35
Rivka Halevy The Grammaticalization 'Chains' of the Subject-Coreferential Dative in Semitic and Elsewhere 1. Introduction One of the characteristics of Hebrew is the employment of the dative case as an extra marking on the action described in the sentence or on the agent performing this action. Of special interest is the greatly expanded use of the construction verb + the dative preposition l- suffixed with a personal pronoun agreeing with the verb-incorporated Subject. This construction is found in Biblical Hebrew but only on a limited scale, while in Modern Hebrew it is widespread. The distinguishing characteristics of this construction are the optional nature of the Subject- coreferential dative, as well as its special stylistic and pragmatic effect, which will be explained below. In Semitic languages, the dative case is generally marked by the ex-allative preposition l-, meaning 'to'. As in many other language groups, this dative case-marker has a multitude of functions: it indicates not only the obligatory dative case but also direction towards a goal, 1 as well as the meaning 'for' (i.e. the benefactive) 2 and 'of' (i.e. the possessive). 3 In Biblical Hebrew, the preposition l- suffixed with a Subject-coreferential pronoun was also used to indicate the reflexive. (An independent, unambiguously reflexive pronoun based on the stem etsem – 'bone', 1 In the latter meaning, it alternates with more expanded locative forms, e.g. Hebrew el ('to', 'towards') and its variations. 2 In post-Biblical Hebrew, this type of dative pronoun alternates with the benefactive preposition bišvil (lit. 'in the path of'). 3 E.g. Biblical Hebrew mizmor le-dawid 'a psalm of David' (Psalm 22:11).This type of dative pronoun alternates with the unmarked possessive marker šel ('of') in post-Biblical Hebrew. In other Semitic languages (such as Aramaic and Arabic), it also commonly signals the genitive (possessive), e.g. in the Aramaic verbal pattern ī leh, lit. has to-him ('he has'), as well as in nominal possessive constructions.

R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

Embed Size (px)

Citation preview

Page 1: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

Rivka Halevy

The Grammaticalization 'Chains' of the Subject-Coreferential

Dative in Semitic and Elsewhere

1. Introduction

One of the characteristics of Hebrew is the employment of the dative case

as an extra marking on the action described in the sentence or on the

agent performing this action. Of special interest is the greatly expanded

use of the construction verb + the dative preposition l- suffixed with a

personal pronoun agreeing with the verb-incorporated Subject. This

construction is found in Biblical Hebrew but only on a limited scale,

while in Modern Hebrew it is widespread. The distinguishing

characteristics of this construction are the optional nature of the Subject-

coreferential dative, as well as its special stylistic and pragmatic effect,

which will be explained below.

In Semitic languages, the dative case is generally marked by the

ex-allative preposition l-, meaning 'to'. As in many other language groups,

this dative case-marker has a multitude of functions: it indicates not only

the obligatory dative case but also direction towards a goal,1 as well as

the meaning 'for' (i.e. the benefactive)2 and 'of' (i.e. the possessive).3 In

Biblical Hebrew, the preposition l- suffixed with a Subject-coreferential

pronoun was also used to indicate the reflexive. (An independent,

unambiguously reflexive pronoun based on the stem ∙etsem – 'bone',

1 In the latter meaning, it alternates with more expanded locative forms, e.g. Hebrew ≤el ('to', 'towards') and its variations. 2 In post-Biblical Hebrew, this type of dative pronoun alternates with the benefactive preposition bišvil (lit. 'in the path of'). 3 E.g. Biblical Hebrew mizmor le-dawid 'a psalm of David' (Psalm 22:11).This type of dative pronoun alternates with the unmarked possessive marker šel ('of') in post-Biblical Hebrew. In other Semitic languages (such as Aramaic and Arabic), it also commonly signals the genitive (possessive), e.g. in the Aramaic verbal pattern ī leh, lit. has to-him ('he has'), as well as in nominal possessive constructions.

Page 2: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

'thing' – came into use only in post-Biblical Hebrew). This fluidity of

function displayed by the dative expression has implications in terms of

the ability to establish clear boundaries between the categories

represented on its grammaticalization 'chains', as will be shown below.

For purely formal reasons, the Subject-coreferential l- pronoun

under discussion will be referred to as a 'Subject-coreferential dative'

(SCD), though, as we shall see, it is in fact a Caseless affix (i.e. a dative

form which does not indicate abstract dative Case).

The following are examples of the construction in both Biblical

and Contemporary Hebrew:4

(1) le l∞ m-≤arş∞ u-mi-molad∞t (Gen 12:1)

go SCD-2SG.M from your-country and-from-your-mother-land

'Leave your country and your homeland'

'Va t'en hors de ton pays, et de ta parenté' (LS Bible)

(2) kaxa stam šotatnu lanu bi-sderot qaqal (Kaxa Stam, a popular

Israeli song by A. Hillel)

so just we-were-walking-around SCD-1PL in-boulevard qaqal

'We were just hanging around on Qaqal Boulevard'

This construction is occasionally found in Spoken Modern Arabic as

well,5 as shown in (3); however, it is quite rare in comparison to its

Modern Hebrew counterpart.

4 In examples from Contemporary Hebrew, I will use a simple transcription reflecting the standard pronunciation of the spoken language. "x" will be used for both heth (˙) and khaf (), ' for both ∙ayin (∙) and ≤aleph (≤), "ts" for şadi (ş), "v" for both vav and weak beth (), and "f" for weak peh (p‡). Furthermore, I will generally ignore gemination (dagesh forte), and use only five vowels (a, e, i, o, u), disregarding length. 5 Cf. Brockelmann 1961:380; Piamenta 1981: 217.

Page 3: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

(3) qa∙adË lahum ∆woye

they-sat SCD-3PL.M a-little

'They were sitting [about viz. leisurely] for a short while'

A construction involving a post-verbal l- + Subject pronominal suffix

evolved in Aramaic as well, most significantly in North-Eastern Neo-

Aramaic (NENA),6 but this construction, which does not stem directly

from the reflexive-benefactive

(as will be explained in some details in Section 4), grammaticalized in the

synchronic system in a different way. Furthermore, a construction similar

to the Hebrew SCD also evolved, apparently independently, in some non-

Semitic languages, as will be discussed below.

In Semitic linguistics, the Hebrew and Aramaic construction of

post-verbal l- + pronominal suffix agreeing with the verb-incorporated

Subject has traditionally been referred to as dativus ethicus.7 This term

(which is problematic and deceptive, since this construction is neither

'ethical' nor 'dative') is one of many technical terms that Semitic

linguistics has borrowed from classical linguistics, especially from

analyses of Greek and Latin.8 The typical 'ethical dative' (ED) is a

personal pronoun in casus obliquo, which, unlike the SCD, is not co-

referential with the Subject, or, in fact, with any other argument in the

sentence. It usually appears in first or second person (but sometimes also

in third), and is typical of the colloquial register and especially of direct

6 North-Eastern Modern Aramaic (NENA) comprises a large number of diverse dialects spoken by Christian and Jewish communities in northern Iraq, south-eastern Turkey, Armenia and Georgia. 7 Regarding Biblical Hebrew, see GKC 1910: 381; BDB 1966: 515b; Jenni 2000:48-53; Regarding Syriac, see Joosten 1989, inter alia. 8 According to The Revised Latin Primer (London 1962) by B. H. Kennedy, "a Dative of a Personal Pronoun, called the Ethic Dative, is used, in familiar talk or writing, to mark interest or call attention, e.g. quid mihi Celsus agit? Horace ('Tell me, what is Celsus about?'), Haec vobis per biduum eorum milita fuit Livy ('This, mind you, was their style of ghting for two days')". About Greek, see Schwyzer & Debrunner 1950:149.

Page 4: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

speech.9 These "datives", very common in Indo-European languages, both

modern and ancient, are very rare in classical Semitic, but widespread in

Contemporary Hebrew (probably owing to the influence of Yiddish). The

following are examples from contemporary French, German and Hebrew:

(4) Les gosses lui ont gribouillé sur tous les murs

the kids ED-3SG have scribbled on all the walls (example taken

from Autier & Reed 1992: 295).

(5) Das ist mir ein feiner Kerl This is ED-1SG a fine lad

(6) tihyeh li bari'

be ED-1SG healthy (calque of Yiddish zai mir gezunt)

When appearing in the first person, these EDs can be characterized as

signifying the emotional interest of the speaker in engagement with his

counterpart, whereas in the second person, they can be said to signal an

intention, on the part of the speaker, to involve the addressee in the

situation being described.

The SCD construction (as in 1-3) and the EDs (as in 4-6) are

similar in that both are non-lexical clitics. They are not arguments of the

verb or even adjuncts, they do not affect the grammatical function of the

verb or add any new participant role to the event structure, and their

insertion does not affect the relationship between the core participants in

the event. In other words, the hallmark of these "dative" pronouns is their

9 According to the Oxford English Dictionary, the ethical dative is "used to imply that a person, other than the subject or object, has an interest in the fact stated". Autier & Reed (1992: 296) state that "ethical datives […], unlike affected datives, denote individuals who are not necessarily interested parties as far as the process denoted by the sentence is concerned, but rather, this type of non-lexical dative refers to individuals whose relation to the process denoted by the sentence is only that of potential witness [emphasis mine]." Borer & Grodzinsky 1986: 211 (working in the framework of Generative Grammar) state: "Ethic Dative must be disjoint from the external argument, and in fact from all the arguments of its clause. While we do not have a complete explanation for this disjointness, we would like to propose that this is due to the pragmatic function of the construction: conveying in essence the effect of an event on a seemingly uninvolved party [emphasis mine]. "

Page 5: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

dispensability, since they can be dropped without affecting the truth

conditions of the sentence. I suggest that both should be analyzed as

elements functioning on the extrasentential level, i.e. on the level of

discourse and pragmatics. In fact, I propose to treat the Modern Hebrew

SCD as a special case of the ED.

Unlike many writers, I will also avoid using the term 'reflexive

dative pronoun' in referring to the Semitic SCD,10 and this for two

reasons. First, in the contexts where it appears, the SCD cannot be

felicitously replaced by the unambiguously reflexive pronoun that is

found synchronically in the language (e.g. by the reflexive etsem pronoun

of post-Biblical Hebrew). Second, in Modern Hebrew, this l- pronoun co-

occurs with verbal patterns of overt reflexive-middle morphology (i.e.

with the verbal templates Hitpa∙el and Nif∙al), as demonstrated in (7):

(7) me'ever la-∆ulxan hi∆tapla lah ha-keres ha-mefuneqet (A.B.

Yehoshua, The Return from India, 1994: 197)

beyond to-the-table went-lower- REFL SCD-3SG.F the-belly the-

spoiled

'Beyond the table the spoiled belly hung down (to herself)'

Another term I will avoid is dativus commodi/incommodi, which

traditional studies of Hebrew (especially of Biblical Hebrew), have

associated with the SCD. In my view, this association is misleading,

firstly because the SCD necessarily involves Subject coreference, while

the benefactive dative does not, and secondly, because the SCD, unlike

10 Borer & Grodzinsky (1986: 185 ff) refer to the Hebrew SCD, for purely morpho-syntactic reasons, as a "reflexive dative". Berman (1982: 51ff) uses the rather loose term "reflexive or coreferential dative". Many studies of Biblical Hebrew likewise use the term "reflexive dative" (cf. Jöuon 2003: 488, König 1897: §35, Waltke & O'Connor 1990: 208, Williams 1976: §272), even though Biblical Hebrew has no other paradigm of reflexive pronouns. One exception I have found in the Semitic literature is that of R. Contini, who prefers the term "coreferential dative" (dativo coreferenziale). This he uses as a purely formal label, not meant to indicate the constructions' function.

Page 6: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

the benefactive dative, cannot be replaced by the explicitly benefactive

prepositions (e.g. the post-Biblical Hebrew preposition bi∆vil 'for').

The provenance of the SCD in Hebrew and in certain non-

Semitic languages such as Spanish, Slavic and Yiddish, could suggest

the following 'chain' of grammaticalization: allative > dative >

benefactive > reflexive-benefactive >SCD (and "genuine" ED).

However, a closer examination reveals that the diachronic 'chain' of

development is not so straightforward.

The purpose of this paper is twofold. From a narrow point of

view, it aims to investigate the internal grammaticalization 'chains' in

which the Semitic SCD developed. From a broader, theoretical

perspective, it aims to show that grammaticalization proceeds in

strictly local steps, and does not necessarily follow any presupposed

universal pattern. To illustrate this point, I will examine the

grammaticalization of the SCD in two Semitic languages: Hebrew

(Classical and Modern), which will be the main focus of the paper, and

Eastern Aramaic, especially Eastern Neo-Aramaic (though

comparative remarks about some non-Semitic languages will be made

as well). I will show the local pathways that diverge independently

from the grammaticalization 'chain' of this common l- pronominal.

2. The grammaticalization of the Semitic SCD construction

Let us first introduce the grammaticalization pathways of the SCD in the

two languages under investigation, namely Hebrew and Eastern Aramaic.

The process of grammaticalization evolved independently in each

language, through both pathways begin with the same construction, i.e.

the ex-allative l-.

Page 7: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

As mentioned above, Biblical and Modern Hebrew have formally

identical SCD constructions. However, I maintain that the modern SCD is

not patterned on the SCD of Biblical Hebrew (or post-Biblical Hebrew),

but is a "reinvention" by native speakers of Modern Hebrew, and its

function is different from that of the Biblical construction.11 The modern

SCD is first attested in the colloquial language of (non-native) Hebrew

speakers during the first decades of Israeli Hebrew, though it was fairly

quite infrequent in their language. The data presented below suggests that

this SCD is a calque of a construction commonly found in Slavic and

Yiddish, the main languages which were in contact with the revived

Hebrew at the time. In this local pathway, the SCD construction

expanded in a process of subjectification, as will be explained below, and

eventually developed, in Contemporary Hebrew, into a "genuine" ED

functioning purely on the pragmatic level.

The second local pathway of the construction's development is the

one attested in Eastern Aramaic, where the post-verbal l- + pronominal

Subject eventually became grammaticalized as part of the verbal system.

Let us first outline the grammaticalization 'chain' of the

construction in both Biblical and Modern Hebrew, and then examine the

independent grammaticalization 'cline' of the post-verbal l- + pronominal

Subject in Aramaic.

2. The grammaticalization of the SCD in Hebrew

2.1 Biblical Hebrew

Like all grammatical words, the l-preposition probably evolved from

lexical words.12 Most likely, the grammaticalization 'cline' took the form

11 For a fully developed analysis of the construction in Modern Hebrew, see Halevy 2007. 12 To be precise, it probably evolved from the unreduced grammaticalized prepositions ∙al ('on') or 'el ('towards'), which, in turn, probably evolved from lexical words such as ∙ala /∙aley- ∙elyon/∙elyonim, ma∙ala, etc., denoting the property of being up or above.

Page 8: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

of a unidirectional shift from a less grammatical, or unrestricted,

morpheme into a more grammatical and restricted one, apparently tracing

the path allative > dative. Subsequently, l- expanded to become the

optional benefactive, the reflexive-benefactive and presumably also the

non-argumental pronominal l- (SCD) under investigation. Thus, the

immediate precursor of our construction in Biblical Hebrew is not the

basic dative itself, but rather the expanded reflexive-benefactive pronoun.

Traces of this diachronic grammaticalization 'cline' can be clearly

seen in Early Biblical Hebrew. For example, in the book of Genesis, the

preposition l- still alternates with allative forms such as ≤el, with the

preposition ≤al and with more expanded and less grammaticalized

prepositions like ∙imad- – which later institutionalized (i.e.

grammaticalized) in the synchronic system, receiving the meanings of

'on/upon', 'towards' and 'with', respectively, e.g.:

(8) va-yyi∙aş ≤el libb (Gen. 6: 6)

and-grieved-REF-3M.SG to his-heart

'and he became grieved at heart'

(9) h-≤iša ≤ašer natta ∙immad- h≤ nana l (Gen.

3:12)

the-woman that gave-2M.SG with-1SG she gave-3F.SG DAT-1SG

'the woman that you gave me, she gave (it) to me'

Eventually, the ex-allative l- became conventionalized as the standard

dative marker.

However, though there is plentiful evidence of the well-governed

use of the obligatory dative in Biblical Hebrew, and of its extension into

the benefactive and reflexive, there are relatively few occurrences of the

Page 9: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

SCD construction. Furthermore, this construction is restricted in its

semantic and syntactic distribution. In early Biblical Hebrew it is

generally confined to intransitive motion verbs such as 'go', 'flee', 'rise',

'ascend', 'pass' and 'turn' (plus one stative verb, namely 'sit'). In this layer

of Biblical Hebrew, the SCD it usually occurs in imperative-hortative-

jussive contexts, rather than in narrative statements. For example:

(10) v∞-qm bra˙ l∞a ≤el ln (Gen. 27:43)

and-stand-up run-away SCD-2SG.M to Laban

'go away and take refuge with Laban'

'Lève-toi, enfuis-toi vers Laban' (LS Bible)

In Late Biblical Hebrew, which, according to the vast majority of

scholars, is represented by the Song of Songs, the use of the SCD

construction is more frequent and much more flexible. Intransitive

motion verbs still predominate, as in (11):

(11) ha-ssv ∙ar ha-ggešem ˙la hla l (SoS 2:11)

the-autumn passed the-rain passed-by went SCD-3SG.M

'La pluie a cessé, elle s'en est allée / s'en alla' (LS Bible)

However, there are also some occurrences with non-prototypical

transitive verbs such as 'know' and 'resemble', as demonstrated in (12)

and (13):

Page 10: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

(12) ≤ im lo≤ td∙ la ha-yypˉ ba-nnšm (SoS 1:8)

if not know-PRET SCD-2-SG.F the-beautiful-SG.F in (among)-

the-women

'don't you know, (you) the most beautiful of (all) women'

(13) u-dm l∞ li-şv (SoS 8:14)

and-resemble-IMPER-2SG.M SCD-2SG.M to-stag

'and resemble (you) the stag'

Moreover, unlike in Early Biblical Hebrew, the SCD is no longer

confined to imperative-hortative-jussive constructions but also appears

in narrative speech, as exemplified in (11) above.

2.2 The grammaticalization of the construction in Modern Hebrew

In Modern Hebrew, the SCD shows an overall pattern of gradual

expansion in terms of its semantic range, i.e., the range of situation

types it expresses. Semantic constraints associated with the reflexive-

benefactive source gradually disappear – especially constraints on the

animacy of the Subject and its volitionality/agentivity, and the

requirement for a basically two-participant event. Although a

coreference still holds between two participants in the event, the SCD

construction has basically become an indication of instigator-

affectedness.

In contrast to its Biblical counterpart, the SCD of Contemporary

Hebrew is very free in its distribution: it is not restricted to intransitive

verbs or to animate Subjects, nor is it confined to the colloquial

register, but appears in various styles and registers of the language. It

should be emphasized, however, that in Hebrew, and probably in all

languages that display this construction, it is prototypically confined to

Page 11: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

verbs whose subjects have properties of control and/or instigation of

the action, and it is therefore precluded with verbs such as 'fall', 'slip' or

'feel'. The following utterances, collected from written Modern Hebrew

texts of various styles, demonstrate the very wide distribution of the

SCD. Note that, unlike in the examples from Biblical Hebrew, in the

following I add possible pragmatic inferences of the construction,

inserted in square brackets.13

(14) ve-hi' qama ve-'azva yom 'exad 'oto ve-'et ∆ney ha-yeladim […]

pa∆ut barxa lah (A. Oz, Story of Love and Darkness 2002: 195)

and-she stood-up and-left day one him and-OM the-two children

[…] simply run-away SCD-3SG.F

'And one day she [just] left him and her two children […] [fancy

that!] just ran away'

(15) 'anaq 'exad mexo'ar, […] haya mitnapel al kfarim, mišpaxot

xo†ef lo (E. Sidon, The Giant Monster, children poem, 1991)

giant one ugly, was attacking villages, families kidnapping SCD-

3SG.M

'One ugly giant, […] used to attack villages, kidnapping families

[for fun]'

(16) be-lev ha-brexa pi'ape'a lah mizraqa xari∆it (A. Oz, ibid. : 356)

In-the-heart (of) the-pool bubbled SCD-3SG.F fountain silent

'In the center of the pool, a fountain was bubbling quietly [to

itself, dissociating itself from its surroundings]'

13 With regards to Biblical Hebrew, I maintain that it is impossible to draw any definite and generally valid conclusions about the construction's pragmatic function, since this is an ancient "dead" language which, moreover, consists of different diachronic layers.

Page 12: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

In terms of its function, the SCD construction basically expresses two

opposite directions of the action, namely movement in and out, flowing

outwards (from the Subject) and going back (i.e. turning the action back

upon the Subject). The main effect of the construction is therefore to

assign reflexivity to the event, in that the event is viewed as reflecting

forward from the Subject-nominal to the pronoun, and then back from the

pronoun to the Subject-nominal. In addition, it also imparts a sense of

benefactivity in that the event is viewed as performed by and for the

Subject. In other words, the Subject is presented as fulfilling both the

agent role and the experiencer role at the same time. It seems, therefore,

that the essential function of the construction is to express a rather intense

involvement of the Subject in the action,14 as well as the autonomy of the

event.15 The Subject is perceived as a free agent, detached from the

surrounding world.16 Accordingly, this construction can often be

associated with a sense of isolation, loneliness, dissociation, egoism or

frivolity.

Thus, we may define this construction as Subject-oriented.

However, in Modern Hebrew, the construction also clearly signals that

the event is presented from the speaker's perspective.17 That is to say, in

addition to increasing the prominence of the Subject, this construction

also functions as an anaphoric reference to the speaker's attitude towards

the Subject or the situation in which he/she/it is immersed. To use Lyons'

14 Cf. Glinert (1989: 224): "[The construction] highlighti[s] that the subject is his own 'free agent' capable of acting on his own, and responsible for his own condition". 15 Cf. Berman (1982: 55): "The use of the SCD pronoun highlights the AUTONOMY of the event, as perpetrated to, by and for the subject noun (even where the subject is nonanimate)". 16 Cf. Muraoka (1978: 497): "The preposition Lamedh followed by the matching pronominal suffix seems to have the effect of creating a self-contained little cosmos around the subject, detached from the surrounding world, an effect of focusing on the subject […] this preposition can be best described as centripetal [emphasis mine]". 17 Jenni (2000: 49) denes the function of the SCD in Biblical Hebrew as "eine Revaluation der Person […] als Re-Lokalisation und Re-Situierung, kurz als Aktualisation: 'x […] in seiner aktuellen Situation'". As indicated above (fn. 13), I feel that, with regards to Biblical Hebrew, we lack sufficient data to determine this. However, regarding Modern Hebrew I completley agree with this characterization of the SCD function.

Page 13: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

term, the SCD under investigation is a case of "empathetic deixis".18 To

my mind, this expressive, or evaluative, meaning of the construction is

not far removed from that of the "genuine" ED, which is essentially a

discourse pronoun that is anchored in the speech situation and co-indexed

with one of the speech participants (speaker or addressee), and as such

has the effect of enlisting the solidarity or complicity of the hearer, or

simply of creating a greater affective closeness between hearer, speaker

and message. That is, its development does not merely involve

desemanticization, or loss of function, but rather change of function. It is

important to note that "grammaticalization", as used here, is not meant

only in the diachronic sense. It refers primarily to the systematic or

conventionalized way in which an abstract grammatical function is

attributed in the synchronic system. Our pronominal SCD is thus a true

"grammaticalizator", since it codes semantic and pragmatic relations that

were not coded in the same way before.

18 Lyons (1977: 452) refers to all elements which express the speaker's attitude towards, or opinion about, the content of the proposition as "modals". He applies the term "emphatetic deixis" (ibid :677) to cases of anaphoric reference where "the speaker is personally involved with the entity, situation or place to which he is referring, or is identifying with the attitude or viewpoint of the addressee". He concludes that "at this point deixis merges with modality".

Page 14: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

2.3 A Cross-linguistic comparison with languages in contact with

Modern Hebrew

A construction similar to the Hebrew SCD is apparently also found in

Slavic languages, though it varies from one Slavic language to another in

a number of parameters.19 Polish has two reflexive markers, one that is

not sensitive to case (się) and another which is case-sensitive (siebie).

Frazjyngier (2000) proposes that while siebie and its related forms code

the identity of the Subject with some other argument, the marker się does

not indicate that the Subject controls the action but only that it is affected

by the action. In other words, the contemporary Polish construction with

the Subject-coreferential pronoun (henceforth SC.PRO) się seems to be

Subject-oriented, just like the Hebrew SCD. This explains why it can

appear not only with transitive verbs, but significantly also with

intransitive ones. The following examples, taken from Frazjingier,

demonstrate its usage with an animate as well as an inanimate Subject:20

(17) Ale serce mi się kraje na myšl o rozstaniu

but heart 1SG.DAT SC.PRO cut on thought about separation

'But my heart hurts when I think about separation'

(18) Zabacz czy woda się podnosi

see whether water SC.PRO is rinsing

'See whether the water is rinsing'

19 According to Geniušenė (1987: 274-5), it varies in the obligatory vs. optional character of the dative pronoun to which the initial subject is demoted; the presence vs. absence, and the obligatory vs. optional character, of modal qualifiers meaning 'well', 'easily', etc.; and in restrictions on the lexical base. 20 Frajzyngier 2000:131.

Page 15: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

The following are examples of the construction in colloquial

Russian,21 conveying a durative state in which the Subject is immersed,

very much like their Modern Hebrew counterparts:

(19) a. R: ya sidel sebe v uglu i molchal

I was-sitting SC.PRO [casually] in (the) corner and was-silent

(keeping silent)

b. MH: ya∆avti li ba-pina ve-∆atakti

I-was-sitting SCD-1SG [casually] in-the-corner and-I-was-silent

(20) a. R: ona ∆la sebe po ulitse i nikogo ne trogala.

she was-walking SC.PRO [casually] on (the) street and didn't

disturb anybody

b. MH: hi' halxa lah ba-rexov ve-lo hifri'a le-'af 'exad

she was-walking SCD-3SG.F in-the-street and not-disturbed to-

anybody

An SCD construction very close in its pragmatic function to that of the

Slavic languages is already present in the Hebrew of the revival period,

for example in Uri Nisan Gnesin's novels from the beginning of 20th

century. As pointed out by Even-Zohar,22 this literary usage is probably a

calque from Russian (while the SCD of current spoken Hebrew may also

be inspired by Yiddish and Polish). The following are examples from

Gnesin:

(21) u-ma? halxa la linqa? (U. N. Gnesin, Be†erem, 1913: 248)

and-what? Went SCD-SG.3F Linka?

'So [is it really true that] Linka went off [just like that]?

21 I thank Ben-Zion Dimersky, a native speaker of Russian, for providing these examples. According to his intuition, they are characteristic of the southern Russian dialect. 22 Cf. Even-Zohar 1986:31.

Page 16: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

(22) Va-'atem be-fo – harey 'atem mazqinim laxem, ah? (ibid. 258)

and-you in-here – actually [emphatic word] you are-getting old

SCD-PL.2M, ah?

'So, you guys are really getting old [just like that/regardless of

others],ah?'

As already indicated, a very similar construction is prevalent in Yiddish,

which was also in contact with revived Hebrew and probably influenced

the Hebrew SCD construction. Many occurrences are found in Sholem

Aleichem's Yiddish novels, which were translated into Hebrew by his

son-in-law, Y. D. Berkovitz. On a single page (the first page) of "Dos

Meserl" ("The Knife"), a story written in highly colloquial Yiddish, there

are four occurrences of the construction. The SCDs appear with both

transitive and intransitive verbs and are very similar to their counterparts

in Contemporary Hebrew:

(23) Y: dos meserl zol zix ligen in kešene

MH: yanuax lo ha'olar ba-kis

'The knife will lay SCD-3SG.M [for a while] in the pocket'

(24) Y: un ven ix vil, zol ix dos mir aroysnemen

MH: u-kše 'ertse, 'otsi' li 'oto

'and when I want, I'll take SCD-1SG it out [as I please]'

(25) Y: ix hob dos mir taki alain gemaxt

MH: 'ani 'atsmi (levadi) 'asiti li 'oto

'[imagine/ fancy that!] I made SCD-1SG it by myself (all alone)'

(26) Y: un hob mir fargeštelt

MH: ve-hirkavti li 'oto.

'and I have assembled it SCD-1SG [by myself/for my own pleasure]'

Page 17: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

SCDs are also ubiquitous in Sh. Y. Agnon's writing, representing the

second phase of revived Hebrew (the 1930s and 1940s), e.g. in his novel

Shira,23 and, as already mentioned, they are generally widespread in

Israeli Hebrew, both written and spoken.24

I argue that the expressive (pragmatic) SCD construction in

Modern Hebrew, and probably in numerous other languages such as those

mentioned above, developed via a process of subjectication, as did the

"genuine" ED. That is, both are instances of a propositional element that

evolved into a non-propositional element in the sentence structure. Thus,

I suggest that the SCD in question illustrates a dynamic, unidirectional

process of grammaticalization whereby lexical items that originally

function on the lower level of structure (viz. predication, locution)

acquire a new status in the synchronic system, that is, come to encode

meanings and relations on the higher level of structure (viz. pragmatics,

illocution). In this sense, the development of the SCD can be

characterized in terms of the grammaticalization 'cline' propositional>

expressive, as defined by Traugott (1982; 1989).25

23 For examples and discussion, cf. Ullendorff 1985. 24 Cf. Halevy 2004; 2007. 25 As stated in Traugott (1989: 31, 35), "meanings with largely propositional (ideational) content can gain either textual (cohesion-making) or expressive (presuppositional, and other pragmatic) meanings, or both"; "Meanings tend to become increasingly situated in the speaker's subjective belief state/attitude toward the proposition"; or, more recently, as a type of semantic change causing meanings "to become increasingly based in the SP(eaker)/W(riter)'s belief or state or attitude toward what is being said and how it is being said" (2003 :125)..

Page 18: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

3. A Cross-linguistic comparison with SCDs in unrelated languages

A similar construction also evolved in languages that are unrelated to

Hebrew and were not in direct contact with it. In Vulgar Latin, there is

evidence of a reflexive (dative) construction conveying similar extra

meaning regarding the doer of the action. To quote Bourciez,26 "On disait

déjà dans la langue classique abstinere ou abstinere se, erumpere ou

erumpere se […] pour indiquer d'une façon plus analitique la part que le

sujet prend à l'action; beacoup de verbes se sont ainsi construits,

notamment des verbes de mouvement: Surrexisse se Deos (Arn. 5, 18);

vadent se unusquisque (Peregr. 25, 7)". Evidently, this construction later

spread to other Romance languages. In Spanish, it is quite evident that

two separate constructions, of different diachronic origins, merged into

one single form, marked for 3rd person by the reflexive pronoun se-. The

first construction, a "genuine" reflexive marked by se- (Lat. sui),

appeared most commonly with transitive verbs but also developed a

middle-voice meaning and later an impersonal-passive meaning. The

second construction is the old dative>benefactive pronoun le-, which

later evolved into ge->se-.27 It is noteworthy that the modern Spanish

SCD conveys no sense of reflexivity. Just as in Biblical Hebrew, it is

typically associated with perfectivity, sudden change, or sudden

departure, as demonstrated by the following examples:

(27) El joven se-fue

the boy SC-PRO-3SG went

'The boy went away [took off]'

26 Bourciez 1946: §118c, 27 Cf. Monje 1955.

Page 19: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

(28) El joven se bebió el vino

the boy SC.PRO-3SG drank the wine

'The boy [just] drank [up] the wine'

Maldonado (1999) shows that the Spanish se- construction does not

necessarily impose a completive interpretation, but rather a sense of full

involvement, i.e. a maximal degree of participation by the Subject in the

execution of the action designated by the verb, whether transitive or

intransitive. An example is (29):28

(29) Tongolele se bailó una rumba inolvidable

'Tongolele SC.PRO-3SG danced an unforgettable rumba'

If the action is done involuntarily or without much interest, the addition

of se- produces an ungrammatical output, e.g. (30):

(30) *Tongolele se bailó una rumba inolvidable sin mayor interés

'Tongolele SC.PRO-3SG danced an unforgettable rumba without

much interest'

However, contrary to the SCDs of Contemporary Hebrew, Slavic and

Yiddish, the Spanish construction does not seem to have developed an

evaluative meaning, i.e. the function of anaphoric reference to the

speaker's attitude towards the Subject and the situation in which the

Subject is immersed. To use Traugott's (1982; 1989; 2003) terminology,

it has not yet undergone a process of subjectification.

For a broader typological comparison, it might be worthwhile to

investigate a parallel construction (namely verb + Subject-repetitive

28 Cf. Maldonado 1999: 153-4.

Page 20: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

dative or genitive pronoun) in Chamito-Semitic languages. Such

constructions exist in Ancient Egyptian,29 as well as in contemporary

African languages, including some Chadic languages and Swahili.30

For instance, a team of Bible translators from Chad translated Gen.

27:43 as follows (see example 1 for the Hebrew original):

(33) col kolo pii tud ki man vi wayanna Laban (taken from Noss 1995:

334)

Arise up flee go away of-you to your uncle Laban

The following is an example in Swahili:

(34) wamekwenda zao (example taken from Ashton 1947:57)

they have gone theirs

'They've gone off'

Significantly, in these languages, the Subject-coreferential pronoun is a

post-verbal suffix identical in its morphological form to a possessive,

rather than dative, pronoun.

29 Cf. Polotsky 1979: 208 fn. 9; Jenni 2000:48. 30 Cf. Noss 1995.

Page 21: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

4. The grammaticalization 'chain' of the SCD construction in

Aramaic

The construction of post-verbal l- + pronominal Subject takes a different

path of development in middle (pre-Christian) Aramaic, Rabbinical

Hebrew (probably due to Aramaic influence), Babylonian Talmudic

Aramaic, Classical Syriac, and finally in the later varieties of Neo-

Aramaic, where it appears as a conventionalized grammatical form

constituting part of the synchronic verbal system. Unlike in Hebrew, in

these Semitic languages it did not develop further to become an element

functioning on a higher level (i.e. not on the sentence level, but on that of

illocution or the universe of discourse).

At the early stage of the grammaticalization 'cline' – as represented

by the Aramaic of the Babylonian Talmud and by Classical Syriac – we

find, in addition to the regular past-tense forms, three variations of a

participle base, primarily passive in meaning, which occur with an agent

expression introduced by l- and appear with both transitive and

intransitive verbs:31

1. The perfect/passive participle (q†il), bare or inflected + l- + explicit NP

representing the agent. E.g. in Classical Syriac:

(35) kol da-re l-alåhå (I Tim. iv 4)

all that create PAST.PART to-God

'All that God has created'

2. The inflected perfect/passive participle + proleptic l- pronominal agent

+ l- + explicit NP representing the agent. This construction is widespread

in the Aramaic of the Babylonian Talmud, e.g.:

31 Cf. Goldenberg 1998 [1992] : 118 [610]; Bar-Asher 2007: 378.

Page 22: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

(36) svira leh l-rabbi yehuda (Kritut 10:a)

think-PAST.PART. SCD-3SG.M to-Rabbi Yehuda

'Rabbi Yehuda thought'

3. The bare or inflected perfect/passive participle + l- + pronominal agent.

This construction is frequently found in Classical Syriac, e.g.: 32

(37) w-en ˙a†åhe ‡ ∙bidin le ‡h (Jas. V 15)

and-if sins make- PAST.PART 3PL. SCD-3SG.M

'And if he has committed sins'

These three variations apparently represent three consecutive stages in the

grammaticalization 'chain' of the Aramaic SCD, culminating in the third

construction. In this last grammaticalized construction, common in

NENA languages, the doer of the action denoted by the passive participle

base is expressed by the enclitic preposition l- + a pronominal suffix

obligatorily affixed to the bare past-tense (passive participle) base. In

these languages, then, the bare passive participle form known as q†il and

the pronominal li have become – formally and syntactically – a single

verbal expression, "contracted or shriveled up into a single word",33 e.g.

(Urmi dialect) ptixle 'he opened', ptixla 'she opened'. The l- pronominal

suffix agrees with the Subject nominal and actually replaces the Subject

pronominal suffix of the historically inflected verb, e.g. (Sulemaniyya)

gorā pli†le, lit. 'man went out to-him'; ≤āna pli†li, lit. 'I went out to-me'.

32 It should be noted that this construction with the passive participle can also introduce actants other than the agent, namely the 'beneficiary' (when the agent is not mentioned). Cf. Goldenberg 1998 [1992]: 613 [117]. 33 Cf. Goldenberg 1998 [1992] : 614 [122].

Page 23: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

In NENA languages, the passive participle base thus displays a

discrepancy between 'logical' function and grammatical form. The

erstwhile dative, which normally represents the undergoer role, has

spread into the Subject-pronominal paradigm, where it has the inverse

function of actor. It is noteworthy that the phenomenon of dative-marked

agents or agent-phrases – expressed as instrumentals, locatives or

genitives in both active and passive – is well known in many other

languages throughout the world,34 as are agent-phrases that attach to the

(passive or active) verb in the same way as possessors in possessive

constructions. The latter phenomenon is attested, for example, in the

languages of the Philippines and in Malagasy.35

In the Aramaic SCD under discussion, the agent-participant serves

as the anchor of the action, and thus has a dative-possessive meaning. In

line with Polotsky (1979) and Goldenberg (1992/1998), it seems

reasonable to posit that the passive character of the q†il(-li) form is the

key to understanding the whole construction. To quote Polotsky, "the

'logical' undergoer of ptix-lı ['he opened'] is grammatically the nå∙ib al-

få∙il [the undergoer], and its 'logical' actor is the Dative, expressing the

actor as possessor of the accomplished action and its result".36 And

similarly in Goldenberg: "With a perfect/passive form […] of transitive

verbs, […] the agent is expressed as the possessor of the-patient-having-

undergone-the-accomplished-action with the resulting state […]. With a

34 Hebrew has such a construction with perception verbs, e.g. nir'e li (lit. seen to me) meaning 'it appears to me' and nidme li (lit. seem to me) meaning 'it seems to me'. Unlike in Aramaic, however, the Hebrew construction is restricted to a small class of verbs, and thus has not reached the same level of grammaticalization in the synchronic system. More significantly, in the Hebrew construction, the referent of the dative is basically viewed as the 'experiencer', i.e. as non-agentive. 35 Cf. Keenan 1985: 259, 263-265. 36 Polotsky 1979: 208.

Page 24: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

perfect/passive form of intransitive verbs, the agent is accordingly

expressed as the possessor of the accomplished action and its result".37

In Western Neo-Aramaic languages, this l- pronominal has

integrated mainly into the participle form of movement and position

verbs, and seems to express the semantic domain of the middle voice. The

following are examples from Ma∙lula, a modern Western Aramaic dialect

of the Anti-Lebanon region:

(38) †oˉle

come-PRES.PART.SCD-3SG.M

'he comes'

(39) zålle

go-PAST.PART. SCD-3SG.M

'he went away'

(40) q∙oˉle

sit-PAST.PART. SCD-3SG.M

'he sat down'.

In the Eastern Aramaic languages, such as Mandaic, Classical Syriac38

and to some extent Babylonian Talmudic Aramaic,39 the l- agentive

pronominal has extended into other categories of verb, including the

stative and the fientive (i.e. verbs designating a durative and dynamic

action, such as verbs of motion and change of position), and also into the

category of verbs expressing emotions and perceptive content, which are

37 Goldenberg 1998 [1992]: 608 [116]. Lazard (1984:242), addressing the question of how to analyze the case-marking of the actor in passive constructions, suggests a more subtle and flexible analysis. He claims that it depends on whether the action described is potential or completed: "Si celle-ci a un sens potentiel, l'agent est un destinataire … Si la forme verbale exprime l'action accomplie, particulièrement au parfait, l'agent est un possesseur. C'est le cas dans le tour vieux-perse manā krtam. L'expression A(h) urmazdā-šām ayadiya, où le verbe est à l'imparfait, semble occuper une position intermédiaire, mais plus proche du rapport de destination". 38 Cf. Joosten 1989:490. 39 Cf. Schlesinger 1928 §30.

Page 25: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

likewise associated in many languages (e.g. Spanish) with the middle

voice. Recall that in Modern Hebrew, unlike in Aramaic, the SCD

construction is not confined exclusively to verbs of motion or to

intransitive verbs.

When used with stative verbs, the l- pronominal agent often

denotes the ingressive Aktionsart.40 Some examples from Classical

Syriac:

(41) qim leh

arise-PAST.PART. SCD-3SG.M

'he arose '

(42) ∆teq leh

silent-PAST.PART SCD-3SG.M

'he shut up (became silent/stopped talking)'

Thus, the local development of the l- pronominal Subject seems to have

traced the following path: stative > fientive, durative > ingressive. In

NENA, the bound q†il-le form is also grammaticalized with intransitive

verbs that have the semantic properties of a prototypical agent, e.g.

control and/or instigation of the action, e.g. xille 'he ate', pqele 'he shot (a

gun)', or with verbs denoting actions that are controlled by the Subject

and express a reflexive activity, e.g. lwšle 'he dressed' (himself)', ksele 'he

covered (himself)'.41 Another class of intransitive verbs that take the q†il-

le form in the past tense is that of punctual verbs denoting the production

of noise, e.g. nwixle 'it (the dog) barked', šrixle 'he shouted'. Conversely,

40 Joosten ibid. 41 When these verbs are inflected without the l- suffixes, they express actions in which the Subject is the affectee of the action rather than one who controls it, and so have a passive meaning.

Page 26: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

when the action is not perceived as being controlled by the Subject (e.g.

šire 'he slipped', pil 'he fell'), this grammaticalization does not occur; that

is, the verb does not take the l-suffix.42 Interestingly, the same semantic

features characterize the SCD construction in Contemporary Hebrew. The

construction is excluded with verbs denoting actions not controlled by the

Subject, e.g.:

(43) ??hu ma'ad lo barexov

he slipped SCD-SG.3M. in-the-street

The grammaticalization 'chain' in Aramaic thus apparently developed as

follows. The dative preposition l- first spread to the definite-accusative,

and subsequently into the pronominal agreement of all objects, dative and

accusative alike. In addition, as shown above, it also spread into the

Subject-agreement paradigm via a passive participle form indicating the

perfective or past tense.43 Eventually, this pronominal l- construction

became restricted to perfective verb forms with a base derived from the

earlier Aramaic passive participle q†il- type. Verb forms referring to the

present or future, and all imperfect verbs, retained accusative syntax.

Historically, these l- pronominals were not grammatical subjects,

but rather expressions denoting the agent in a passive construction

consisting of the bare form of the passive participle plus an agentive

42 Jastrow (1988) uses the term 'ergative' to describe this construction. Khan (2007) defines it as 'split ergativity', conditioned by the tense/aspect of the verb as well as by its semantic nature. However, the term 'ergative' is not unanimously accepted by Iranists, nor is it normally used in Neo-Aramaic grammar. In Aramaic studies this construction is normally regarded as a preterite and perfect. Iranists mostly refer to it either as passive, or, since the publication of Benveniste (1952), as possessive. 43 Documents from the Achaemenid period – datable to the 5th century BC and published by Driver (1954) – already display (inflected) past-tense verbs suffixed with an l- preposition denoting the agent. Kutscher and Friedrich (1957/8) note the presence of this verbal form and attribute it to the influence of Old Persian.

Page 27: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

prepositional expression.44 The common view among scholars is that this

construction first developed under the influence of Iranian languages,

especially the Kurdish dialects, but subsequently took on a life of its own,

and developed differently from its correlates in the modern Iranian

dialects.45 According to this view, the l- pronominal agent of Neo-

Aramaic corresponds to what Benveniste (1952) analyzes as a possessive

expression in Iranian languages, e.g. Old Persian manā krtam 'I have

done' (literally 'to me/of me done').46 However, in Classical Syriac and

NENA – unlike in Old Persian – the true possessive pronouns are never

used to mark the agent. This suggests that the Neo-Aramaic construction

was not felt to be truly possessive.

Another important difference between the two language-groups –

namely Old Persian and Old Syriac (and subsequently NENA) – concerns

the verbs themselves. In Old Persian, and in later Iranian languages, this

passive (or 'ergative') construction (i.e. the participle form of the verbal

base preceded by a pronoun in an oblique case designating the agent) is

restricted to transitive verbs, whereas in NENA it has also extended to

intransitive past-tense verbs. Therefore, it can no longer be appropriately

described as a passive construction. However, in both language-groups,

an enclitic agent pronoun (from the l- series in Neo-Aramaic and

possessive/adnominal in Persian) eventually became mandatory.

In conclusion, we may assume that in Eastern Aramaic – unlike in

the case of Hebrew and some other modern languages mentioned above –

44 Polotsky (1979:208-209) refers to this as a special case of a more general phenomenon, namely that of an 'objective' expression with l- that may function as the actor of a passive base, in parallel to its function as the undergoer of an active verb. 45 For general accounts of the development of this past-tense form in Eastern Aramaic, see Kutscher (1969), Hopkins (1989), Goldenberg (1992/1998), Kapeliuk (1996) and Khan (1999) , (2004). 46 Polotsky (1979:208 fn. 9) points out that the dative-possessive construction was familiar to scholars of Ancient Egyptian long before the publication of Benveniste's article 'La construction passive du parfait transitif' BSL 48 (1952). The Ancient Egyptian verbal base has been characterized as an undifferentiated 'perfect passive participle', and compared both with the Syriac construction and with the European Perfect with 'have'. For references see Polotsky, ibid.

Page 28: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

the direct precursor of the post-verbal agentive l- pronoun was the dative-

possessive rather than the dative-benefactive. In other words, we may

posit that, due to its essential nature as an element signifying movement

towards a goal, the l-pronoun in this construction signifies the direction

of the action towards the agent as possessor of the accomplished action

and its result. In light of this, it seems reasonable to assume that this

Aramaic syntagm traced the following grammaticalization 'chain':

allative>dative> possessive>perfect/past tense. As regards the

universally defined grammaticalization 'cline' , we have seen that it is a

manifestation of the course of content word> grammatical

word>clitic>inflectional affix.47

5. Conclusions

The different local paths of grammaticalization taken by the originally

allative-dative pronoun has yielded some radical typological changes in

the Semitic languages under investigation, namely Hebrew and Neo-

Aramaic. The evolution of this pronoun has significant impact on the

typological character of Modern Hebrew, since it is part of the shift from

an essentially VSO language to an increasingly 'dative-orientated' one. As

for Neo-Aramaic, we have seen that the qil-li construction

grammaticalized in specific linguistic environments (namely in the

context of the l- pronominal suffixed to the historical passive participle

base), and eventually became a grammatical operator coding the

perfective (of both transitive and intransitive verbs) in the synchronic

system of the language. While the SCD of Biblical Hebrew

grammaticalized almost exclusively with motion verbs, the Neo-Aramaic

construction also grammaticalized with stative verbs, and thus represents

47 According to Hopper & Traugott 2003:7.

Page 29: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

a decategorization of both the passive form and the dative meaning of the

l-pronominal.

As a conclusion, I propose that the Aramaic q†il-li syntagm

represents a local reanalysis of the possessive construction, probably

under the influence of Old Persian. This grammaticalized Neo-Aramaic

construction is part of a radical reorganization and differentiation of the

verbal system in terms of tense, mode and aspect, categories that are very

seldom expressed in other Semitic languages. This reorganization, which

is undoubtedly the result of close contacts with Old Persian-Iranian

languages,48 represents a typological shift from the inflectional

morphology of verb forms, which characterizes the earlier stages of

Aramaic, to an "Indo-Europeanized" system of syntactic morphology (i.e.

periphrastic verbal forms similar to those found in English, German,

French, etc.). However, the strong semantic and morpho-syntactic

interaction between the past participle verb and the agent participant of

the event – the owner of the accomplished action and its result –

prevented the construction from being 'subjectivizable', as it was in

Modern Hebrew.

As shown above, data from African languages like Chadic and

Swahili provide clear evidence that the grammaticalization 'chain'

apparently traced by the Semitic l- pronoun – and thus by implication

perhaps all grammaticalization 'chains' – is accidental (rather than

predetermined by universal principles), for in Chadic and Swahili the link

to the dative is completely absent. One conclusion that may be drawn

from the data presented above is that even when two related languages

display constructions similar in form, such as the l- pronominal agent in

various Semitic languages, the grammaticalization 'chains' of the

construction may have in fact proceeded in different local steps, each

48 Cf. Kapeliuk 1996.

Page 30: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

independent of its counterparts elsewhere. The grammaticalization of the

Semitic SCD pronoun may thus suggest that the apparent 'chains' are in

fact accidental and epiphenomenal, arising only because of the high

probability of two or more strictly local steps.

This suggests that it is not possible to develop a predictive

hypothesis about the grammaticalization 'cline' of a given construction

based on broad cross-linguistic processes or supposed language

universals, just on the basis of surface identity between two or more

languages. Rather, there seem to be language-internal constraints which,

for example, block grammaticalization from proceeding beyond a certain

point. This is clearly demonstrated by the divergent grammaticalization

'chain' of the l- pronominal agent in Semitic languages, which in Hebrew

evolved beyond the level of sentence and grammar via a process of

subjectification, eventually becoming a "genuine" ED, while in NENA,

due to language-specific typological developments, remained on the level

of sentence and grammar.

Finally, I hope to have shown that observations deriving from the

study of Semitic languages may be instructive for typological studies in

general, and may shed light on processes and mechanisms – both

universal and language-specific – which underpin grammaticalization.

References

Autier, J.-Marc and Reed, Lisa. (1992), On the syntactic status of French

affected datives, in: The linguistic review 9.2, 295-311

Ashton, Ethel O. (1947). Swahili, London: Longman's

Bar-Asher, Elitzur A. (2007). The origin and typology of the pattern q†il

li in Syriac and Babylonian Aramaic, in: Maman, Aharon,

Fassberg, Steven E., Breuer, Yochanan (eds.), Sha'arei lashon II

Page 31: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

presented to M. Bar-Asher, Jerusalem: Byalik Institute, 360-392

[In Hebrew]

BDB = Brown, Francis, Driver S. R. and Briggs Charles A. (1966). The

new Brown-Driver-Briggs-Gesenius Hebrew and English lexicon

of the old testament. Oxford: Clarendon

Benveniste, Émile (1952). La construction passive du parfait transitif in:

BSL 48 (1952) 52-62 [reprinted in Benveniste, Émile (1966).

Problèmes de linguistique générale, Paris: Gallimard 176-186]

Berman, Ruth (1982). Dative marking the Affectee role: Data from

Modern Hebrew, in: Hebrew annual review 6, 35-59

Borer, Hagit and Grodzinsky, Joseph (1986). Syntactic cliticization: The

case of Hebrew dative clitics, in: Hagit Borer (ed.), The syntax of

pronominal clitics [Syntax and semantics 19], New York:

Academic Press, 175−217.

Bourciez, Édouard (1946). Éléments de linguistique romane4, Paris: C.

Klincksieck

Brockelmann, Carl (1961). Grundriss der verleichenden Grammatik der

semitischen Sprachen, vol. II, Hildsheim: G. Olms

Contini, Riccardo (1998). Considerazioni sul presunto "dativo etico" in

Aramaico pre-cristiano, in: Amphoux, Christian-Bernard, Frey,

Albert and Schattner-Rieser, Ursula. (eds.), Études sémitique et

samaritaines offertes à Jean Margain [Histoire du texte biblique

4], Lausanne: Editions du Zèbre, 83-94

Driver, Godfrey Rolles . (1954), Aramaic documents of the fifth century

B.C., Oxford: Clarendon Press

Even-Zohar, Itamar (1986). The dialogue in Gnesin and the issue of

Russian models, in: Miron, Dan and La'or, Dan (eds.), Uri Nisan

Gnesin – Me˙qarim u-te'udot [Uri Nisan Genesin - Studies and

Documents], Jerusalem: Byalik Institute, 11-41 [In Hebrew]

Page 32: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

Frazjyngier, Zygmunt (2000). Domains of point of view and

coreferentiality, in: Frajzyngier, Zigmunt & Curl, Tarsi S. (eds.),

Reflexives: Form and functions, Amsterdam and Philadelphia:

Benjamins, 125-152

Geniušienė, Emma S. (1987). Typology of reflexive construction, Berlin:

Mouton de Gruyter

GKC = Gesenius, Wilhelm (1910). Gesenius' Hebrew Grammar as

edited by E. Kautzch, translated into English by A. E. Cowley,

Oxford: Clarendon Press

Glinert, Lewis (1989). The grammar of Modern Hebrew, Cambridge:

Cambridge University Press.

Goldenberg, Gideon (1992). Aramaic perfects, in: Israel oriental studies

12, Tel Aviv: Tel Aviv University, 113-137

Halevy, Rivka (2004). 'Shoteq lo ha-'adon': The function of the

construction 'verb + l + co-agent pronoun' in Contemporary

Hebrew, in: Leshonenu 65 (the academy of Hebrew language) 113-

142 [in Hebrew]

Halevy, Rivka (2007). The subject co-referential l- pronoun in Hebrew,

in: Bar, Tali & Cohen, Eran (eds.), Studies in Semitic and general

linguistics in honor of Gideon Goldenberg, Münster : Ugarit

Verlag, 299-321

Hopkins, Simon (1989). Neo-Aramaic dialects and the formation of the

preterite, Journal of Semitic studies 34, 413-432

Hopper, Paul J. and Elizabeth Closs Traugott (2003).

Grammaticalization. 2nd ed. Cambridge: Cambridge University

Press

Jastrow, Otto (1988). Der neuaramäische Dialekt von Hertervin,

Wiesbaden: O. Harrassowitz

Page 33: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

Jenni, Ernst. (2000). Die hebräische Präpositionen, band 3: Die

Präposition Lamed, Stuttgart-Berlin-Köln: W. Kohlhammer, 48-53

Joosten, Jan (1989). The function of the so-called Dativus Ethicus in

classical Syriac, Orientalia 58 [new series], 473–492

Jöuon, Paul (2003). A grammar of Biblical Hebrew, vol. II, Translated

and Revised by T. Muraoka, /Pontifico Instituto Biblico: Roma

Keenan, Edward L. (1985). Passive in the world's languages, in: Shopen

Timothy (ed.), Language typology and syntactic description, I:

Clause structure, Cambridge-London-New York etc.: Cambridge

University Press, 243-281

Khan, Geoffry (1999). A grammar of Neo-Aramaic: The dialect of the

Jews of Arbel, Leiden & Boston: Brill

Khan, Geoffry (2004). The Jewish neo-Aramaic dialect of Sulemaniyya

and Halebja, Leiden & Boston: Brill

Khan, Geoffry (2007). Ergativity in North-Eastern Neo-Aramaic dialects,

in: Bar, Tali & Cohen, Eran (eds.), Studies in Semitic and general

linguistics in honor of Gideon Goldenberg, Münster : Ugarit

Verlag, 147-157

Kutscher, Eduard Yechezkel and J. Friedrich (1957/8). Zur passivischen

Ausdruckweise im Aramäischen, in: Archiv für Orientforschung

18, 124-25

Kutscher, Eduard Yehezkel (1969). Two 'passive' constructions in

Aramaic in light of Persian, in: Proceedings of the international

conference on Semitic studies in Jerusalem 1965, Jerusalem:

Hebrew University, 132-151

Lazard, Gilbert (1984). Deux questions de linguistique iranienne, in: E.

Benveniste aujourd'hui - Actes du colloque international du

C.N.R.S., Tours 28-30 septembre 1983, Paris: Peeters, 239-248

Page 34: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

Lazard, Gilbert (2004). Légitimité des approches multiples en typologie,

un exemple: Actance et possession, in: Bulletin de la Société de

Linguistique de Paris 99, Paris 107-128

Lyons, John (1977). Semantics, Cambridge: Cambridge University Press

Maldonado, Ricardo (2000). Conceptual distance and transitivity

increase in Spanish reflexives, in: Frajzyngier, Zigmunt & Curl,

Tarsi S. (eds.), Reflexives: Form and functions, Amsterdam and

Philadelphia: Benjamins, 153-185

Monje, Fernando. (1955). Las frases pronominales de sentido

impersonal en Español, in: Archivo de filología áragonesa, VII: 7-

102

Muraoka, Takamitsu (1978). On the so-called dativus ethicus in Hebrew,

in: Journal of theological studies 29, 495-498.

Noss, Philip.A. (1995). The Hebrew post-verbal lamed preposition plus

pronoun, in: The Bible Translator 46.5, 326–335.

Piamenta, Moshe (1981). Selected syntactic phenomena in Jerusalem

Arabic narrative style in 1900, in: Morag, Shlomo, Ben-Ami,

Issachar, Stillman, Norman A. (eds.), Studies in Judaism and Islam

(in honour of S. D. Goitein), Jerusalem: Hebrew University, 203-

230

Polotsky, Hans Jacob. (1979). Verbs with two objects in modern Syriac

(Urmi), in: Israel oriental studies 9, Tel Aviv: Tel Aviv University,

204-227

Schlesinger, Michel. (1928). Satzlehre der aramäischen Sprache des

babylonischen Talmuds, Leipzig: Asia Major

Schwyzer, Eduard and Debrunner, Albert (1950). Griechische

Grammatik, Vol. II: Syntaktische Stilistik, München: C. H. Beck

Traugott, Elizabeth C. (1982). From propositional to textual and

expressive meaning: Some semantic-pragmatic aspects of

Page 35: R halevy Grammaticalization of SCDl - pluto.mscc.huji.ac.ilpluto.mscc.huji.ac.il/.../R_halevy_Grammaticalization_of_SCDl.pdf · Rivka Halevy The Grammaticalization 'Chains' of the

grammaticalization, in: Lehman Winfred P. & Malkiel Yakov

(eds.), Perspectives on historical linguistics, Amsterdam &

Philadelphia: Benjamins, 245-271

Traugott, Elizabeth C. (1989). On the rise of epistemic meanings in

English: an example of subjectification in semantic change.

Language 65: 31-55

Traugott, Elizabeth C. (2003). From subjectification to

intersubjectification. In Joseph, Brian D. & Janda Richard D.

(eds.), The handbook of historical linguistics. Oxford: Blackwell,

624-647

Ullendorff, Eduard. (1985). Along the margins of Agnon's Novel Shira,

in: Robin, Christian (ed.), Mélanges linguistique offerts a Maxime

Rodinson par ses élèves, ses collègues et ses amis , Paris: G.

Geuthner, 393-400

Waltke, Bruce K. and O'Connor Michael P. (1990). An Introduction to

Biblical Hebrew Syntax, Winona Lake Indiana: Eisenbrauns

Williams, Ronald J. (19762). Hebrew Syntax: An Outline, Toronto:

University of Toronto Press