+ All Categories
Home > Documents > Nominal Attribution in Semitic. Typology and Diachrony

Nominal Attribution in Semitic. Typology and Diachrony

Date post: 05-Feb-2023
Category:
Upload: gu-se
View: 0 times
Download: 0 times
Share this document with a friend
33
1 Journal of Semitic Studies Supplement 25 ed. J.C.E Watson & J.Retsö. Oxford 2009 Nominal Attribution in Semitic. Typology and diachrony Jan Retsö Introduction: General Typology In the following, a tentative sketch of the different devices used in the most well- documented ancient Semitic languages to mark nominal attribution will be presented. The main aim is to sketch a morpho-syntactic typology for Semitic which may shed light on the modern Semitic languages as well (cf. Lehmann 1984:412). Some of the modern languages (Neo-Aramaic, Modern South Arabian, Amharic, Israeli Hebrew) are treated in this volume and a survey of the modern Arabic dialects is included in this paper. The diachronic aspect will not be treated in any detail apart from some tentative remarks at the end. By nominal attribution is meant the adding of an independent semantic-syntactic complement to a noun which then becomes the head of a nominal phrase. Semantically, the complement functions as a modifier of the meaning of the head noun (Lehmann 1984:434; 17397; Goldenberg 1995:12). The three following instances are relevant for this sketch: (1) head noun + noun = the traditional genitive construction (2) head noun + adjective = the adjectival attribute construction (3) head noun + clause = the relative clause That it is meaningful to investigate these three syntagms together in Semitic has been shown by Pennacchietti and more recently by Goldenberg who has emphasized the common semantic properties of the three syntagms (Pennacchietti 1968; Goldenberg 1995, cf. also Lehmann loc.cit.). Pennacchietti’s study is ground-breaking but does not treat all the specific mophosyntactic properties. We shall here concentrate on these properties. Languages can mark these in different ways. Some languages use the same device for all three, others differentiate between all three (as in most European languages), and some differentiate between one and the rest. Consequently we end up with four possibilities: 1=2=3
Transcript

1

Journal of Semitic Studies Supplement 25 ed. J.C.E Watson & J.Retsö. Oxford 2009

Nominal Attribution in Semitic. Typology and diachrony

Jan Retsö

Introduction: General Typology

In the following, a tentative sketch of the different devices used in the most well-documented ancient Semitic languages to mark nominal attribution will be presented. The main aim is to sketch a morpho-syntactic typology for Semitic which may shed light on the modern Semitic languages as well (cf. Lehmann 1984:41–2). Some of the modern languages (Neo-Aramaic, Modern South Arabian, Amharic, Israeli Hebrew) are treated in this volume and a survey of the modern Arabic dialects is included in this paper. The diachronic aspect will not be treated in any detail apart from some tentative remarks at the end.

By nominal attribution is meant the adding of an independent semantic-syntactic complement to a noun which then becomes the head of a nominal phrase. Semantically, the complement functions as a modifier of the meaning of the head noun (Lehmann 1984:43–4; 173–97; Goldenberg 1995:1–2). The three following instances are relevant for this sketch:

(1) head noun + noun = the traditional genitive construction

(2) head noun + adjective = the adjectival attribute construction

(3) head noun + clause = the relative clause

That it is meaningful to investigate these three syntagms together in Semitic has been shown by Pennacchietti and more recently by Goldenberg who has emphasized the common semantic properties of the three syntagms (Pennacchietti 1968; Goldenberg 1995, cf. also Lehmann loc.cit.). Pennacchietti’s study is ground-breaking but does not treat all the specific mophosyntactic properties.

We shall here concentrate on these properties. Languages can mark these in different ways. Some languages use the same device for all three, others differentiate between all three (as in most European languages), and some differentiate between one and the rest. Consequently we end up with four possibilities:

1=2=3

2

1#2#3 1#2=3 1=3#2

The order between the elements is dependent on the general typological classification of the language: SVO/VSO or SOV and does not seem to have any implication for the meaning. The different word-orders are thus left out of this discussion.

The categories established above will first be illustrated by examples from some non-Afroasiatic languages, beginning with examples from Modern Chinese (Li/Thompson 1981:113–23; 579–85):

(1) jiào shòu de fáng zi

professor house = ‘the professor’s house’

(2) hóng de huā

red flower = ‘a/the red flower’

(3) xué hànyŭ de waiguórén

study Chinese foreign people = ‘the foreigners who study Chinese’

In all three cases Chinese uses the particle de as a marker of the attributive relationship between the elements. Chinese is thus an example of category (1) above. It should be remarked, however, that the particle de is not always obligatory (Li/Thompson 1981:115).

The second possibility, where a distinction is made between all three categories, is what is normally found in the European languages and will not be discussed here. Instead we shall turn to another Indo-European language, viz. Modern Persian (Boyle 1966:25, 58–60):

(1) ketab-e pesär

book boy = ‘the boy’s book’

(2) ketab-e bozorg

book big = ‘the big book’

(3) a. ketab-i ke män u-ra xäridäm

book I it-ACC buy.PRET = ‘the book which I bought’

b. pesär-ke omäd kuček bud

boy come.PRET small be.PRET = ‘the boy who came was small’

3

Modern Persian has identical marking of (1) and (2) which are differentiated from (3), thus a 1=2#3 system. In the case of Persian we know something of the diachrony. In Middle Persian it looked as follows (Boyce 1964):

(1) dar ī čašmān

gate eye.PL = ‘the gate of the eyes’

(2) pid ī dōšarmīgar ud frihrōd

father loving and compassionate = ‘a loving and compassionate father’

(3) pad ān anāgīh ī-š az xwēš kunišn ǧast (Gignoux 1993 10:4)

‘because of the harm which happened from his own action’

In Persian we can observe a diachronic development. Middle Persian belonged to the same type as modern Chinese. In modern Persian the marking of the relative clause has taken a new direction and become differentiated from the other two syntagms. There is thus a development from 1=2=3 > 1=2#3.

Without entering into all the intricacies of Turkish syntax it could be pointed out that Turkish has similar marking of (2) and (3) which distinguishes them from (1) (Lehmann 1984:52–5):

(1) ev kapı-s-ı

house door-POSS = ‘the door of the house’

(2) kücük ev

small house = ‘a small house’

(3) adam-ın git-tiğ-i mektep

man-GEN go-NR-POSS school = ‘a school to which a man goes’

Nouns attributed to head nouns are postposed whereas attributive adjectives and the equivalent of relative clauses are preposed. In Turkish an attributive clause is transformed into a nominal construction, basically a construction of type (1). But the whole construction is attached to a head noun in the same manner as an adjectival attribute, viz. by preposition. The example in (3) is thus literally: *ʻa man’s-going-his school’. Turkish thus has a system 1#2=3.

Semitic (excluding Arabic)

4

When we turn to Semitic we find two alternating ways of attaching the modifier to the head with quite different distribution in the languages. We shall use the terms annexation and juxtaposition in this paper. By the latter is meant a relationship between the constituents that is similar to what is conventionally called apposition. There is no formal indicator of the link between the constituents; the modifier is simply placed after the head as a comment. One might imagine a comma between them. The internal structure of the modifier, however, may be more complex and, as will become apparent, may contain an element pointing back to the head noun. Annexation denotes a tighter connection between the constituents with a clear morphological and/or syntactical marking of the link.

The distinction between juxtaposition and annexation is used to differentiate between the three categories in various combinations in the separate languages. According to how juxtaposition and annexation are distributed among the three categories of nominal attribution Semitic can be divided into two larger groups:

Group A: 1 = 3 # 2 Group B: 1 # 2 = 3

As will become evident, the devices used are variegated. Even when a language employs identical morphosyntactic means of marking two of the categories there are often different options. We shall start with a survey of the main languages of group A. Among the ancient languages we find Akkadian, Geez and Old/Middle Aramaic belonging to this group.

Akkadian

(1) a. bīt awīlim ‘a/the man’s house’

b. bītum ša awīlim do.

(2) bītum dannum ‘a/the big house’

(3) a. bīt īpušu imqut ‘the house which he built fell down’

b. bītum ša īpušu imqut do. (Codex Hammurabi 19:69)

Akkadian marks annexation by a form of the head noun characterized by the absence of case-markers in the singular. Akkadian nouns in the singular have two basic forms: one with case-marking and one without. This alternation is traditionally called state/status. We shall here use the terms construct state for the annexation form and and non-construct state for the other one, terms applicable to Semitic as a whole. The

5

non-construct is the ‘normal’, quotation form of the Akkadian noun when it occurs without an annexed complement (noun or clause) or a pronominal suffix. So in examples (1) and (3) we have annexation of the modifier to the head noun which marks this relationship by its morphemic shape (von Soden 1969:189 ff.; 219 ff.; cf. Ravn 1941:36–40). But in both there is an alternative structure where the head nouns keep their case-inflected form, i.e. the non-construct state, which means that the following complement is juxtaposed. In category (1) and (3), but not in (2) Akkadian can then use a particle ša to introduce the modifier (von Soden 1969:191 ff.; 126 ff.; Ravn 1941). It can be argued that this particle should be seen as a status constructus form of the demonstrative šu (Pennacchietti 1968:58 with references). For category 1) and (3) Akkadian thus uses both possibilities: annexation and juxtaposition. With an attributive adjective, category (2), only juxtaposition is found (von Soden 1969:187). The syntactic structure of the ša-construction in (1) and (3) in these languages thus appears as a kind of hybrid: a construction consisting of a pronominal element annexed to a following modifier, a noun or clause, thus a head followed by a modifier. This construction in turn been placed in juxtaposition to a preceding nominal head.

Geez

(1) a. betä gäbr ‘a/the man’s house’

b. bet(u) zä-gäbr do.

(2) a. ḥaddis bet ‘a/the new house’

b. mayä ṭəәʕum ‘fresh water’

(3) a. bəәʔəәsi zä-yəәḥäwwəәr ‘a/the man who walks’

b. bä-mäwaʕəәlä yəәkwennəәnu mäsafəәnt ‘in the days when the judges ruled’ (Ruth 1:1)

In Geez one finds examples of annexation of a clausal attribute by a head noun (3b). This construction is not frequent and seems to be an archaism (Dillmann/Bezold 1907:§ 201). The genitive construction, (1), is either an annexation where the noun has the suffix -ä (< ă) indicating the construct state (not the accusative, pace Rubin 2005:48), or a construction with the particle zä. The particle ʔəәntä may be used if the head is feminine singular and ʔəәllä if it is in the plural (Dillmann/Bezold 1907:§§145, 184–86). These particles also mark the relative clause (Dillmann/Bezold 1907:§§201–2). The demonstrative pronoun for masc. sg. is zəә, fem. sg. zä, pl. comm. ʔəәllu. One therefore has to assume that zä (< ză) and ʔəәllä are in fact status constructus forms of

6

zəә and ʔəәllu respectively. In both (1) and (3) Geez thus has the hybrid juxtaposition construction similar to the one in Akkadian. Unlike Akkadian this construction is the rule in (3), where Geez shows only isolated cases of annexation. Adjectival attribution (category 2) is predominantly expressed by juxtaposition (2a) although cases of annexation (2b) are found (Dillmann/Bezold 1907:461–2). Only nominal attribution, i.e. the genitive construction, can regularly be expressed by both.

Aramaic

The Aramaic complex also has a common construction for 1) and 3). Like in the other languages there are two alternating constructions, annexation or juxtaposition with the complement marked by a particle viz. d(ī/ă). In most forms of Aramaic the tendency towards predominance of the latter construction is evident and in e.g Syriac it dominates completely:

(1) bayto ḏ-gaḇro ‘a/the man’s house’

(2) bayto rabo ‘a/the big house’

(3) gaḇro ḏ-qaṭleh ‘a/the man who killed him’

The annexation in (1) occurs frequently in Old Aramaic (Degen 1969:82–9), continues in Imperial Aramaic (Kaddari 1969:104, Bauer/Leander 1927:309ff., Muraoka/Porten 2003:228–33, Folmer 1995:259–325) and is also quite frequent in Targumic Aramaic (Fassberg 1990:251ff.; cf. R. Kuty in this volume). In other dialects, especially in the eastern ones, it recedes (Syriac: Nöldeke 1898:154–60; Mandaean: Nöldeke 1875:308–16, but cf. Macuch 1965:390 ff.); Talmudic: Schlesinger 1928:62–73, cf. Juusola 1999:140–41) and tends to survive in lexicalised expressions. This tendency is also found in Galilean Aramaic (Odeberg 1939:79–82). Category 3) uses juxtaposition from the beginning (Degen 1969:60, 128ff., Segert 1975:434, Muraoka/Porten 2003:167–71) which remains the predominant pattern in Aramaic. Very few cases of annexation are found in early Syriac texts like the Book of the Laws of the Countries: sim ʔares w-ḇelaṯy b-emro ʔaṯar da-ḵṯiḇ d-meṯyalḏin gaḇre ḥliṣe ‘Mars and Venus are posed in the Ram where it is said that brave men are born’ (Nau 1931:21:24–5, cf. 22:11–12), cf. l-ḵul ʔaṯar d-nizal ‘to every place it will go’ and b-ḵul ʔaṯro ḏ-ezalw ‘at every place where they went’ (Nöldeke 1898:273). One observes that the noun is annexed to a clause in its turn introduced by d-. Otherwise there seem to exist cases of pure juxtaposition of clausal complements in

7

some dialects like Mandaean GʔBRʔ RʔM ŠWMH W-ʔNTH RWD ŠWMH ‘a man whose name is R. and a woman whose name is R.’ (Nöldeke 1875:460; Macuch 1965:453, cf. Muraoka/Porten 2003:171), Talmudic: BY RDW YWMʔ ‘a piece of land which can be ploughed in one day’ (Baba bathra 12a = Schlesinger 1928:238), Palestinian Targumic (Neophyti): ʔRʕʔ ʔT DMK ʕLH ‘the land on which you are sleeping’ (Golomb 1985:59 to Gen. 28:13 MT haʔareṣ ʔašer ʔatta šoḵeḇ ʕaleha), cf. Targum Neophyti to Gen. 27:27), MN YWM DY NPQTH ‘from the day you went out from it’ (Schattner-Rieser 2004:131, Genesis Apocryphon 22:28). The last example is somewhat doubtful since it could also be a status absolutus form like in KL YWMYN DY THWʔ ʔRMLH ‘all the days when you will be a widow’ (ibid., from Nahal Hever). One could compare the two different Aramaic renderings of Ex. 5:23 ‘from the day I went in before Pharao’: Targum Onqelos: me-ʕidān d-ʕaliṯ lwaṯ parʕō, Targum Ps. Jonathan: min šaʕṯā d-ʕiliṯ lwaṯ parʕō.

In the Syriac version of the Aḥiqar novel there are a few cases which could be hybrid juxtaposition of adjectives (category 2): ṭoḇo neqyo ḏ-qariḇo men turṯo ḏ-raḥiqo ‘better a sheep which is close than a cow which is far away’ (Rendel Harris 1913:Ahiqartext 45:15, cf. 45:8 ṭoḇ roḥmo d-qariḇ men ʔaḥo d-raḥiq’better a friend who is close than a brother who is far away’). The interpretation of these examples as belonging to 2) rather than 3) is supported by the documentation of this from Galilean Aramaic: ʔoraḥ bīšā ‘evil way’ (Odeberg 1939:83–84) and Mandaean ḤYWʔT NWQWBTʔ ‘the wild animal’ (Nöldeke 1879:311). The expression ʕattîq yômîn (Dan. 7:9) could belong to this category (Segert 1975:415). Galilean Aramaic also uses the hybrid juxtaposition with d: senaṯore d-romai ‘Roman senators’ (Odeberg loc. cit.).

In spite of the examples quoted in the preceding paragraph the distinction between (1) and (3), on the one hand, and (2), on the other, has been quite strict all through the history of Aramaic. The construction with the particle d(a/i) should probably be seen as a construction of the type: ‘the house, viz. that of the man’, thus with the demonstrative particle in a construct state, i.e. a construction similar to the corrsponding one in Akkadian and Geez. This means that we have here the hybrid juxtaposition since the head noun stands in the non-construct state. In other forms of Aramaic (e.g. Old Aramaic) the particle has the form dī. The particle itself is most likely the same morphological element as zä in Geez, thus going back to a common Semitic ḏ-. In this case, even in Aramaic the particle is a deictic element in the construct state. Evidence in favour of this is the parallel adding of a suffixed -ī to biradical nouns (and occasionally some triradical ones) in the construct state like ʔabī,

8

ʔaḥī etc. in both Aramaic and Hebrew which, contrary to what is often claimed (Rubin 2005:49), should not be seen as a genitive case. One could also adduce the Arabic particle ḏū/ḏī/ḏā which also should be seen as a construct form and which is actually documented as a marker of relative clause in certain forms of Arabic (Wright I:272–73).

The typological classification of Aramaic is evident. Only (2) is marked with pure juxtaposition while (1) and (3) are marked with hybrid juxtaposition, i.e. with an introductory deictic particle in the construct state or annexation. Aramaic thus undoubtedly belongs to the same type as Akkadian and Geez.

Sabaean

It is worth looking at the conditions in the Ancient Epigraphic languages of South Arabia since the marking of the three syntagms is quite variegated there. The examples are taken from Sabaean which is the one best documented.

The Sabaean nominal declension is characterized by a three-state system which in the singular/’broken plural’ looks as follows (Beeston 1984:30–1):

0-state m-state n-state

(= status constructus) (= status absolutus) (= status determinatus)

BYT BYTM BYTN

Of these, the n-state functions as a definite article. Nominal attribution looks as follows (Stein 2003:82–90;145–51):

(1) a. BYT ʔS1M ‘a man’s house’

b. BYT ʔS1N ‘the man’s house’

c. ṢLMM Ḏ-ḎHBM ‘a statue of bronze’

d. ṢLMN Ḏ-ḎHBN ‘the statue of bronze’

(2) a. ʔWLDM HNʔM ‘healthy children’

b. ṮMR ṢDQM ‘good harvests’

c. NʔD ʔṮMRM ‘abundant crops’

(3) a. BYT BNY ‘a house which he built’

b. BYTM BNY ‘a house which he built’

c. *BYTN BNY ‘the house which he built’

d. BYTM Ḏ-BNY ‘a house which he built

e. BYTN Ḏ-BNY ‘the house which he built’

9

It is clear that in (1a) and (3a) we have annexation of the complement to a head noun in the construct state. Category (2) employs juxtaposition (2a) which might be the most common. In this case, Sabaean belongs with the three languages already studied. Unfortunately there are only a few cases of adjectival attributes or even adjectives in general in the Sabaean material. (2b) seems to be an annexation (Beeston 1984:33) but the question is whether the annexed complement is a noun (ṣidq) or an adjective (ṣad(d)īq). 2c could be an ʔiḍāfa ġayru ḥaqīqiyya if NʔD is an adjective (’abundant as far as crops are concerned’, Beeston 1984:33). But both 2b) and 2c) are doubtful cases.

In both (1) and (3) we find a construction with a particle Ḏ as an alternative to annexation (Beeston 1984:41–4). This particle has masculine singular Ḏ, feminine singular ḎT, dual ḎY, and plural ʔL.

The particle Ḏ etc. is also used in syntagms like: Ḏ-BYTN ‘the one (= the possessor) of the house’ i.e. ‘the owner of the house’. The demonstrative pronoun in Sabaean is masculine singular ḎN, feminine singular ḎT, dual ḎYN, and plural ʔLN. This comparison shows that we should assume that the particle used in the constructions (1) and (3) in Sabaean is in fact the construct form of the demonstrative pronoun, just as we have found in Akkadian, Geez and Aramaic. The n-suffix in the pronoun is identical to the suffix of the n-state of the noun. We thus have the hybrid juxtaposition as an alternative to annexation.

In Sabaean we find a new construction for category (3), viz. pure juxtaposition of the clause to the head noun (Beeston 1984:43), without any demonstrative particle. The head noun thus remains in the absolute state which in Sabaean is represented by the –m and the –n-states (3b). This alternative is not found in the other languages of this group but is a main device in the second large group of Semitic (see below) which thus links Sabaean to it although typologigally Sabaean tends to belong to the group we have treated here.

Modern Ethio-semitic as well as Modern Aramaic reflect the system inherited from their older relatives. One might, however, point to the development in e.g. Amharic where the marking of nominal and clausal attributes by yä is supplemented by the tendency to use clausal attributes even as adjectival attributes in expressions like y-alläfä-w samməәnt, ‘last week’ (lit. ‘which-passed-the week’, cf. Yri, this volume). It could be interpreted as a tendency towards identical marking of all three categories

In the modern South Arabian languages we find category (1) and (3) marked with the particle d (cf. Watson, this volume). In category (1) there are a few cases of pure

10

annexation in Mehri, but in Soqotri it is said to be more frequent. The relative clause is marked by the same particle, or by pure juxtaposition. The adjectival complement (category 2) is almost always introduced by (pure) juxtaposition only (Johnstone 1975:27, Wagner 1953:66f, 72f, 116ff., but cf. Watson, this volume). Both (1) and (3) thus employ the hybrid juxtaposition as the dominating pattern. It seems though that we can discern a differentiation between (1) and (3) in that the latter also uses pure juxtaposition, thus siding with Sabaean. But on the whole, modern South Arabian seems to belong to the same group as the others treated in this section.

Hebrew

Of the ancient languages Biblical Hebrew and the Arabiyya belong to group B. We shall start with a survey of the variegated picture in Biblical Hebrew (Gesenius/Kautzsch 1910:414–23, 427–9, 436, 485–9).

(1) a. beṯ ha-ʔiš ‘the man’s house’

b. beṯ ʔiš ‘a man’s house’

c. miṭṭaṯo še-lli-šlomo ‘the bed of Solomon’ (Song 3:7)

(2) a. ha-bbayiṯ ha-ggaḏol ‘the big house’

b. bayiṯ gaḏol ‘a big house’

c. šnaṯ ha-ttšiʕiṯ ‘the ninth year’

d. msillā ha-ʕolā ‘the upper path’ (Judg. 21:19)

(3) a. ʕaḏ ha-maqom ʔašer hayā šam ʔoholo ‘to the place where his tent was’ (Num. 10:29)

b. maʕasim ʔašer lo yaʕasu ‘deeds which one does not do’ (Gen. 20:9)

c. ʔelohim lo yeḏaʕum ‘gods whom they did not know’ (Deut. 32:17).

d. wa-yyizbeḥu l-YHWH ba-yyom ha-hu’ min ha-ššalal heḇiʔu ‘They sacrificed to the Lord on

this day from the booty which they had brought’ (2 Chron. 15:11)

e. wa-yyomer [yehošuaʕ] ʔel qeṣine ʔanše ha-mmilḥamā he-halḵu’ ʔitto ‘then Joshua said to the

officers of the warriors who had gone with him’ (Jos. 10:24)

f. qiryaṯ ḥanā ḏawiḏ ‘the town where David camped’ (Isa. 29:1)

g. kol ha-hiqdiš šmuʔel ‘all that Samuel had sanctified’ (1 Chron. 26:28)

h. šemaʕ le-ʔaḇiḵa ze yelaḏeḵa ‘listen to your father who has born you’ (Prov. 23:22)

i. ʔel meqom ze yasaḏta lahem ‘unto the place which you have founded for them’ (Ps. 104:8)

j. wa-yyittnehu ʔel beṯ ha-ssohar, meqom ʔašer ʔasire ha-mmeleḵ ʔasurim ‘and he delivered him

to the prisonhouse, the place in which/where the prisoners of the king were bound’ (Gen. 39:20).

k. kol ha-maʕasim še-nnaʕasu ‘all deeds which are done’ (Eccl. 1:14)

11

Category (1) is in Biblical Hebrew almost totally expressed by annexation. In (1c) we have a juxtaposition in which the complement is marked by the particle še and the dative preposition l-. This syntagm occurs in only a few texts such as the Song of Songs, and even there it is not frequent.

Category (2) is also dominated by pure juxtaposition of the adjectival complement. There are, however, cases of annexation as in (2c) especially with ordinals. This construction should be kept apart from another one, viz. one with the definite article on the complement, where the head noun remains in the absolute state according to the Tiberian vocalisation (2d): ḥåṣer hå-ʔaḥæræṯ, qånǣ haṭ-ṭoḇ, li-msillå hå-ʕolå. (cf. Brockelmann 1906 II:208–9). Judging from the form of the head noun this should be seen as a pure juxtaposition.

The clausal complement shows the most variegated devices although constructions (3a) and (3c) are the dominant ones and should be considered productive. Judging from the morphology of the head noun this is a juxtapositional construction. The distribution is mainly regulated by the definiteness (3a) or indefiniteness (3c) of the head noun although there are several exceptions (3b&d). In 3c) we find a pure juxtaposition of the clausal complement which is thus predominantly found with an indefinite head noun. (3a) is preferred when the head noun is definite. (3e) is noteworthy in that the juxtaposed clausal complement is marked with the definite article in agreement with the definiteness of the head noun and in the same manner as the adjectival complement in (2a). This is an exceptional construction found in a few cases in the Biblical texts and is reminiscent of a similar one documented in the Arabiyya (see below).

The juxtaposed clausal complement in case (3a) is marked with the introductory word ʔašer. There is little doubt that this should be seen as a construct state to an otherwise undocumented abolute state *ʔešer (Huehnergard 2006:107, 123). One should compare the absolute state hebel, ‘wind’, which has a construct state habel. The word is well documented in other Semitic languages (Aramaic, Akkadian). Its occurrence in Akkadian should be noted: ašar, status contructus of ašru(m), ‘place’, often stands as a locative indicator to a following clause: ‘the place where…’: šumma ašar illiku ... lā ītamar ‘when he does not find at the place where he went’ (von Soden 1969:231). We thus have a word meaning ‘place’ standing in the construct state to a following clause which is attributive to a head noun. This is a variant of what we have called the hybrid juxtaposition. It is reminiscent of the word ʔəәntä in Geez which also means ‘place’ and often stands in attributive clause complements to a feminine head noun. The use of a locative element in this position is not infrequent in the world’s

12

languages; one could compare the general relative marker pou in modern Greek (cf. Rubin 2005:50 n. 179).

This means that Biblical Hebrew in general has juxtaposition with categories (2) and (3) and annexation with (1). (3) has both pure and hybrid juxtaposition. As is clearly seen from the examples, the Biblical Hebrew corpus contains many cases where we find other constructions. Both (3f) and (3g) are annexations of the clausal complement, a not infrequent device especially with words with temporal meaning as head nouns, like b-yom dibber YHWH ‘on the day on which/when the Lord spoke’ (Ex. 6:28). In these cases the annexation has the same function as a temporal conjunction in European languages. This syntagm is quite frequent in Sabaean and, as we shall see, also in different forms of Arabic. An interesting comparison can be made between Isa. 15:7 yiṯrā ʕasā and Jer. 48:36: yiṯraṯ ʕasā ‘the riches that he has got’: in the first case we find a pure juxtaposition, in the second a pure annexation with, as it seems, identical meaning.

Annexation is also sporadically found with the demonstrative particle ze introducing a clausal complement (3i). This particle is, however, more frequently found in juxtaposition to the head noun (3h). It can also appear in the forms zo and zu. The particle can thus be seen as originating from *ḏī, *ḏā and *ḏū which looks very much like a similar deictic element in the Arabiyya and, more important here, is identical to Ḏ, etc. in Sabaean and zəә etc. in Geez. Whether the Hebrew form is a construct cannot be deduced from the actual shape of the particle. The comparative evidence is an argument in favour of this assumption although it is not decisive.

Example (3k) shows a variant of the marker of clausal complement which occurs in some texts in the Hebrew Bible, although not consistently. The book of Ecclesiastes, in which the quoted example is found, also has the more expected form: ha-mmaʕasē ʔašer ʕasā ha-ʔelohim (Eccl. 3:11) ‘the deed that God has done’. In Mishnaic Hebrew the particle še is the rule with clausal complements, alternating with pure juxtaposition, seemingly regardless whether the head noun is definite or not (Segal 1927:204, 225; Ridzewski 1992:57–9). Construction (1c) is frequent in Mishnaic Hebrew although construction (1a&b) seems to be the dominant pattern (Segal 1927:185–91; Ridzewski 1992:183–7). The relationship between the complements with še and the head noun is undobtedly that of juxtaposition which in Mishnaic Hebrew is the rule in categories (2) and (3) and occurs quite frequently also in category (1) although annexation is more dominant.

As far as category (2) is concerned Mishnaic Hebrew differs from Biblical Hebrew by the high frequency of definite head nouns without the definite article like

13

kneset ha-gdolā ‘the Great Congregation’ (Segal 1927:182 ff.). Judging from the examples quoted by Segal, this is construction (2d), i.e. juxtaposition of the adjective: ʕeṣā ha-hogenet ‘a (sic) suitable counsel’, neḵasim ha-meyuḥadin ‘property with specified owner’.

In both its main dialects, Hebrew thus predominantly employs juxtaposition for categories (2) and (3) and annexation for category (1). One should however notice the sporadic occurrence of annexation devices in category (3) and in a few cases in category (2) and juxtaposition in category (1) which is quite often found in Mishnaic Hebrew.

Canaanite

It is now worth looking at the close relatives of Hebrew. The evidence of the language of Canaan from the Amarna letters shows the following picture (Rainey 1996:99ff.):

(1) a. ṭi-ṭù ša ka-pa-ši-ka ‘the clay of your trading’

b. ṭi4-iṭ ša ka-pa-ši-ka do.

(2) a-wa-tu ša-ru-tu ‘evil word(s)’ (Rainey 1996:130).

(3) a. ip-šu ša-a la a-pí-iš-ta da-ri-ti ‘a deed which has never been committed’

b. a-wa-at ša qa-bi LUGAL EN-ia ‘the word which the king my lord spoke’

The language of the Amarna letters is basically Akkadian but with a strong influence from the local Canaanite dialects of Palestine and Phoenicia. To the latter undoubtedly belong the variants 1b) and 3b) which are not found in Akkadian. Unlike the usual pattern where we find a hybrid juxtaposition with ša we have here annexation plus the ša in construct to the following clause. Syntactically it is a construction identical with 3i) and 3j) documented in Biblical Hebrew. The genuineness of these latter is thus corroborated by the Amarna evidence. As we shall see, a similar construction is also documented from Arabic.

Ugaritic

The Ugaritic evidence looks as follows (Tropper 2000:840–5):

(1) a. bt mlk ‘the house of the king’

b. gpnm dt ksp ‘horse-covers of silver’

(2) a. mlk rb ‘the great king’

14

b. yrq ḥrṣ ‘yellow gold’

(3) a. il mlk d yknnh ‘Il, the king who has created him’.

b. rgm l tdʕ nšm ‘a word which people do not know’.

c. aṯt itrẖ // y bn ašld ‘O you two women whom I have married // you two sons whom I have

raised’.

The Ugaritic evidence is somewhat uncertain due to the absence of vowels which makes it difficult to see how the state-system works. But it seems clear that Ugaritic has both annexation and juxtaposition in category 1) with the former as the more dominant pattern, pure juxtaposition in category (2) (the complement is often preposed). The problem is category (3). The pattern with the particle d etc. (3a) is quite frequent but so is also the one(s) without particle (3b&c). There is only one certain documented case (3c) where the annexation of the clausal complement to the head noun(s; both are in the dual according to the context, thus *ʔanṯē and *bi/anē) seems indisputable. The latest investigator inclines to assume that annexation was the rule (Tropper 2000:899). If we take into account the Canaanite evidence from the Amarna letters, which is a corpus roughly contemporary with the Ugaritic texts, one could assume a similar situation in Ugaritic: both juxtaposition and annexation were used for category (3).

Phoenician/Punic

Phoenician is the closest neighbour to Hebrew in time, space and structure. They also treat nominal attribution in a similar way:

(1) a. HSLMT ʔŠ LMPQD Z ‘the storerooms of this depository’ (Krahmalkov 2001:101).

b. QBʕ KSP ‘cups of silver (Krahmalkov 2001:104).

c. QBʕM ŠLKSP do.

(2) ŠM Rʕ ‘a bad name’ (Krahmalkov 2001:144).

(3) a. LBNT ʔYŠ BN MḤLY ‘the incense altar which Mahli built’ (Krahmalkov 2001:94).

b. ʔRN ŠN MGN ʔMTBʕL ‘the ivory-box which Amotbaal gave’ (Krahmalkov 2001:100–1).

For category (1) Phoenician uses annexation (1b) or juxtaposition with the particle ʔŠ (1a), in Latin and Greek transcriptions is, ys, es, plus the preposition L introducing the complement (1a). Unfortunately there is no indication of differences in frequency. In a few cases we find the form ŠL (1c) which is reminiscent of the Hebrew šel. Whether

15

ʔŠ (L) and Š(L) are two variants of the same particle or have different origins is still discussed (Huehnergard 2006:109).

As a rule the verbal clausal complement is marked by juxtaposition with the particle ʔ(Y)Š introducing the clause (3a). Pure juxtaposition (3b) occurs but rarely. There are no certain cases of annexation here (Schuster 1965:433–7 but cf. Friedrich/Röllig 1970:148–9). The introductory particle seems to be the same as in category 1. It has been suggested that the particle ʔŠ in fact is a noun in the construct state (Schuster 1965:437, cf. Rosén 1959:191). In that case we would have a hybrid juxtaposition similar to the Hebrew one with ʔašer. It has also been suggested that the Phoenician and Hebrew particles are indentical. In Punic there is a differentiation between Š, in transcription su, sy, si, for nominal complements as well as verbless clauses, and ʔŠ, i.e. es, is, ys for verbal clausal complements (Schuster 1965:438; Krahmalkov 2001:93, 103). This could be adduced as an argument in favour of assuming a different origin of ʔŠ and Š, although arguments have been raised for a common origin (Schuster 1965:469, cf. Israel 2003 for a general survey of the etymological question). The general picture of Phoenician-Punic is that of a language which treats nominal attribution largely in the same manner as Hebrew with annexation mainly occurring with category (1) and different kinds of juxtaposition with categories (2) and (3). Compared with the other languages of this group Phoenician-Punic stands closest to Mishnaic Hebrew.

Arabic

We shall now turn to the Arabic linguistic complex. This consists of the Arabiyya, i.e. the language of the ancient poetry, the Qur’ān and, at least in principle, the medieval literature employing this language as well as the modern vernaculars of the present-day Arab countries. The devices used for nominal attribution are more variegated than usually assumed and should be analysed in the light of the comparative Semitic evidence presented above.

Arabiyya

(1) a. baytu r-raǧuli ‘the man’s house’

b. baytu raǧulin ‘a man’s house’

c. ġaḍibta min ġayri mā ǧurmin ‘you are angry without any offence (being made)’

(2) a. baytun kabīrun ‘a big house’

16

b. al-baytu l-kabīru ‘the big house’

c. baytu l-muqaddasi ‘the holy house’ (= Jerusalem)

(3) a. al-baytu l-laḏī askunu fīhi ‘the house in which I live’

b. baytun askunu fīhi ‘a house in which I live’

c. mā ʔanta bi-l-ḥakami t-turḍā ḥukūmatuhu ‘you are not the judge whose sentence is approved’

(Wright I:269–70)

d. xīfata yaḥmīhā (yaḥmiyahā?) banū ʔummi ʕaǧrada ‘from fear that the ʔUmm ʕAǧrad tribe

would protect them’ (ʔAxṭal: Reckendorff 1921:389)

e. mā tarā raʔya mā narā ‘you do not think what we think’ (Zuhayr: Reckendorf 1921:436)

The Arabiyya regularly employs annexation for nominal complementation of category (1), i.e. the traditional ʔiḍāfa, and juxtaposition for categories (2) and (3). It thus constitutes the classic example in group B. With definite head nouns the complement regularly receives the definite article l-. In this case the clausal complement is introduced by a deictic element la-ḏī which then carries the article. There are, however, in the Arabic linguistic tradition traces of other devices. Occasionally the article is prefixed directly to the clausal complement without the deictic element (3c), thus a parallel to example (3g) in Biblical Hebrew. Also worthy of note is the not infrequent use of annexation for category (2) (2c; Wright II:232–4), a parallel to example (2c) in Biblical Hebrew. Further, there are numerous examples of pure annexation of clausal complements (3d). Finally, the annexation may appear with a linking particle mā both in category (1) and (3) (1c, 3e). These two latter devices live on in the language until today in the particles introducing temporal subordinate clauses like ḥīna/ḥīna-mā etc. which should be analysed as originally head nouns in annexation to clausal complements with or without the linking mā. This is thus a parallel to examples (3f) in Biblical Hebrew, (3a) in Sabaean and (3a) in Akkadian.

There is, however, a small problem with construction 2c. In a footnote to Wright’s grammar (Wright II:§95.f) De Goeje claimed that baytu l-muqaddasi actually is an artificial creation by the grammarians who were confronted with bayt al-muqaddas without final vowels, i.e. a juxtaposed adjectival complement to a definite head noun which lacks the article, thus *baytu l-muqaddasu. De Goeje adduced the existence of such a construction in modern Arabic dialects (see below) and one could also refer to expressions 2d) in Biblical Hebrew like ḥåṣer hå-ʔaḥæræt ‘the other court’(1 Kings 7:8), qånǣ haṭ-ṭôḇ ‘the good willow’ (Jer. 6:20, M lectio difficilior?) etc. which, at least according to the Tiberian vocalisation, are clear juxtapositions (cf. Brockelmann 1913:208–9). From a comparative viewpoint, however, the existence of both

17

annective and juxtapositive constructions in which adjectives are joined to definite head nouns without article in the Arabiyya would be no anomaly, since it is documented in Biblical Hebrew. As will be evident below, both are also clearly attested in Arabic dialects. The evidence from the Arabiyya is, however, ambiguous since the frequent occurrences in medieval texts of article-less head nouns with definite adjectival complements occur in unvocalised texts. Furthermore, we do not find cases with plural or dual-suffixes mentioned in the studies (Blau 1966:359–60, id. 1980:161; Hopkins 1984:182–3).

Eastern Arabic dialects I

In spite of the occurrence of constructions (2c) and (3d–e) the Arabiyya definitely belongs to our group B, characterized by similar syntactic marking of (2) and (3) (juxtaposition), differentiated from the marking of (1) (annexation). Most modern Arabic vernaculars belong to the same type, although with somewhat different means of marking. The frequent use of annexation with category (1) and juxtaposition with categories (2) and (3) is basic (Retsö 2004). At the same time just as in the Arabiyya there is a tendency in many dialects to differentiate between (2) and (3) as well when the head noun is definite. Clausal complements are widely marked by a particle illi which is not used for category (1) or (2). Instead, many Arabic dialects possess a variant of (1) with a particle which, however, is different from the one marking (3). The particle often seems originally to have been a noun: matāʕ, tabaʕ, šay, šuġl, māl, ḥaqq etc. (Eksell Harning 1980, Eksell 2006, Vicente 2009) and in these cases it could be argued that the noun is in fact in the construct state to the following clause (Lentin 1997:753). The so-called ‘analytic genitive’ is thus a hybrid juxtaposition. The general picture in Arabic dialects is thus that of annexation or (hybrid) juxtaposition for category (1), pure juxtaposition for (2) and (3) but with a differentiation between l- and illi in the two latter.

It should once again be kept in mind that the variation in marking nominal attribution in Arabic dialects is mainly found with definite head nouns. With indefinite head nouns the marking is strikingly similar: usually pure juxtaposition with (2) and (3) and annexation and/or hybrid juxtaposition with (1). In many modern Arabic dialects there are, however, traces of other constructions which can be connected with some of those treated earlier and which, potentially, are of considerable interest for typology and diachrony. Cases where the differentiation between (2) and (3) with definite head nouns is absent appear in some eastern

18

dialects. The common marker is usually the definite article l-, e.g. Dēr iz-Zōr: il-ḅāmye d-dēriye ‘the ḅāmiye dish from Dēr ez-Zōr’ (Jastrow 1981:XI.4.11), hāḏa flān ādami iḍ-ḍarab il-walad ‘this is the person who beat the boy’ (Jastrow 1981:426, cf. n. 11 cf. 428 §16), hal-balad iǧ-ǧīna minha ‘the city we came from’ (Jastrow 1981:XI:3.34; cf. probably Antakya: Arnold 1998:114, 186).

The syntagm is thus identical with (3c) in the Arabiyya and 2d) in Hebrew, i.e., juxtaposition. It is often claimed that the clausal marker and the definite article still should be seen as different morphemes, shown by their sometimes slightly differing forms (Jastrow 1978:124). It seems, however, that the differences often can be explained by juncture phenomena. Another more important observation is that these dialects often have illi as an alternative marker of the juxtaposed clausal complement. From this it could be argued either that the identical marking of juxtaposed (2) and (3) is due to the intrusion of the definite article from (2) to (3), or that illi is a secondary development of l- due to a tendency to differentiate between all three main categories. This will be discussed in the final section of this article.

Eastern Arabic dialects II

In the same context as the example from Dēr ez-Zōr quoted above we find a variant without the definite article on the head noun in category (2): ḅāmiye d-dēriye (Jastrow 1981:XI.4.11). This is similar to the construction ḥåṣer hå-ʔaḥæræt etc. in Hebrew (2d) which is a pure juxtaposition.

We do, however, occasionally find clear cases of annective constructions for category (2) in many areas. The annexation is only clearly discernible when the head noun has the feminine suffix (Wild 1973:58; cf. Lentin 1997:726-28). Clear cases of annexation are documented from Malta: Ħaġret il-Kbira ‘the large stone’ (Borg 1989:65 cf. Ħaġra Is-Sewda ‘the black stone’ ibid. 71), Djidjelli: līlt es-sʕīda ‘happy night’ (Marçais 1956:416), Tripoli: barḥit l-ūla ‘the night before last’ (Borg 1989:65), Jerusalem: sāʕat es-sōda ‘the black hour’ (Borg 1989:65), Lebanon: sint iž-žāyi ‘next year’ (Jiha, Bišmizzīn 172), marrt il-žāy ‘next time’ (Abu Haidar 1979:86), and Damascus: səәnt əәlmāḍye ‘last year’ (Grotzfeldt 1965:93). This list can probably be extended with examples from further areas. This syntagm is mostly found with toponyms and temporal expressions and always with a definite head noun (Borg 1989 passim) and should not be classified as a productive grammatical category.

It seems clear that we have to do with two different syntagms, one with juxtaposition and another with annexation. This difference has not always been

19

observed and many of the cases adduced e.g. Borg (1989) remain ambiguous. Examples like borž el-kbīr (Tunisia) ‘the large tower’ (Borg 1989:75, cf. Singer 1984:441–2), tall il-kibīr (Egypt) ‘the big hill’ (Borg 1989:75) etc. could represent either construction. Cf. Tlemcen: līl et-tālya (Marçais 1902:154).

Eastern Arabic dialects III

Within a large area stretching from Cilicia in southern Turkey through Anatolia into Iraq we find in certain regions that when it is definite the head noun not only may lack the article but also may occur in the construct state, not only in (2) but also in (3). Otherwise pure juxtaposition is used. The exact delineation of this group is uncertain since it cannot be excluded that clause annexation may sporadically occur in several of the dialects discussed in the preceding paragraph, although unattested so far. A characteristic of these dialects, however, seems to be that annexation is of high frequency for all three categories (1), (2) and (3), often being the rule and thus a productive category. Cilician Arabic, the Mḥallamiye complex east of Mardin, the Jewish dialects of Aqra and Arbil together with the Muslim, Christian and Jewish Baghdadi dialects all have similar marking of the complement to a definite noun, viz. l- which is identical with the definite article while the head noun is in the construct state: ʕarabāyt il-xaḍra ‘the green car’ (Procházka 1999:152), ʕarabāyt il-xadta ʕǧabta ‘I like the car I bought’ (Procházka 1999:158); Mḥallamiye: ktēbəәt l-ʕarābīye ‘the Arabic script’, zalaməәt ləә-maʕu ẖanǧar ‘the person who has a dagger’ (Sasse 1971:88, 131); Azəәẖ: bērməәt ləә-gbīre ‘the large pool’ (Wittrich 2001:148), ḥakkoyəәt ləә-staẖbartūwa ʕalaya ‘the story which I had asked her about’ (Wittrich 2001:167); ʕAqra: waxt əәl-maṭar ‘the rainy season’ (Jastrow 1990:B.89), šəәrīkəәt əәẖti ‘my sister’s friend’ (ibid. B.160) but cf. l-māy əәl-ʕəәnəәb ‘grapejuice’ ibid. B.57), kəәbbəәt ḥāməәḍ ‘sour kəәbbe’ (ibid. B.92, cf. 340 s.v. ḤMḌ), hāk əәš-šīšəәt əәl-ǧarašna ‘this wheat porridge that we had crushed’ (ibid. B.133) 405). Arbil: dakēkīn māl yəәhūd ‘the shops of the Jews’ (ibid. C.1), ), sant əәǧ-ǧēye ‘next year’ (ibid. C.43) sant əәǧ-ǧəәdīde ‘the new year’ (ibid. C.46), ōḍəәt əәl-gəәbīre ‘the large room’ (ibid. C.112), šay l-kū-ṛəәbəәḥtu ‘that which I have gained’ (ibid. C.11 cf. əәl-ḥəәnṭa əәl-yəәštarīya ‘he who buys wheat’ (ibid. C.44), Baghdad Muslim: ʕēn iz-zarga ‘blue eye’ (Erwin 1963:307), sant il-fātat ‘the year that has passed’ (Erwin 1953:386, cf. Blanc 1964:126–8); Baghdad Jewish: sant el-leẖ ‘next year’, sant eǧ-ǧīna ‘the year we came’ (cf. ḥǧāġt/ḥǧāġa el-ma-teġḏāha ‘a (?) stone you do not like’ Blanc 1964:127–8); Baghdad Christian: qōndart el-leẖẖi ‘the other shoe’ (cf. ṣbāʕa le-kbīġi ‘the big finger’ Blanc 1964:127; cf. ibid. n.

20

142). The dialect of Mardin belongs to this type but with one characteristic feature of its

own: it differentiates between adjectival and clausal complements: rīḥat əәl-kwayyəәse ‘the good smell’, rīḥat la-təәqləәb maʕdati ‘the smell that makes my stomach turn’ (Grigore 2007:310–12, cf. Jastrow 1978:123–6).

Most of these dialects also use juxtaposition of different kinds for (2) and (3). All these dialects predominantly use annexation for category (1), although hybrid juxtaposition is also found as a less frequent variant (Eksell Harning 1981:41 ff., 158-62; Procházka 2002:153, Wittrich 2001:148–50). The distribution of these alternatives is still largely unexplored. It is thus characteristic of these dialects that annexation has a wide distribution and may be employed for all three categories.

Eastern Arabic dialects IV

In another dialect group belonging to the qəәltu-complex, viz. Siirt, Kozluk, and Diyarbakır, we find yet a different system. They employ a particle lay/lē to mark nominal and clausal complements while the head noun remains in the absolute state. Unlike all the others, they thus have identical marking of category (1) and (3), leaving out (2) where annexation seems to dominate. According to Jastrow (1981:218) category (2) is expressed by annexation even with indefinite head nouns. Siirt: baṭṭānīyəәt sawda ‘a black blanket’, čarčaf lē mara ‘a woman’s dress’ (Jastrow 1978:VII.2.10), āva Kōčaro lē kəәṭ-ṭalaʕlu əәsəәm ‘this K. who has become famous’ (Jastrow 1978:VII.1.1), Tillo: ṣəәḥbat əәṭ-ṭaybe ‘the nice company’ (Lahdo 2003:135). Daragözü (= Kozluk) ḥabs lē sērtu ‘the prison of Siirt’ (Jastrow 1973:94, id. 1978:126), ōzi lē yaqṭaʕo fəәl-məәnǧəәl ‘those who cut with the sickle’ (Jastrow 1973:8.7.3), Kaʕbīye (= Diyarbakır) ōḍa lē l-bīʕa ‘the room at the church (Jastrow 1978:IX.1.11), awde arman … lē k-baqu ‘those Armenians who were left’ (Jastrow 1978:IX.1.29).

Thus in one group (Cilicia, Mardin-Mḥallamīye, Baghdad) annexation may be used for both adjectival (2) and clausal complements (3). In another (Siirt, Kozluk, Diyarbakır) annexation is limited to adjectival complements, whereas nominal and clausal ones (1 and 3) are marked with the same particle. All this variation is predominantly found with definite head nouns. With indefinite ones, the picture is more in ‘agreement with Arabic dialects in general: juxtaposition in (2) and (3), annexation and/or hybrid juxtaposition in (1). In some cases the pattern of definite head nouns also appear sporadically with undefinite ones. Thus, in AzÉẖ we find

21

cases like maqbəәrəәt masīḥīye ‘a Christian cemetery’ in contrast to maqbəәrəәt əәl-masīḥīye ‘the Christian cemetery’ (Wittrich 2001:148). Such a noun phrase should, however, be seen as one word rather than a noun plus attribution.

In the Syrian desert we find a case like Palmyra: sqāq aṭ-ṭawīl ‘the long alley’ (Cantineau 1934:116) which should be compared to neighbouring Sukhne: bī ʔālet tinxol ṭaḥīn ‘there is a machine which sieves the flour’ (Behnstedt 1994:178). The first example can be interpreted in two ways: juxtaposition or annexation. The second example, a clear case of clausal annexation, might indicate that also adjectival annexation exists in the dialects of the oases in Central Syria. Clausal annexation (together with adjectival annexation) is further found in Feghali’s material from Kfar ʕAbîda in Lebanon: səәnt əәl-wāṣle ‘next year’ (Feghali 1928:136), sāʕt əәl-maʕak ‘the watch which you have’ (Feghali 1928:312) which shows that this syntagm might have a larger spread in Syrian Arabic than is apparent from the published material. Clausal annexation is sporadically documented from more unexpected areas: Central Arabia: sāʕat bidēt ar-riǧm ‘the time when I climbed the rock’ (Kurpershoek 1999:7:4), ḥarbīyet ṯāniye ‘a second campaign’ (Socin 1900–1:III:§153), Bahrayn: min sanat ǧābat fiḍḍaw brahīm ‘from the year when F gave birth to B’ (Holes 1984:28). From Baḥrayn we even have a case of annexation of a clause introduced by illi: min sanat illi tuʕṭīha ziyāra ‘the year when she would pay for her to visit the Holy Places’ (Holes 1984:28). Similar cases are documented from Lebanon: baṭṭit illi ʔakalnāha ‘the duck we ate’ (Mughazy 2009:64), the Syrian oases: laḥḏet il biddu ‘the moment when he wanted’ (Sukhne, Behnstedt loc. cit.) and Central Arabia: barrīyet elli mā bah moye ‘the desert in which there is no water’, hal-ẖubzit all-ana baʕaṯt-luh ‘this bread which I sent him’ (Socin loc. cit.).

Arabic dialects: Central Asia

The Arabic spoken in Central Asia has identical marking of (2) and (3): juxtaposition, and annexation with (1). There is also a variant (1b) which is due to Turco-Iranian influence (Vinnikov 1962:21–2; Fischer 1961:241–2, 244–5):

(1) a. mart abūy ‘my father’s wife’

b. maratu dewin aḥmar/dewin aḥmar maratu ‘the wife of the red demon’ (lit. ‘wife-his demon

red/demon red wife-his).

c. nuṣṣin lēl ‘midnight’ (Fischer 1961:244).

(2) martin ʕaǧūza ‘an old woman’

22

(3) kabšin šarinah ‘a sheep I bought’

The absence of variation in constructing (2) and (3) in Central Asian Arabic is connected with the lack of definite article in this dialect. The juxtaposition used for (2) and (3) parallells the same construction in all other Arabic dialects with indefinite head nouns. The use of -in also for (1) is remarkable. This use has been explained as influenced by the Persian ezâfe-element (Fischer 1961:244f.). It looks like the traditional Arabic tanwīn found in many modern dialects on the peninsula, marking juxtaposition of adjectival and clausal complements (Blau 1981:167–212) rather than a derivation fom ʔayna ‘where’ (Fischer loc.cit.). Central Asian Arabic is thus unique in all forms of Arabic in its tendency towards identical marking of all three categories.

Arabic Dialects: Maghrib

The predominant pattern in the sedentary Maghribi Arabic dialects is that of differentiation between the three categories, following the general trend in the Arabic complex. Thus, e.g. standard Moroccan: bāb əәd-dār ’the threshhold’ (Caubet 1993:II:301), sōt dyāl bəәnt ‘the voice of a girl’ (ibid. 306), əәl-ẖəәdma lli ka-tẖəәdmi ‘the work you do’ (ibid. 308). The pure annexation of (1) is of limited occurrence (Eksell Harning 1980:102, 119, 135–6; Eksell 1985; Caubet 1993 II:301), category (2) employs pure juxtaposition and (3) juxtaposition with the particle (i)lli introducing the clause complement. According to Heath (2002:265) most nouns do not occur at all in the construct state in the Moroccan urban belt as well as in the northern and Jebli Muslim dialects. In true Saharan dialects virtually any noun can occur in construct (ibid. 494–5; Eksell Harning loc. cit.).

In some dialects in the Maghrib, however, we find a system with identical marking of (1) and (3), this time by the particle (d)di, which differs from the marking of (2) (Eksell 1985):

Tlemcen

(1) a. əәl-ktsāb əәddi xāy ‘my brother’s book’ (Marçais 1902:172)

b. qāḍe-l-blād ‘the judge of the town (ibid. 148, ‘assez rare’ ibid. 147)

(2) a. līl et-tālya ‘the second night’ (ibid. 154)

b. el-bnāt el-melāḥ ‘the pretty girls’ (ibid. 213)

(3) əәl-məәfts əәddi ssemma ‘the mufti who was called’ (ibid. 175)

Sefrou (Jewish):

23

(1) li-ʔabura di rabbi Müsi ‘to the grave of Rabbi Moshe’ (Stillman 1988:52)

(2) a. l-zbil l-kbir ‘the big mountain’ (ibid. 68)

b. mzuza mxebbeʔa ‘a hidden mezuza’ (ibid. 69)

c. ʔabīlets mhaws ‘the tribe of M’ (ibid. 68)

(3) l-bluġat di xdimts ‘the slippers I made’ (Stillman 1988:50)

Fes (Jewish):

(1) l-məәllāḥ di fās ‘the məәllāḥ (Jewish quarter) of Fes’ (Brunot/Malka 1939:3:1)

(2) ṣ-ṣla l-kbīra ‘the great synagogue’ (Brunot/Malka 1913:5:16–17)

(3) l-məәllāḥ di kān ka-ittbna ‘the məәllāḥ which was being built’ (Brunot/Malka 1939:3:14)

In some of these dialects, such as the Jewish ones, the 1=3#2 marking is the rule. In others it is more irregular, (3) often marked by illi, ntāʕ, etc.

In at least one area in the Maghrib we find dialects in which similar marking of all three categories are found (Eksell Harning 1980:112):

Djidjelli:

(1) a. l-ḥāl əәddi xūyi ‘the condition of my brother’ (Marçais 1956:418)

b. qəәbb əәd-dār ‘the interior of the house’ (ibid. 409)

(2) a. lilt əәs-sʕīda ‘the happy night’ (ibid. 416)

b. əәr-rāžəәl əәddəә ṣḥīḥ ‘the solid (built) man’ (ibid. 505)

(3) a. əәr-rāžəәl əәddi ža ‘the man who came’ (ibid. 493)

b. əәn-nās əәlli žāw ‘the people who came’ (ibid. 505)

According to the investigator (3b) is a variant which is a recent phenomenon in this dialect (ibid. 493). The annexation of (2) is of low frequency, characterized as ʻpartie de lexique traditionnel’. Remarkable is the occurrence of (2b). According to the investigator (2b) is the preferred construction with an adjectival complement in the dialect of Djidjelli (505).

Andalusian Arabic

A few remarks on the Arabic once spoken in Spain are in place here. The evidence shows that (1) was mostly marked with particles such as mataʔ. For (2) we have the still existing cases like Guadalquivir = wād al-kibīr etc. which is ambiguous. But cases like tarbiat ahade ‘one child’ and thamárat mouareq ‘holy fruit’ show that

24

annexation existed for this category (Corriente 1977;125 n. 213). Besides, ample use of tanwīn with the head noun is found: bi cálben géid ‘wholeheartedly’ (Corriente 1977:121). 3) could be marked with an indeclinable allaḏi as in Ṣanʕā in Yemen but alli, addi/aḏḏi etc. are also documented (Corriente 1977:98–9). Worth noting is the scanty documentation of the particle d for category (1) (Corriente 1977:124 n. 215).

Sudan

In the Arabic spoken in a belt from northern Nigeria, through Chad to Sudan we find identical marking of (2) and (3) (for Sudanese, cf. Dickins, this volume):

(1) a. bagarat abuí ‘my father’s cow’ (Owens 1993:60)

hikāya(t) rāǧil mā wilēda ‘a story of a man and his child’ (Roth 1979:143)

b. al-ʕādāt hine al-bitt ‘the customs of the girl’ (Owens 1993:64)

ẖulgān hana al-faki ‘the clothes of the faqīh’ (Roth 1979:144)

(2) a. ar-rāǧil aṭ-ṭayīl ‘the tall man’ (Owens 1993:48)

bagaratan kabīre ‘a big cow’ (Owens 1968:67)

dar aẖärä ‘another country’ (Roth 1979:232)

b. sanat al-ǧadīd ‘the new year’ (Roth 1979:143)

(3) dá al-maktūb al-ankatab ‘this is the letter that was written’ (Owens 1993:88).

ar-rāǧil al-maʕāya ‘the man who is with me’ (Roth 1979:176)

With a definite head noun we thus find both the adjectival and the clausal complement marked with the definite artcle. This is system (3c) in the Arabiyya and (2d) in Hebrew and identical to the one found in e.g. Dēr ez-Zōr. Otherwise (1) is clearly distinguished from the two others but one should notice the many examples of annexation with (2) documented in Chadian Arabic.

Arabic: The General Picture

The Arabic linguistic complex turns out to present a most variegated picture of nominal attribution. From a strictly syntactic viewpoint we can discern three basic patterns:

(i) 1#2=3. (1) employs annexation and (2) and (3) juxtaposition. This group consists of the Arabiyya, the Sudanese complex and the majority of Arabic dialects of North Africa, the Middle East and the Arabian peninsula. We should notice the completely

25

identical marking of (2) and (3) in the Sudanese group, a pattern sporadically documented in the Arabiyya tradition, and the tendency to differentiate between (2) and (3) discernible in the other dialects as well as in the Arabiyya (the introduction in (3) of the ‘relative particle’ la-ḏī etc.).

(ii) 1=3#2. (1) and (3) use identical means of marking, viz. juxtaposition. This group consists of dialects in two different regions. In the Siirt-Kozluk-Diyarbakır complex the juxtaposed complement is marked by a particle lay/lē, and in some sedentary dialects in the Maghrib by the particle d(di).

(iii) 1=2=3. This group is found in the southern Anatolian qəәltu-areas and in Central Asia. In the former we find annexation as the dominant pattern for all three categories. In Central Asia the opposite construction, viz. pure juxtaposition, is used. This is at least partly due to Iranian influence and the absence of the definite article.

Few of these different language forms show a completely homogeneous pattern. In almost all variants there are cases of annexation or juxtaposition in constructions where the opposite is predominant. Especially worth noting is the widespread occurrence of sporadic annexation with (2) mostly with a word denoting time even outside group (ii). It can also be orbserved that the Arabiyya tradition possesses almost all the variants found in the modern dialects. Even from this fact it is natural to conclude that the variation within the Arabic complex is a development of a common multifarious heritage and not a deviation from a unitary system identical with the predominant pattern of the Arabiyya.

Arabic and Semitic

A comparison between the devices of nominal attribution in the Arabic complex and the other Semitic languages gives a quite interesting picture. Most forms of Arabic (the Arabiyya, the mainstream Arabic dialects) sides with Hebrew in having juxtaposition for (2) and (3) and annexation for (1). In both we also find the tendency to differentiate between (2) and (3) while preserving the juxtapositional structure. Typical for both is the distribution of the two different types of juxtaposition in (3): BH has pure juxtaposition when the head noun is indefinite, hybrid juxtaposition (with ʔašer, a construct form) when it is definite (not 100%). Arabiyya has pure juxtaposition in both but introduces the complement of a definite noun with the particle la-ḏī etc.) which carries the definite article. In fact, the Arabiyya has identical

26

construction of (2) and (3) both with definite and indefinite heads whereas BH has a hybrid juxtaposition in (3).

Phoenician-Punic definitely belongs to the same type as Hebrew with juxtaposition for (2) and (3), and annexation for (1). Ugaritic and Canaanite are more difficult to judge even if the former seems to follow the Phoenician-Hebrew type. But the frequent use of juxtaposition even for (1) should be noted as well as the employment of the same device for both (1) and (3), viz. the particle D(T). This could thus be a hybrid juxtaposition approaching Ugaritic to Aramaic.

The system 1# 2=3 (languages of type B) appears as a contrast to that found in group A where we have 1=3#2. In this group we find two variants, one represented by Akkadian with annexation/hybrid juxtaposition (particle ša) for (1) and (3), and pure juxtaposition for (2), and the other by e.g. Syriac with hybrid juxtaposition (particle d) for (1) and (3), and pure juxtaposition for (2). But typologically both belong to the same category. On the whole the languages of group B seem to have a quite stable pattern of pure juxtaposition for (2) whereas (1) and (3) show variation between hybrid juxtaposition and annexation.

Even the modern South Arabian languages seem to belong to group B despite the fact that the clausal complement of an indefinite head is juxtaposed as in BH and the Arabiyya. A remarkable fact is that among the modern Arabic dialects, the Siirt-Kozluk-Diyarbakır group in Anatolia and some sedentary Maghrib-dialects belong to group A. The latter have a system with definite head nouns that is almost identical to that of Syriac. Both dialect-complexes use juxtaposition, the Anatolian one with the particle lay, the Maghrib a morpheme that looks identical to (and most likely is, pace Eksell 1985:24–7, cf. Retsö 2004:269–71) the same morpheme as in Aramaic: d.

It appears from this investigation that it is useful to make a distinction between annexed clauses without demonstrative particle, and juxtaposed clauses without demonstrative particle. Both are often called asyndetic relative clauses but this term, if used at all, suits the latter construction better that the former. The usefulness of the term as well as its positive counterpart, syndetic, could be doubted since the word syndēsis actually means ’binding together’. From what has been said here it seems that viewing the particle as a kind of link or connection between two elements is not an exhaustive or even satisfactory description of the syntactic relationship. The terms ’analytic’ and ’synthetic genitive’ are not quite satisfactory either. They do not cover the whole field of nominal attribution and also lack morphosyntactic precision.

27

Diachronic Perspectives

A final diachronic evaluation of these facts cannot yet be made. There are many aspects that demand further research such as e.g. the distribution and historical development of the alternating constructions, the semantic factors etc. Many of these aspects are pointed out in the papers in this volume. At this stage, a few tentative diachronic suggestions can be made:

1. It can be observed that there seems to be more variation in the languages of group B than in those of group A. It is striking that the Arabiyya and Biblical Hebrew both show ample traces of the system found in group A whereas the latter do not exhibit parallels to the B-languages to the same extent. It is generally striking that upon closer observation both Arabiyya and BH show innumerable traces of e.g. markings of (3) by hybrid juxtaposition or annexation, which give the impression of being survivals of earlier stages of the languages. This could be an indication of diachrony.

2. A common feature of the B-languages seems to be the use of a pre-posed definite article. Since this phenomenon undoubtedly is not an original feature in Semitic we have yet another argument for the innovative character of the B-group. This has been emphasized by Pennacchietti (1968: 55–93) and his conclusions should be retained. According to him the constructions with d- or š- are as ancient as the pure annexation but the system was reshuffled when the preposed definite article was introduced. Pennacchietti sees the two particles as having a function similar to the so-called determinative pronouns in some European languages. He suggests that they are general indicators of attribution much like the definite articles in the B-languages (Pennacchietti 1968:65, 71 ff.). An interesting consequence is that that adjectival attributions in B-languages with a definite article on the complement only, well documented fom e.g. Hebrew and widely spread in modern Arabic dialects, could be more archaic than the one with article on both units (cf. Pennacchietti 1968:76–7). A full evaluation of this analysis cannot be made here but one of the conclusions is that there is no immediate reason to see the system of the Arabiyya as the original Semitic one. The statement often made that ’analytic’ constructions developed because of loss of case endings (Rubin 2005:51–2, cf. also Pennacchietti 1968:82) does not hold.

3. The system or rather systems of nominal attribution in the modern Arabic dialects is quite variegated and in light of point 1) there are no good reasons to assume that they represent a straight development from that of the Arabiyya. For example, the

28

Maghribinian one gives the impression of standing somewhere between Aramaic and the Arabiyya. In the light of comparative Semitics the common marking of (1) and (3) is an archaic feature which also holds for the annexation with 1) and 3) found in some qÉltu-dialects. Its origins should not be searched for in an Arabiyya-like forebear but this trait should rather be seen as ancient heritage (Pennacchietti 1968:81ff., 147–52, pace Grigore 2007).

4. The ample use of the construct form with syntagms (1) and (3) in the ancient and, to a certain extent, peripheral Semitic languages which also survives in fossilized form into BH and the Arabiyya should lead to posing question about the origin of the state-system in Semitic. In Arabic in particular this is a problem since the concept of state is not consistently used or even properly understood in Arabic linguistics (Retsö forthcoming). There are, however, arguments in favour of assuming a) that the construct state in Semitic was originally characterized by the absence of case inflexion, b) that the non-construct state is a secondary category, diachronically speaking in Semitic (Retsö 1997). Since case inflexion is a characteristic of the latter, not the former, the full case inflection is a secondary phenomenon. It is not a proto-Semitic feature.

5. In this light there is no reason to assume that all the ancient Semitic languages which do not possess case-inflection ever had it. The rudimentary case system of Geez might well be older than the more developed one in Akkadian let alone the one in the Arabiyya. It can be argued, based on the evidence discussed here, that the absence of case inflection in Hebrew, Aramaic and the modern Arabic dialects is due to the fact that they never had it. A proper understanding of the typology of the nominal attribution in Semitic lends support to this claim.

REFERENCES

Arnold, W. 1998. Der arabische Dialekt Antiochiens. (Wiesbaden) Bauer, H/Leander, P. 1927. Grammatik des Biblisch-Aramäischen. (Halle/Saale) Baumann, V. 1894. Hebräische Relativsätze. (Leipzig) Beeston A.F.L. 1984. Sabaic Grammar. (Manchester) Behnstedt, P. 1994. Der arabische Dialekt von Souchné (Syrien) Teil 2: Phonologie,

Morphologie und Syntax Teil 3: Glossar. (Wiesbaden) Blanc, H. 1964. Communal Dialects in Baghdad. (Cambridge Mass.)

29

Blau, J. 1966. A Grammar of Christian Arabic Based Mainly on South-Palestinian Texts from the First Millennium. (Louvain)

— 1981. The Emergence and Linguistic background of Judaeo-Arabic. A Study of the Origins of Middle Arabic. 2nd ed. (Jerusalem)

Borg, A. 1989. ‘Maltese Toponyms’, in P. Wexler, A. Borg, S. Somekh (eds.), Studia linguistica et orientalia memoriae Haim Blanc dedicata (Wiesbaden), 62–85

Boyce, M. 1964. ‘The use of Relative Particles in Western Middle Iranian’, in Indo-Iranica. Mélanges présentés à Georg Morgenstierne. (Wiesbaden), 28-47

Boyle, J.A. 1966. Grammar of Modern Persian. (Wiesbaden) Brockelmann, C. 1913. Grundriss der vergleichenden Grammatik der semitischen

Sprachen Bd II: Syntax. (Berlin) L. Brunot/E. Malka 1939. Textes judéo-arabes de Fès par L. Brunot et E. Malka. Textes,

transcription, traduction annotée. (Rabat) Corriente F. 1977. A Grammatical Sketch of the Spanish Arabic Dialect Bundle. (Madrid) Dillmann, A./Bezold, C. 1907. Ethiopic Grammar 2nd ed. transl. by J. A. Crichton

(London) Eksell Harning, K. 1980. The Analytical Genitive in the Modern Arabic Dialects.

(Göteborg) Eksell, K. 1985. ‘On the Development of d-particles as Genitive Exponents in Arabic

Dialects’, Acta Orientalia 46, 21–42 — 2006. ‘Analytic Genitive’, in K. Versteegh (ed.), Encyclopedia of Arabic Language

and Linguistics vol. I (Leiden/Boston), 82-5 Erwin, W. 1963. A Short Reference Grammar of Iraqi Arabic. (Washington D.C.) Fassberg, S.E. 1999. A Grammar of the Palestinian Targum Fragments from the Cairo

Genizah. (Atlanta) Féghali, M. 1928. Syntaxe des parlers arabes actuels du Liban. (Paris) Fischer, W. 1961. ‘Die Sprache der arabischen Sprachinsel in Uzbekistan’. Der Islam 36,

231–76 Folmer, M.L. 1995. The Aramaic Language in the Achaemenid Period. A Study in

Linguistic Varation. (Louvain) Friedrich, J./Röllig, W. 1970. Phönizisch-Punische Grammatik (Analecta Orientalia 46).

(Roma) Gesenius/Kautzsch 1910 = Gesenius’ Hebrew Grammar as edited and enlarged by the

late E. Kautzsch, 2nd ed. by A. E. Cowley. (Oxford) Gignoux 1993 = Wizīdagīhā ī Zādspram (Anthologie de Zādspram) ed., trad. et

commenté par Ph. Gignoux et A. Tafazzoli (Studia Iranica. Cahier 13). (Paris)

30

Goldenberg, G. 1995. ‘Attribution in Semitic Languages’. Langues orientales anciennes:

Philologie et linguistique 5-5, 1–20 Golomb, D.M. 1985. A Grammar of Targum Neofiti. (Chico, Calif.) Grigore, G. 2007. L’arabe parlé à Mardin. Monographie d’un parler arabe

”péripherique”. (Bucŭresti) Heath, J. 2002. Jewish and Muslim Dialects of Moroccan Arabic. (London/New York) Holes, C. 1983. ‘Bahraini Dialects: Sectarian Differences and the Sedentary/Nomadic

Split’. Zeitschrift für arabische Linguistik 10, 7–38 — 1984. ‘Bahraini Dialects: Sectarian Differences Exemplified through Texts’.

Zeitschrift für arabische Linguistik 13, 26–67 Hopkins, S. 1984. Studies in the Grammar of Early Arabic Based upon Papyri Datable to

Before A.H. 300/A.D. 912. (Oxford) Huehnergard, J. 2006. ‘Etymology of Relative šε’, in S. Fassberg and A. Hurvitz (eds.),

Biblical Hebrew in its Northwest Semitic Setting Typological and Historical Perspectives. (Winona Lake), 103–25

Israel, F. 2003. ‘Il pronome relativo nell’area cananaica’, in J. Lentin and A. Lonnet (eds.), Mélanges David Cohen. Études sur le langage, les langues, les dialectes, les litteratures offerts par ses élèves, ses collègues, ses amis présentés à l’occasion de son quatre-vingtième anniversaire (Paris). 331–46

Jastrow, O. 1973. Daragözü, eine arabische Mundart der Kozluk-Sason-Gruppe (Südostanatolien). Grammatik und Texte. (Nürnberg)

— 1990. Der arabische Dialekt der Juden von cAqra und Arbil. (Wiesbaden) — 1978. Die mesopotamisch-arabischen Qəәltu-Dialekte I: Phonologie und Morphologie.

(Wiesbaden) Johnstone, T.M. 1975. ‘The Modern South Arabian languages’, Afroasiatic Linguistics 1,

93–121. (Malibu) Juusola, H. 1999. Linguistic Peculiarities in the Aramaic Magic Bowls. (Helsinki) Kaddari, M.Z. 1969. ‘Construct State and dī-Phrases in Imperial Aramaic’, Proceedings

of the International Conference on Semitic Studies Held in Jerusalem 19-32 July 1965. (Jerusalem), 102–15

Krahmalkov, Ch.R. 2001. A Phoenician-Punic Grammar. (Leiden/Boston/Köln) Kurpershoek, P.M. 1999. Oral Poetry from Central Arabia III: Bedouin Poets of the

Dawāsir Tribe Between Nomadism and Settlement in Southern Najd. (Leiden) Lahdo, A. 2003. The Arabic Dialect of Tillo in the Region of Siirt (South-eastern Turkey).

Unpublished thesis, Uppsala university

31

Lehmann, Chr. 1984. Der Relativsatz. Typologie seiner Strukturen. Theorie seiner Funktionen. Kompendium seiner Grammatik. (Tübingen)

Lentin, J. 1997. Recherches sur l’histoire de la langue arabe au Proche-Orient à l’époque moderne. (Lille)

Li, Ch. N./Thompson, S. 1981. Mandarin Chinese. A Functional Reference Grammar. (Berkeley/Los Angeles/London)

Macuch, R. 1965. Handbook of Classical and Modern Mandaic. (Berlin) Marçais, Ph. 1956. Le parler arabe de Djidjelli (Nord constantinois, Algérie). (Paris) Marçais, W. 1902. Le dialecte arabe parlé à Tlemcen. Grammaire, textes et glossaire

(Paris) Mughazy, M. 2009. ‘Relative Clause’, in K. Versteegh (ed.), Encyclopedia of Arabic

Language and Linguistics vol. IV (Leiden/Boston) 60-70 Muraoka, T./Porten, B. 2003. A Grammar of Egyptian Aramaic. 2nd ed. (Leiden) Nau 1931 = Bardesane: Le livre des lois des pays. Texte syriaque publié avec une

avertissement et une analyse par F. Nau. Paris Nöldeke, Th. 1898. Kurzgefasste syrische Grammatik. 2nd ed. (Leipzig) — 1875. Mandäische Grammatik. (Halle/Saale) Odeberg, H. 1939. The Aramaic Portions of the Bereshit Rabba with Grammar of

Galilaean Aramaic II: Short Grammar of Galilaean Aramaic. (Lund/Leipzig) Owens, J. 1993. A Grammar of Nigerian Arabic (Wiesbaden) Pennacchietti, F. 1968. Studi sui pronomi determinativi semitici. (Napoli) Procházka, S. 1999. Der arabische Dialekt der çukurova (Südtürkei). (Wiesbaden) Rainey, A.F. 1996. Canaanite in the Amarna Tablets. A Linguistic Analysis of the Mixed

Dialects Used by the Scribes from Canaan vol. 1. (Leiden/Boston/Köln) Ravn, O.E. 1941. The So-Called Relative Clauses in Accadian or the Accadian Particle ša. (Kjöbenhavn)

Reckendorf, H. 1921. Arabische Syntax. (Heidelberg) Rendel Harris 1913 = The Story of Aḥiḳar from the Aramaic, Syriac, Armenian, Ethiopic,

Old Turkish, Greek and Slavonic Versions ed. F.C. Conybeare, J. Rendel Harris, A. Smith Lewis. 2nd ed. Cambridge

Retsö, J. Forthcoming. ‘Status – eine vernachlässigte Kategorie der arabischen Grammatik?’ (Papers presented at the Wolfdietrich Fischer Tagung, Erlangen June 2008)

— 2004. ‘Relative-clause marking in Arabic Dialects: A Preliminary Survey’, in M. Haak/R. De Jong/K. Versteegh (eds.), Approaches to Arabic Dialects. A Collection of

32

Articles presented to Manfred Woidich on the Occasion of his Sixtieth Birthday. (Leiden-Boston), 263–74

— 1997 ‘State and Plural Marking in Semitic’, in E. Wardini (ed.), Built on Solid Rock. Studies in Honour of Professor Ebbe Egede Knudsen on the Occasion of his 65th Birthday April 11th 1997. (Oslo), 268–82

Ridzewski, B. 1992. Neuhebräische Grammatik auf Grund der ältesten Handschriften und Inschriften. (Frankfurt/M)

Rosén, H. B. 1959. ‘Zur Vorgeschichte des Relativsatzes im Nordwestsemitischen’. Archiv orientální 27, 186–98

Roth, A. 1997. Esquisse grammaticale du parler arabe d’Abbéché (Tchad). (Paris) Rubin. A. 2005. Studies in Semitic Grammaticalization. (Winona Lake) Sasse, H.-J. 1971. Linguistische Analyse des arabischen Dialekts der Mḥallamīye in der

Provinz Mardin (Südosttürkei). (München) Schlesinger, M. 1928. Satzlehre der aramäischen Sprache des babylonischen Talmuds.

(Leipzig) Schuster, H.-S. 1965. ‘Der Relativsatz im Phönizischen und Punischen’. Studies in Honor

of Benno Landsberger on His Seventy-fifth Birthday April 21, 1965. (Chicago), 431–48

Segal, M.H. 1927. A Grammar of Mishnaic Hebrew. (Oxford) Segert, S. 1975. Altaramäische Grammatik mit Bibliographie, Chrestomathie und

Glossar. (Leipzig) Singer, H.-R. 1984. Grammatik der arabischen Mundart der Medina von Tunis. (Berlin-

New York) Socin, A. 1900-01. Diwan aus Centralarabien gesammelt, übersetzt und erläutert I-III,

hrsg. H. Stumme. (Leipzig) von Soden, W. 1969. Grundriss der akkadischen Grammatik. (Roma) Stein, P. 2003. Untersuchungen zur Phonologie und Morphologie des Sabäischen.

(Rahden) Stillman, N. 1988. The Language and Culture of the Jews of Sefrou, Morocco: An

Ethnolinguistic Study. (Manchester) Tropper, J. 2000. Ugaritische Grammatik. (Münster) Vicente, A., ‘Relative pronoun’, in K. Versteegh (ed.), Encyclopedia of Arabic Language

and Linguistics vol. IV (Leiden/Boston), 70-2 Vinnikov, I.N. 1962. ‘Slovar’ dialekta buxarskix arabov’. Palestinskij Sbornik 10 (73) Wagner, E. 1953. Syntax der Mehri-Sprache unter Berücksichtigung auch der anderen

neusüdarabischen Sprachen. (Berlin)

33

Wild, S. 1973. Libanesische Ortsnamen. Typologie und Deutung. (Beirut) Wittrich, M. 2001. Der arabische Dialekt von �Azəәx. Wiesbaden Wright, W. 1896, 1898. A Grammar of the Arabic Language I–II, 3rd ed. (London)


Recommended