create” (OInd. dhā-) was formed dā-tar- “giver; creator”; the formant -a, however, is no more than a “thematic” vowel added to the stem, with the function of marking masculine (and neuter) vs. feminine , etc. (cf. Latin masc., neut. o-stems and fem. ā-stems).

The inherited case endings were fusional, that is, one and the same ending was used to distinguish not only cases but also number: singular, plural, and dual. They were attached to the root or stem, with considerable variation depending on the phonological and morphological characteristics of the root or stem, and when attached to stems ending in vowels there was further fusion of case and number with gender, as in the a- and ā-stems.

In the following discussion of the development of the case systems in Iranian the focus will be on the thematic a- declension, as this declension spread to nouns of originally different declension classes and was transformed at a slower rate than the other declensions.

Origins of Indo-European nominal inflection. The Indo-European mother language in its early stages probably had no or just a few case markings, and gender, number, and case endings developed only gradually. It is likely that the development of the case endings from pronouns, adverbs, or postpositions went through three principal stages (cf. Shields; Baldi, 1987, p. 53): First these words in some or all contexts lost their independent meaning and were no longer used freely in the sentences but only in conjunction with nouns (“grammaticalization”; cf. the use of his to denote possession in early modern English). Next they became even more closely attached to the nouns preceding or following them and, losing their own stress, became enclitics (or proclitics; “clitization”; cf. Eng. his > ’s, identical with the regular plural ending). Finally they could become phonetically fused with other markers, for instance, some endings would mark gender and number, as well as cases. This development, which proceeded differently in different Indo-European languages, resulted in the many attested declensional classes with their varying paradigms and irregularities of nominal endings well known from languages such as classical Latin, Greek, Sanskrit (Old Indian), and so on. An example: the original nominative plural ending can be reconstructed as *-es and is seen clearly in consonant stems such as IE. *pod + es > OInd. pādas, OIran. *pādah “feet”; in o-stems, however, it merged with the stem vowel o to produce the ending *-ōs, as in IE. *deḭṷo + es > *deḭṷōs “heavenly ones” > OInd. devās “gods,” OIran. *daḭṷāh “demons.”

This process was in some instances accompanied by changes in the vowels of the noun itself, either in vowel quantity or quality, often depending upon the position of the IE. accent: long/short/no vowel (e.g., IE. *pēd-/ped-/pd- > OIran. *pād-/pad-/bd-), or variation in vowel quality between e and o (e.g., IE. *ped-/pod- and *pēd-/pōd- > Indo-Iran. *pad-/pād-, with partly new distribution of short and long vowels). The declension of a few “irregular” nouns, as well as that of pronouns (see below), could involve two different, suppletive, stems, as, for instance, in Av. karšuuarə-karšuuąn “country-countries,” with variation between r-stem and n-stem (also called heteroclitic declension), or from rāzar- “commandment”: nom. sing. rāzarə̄, gen. sing. rāzə̄ng ( < -anh), gen. plur. rāšnąm.

Old Iranian.

The case system of the original common language of the ancestors of the Indian and Iranian tribes can be reconstructed from the Old Indian and Old Iranian languages. The system so reconstructed is among the most elaborate of those attested in Indo-European languages, containing three numbers (singular, dual, and plural), three genders (masculine, neuter, and feminine), and from four to eight cases.

Depending upon the declension, a maximum of eight cases were distinguished in the singular, six in the plural, and four in the dual. The differences reflect a typological tendency to have most distinctions in the singular, fewer in the plural, and fewest in the dual. Table 4 shows the basic Old Iranian case endings (reconstructed from Avestan and Old Persian) and Table 5 the endings of the masculine thematic (a-stem) declension, from which the distinction between direct and oblique case in most modern Iranian dialects derive (see below). Original IE. s (marked S in the table) developed into OIran. -s, -h or depending on the preceding phoneme.

Case distinction in pronouns is characterized by suppletive paradigms. In particular, the forms in the three numbers of the first and second persons show many peculiarities, for instance, Av. azəm - mąm, vaēm - ahma-, cf. Eng. “I - me,” “we - us.” Case is less consistently distinguished in enclitic personal pronouns. Many adverbs reflect old (“frozen”) case forms of pronouns, for instance, Av. āaṱ “then,” originally ablative of the pronominal stem a- “that.”

Examples of the singular and plural cases of the thematic a- and ā- declension, yasna-masc. “worship, sacrifice” with ahura-masc. “lord,” and daenā- fem. “religion,” as well as derivative stem in -tar, dā-tār-masc. “creator,” showing stem variation, are given in Table 6.

Middle Iranian.

In Middle Iranian the highly differentiated Old Iranian case system was simplified to various degrees in

The different languages. The first major change, documented already in Old Persian, was the extension of the genitive to assume also dative function, which resulted in the loss of the dative marker in all three numbers (except occasionally in personal pronouns, e.g., Tumshuqese tivya “you,” Emmerick, in Schmitt, ed., p. 220).

West Iranian. The change from Old Persian to Middle Persian, seen as early as in the Old Persian inscriptions from the 4th century b.c., was radical. The dual and the distinction of gender were completely lost. In the singular and plural the endings were initially reduced to a binary opposition between direct and oblique case: the nominative, vocative, and accusative coalesced into one endingless case both in the singular and the plural, whereas the remaining cases coalesced with the genitive-dative (sing. *-ahya > ), which was then also lost in the attested West Iranian languages but must have survived in many areas, as it resurfaces in modern dialects. As to the nominative plural, there is some evidence that at least in the very early stages it was marked by -e/-i, since such endings are used in many modern dialects (see below). The genitive plural ending *-ānām became the oblique plural ending-ān (-īn and -ūn in a few nouns from the Old Iranian i- and u-stem endings). This is the stage of the language of the Middle Persian and Parthian inscriptions (Skjærvø, esp. pp. 176-77). Only in the terms for family members (“father,” etc.) do the Middle Persian inscriptions (and also Manichean Middle Persian) still preserve case distinction in the singular (Skjærvø, loc. cit.; Sims-Williams). Subsequently, -ān became the general plural ending, which resulted in the total loss of inflectional case distinctions. Later still, in Middle Persian -īhā, originally an abstract marker, developed as a plural marker in inanimate words (the earliest example is anjomanīhā in the Pahlavi Psalter, Skjærvø, loc. cit.). Table 7 shows the declension of yazd “god” and pid “father” according to the (fragmentary) evidence of the Middle Persian inscriptions and the Pahlavi Psalter (cf. Skjærvø, p. 176).

East Iranian. In the better documented eastern Middle Iranian languages, Choresmian, Sogdian, and Khotanese, the coalescence and loss of case endings and cases was slower. Sogdian and Khotanese preserved the six Old Iranian cases in the singular, in Choresmian nominative = accusative, and the vocative was lost. Both older Khotanese and Sogdian retained a special form of the nominative-accusative singular neuter in -u < -am. The dual was lost almost everywhere: only traces of it remained in Choresmian (q.v.; and see H. Humbach in Schmitt, ed., p. 197) and in the use of special forms after numerals in Sogdian (“numerative,” Sims-Williams, 1979, and idem, in Schmitt, ed., p.183). Khotanese originally preserved four cases in the plural: nominative-accusative, genitive-dative, vocative(!)-instrumental, and locative, but in the latest texts only two or three cases are distinguished. In Sogdian and Choresmian the case system in the plural was reduced to a binary opposition: nominative-accusative versus genitive-dative. In Sogdian two new plural markers developed: -t(a) from the Old Iranian collective or abstract marker -tāt- and -īšt (from an obsolete plural *-īš of i-stems with the addition of -t), to which the oblique singular endings were added (Sims-Williams, 1979, 1982). Table 8 shows the endings of a- and ā-stem nouns in Khotanese and Sogdian (light stems).

Modern Iranian.

Most modern Iranian dialects have no known direct predecessors. The exceptions are Persian which derives from Middle Persian, and Yaghnobi which is descended from Sogdian, although considerable differences are evident in both instances.

Direct and oblique cases. As shown above, inflectional case distinctions had already been lost in the Western Middle Iranian languages Parthian and Middle Persian, and this is also the case in Modern Persian, the best known and most influential of the modern Iranian languages, as well as in the modern Perside dialects, such as Lori, and in the modern southwestern and southern dialects of Fārs, Lāristān, and Bašākard (see baškardī), and in most of the Central dialects.

However, as in other areas of the morphology of modern Iranian, Persian is the least typical, and most of the modern Iranian languages have preserved at least a two-case system. Note also that, in general, pronouns tend to preserve case distinctions longer than nouns (cf. the Romance languages, which have preserved the dative in the pronouns, and Germanic languages such as English and Scandinavian, which have an oblique case only in pronouns) and may even develop additional case distinctions (see below).

Two-case systems are found in the majority of the (north)western Iranian dialects: in the Dimili (Zaza) dialects in eastern Turkey, in the Gōrāni dialects in the middle Zagros near the Iraqi border and around Mosul in Iraq, and in the Kurdish dialects (major exceptions are Solaymānīya, Warmāwa, and several dislocated subdialects), Ṭāleši on the west coast of the Caspian, Tati in individual villages west of Tehran surrounded by a generally Turkophone population, in the Caspian dialects of Gīlakī and Māzanderāni, northern dialects of central Iran such as Semnāni and the surrounding dialects east of Tehran, Ḵūri in the central desert, and in several dialects of Baluchi, stretching from Central Asia to Pakistan, as well as in most eastern Iranian languages: Ossetic in the central Caucasus (which has developed a complex case system, see below), Pashto, Wakhi in northeast Afghanistan, Yaghnobi in Central Asia, Yidgha and Munji, Yazghulami, Ishkashmi, Roshani, and Sarikoli (only in the plural).

Direct case. The direct case of the singular is generally unmarked (no case ending). In the plural it is marked variously. It is zero (i.e., no ending) in such dialects as northern Kurdish, Sangesari, Baluchi, resulting from the loss of the earlier nominative-accusative ending. In a number of dialects it is a front vowel (-i/-e, perhaps based on the descendants of the OIr. i-stem endings nom. -ayah, acc. *-īš), such as Dimili, Semnāni, and Munji -i; Gōrāni and Tati -e, and Ṭāleši -en. Both in dialects with and without case distinction it is quite common for the direct plural to be marked either by -ān, which results from the extension of the plural oblique ending into the direct case (e.g., Central Kurdish -ān, Ṭāleši -on, Māzanderāni and Gīlakī -un, -an), or by endings derived from original abstract markers, for instance, Yazghulami -āθ, Yaghnobi -t, Ossetic -ät, Wakhi -īšt (probably related to Sogdian -t[a] and -īšt, see above), or words meaning “group,” for instance: Persian -hā (of disputed origin but probably from a form such as *-iyaθwa), the western Iranian common plural ending -gal (southern Kurdish dialects, the Lor dialects, and other dialects of central Persia; cf. Pers. galla “flock, herd”), -bar in the Koroši dialect of Baluchi in Fārs, Sarikoli -xeyl. The ending -gal may be combined with the endings -ān or -hā.

Oblique case. In most dialects the singular and plural oblique cases derive from the older genitive-dative. The origins of the oblique singular endings containing n in many dialects (e.g., Yidgha -en, Munji -ān: sing. obl. masc./fem.; Wakhi -ən, “from”) cannot be definitely ascertained; they may be borrowed from pronouns or be descended from an Old Iranian form, possibly related to the Khotanese instr.-abl. sing. ending -äna. Some eastern dialects have oblique plural endings deriving from other old case forms (see also pronouns, below), for instance, the ablative-dative ending *-aibyah (or a similar form) is the origin of Munji -āf; Wakhi -ev, Sarikoli -ef; Pashto and Ishkashmi -o (cf. Khot. -yau). In a number of dialects the oblique marker of the singular was added to the plural marker to mark the oblique case (cf. Sogdian, above), for instance, Sangesari (-un-i), Māzanderāni (-un-e), Gīlakī (-an-e), and Baluchi (-ān-i), Ishkashmi (-o-yi), Wakhi (-ev-i or -ev), Yazghulami (-āθ-i), Yaghnobi (-t-i), and Ossetic (-ät-i).

Kinship markers. The Middle Persian and Parthian special case distinction in nouns denoting kinship has been maintained in a large number of western dialects, including Dimili, Ṭāleši, Tati, Semnāni, and Sangesari and neighboring dialects. Here the oblique case of kinship nouns is marked by -r. Originally limited to r-stem nouns this marker has spread within this semantic group and even outside of it, as in southern Tati, where it has become the general marker of human objects (see, e.g., čāl and čāli; čarza), and in Dimili, where it is an optional marker of animate femininess.

Morphophonemic variation. There is some variation in the form of the case markers depending on the stem finals and gender, but only in Pashto do we find a proliferation of declensional classes, both of nouns and adjectives, some of which exhibit umlaut (see, e.g., Skjærvø, in Schmitt, ed., pp. 389-91).

Table 9 illustrates some common inflectional patterns: Semnāni (asp “horse,” piyá “father”; dāra “tree,” me “mother,” after Christensen, pp. 39-40; Majidi, p. 95); Avromani dialect of Gōrāni (har donkey/māhar-a “she-ass,” piā “father,” aḏā “mother,” after MacKenzie, pp. 14-15), and Dimili (mār “snake,” “father,” “mother,” after Todd, pp. 49-57); Pashto (ās “horse,” aspa “mare,” soṛ “cold” see, e.g., Skjærvø, in Schmitt, ed.); and Munji (pūr “son,” wula “wife,” see, e.g., Skjærvø, ibid.).

Differentiation of cases. The loss of the syntactically crucial formal distinction between nominative and accusative resulted in a need to make the direct object unambiguous by other means already in the western Middle Iranian languages (see, e.g., Skjærvø, rev. of Bossong; and Sundermann, in Schmitt, ed., pp. 134, l60). In the modern languages and dialects this distinction is commonly achieved by three major strategies: by using the marked oblique case (the genitive-dative), the dative marker -rā, or directional pre- or postpositions; combinations of these three are also used. This extension of the use of case markers has in turn spread to the indirect object (dative) and, to a lesser degree, to the possessor (genitive).

The typological extreme is in the Iron dialect of Ossetic, where, on the basis of the Middle Iranian oblique combined with postpositions, a paradigm of nine cases in both singular and plural has developed: nominative, genitive, dative, allative (“to”), ablative (“from”), inessive (“being in”), adessive (“being at”), equative (“as . . . as”), comitative (“with”; see, e.g., F. Thordarson in Schmitt, ed., pp. 469-71; and caucasus ii).

Direct object (accusative). All Iranian dialects except Pashto (which has preserved the passive construction of past transitive verbs in its pure form, see below) developed differential markers to mark the direct object (or some direct objects, see below).

The oblique case is used in such dialects as Ṭāleši, Tati, Semnāni, Dimili, Gōrāni, Northern Kurdish, Ishkashmi (where it has lost its other functions), Wakhi, Yaghnobi, and Ossetic.

The remaining eastern dialects employ pre- or postpositions which derive from the merger of Old Iranian directional pre- or postpositions. The Shughni group, including Sarikoli, has as (< OIr. hača “from”); Yazghulami has na(): (na- also ablative, < OIr. anā “along, upon,” and -ž- from *hača); Munji-Yidgha and Sanglechi have va- ( < OIr. *upā “toward”); Ormuri has ku-,probably related to the indirect object/dative marker kun-, whereas the closely related Parachi has ma-, possibly connected with Pers. mar = ; Morgenstierne, p. 53). In the west only Sangesari has -də (from dar, andar “in, into”), which is added to the oblique form. As a rule, these affixes retain their function as ablative or locative-directional markers, respectively.

The postposition -rā is used as an accusative marker in a large number of western dialects. It developed from OPers. rādiy “because of, on account of” (see Skjærvø, 1985, pp. 215-16), via rāy, dative marker in western Middle Iranian (also in modern Dimili -re), to dative and accusative or only accusative marker in the modern dialects (e.g., both dative and accusative in classical Persian and Baluchi; accusative only in contemporary Persian and most of its dialects).

The marking of the direct object in many dialects depends on the nature of the object: whether it is human or non-human, animate or inanimate, definite or indefinite, specific or nonspecific, etc., with considerable variation depending on the dialect group and subgroup (see further, below; cf. the similar rules in Spanish for marking direct objects by the preposition a).

Indirect object (dative). The marking of the indirect object shows even greater variety than that of the direct object. Dimili has preserved -rē ( < -rā) in this function. In dialects where the accusative became identical with the dative the latter was often additionally marked by pre- or postpositions. Exceptions to this general tendency are Tati (where the oblique has preserved its original dative function with verbs of saying and giving), Semnāni, Sangesari, and several northern Central dialects. In dialects that use -rā (or-(n)-a<-rā as in Lori) for the direct object, the dative is generally expressed by the preposition ba-/be-, even in some dialects with inherited oblique forms, for instance, Gōrāni ba-. In Baluchi the function of the inherited oblique in -i is reduced to that of genitive, and -ra has retained the old accusative and dative functions. A more developed form of -rā, , has the function of a general oblique, to which in many Baluchi dialects the accusative-dative -rā may be added. The eastern dialects show much variety, using developed forms of -rā (Yazghulami and Sarikoli), ba- (Yaghnobi, Sanglechi, and Ishkashmi), -ta(r) (Pashto, in the pronominal adverbs, and the Shughni group), or na- (Munji), -än (Ossetic). Both Parachi and Ormuri use a preposition ku(n)-, which in Ormuri may be accompanied by to distinguish it from the direct object.

Other uses of the indirect object (dative). In a number of dialects the experiencer (our subject) with verbs of sensation such as “to feel hungry” and with modals “must,” “can,” “want,” has the form of the indirect object, a construction found as early as in Old Persian (see, e.g., Bossong, p. 17, and Skjærvø, rev. of Bossong, p. 70). “To have” is commonly expressed by “to be” plus indirect object (cf. Cl. Pers. marā hast “I have”).

Use of the cases in past transitive verbal constructions. Ergative. Iranian inherited the Indo-European marking of the agent/subject and patient/object by the nominative and accusative, respectively, and the verbal ending agreeing with the agent/subject, the so-called nominative-accusative pattern. However, already in Old Iranian there was a tendency to develop new past tense constructions based on the past/perfect participle in -ta plus the copula “to be.” Thus in the present tense case marking follows the inherited nominative-accusative pattern, and the verbal ending agrees with the subject; but in the past/perfect tense the agent/subject is morphologically in the oblique case and the patient/direct object in the direct case, and the verbal ending/ marker agrees with the patient/direct object. The origin of this construction is a participial passive construction, “which was done by me,” and it therefore used to be called the “passive construction” in the tradition of Iranian linguistics. It soon became incorporated into the verbal system, however, with active function (as opposed to the regular derivative passive marked by -y- in both present and past tenses). This pattern is illustrated in Table 10 with two examples from Old Persian (taya “which,” both nominative and accusative singular neuter; PR = present stem of verb; PF = perfect participle):

In modern Iranian this is the common past tense construction of transitive verbs, with the exception of the imperfect in Gōrāni and Ṭāleši in the west and Yaghnobi in the east, in which the imperfect stem is still based on the old present stem, rather than on the past participle.

The original passive construction of past transitive verbs is today only found in its pure form in Pashto (zə dā lidə/əm “he saw me,” lit. “I [by] him seen-am”). The assimilation of this “passive” system to the “active” one began with the marking of the direct object as in the present (already in Manichean Middle Persian and Parthian and Book Pahlavi, see Sundermann, in Schmitt, ed., loc. cit.), resulting in a system where neither the subject nor the direct object was in the direct case; this stage is found in many modern dialects. Another typical change is the change occurring in the endings of the verb, which originally and still in many dialects agree with the direct object (logical subject). In other dialects, however, the endings are completely lost, and in the most developed instances, the verb agrees with the subject (agent), when in the ergative case.

In modern Persian and some other dialects in central Iran the old ergative construction has been completely ousted by the nominative-accusative construction, probably as a result of the loss of distinct nominative case forms in the personal pronouns (e.g., man “I, me” for early Mid. Pers. an-man) and the phonetical merger of the copula with the pronominal suffixes (e.g., -am both “I am” and “me, my”; note the reverse analogy, which allows the 3 pers. pronominal suffix -aš/-eš to be attached to any 3 pers. verbal form in popular speech in some areas, e.g., mīkonad-eš “he does,” raft-eš “he went” in Tehrani (see Table 12).

Ergative subject marking. Similar to object marking the marking of the ergative in many dialects depends upon the degree of animacy, definiteness, and thematic polarization. There is considerable variation among dialects and even subdialects belonging to the same dialect group, but as a rule marked direct objects are human and/or definite and/or thematic, whereas non-concrete, indefinite, non-thematic objects are unmarked (see Bossong, passim).

Agentive or volitional construction. The single, and typologically most noteworthy, exception to this overall Iranian pattern and trend is found in the system of subject marking in the northeastern Pamir language Wakhi, the semantics of which has only recently been recognized (Bashir). Whereas in the other Iranian dialects the oblique marking of the agent/subject in the past tenses, at whatever stage of simplification, is determined by the transitivity of the verb, in Wakhi it is not. Instead, oblique marking of the agent/subject is found with both transitive and intransitive verbs. The marking of the agent/subject by the oblique is instead determined by semantic and discourse-related factors, including volitional action and change of focus in discourse, irrespective of whether the verb is transitive or intransitive. As such the “logic” of this system has radically diverged from the common Iranian one, most probably under the influence of the neighboring non-Indo-European language Burushaski. Note the following examples in Chart 1 (Bashir, pp. 16, 19).

Possessor/genitive. In the vast majority of dialects the genitive is marked by the oblique, though several eastern dialects have innovated by the use of pre- or postpositions, usually with the oblique if distinct, such as Parachi, Munji, Sarikoli -an, while Ishkashmi has -no-y, where -y is a reflex of the oblique, parallel to Baluchi -ā-i, where the original oblique -i is confined to this function. Ormuri has -tar, Pashto has da-.

As for word order, two main alternatives can be distinguished: “possessed(NI)-connector-possessor(N2),” as in Dimili, Tati, Gōrāni and Kurdish, Persian, and the Central and southern dialects (e.g., Pers. pesar-e šāh, Kurdish of Solaymānīya (kuṟ-a pāšā “the king’s son”), and “possessor (N2)-possessed (N1)” in the remainder of Iranian (e.g., Māzandarāni kijā-i pēr “the girl’s father,” Baluchi manī pissay brās “my father’s brother,” Pashto da saṛī kōr “the man’s house”). The predicative possessive may be expressed by the oblique and/or special markers (e.g., Pers. by the dem. pron. and eżāfa: ān-e man ast “it is mine,” mainly literary, for common māl-e man ast).

Dimili is unique in having not only gender and number distinction in the eżāfa, that is, the vowel connecting two nouns in an eżāfa construction, like neighboring Northern Kurdish and Gōrāni, but also distinguishing an oblique case of the eżāfa, which is obligatory when a possessive construction is governed by either another noun or a postposition; examples are given in Chart 2 (Todd, pp. 136-43; EZ = eżāfa).

Other cases. A number of dialects, such as Gōrāni, Dimili, Sangesari, Kurdish, and Ormuri, have a locative (with restrictive use), also marked by a front vowel ending and probably identical with the oblique. Parachi has an ablative in -i. Pashto has in the singular masculine an oblique form in -a ( < OIran. ablative -āt) that is used with a few prepositions, mostly meaning “from.” In addition, a number of dialects have vocative markers, most of which appear to be innovations. In Pashto the vocative plural is identical with the oblique (cf. Khotanese, where in the plural vocative = instrumental).

The remaining case functions, such as direction and time, are expressed either by the direct or oblique alone or accompanied by pre/postpositions. Finally, in several central western dialects word order, that is, the position of a noun relative to another may serve to indicate location or direction.

Pronouns. Most dialects with an oblique case in nouns also have oblique cases in the pronouns. The opposition between direct and oblique case forms is usually better preserved in the singular than in the plural, and especially in the first and second persons singular. Diachronically, the history of these oblique forms is not uniform, evidently due to the survival of forms morphophonemically more marked than others. In many dialects the oblique forms derive from Middle Iranian forms derived from Old Iranian dative-genitive forms (New and Mid. Pers. man < OPers. gen.-dat. manā), but other old cases (ablative, instrumental, locative, and even dative and accusative) are also represented, at least in some eastern dialects (see, e.g., Pakhalina, 1982). In addition, new forms were created by the coalescence of enclitic pronouns with pre- or postpositions. Thus, several dialect groups have distinct forms for the possessive (genitive), resulting in a tripartite system of direct, oblique, possessive, as in Gīlakī and Māzanderāni in the west, and Sanglechi in the east. In Pashto and Yazghulami in the east and Ṭāleši and Tati in the west, the possessive/genitive forms derive from OIr. hača “from” plus oblique pronoun. As these possessive forms are distinctly marked as “non-direct” case forms, they have come to replace less distinctly marked oblique forms in the northwestern dialects (e.g., Semnāni `ønaukär “his servant” and `øhākärd “he did,” with < *hača plus an oblique form of the pronoun; Christensen, pp. 26, 43). This, in turn, has created a need for better marked possessives. The typological extreme is represented on one hand by Ossetic, where the pronouns are inflected like nouns (nine cases in Iron), and, on the other, by the Tati dialects of Hazārrūd in eastern Azerbaijan, in which the singular pronouns distinguish five cases: 1. subject of present and past intransitive verbs and present transitive verbs; 2. subject of past transitive verbs; 3. direct object of present transitive verbs; 4. direct object of past transitive verbs; and 5. possessive and postpositional; example 1st singular: 1. az, 2. men, 3. adem, 4. ačem, 5. čemen.

Pronominal suffixes (enclitic pronouns). The history of the modern pronominal suffixes can be traced back to the Old Iranian enclitic pronouns. These were declined for person, gender, and number but did not usually have distinct forms for all the cases. In Old Persian only the opposition between accusative (-mā “me”) and genitive-dative (-maiy) was still regularly maintained, although ablative forms were also found (hačā-ma “from me” < *-mat; Brandenstein and Mayrhofer, pp. 66-67). In Middle Iranian, we find a single form used for all non-direct cases in western Iranian and Khotanese, while Sogdian appears to have maintained a distinction similar to that in Old Persian (Sims-Williams, in Schmitt, ed., p. 186). In this respect, too, the typological extreme is Ossetic, where the pronominal suffixes are inflected as to case similar to the independent pronouns. The Lārestāni dialects are noteworthy in that the agentive affix is not obligatorily enclitic (i.e., prefixed to the verb or affixed to a preceding noun), but may be separated and thus function as a quasi-independent agentive pronoun as opposed to the independent pronoun; examples are given in Chart 3 (Eqtedārī, p. 190).

In Harzani (Mortażawī), in the perfect the original endings of intransitives have been replaced by the agentive suffixes of transitives, except for the 1st and 2nd singular, which optionally retain the intransitive endings; examples “I, you, he, she, we, you, they have been” and “I, you, he/she, we, you, they said” (Chart 4; note r < t in ber- < būta-, ra/runa = Pers. -at/etān):

In the Pamir languages the distinction between transitive and intransitive sets of affixes has been leveled in all past tenses almost everywhere (Payne).

Pronominal suffixes are confined to the position after “self” (cf. Pers. ḵᵛodaš) in Gīlakī and Māzanderāni and are not found at all in Dimili, Northern Kurdish, Sangesari, and in the Pamir languages.

In the modern dialects the pronominal suffixes generally mark the oblique case, being used for the subject (agent) of past transitive verbs (e.g., be-m-ke (rd) “I did,” in several Central dialects, see, e.g., P. LeCoq in Schmitt, ed., pp. 319, 321; Pashto dáy me wə́lid “I saw him”), direct object, indirect object, and possessor (e.g., Pers. dīdam-aš “I saw him,” az ān ḵᵛoš-am mīāyad “I like it,” ketāb-am “my book”).

Synopsis of the development of case markers.

Table 11 contains a synopsis of the development of the endings of the masculine singular and plural endings of the thematic a-declension, from Indo-European through proto-Iranian (OIran.), Avestan, Old Persian, early (EMWI) and standard Western Middle Iranian (Middle Persian and Parthian: MWI), as well as Eastern Middle Iranian Khotanese and Sogdian, including their late forms.

As seen above, the Iranian languages have varied considerably in the development of their case systems. Nevertheless, there is a common drift to be observed that is shared by both West and East Iranian, in terms of the merger and loss of cases.

While Avestan continues the inherited pattern, Old Persian innovates by the loss of the ablative-dative -aibyah in the plural, resulting in merger of the dative with the genitive, and of the ablative with the instrumental. In the singular the dative likewise merged with the genitive and the ablative with the instrumental after the loss of final -t. Exactly this pattern, rather than that of Avestan, is also found in the developments in Khotanese and Sogdian. In these two languages in addition the nominative merges with the accusative in the plural.

The next stage of the simplification of the case system is the development of an opposition between direct and oblique case only, as found in early Middle Persian. Similarly, in Late Khotanese there are only two distinct case forms in the plural. In Sogdian the binary opposition in the plural of -ta vs. -tya, is an innovation, as is also the optional marking of the oblique singular and plural in late Sogdian.

Overall, then, the major division that emerges is that between vocative, nominative, and accusative on one hand, and the remaining cases on the other.

The final stage is the loss of distinction other than singular vs. plural.

The evidence of the modern Iranian languages suggests that all of them, with the exception of Pashto, must have gone through a stage with the binary distinction between direct and oblique in both singular and plural, similar to that of early Middle Iranian. Table 12 shows the basic patterns of the marking of the major marked cases in nouns (Note that the oblique marker cited is that of the 3rd singular masculine, only):

OBL = oblique;

ERG = ergative (subject in transitive past not in the direct case);

GEN = genitive/possessor (dependent noun and possessive);

DAT = dative/indirect object;

ACC = accusative/direct object;

P = oblique in pronouns only; D = direct case; O = oblique case;

A = accusative case in dialects where the direct object is marked by adpositions and/or the oblique;

PR = verbal forms from the present stem;

PT = forms from the past stem.

Example: DO/OD = in present tense constructions the subject is in the direct case and direct object in an oblique case, but in transitive past tense constructions the subject [ = agent] is in an oblique case and direct object in the direct case. The focus is on dialects that have preserved at least some of the inherited inflectional case markers. Ṭāleši, Tati, including Eštehārdi, Semnāni, Dimili, Gōrāni, Northern Kurdish, Sangesari, Gīlakī, Māzanderāni, Baluchi, Ormuri and Parachi, Munji, Sanglechi, Ishkashmi, Yazghulami, Roshani of the central Pamir group, Sarikoli, Wakhi, Pashto, Yaghnobi, and Ossetic. Lori and classical and modern Persian are chosen to represent western dialects which have lost inflection. The list is necessarily selective and omits many dialects, and thus much of the variety of inherited and adpositional case marking actually found in both western and eastern dialects.

This synoptic table suggests a common drift. We see that in the majority of the languages the marking of the accusative/direct object has been renewed by the extension of the dative, either of the oblique, that is, the old genitive-dative, and/or of the pre- or postposition marking the dative. The resulting formal identity of the two cases (cf. modern Spanish) then triggers the differentiation of the dative from the accusative by the use of pre- or postpositions in a number of dialects.

Overall, the West Iranian languages show considerable uniformity in terms of case marking, while the East Iranian languages show great variety not only in the marking of the accusative and dative but also of the genitive.

As to the use of the direct and oblique cases for the subject and object in the past of transitive verbs, the development varies considerably in the different dialects. The general drift of the development is:

Subject-Object: “nominative-accusative construction”

➝ OBL-DIR: “pure ergative construction”

➝ OBL-OBL and/or affix: “double oblique construction”

➝ DIR-OBL: “nominative-accusative construction”

The exception is Wakhi, as already mentioned, where the oblique has been extended from transitive to intransitive verbs, functioning no longer as “ergative,” but as volitional/topical agent.



Bibliography: (For complete bibliographies see Schmitt, ed.) Ph. Baldi, “Indo-European Languages,” in B. Comrie, ed., The Major Languages of the World, London and Sydney, 1987, pp. 31-67.

E. Bashir, “Beyond Split-Ergativity. Subject Marking in Wakhi,” in Proceedings of the 22nd Regional Meeting of the Chicago Linguistics Society, 1986, pp. 15-35.

G. Bossong, Empirische Universalienforschung. Differentielle Objektmarkierung in den neuiranischen Sprachen, Tübingen, 1985 (rev. by P. O. Skjærvø, in Kratylos 34, 1989, pp. 65-71).

W. Brandenstein and M. Mayrhofer, Handbuch des Altpersischen, Wiesbaden, 1964.

A. Christensen, Le dialecte de Sämnān, D. Kgl. Danske Vidensk. Selsk. Skrifter, 7/II.4, Copenhagen, 1915.

R. E. Emmerick, Saka Grammatical Studies, London, 1968.

Idem, “The Transition from Old to Late Khotanese,” in Societas Iranologica Europaea, Transition Periods in Iranian History. Actes du symposium de Fribourg-en-Brisgau (22-24 mai 1985), Studia Iranica, cahier 5, Louvain, 1987.

I. Gershevitch, A Grammar of Manichean Sogdian, Oxford, 1954.

J. Kurylowicz, The Inflectional Categories of Indo-European, Heidelberg, 1964.

M.R. Majidi, Strukturelle Beschreibung des iranischen Dialekts der Stadt Semnan, Forum Phoneticum 22, Hamburg, 1980.

M. Mortażawī, “Feʿl dar zabān-e Harzanī,” NDAT, 1341-42 Š./1962-63, no. 14, pp. 453-88, no. 15, pp. 61-97; also separate, Tabrīz, 1342 Š./1963, with errata.

T. N. Pakhalina, “On the History of the Forms of the Demonstrative Pronouns in the Pamir Languages,” in Monumentum Georg Morgenstierne II, Acta Iranica 22, Leiden, 1982, pp. 141-52.

J. R. Payne, “The Decay of Ergativity in Pamir Languages,” Lingua 51, 1980, pp. 147-86.

V. S. Rastorgueva, ed., Opyt istoriko-tipologicheskogo issledovaniya iranskikh yazykov I-II, Moscow, 1975.

Idem, ed., Osnovy iranskogo yazykoznaniya I-III, Moscow, 1979-87.

R. Schmitt, ed., Compendium Linguarum Iranicarum, Wiesbaden, 1989.

K. Shields, Indo-European Noun Inflection: A Developmental History, University Park, 1981.

N. Sims-Williams, “On the Plural and Dual in Sogdian,” BSOAS 42, 1979, pp. 337-46.

Idem, “Notes on Manichaean Middle Persian Morphology,” Studia Iranica 10, 1981, pp. 165-76.

Idem, “The Double System of Nominal Inflection in Sogdian,” TPS, 1982. pp. 67-76.

P. O. Skjærvø, “Case in Inscriptional Middle Persian, Inscriptional Parthian and the Pahlavi Psalter,” Studia Iranica 12, 1983, pp. 47-62, 151-81.

Idem, “Remarks on the Old Persian Verbal System,” in Festgabe für Karl Hoffmann, pt. 2, MSS 45, Munich, 1985, pp. 211-27.

Idem, rev. of Bossong (see above). T. Todd, A Grammar of Dimili (Also known as Zaza), Ph.D. dissertation, University of Michigan, 1985.

E. Yarshater, A Grammar of Southern Tati Dialects, Median Dialect Studies I, the Hague and Paris, 1969.

Idem, “The Tati dialects of Tarom,” in M. Boyce and I. Gershevitch, eds., W. B. Henning Memorial Volume, London, 1970,pp. 451-67.

Table 4. Old-Iranian Case Endings

Table 5. Thematic Endings in Old Iranian, Avestan, and Old Persian

Table 6. Examples of Avestan Thematic Declension

Table 7. Middle Persian Declension

Table 8. Thematic Declension in Khotanese and Sogdian

Table 9. Examples of Case Forms

Table 10. Transitive Verbal Constructions in Old Persian

Table 11. Synopsis of the Singular Masculine A-Declension in Old and Middle Iranian

Table 12. Survey of Case Markers

(Gernot L. Windfuhr)

Originally Published: December 15, 1990

Last Updated: December 15, 1990

This article is available in print.
Vol.V, Fasc. 1, pp. 25-37