Encoding of the Direct Object throughout the History of the Aramaic Language* Maksim Kalinin [email protected] Sergey Loesov [email protected] Russian State University for the Humanities, Moscow The present paper is the first part of a comprehensive inquiry into direct object encoding within the whole of Aramaic, including Modern Aramaic. Both nominal and pronominal direct objects are considered. The paper starts with a theoretical introduction dealing with differential object marking. It is followed by sections analyzing the expression of the direct object in eighth-century B. C. texts (Old Western Aramaic), Mesopotamian Aramaic of the Neo-Assyrian period, and Egyptian Aramaic of the Achaemenid period. The study has reached new results along the well-trodden research paths through the early monumental inscriptions and Egyptian Aramaic. In particular, the functions of the pre-nominal direct ob- ject markers !yt and l- have been redefined. Keywords: Differential object marking, history of Aramaic, Sefire inscriptions, Mesopotamian Aramaic, Egyptian Aramaic, direct object markers !yt and l- 0. Introduction The direct object (= DirObj) of a verb can be expressed by a noun or a pronoun. In this study we will describe the ways the verb interacts with both kinds of DirObjs in all pre-modern Aramaic varieties and touch upon the situation in Modern Aramaic. We will not discuss clausal com- plements. A speculative calculus of syntactic slots and their fillings, based on one’s general acquaintance with Aramaic, looks as follows: 1) DirObj is a noun: it is introduced by a prefixed differential object marker (= DOM) or by Ø. The choice may depend (among other things) * We thank Peter Arkadiev, Ilya Arkhipov, Eitan Grossman, Stephen Kauf- man, and Aaron Rubin for their comments and suggestions. We are grateful to Boris Aleksandrov, Ilya Khait, and Olga Vinnichenko who provided us with some otherwise unavailable scholarly materials. A special thanks goes to Kenny Bendik- sen for helping us improve our writing. 2 Articles: Semitic Studies on animacy and/or definiteness of the DirObj. The verb may take a bound object pronoun indexing the nominal DirObj that is marked with a DOM (the qatl-eh l-malkā ‘he killed-him to-the-king’ construction). For the purposes of this study, DOM is an analytical marker that flags certain nominal DirObjs to the exclusion of others, which remain unmarked. The concept was first introduced in Bossong 1983–1984 and Bossong 1985. (Note that in the linguistic literature DOM often means “differen- tial object marking” as well and may refer to various kinds of markers, e. g., to case endings.) 2) DirObj is a pronoun: it is bound (or “hosted”) on a DOM or on the verb itself. In the latter case, in certain Aramaic varieties there may ap- pear linking elements between the verb and the pronoun. The best- known and most controversial of them is “the presuffixal nun,” usually called “nun energicum.” In the study, we look at the ways these possibilities take shape in indi- vidual Aramaic idioms. One last word before we start: pre-modern Aramaic had two analytical DOMs, *!iyāt and l-. This fact invites two comments. First, we will see that the functions (or “meanings”) of the two Aramaic DOMs are different, i. e., they are not synonymous syntactic devices. This is important for the study of differential object marking in the world’s languages, as well as for the understanding of individual Aramaic varieties. Second, *!iyāt and l- do not co-occur as productive prenominal markers in the same dialect. For this reason, their dialectal distribution is significant for research into the historical dialectology of Aramaic, in particular into the genealogical classification of Aramaic languages. 0.1. Theoretical Preliminaries on DOM The inquiry will not make a lot of sense should we not wonder why a language may ever need an analytical DOM. Those languages whose nouns possess the accusative case (= ACC) rarely develop an analytical ad- nominal DOM, and the complementary distribution of ACC and DOM is the traditional wisdom in Semitic philology. Thus Rebecca Hasselbach notes as a matter of course that the appearance of a DOM in Old West- ern Aramaic “indicates that at least nominative and accusative must have been indistinguishable morphologically (Hasselbach 2007:103, and see now Hasselbach 2013). We have found so far only one clear-cut counter- example in the world’s languages. According to a p. c. of Wolfgang Schulze, “in Old Armenian, the accusative plural was marked for the M. Kalinin, S. Loesov, Encoding of the Direct Object … 3 morphological case (-s). The preposition z-1 was normally added to defi- nite nouns in DirObj function, e. g. (1) NOM.PL osker-k ‘bones’ ACC.PL osker-s [–def.] z-osker-s [+def.].” Schulze describes the situation as follows: “The main function of the ‘nota accusativi’ z- in Old Armenian was to indicate definiteness/specificity/typ- icality. It does not seem to be restricted to animate, human nouns, as can be seen from the following example: (2) yev sksav l3vanal z-ot-s ašakert-ac-n and he began to wash (l3vanal) z-foot-PL.ACC pupil-PL.GEN-DEF.3 ‘And he began to wash the feet of the disciples’ (John 13:5).” According to a p. c. of Giorgio Iemmolo, who is working on a typologi- cal monograph “Differential Object Marking,” the situation in Old Arme- nian may be unique. He mentions Geez as another language which pos- sessed both prenominal DOM (la-) and the ACC case ending (-a) on the same token of DirObj, yet we believe that the examples that may have shown up in the sources are errors against good usage: DOM la- and the ACC case ending -a ought not to be compatible, because Geez considers la- to be syntactically a preposition (cf. I.3.0 below, an introduction to l- in Egyptian Aramaic), and genuine prepositions do not govern the ACC in Semitic. Cf. also Schneider 1959:3: “Le complément d’objet du verbe se met normalement à l’accusatif. Mais dans de nombreux cas cet accusatif est remplacé par une construction analytique: le verbe s’adjoint un pro- nom suffixe qui se rapporte au complément et celui-ci est introduit par la préposition la-.” Thus what is characteristic of Geez (and probably unique to it within Semitic) is that this language uses in the DirObj slot either the DirObj-indexing qatal-o la-n3guś (‘he killed-him to-the king’) construction or the accusative case, and their distribution depends on se- mantic/pragmatic features of the respective DirObj. (Note that the nomi- nal case in Geez is residual: -a accusative vs. -Ø non-accusative.) The situa- tion in Geez is special due to this distribution of a retention (a case ending) and an innovation (la-flagging coupled with DirObj-indexing on the verb). As we will see in the course of our inquiry, in Aramaic the DirObj-indexing (once more coupled with the use of an analytical DOM) appears as late as in 1 Lazard 2001:875 notes, “la valeur ancienne de z- en arménien classique était sans doute ‘par rapport à’” (Meillet 1936:94), and this is a good source for a DOM. 4 Articles: Semitic Studies certain Middle varieties (e. g., Syriac and Jewish Babylonian Aramaic), while the Aramaic noun had lost its case endings already in pre-historic times.2 Thus the DOM usually appears due to a loss of synthetic ACC but does not replace the latter as a syntactic marker. In comparison to the lost syn- thetic ACC, a younger analytical DOM has a more concrete and language- specific semantic load. The universals of DOM semantics are well-known. The key word will be salience of the flagged DirObj, routinely mani- fested as animacy and definiteness.3 As for Semitic, the oldest varieties of Akkadian and Arabic had the ac- cusative case and no DOMs, while after the loss of the NOM–ACC case dis- tinction both groups of languages did develop DOMs.4 Little is known about the Akkadian DOM ana, whose primitive mean- ing is ALLATIVE–DATIVE ‘to(wards), for.’ Cf. a literature review in Rubin 2005:105, to which add Luukko 2004:169, and see I.3.1.2.2 below. It is generally believed that the DOM ana served to oppose the Agent to the semantic Object, yet the reader will notice that our reference tools on Neo-Assyrian content themselves with reproducing the same set of three to five examples, without much detail or analysis. Hämeen-Anttila 2000: 77 notes: “When the parts of the sentence are not sufficiently indicated by word order and congruence, or when the object is particularly empha- sized, the preposition ana may be used as a nota accusativi.” Luukko 2004: 2 In our study, the DirObj-indexing is an umbrella notion for what descriptive grammars call “resumptive pronouns” and “anticipatory pronouns.” See now a short essay by Eitan Grossman on double accusative marking in general and on the importance of distinguishing between DirObj flagging (= “marking”) and in- dexing: http://dlc.hypotheses.org/446. 3 Eitan Grossman comments on this text in a p. c.: “Salience is a pretty fuzzy no- tion. Again, according to Iemmolo, object markers tend to develop starting from topical objects. Topic is a more specific and better defined notion than salience.” The first and the third sentences of this comment are obviously correct. Yet it looks like what the present writers need at this stage of research is a fuzzy umbrella-con- cept, not a well-defined one. (And it comes to mind that salient DirObjs will often be topical.) Note that “animacy,” unlike both “salience” and “topicality” (and partly unlike “definiteness”) is a binary lexical-semantics feature of a hardware nature, i. e., it does not depend on syntax and pragmatics. To understand our Aramaic data, we have to start with something more flexible than “animacy.” To see what we mean the reader may want to have a look at exx. (42)–(43) and (45)– (46) in Section I.3.1.3. 4 On differential object marking in Tigrinya, see now Kievit 2009. On differ- ential object marking in other Semitic languages not mentioned in this paper, see the respective sections in Rubin 2005. M. Kalinin, S. Loesov, Encoding of the Direct Object … 5 169 claims that “the use of /ana/ was not merely restricted to necessary con- texts, but was possibly (sic!—M. K., S. L.) spreading and becoming the gen- eral marker of the accusative in Neo-Assyrian.” This big claim is supported by no single example. Finally, it is often said that “this use of ana can un- doubtedly be attributed to Aramaic influence” (Rubin 2005:105), yet no piece of evidence in support of this idea has been ever produced. The Arabic picture is known in broad strokes; it has hardly ever been seriously studied in terms of syntactic semantics. On li-/la- as a DOM in spoken Arabic varieties, see a survey in Rubin 2005:105ff. (with lit.). On fi-, cf. the following p. c. of Stephan Procházka: “A characteristic of most dialects of the Maghreb is that direct objects can be marked by the prepo- sition fi- (Procházka 1993:126f.), e. g. Tunis: µắll f-ǝ̆lfnîq ‘he opened the jewelry box (gingerly).’ To my knowledge, there are no in-depth studies of this syntactical phenomenon for Contemporary Tunisian Arabic. The two comprehensive grammars on Tunisian Arabic present relatively vague explanations: Singer 1984:624 maintains that fi- is used for imper- fective actions, while Cohen 1975:248 says: ‘Il [fi-] peut introduire le complément de certains verbes normalement transitifs. Dans ce cas, il apporte une nuance d’action habituelle, durable ou répétée.’ For Moroc- can Arabic see Harrell 1962:209. There is also evidence for fi- as a direct object marker in Palestine and Cairo.” The pristine repertory of nominal cases in Semitic is rudimentary. Nouns in the singular have three cases: NOM encodes the subject, ACC en- codes DirObj, the rest of the syntactic participants are encoded by prepo- sitional phrases, and all prepositions govern their dependent nouns in GEN.5 (GEN codes all kinds of adnominal arguments as well.)6 Hence nom- inal DirObj in older Semitic varieties gets compulsory marking whatever the semantic relationship between the finite verb and the entity encoded by DirObj. In other words, in Semitic languages that preserve case end- ings, the syntax of the transitive clause is insensitive to its contents in so far as the choice of case goes. Consider the verb phrase ‘to cultivate a/the field,’ ubiquitous in Old Babylonian letters, eqlam šipram epēšum , e. g. inf. 5 Personal pronouns of the most archaic Semitic varieties may keep traces of a more ancient picture. Thus Old Babylonian personal pronouns have tonic (non-cliti- cized) DAT shapes, which are still sometimes used by themselves rather than governed by the semantically “dative” preposition ana (Huehnergard 1997:272f., 606). 6 I. e., if we disregard the noun phrases headed by an adjective which is de- fined by a substantive in the accusative, as in the case of the Arabic tamyīz/accusa- tive of specification: µasanun waǧh-an ‘pretty-faced ’. ACC 6 Articles: Semitic Studies (3) eql-am šipr-am ul īpuš field-ACC work-ACC NEG (ul) do-PRET3cs. ‘He has not done the work on the field/he has not plowed the field’ (see CAD Š 77). 3 Epēšum ‘to do’ is a “light verb,” which means its contribution to the lexical semantics of the verb phrase is modest: the burden of the predi- cate’s lexical meaning is carried by the complement šiprum ‘work’ in ACC. Thus both the nominal part of the compound verb šipram-epēšum and the genuine semantic object eqlum ‘field’ indiscriminately receive the ACC marking. Semantic bleaching of ACC is plainly manifest in the “adverbial accusative” of the Semitic languages that still preserve their case endings: the “adverbial accusative” encodes various kinds of non-core participants (or “adjuncts”), e. g., those associated with a verb of intransitive motion.7 After the loss of case markers (say, due to phonetic erosion), the lan- guage may undertake something to tell its subjects from its DirObjs. The thriftiest method is to keep a rigid word-order whatever the semantics of the verb and DirObj, the way it happens in English. Other languages (even those whose word-order is impenetrable to pragmatics) supply cer- tain types of DirObjs with a phonetically embodied DirObj marker. In other words, a cliticized DirObj marker is hardly ever a 100% functional replacement of the lost ACC, therefore its appearance opens a possibility for the form of the transitive clause to be fine-tuned to its semantics. Since the prototypical subject of a transitive clause is thematic (i. e., definite) and volitional (i. e., animate), it is definite animate DirObjs that most likely need a DOM in order to appear as formally distinct from the subjects (cf. Lazard 2001:879f., with lit., the now classical expression of this view being Comrie 1979). Peter Arkadiev and Eitan Grossman tell us that nowadays this “distinguishing arguments” theory does not hold sway in typological studies and that some people think it is topical DirObjs that require special marking more often that anything else. Eitan Gross- man informs us that sentences with definite animate Agents and equally definite animate Patients constitute around 3% or less in most corpora checked. Yet the “distinguishing arguments” approach squares well with part of our Aramaic data, so we will stick to it for the time being. As we observed above, definite animate DirObjs are but the best ex- amples of a cognitively salient patient-like participant. Salience is a prag- matic rather than an ontological feature (cf., e. g. Givón 2001:373f.), so 7 See Fischer 1987:173 for examples from Classical Arabic. M. Kalinin, S. Loesov, Encoding of the Direct Object … 7 so we expect the speaker to manipulate DOM (within certain boundaries) depending on what she wants to get across to the interlocutor. A good example is Standard Spanish with its veo a un niño classroom rule, accord- ing to which the DOM a is used to introduce each and every animate DirObj (i. e., the latter’s definiteness value is supposed to be irrelevant). Yet there are enough examples for animate DirObjs introduced without a, as well as those for inanimate DirObjs introduced by a. The use of the DOM a in Spanish has received lots of scholarly attention. Bernard Pot- tier showed back in 1968 (Pottier 1968) that the use of the DOM a may depend on the semantic properties of the verb (i. e., on what we now call the degree of its semantic transitivity), on the way the situation affects the subject and DirObj, and on the “singularisation dans l’intention” of the DirObj, i. e., on its pragmatic salience. Consider an example: (4) Porque yo perdí a mi madre el año pasado y ese dolor todavía lo tengo igual, me pregunto como será si en el fu- turo perdiera un hijo lo soportaría?8 The question received several answers. They show that the asker’s ut- terance is acceptable for other users of the question-and-answer site, since they employ similar turns of phrase in their comments. The an- swerers seem to understand “un hijo” as non-referential. Now consider a contrasting pair of examples with an inanimate DirObj (and inanimate subjects which are synonyms in the two sentences): (5) a. En 2005 el huracán Stan atacó la región del Lago de Ati- tlán y los efectos fueron enormes. Muchas vidas y hoga- res fueron destruidos.9 b. Hoy domingo 17 de abril a las 14 en Av. de Mayo y Bolí- var se realizará un festival solidario por el pueblo japo- nés, víctima de terremoto y tsunami que atacó a la región el pasado 11 de marzo.10 Both kinds of usage are frequent in regional varieties of Standard Spanish, so we may to surmise that in this particular case the choice lies with the speaker and is pragmatically conditioned. “The speaker’s choice” is the catchword of this Introduction. If we stay with Spanish, in certain cases it leaves the speaker no choice at all. 8 http://es.answers.yahoo.com/question/index?qid=20100125144502AAnbXTv. 9 http://weguatemala.org/es/ong/mayan-families?page=5. 10 http://agendagratisenbuenosaires.blogspot.ru/2011_04_01_archive.html. 8 Articles: Semitic Studies Thus ‘I saw Maksim’ may be translated into Standard Castilian only as ‘Vi a Maksim,’ ‘Le vi a Maksim,’ or ‘A Maksim le vi,’ etc.; i. e., in this case the DOM a may not be omitted (Nueva gramática 2009:2631, § 34.8e). This is ontology as yesterday’s pragmatics, to paraphrase Givón’s dictum today’s morphology is yesterday’s syntax. (The emergence of this hard-and-fast rule in Spanish is easily explainable by semantic and pragmatic properties of referential personal names.) Yet in other cases the speaker of Spanish has the discretion to use or omit the DOM a, she is able to manipulate it to her own satisfaction (see, e. g., Nueva gramática 2009:2630–2637 and von Heusinger–Kaiser 2007, with lit.). As we will see, this pragmatic rather than purely ontological/hardware approach helps explain the use of various DOMs across the history of Aramaic. Part I: Western Aramaic In the sections that follow we will discuss the encoding of DirObj in the below corpora and languages: 1. Eighth-Century Inscriptions in Old Western Aramaic11 2. Mesopotamian Aramaic of the Neo-Assyrian period 3. Egyptian Aramaic of the Achaemenid period 4. Biblical Aramaic 5. Nabataean Aramaic 6. Palmyrene Aramaic 7. Dead Sea Scrolls Aramaic 8. Targum Onqelos-Jonathan Aramaic 9. Jewish Palestinian Aramaic 10. Samaritan Aramaic 11. Christian Palestinian Aramaic 12. Modern Western Aramaic Mini-Review of Literature Our study is meant to be a part of a future “History of Aramaic.” We dis- cuss the synchronic DirObj encoding in each of the above corpora in some philological detail, while our ultimate concern is to understand the 11 An earlier version of this section was published in “The Classification of Se- mitic Languages: Archaism and Innovation. Proceedings of the V Meeting on Comparative Semitics, Córdoba, 06/6-8/2012,” ed. by Juan Pedro Monferrer-Sala and Wilfred G. E. Watson (pp. 45–57). M. Kalinin, S. Loesov, Encoding of the Direct Object … 9 evolution of Aramaic. In this respect, the inquiry does not have exact predecessors. What comes nearest in terms of scope and subject-matter is Rubin 2005:94–105. Aaron Rubin describes in general features the nomi- nal (and partly pronominal) DirObj encoding in the whole of pre-mod- ern Aramaic, as well as DirObj markers in other Semitic languages. The piece is a part of the author’s monograph on “grammaticalization in Se- mitic.” The aim of Rubin’s book is to impart the concept of grammaticali- zation to students of Semitic and to show general linguistic readership that Semitic languages display instances of grammaticalization, in particu- lar of the kinds that are familiar cross-linguistically. Other works which deal with DirObj marking in various Semitic lan- guages including Aramaic are Geoffrey Khan’s “Object Markers and Agree- ment Pronouns in Semitic Languages” (Khan 1984) and Holger Gzella’s “Differentielle Objektmarkierung im Nordwestsemitischen als Konvergenz- erscheinung” (Gzella 2013). Rebecca Hasselbach touches upon the subject in her monograph “Case in Semitic: Roles, Relations, and Reconstruction” (Hasselbach 2013). Of these three, Geoffrey Khan’s contribution is the most informative and rich in detail. As for DirObj encoding in individual Western Aramaic varieties, be- sides reference grammars there exist a number of in-depth studies. Mu- raoka 1992 and Folmer 1995:340–371 are detailed inquiries into the ways DirObj is expressed in Qumran and Achaemenid Aramaic, respec- tively. Both of them have been indispensable in the course of our re- search, as the reader will see in the corresponding sections. Folmer 2008 is a study of DOM in what she calls “Jewish Palestinian Aramaic Inscrip- tions.” On pp. 132ff. of this contribution Folmer offers an insightful lit- erature review, to which we refer the reader for additional bibliographic data. Florentin 1991 is a study of bound object pronouns in Samaritan Aramaic. 1. Eighth-Century Inscriptions in Old Western Aramaic12 1.1. DirObj is a noun The only pre-nominal DOM used in the corpus is !yt. In the corpus, !yt appears twice as a host for pronominal DirObjs (I.1.2), while pre-nominal 12 We do not include here the Tell Fexeriye inscription (KAI 309), since accord- ing to scholarly consensus it represents a different variety of Aramaic. In our view, TF is the first now available textual testimony of Old Eastern Aramaic (Loesov 2012). We will look at the DirObj encoding of this text later in our study. 10 Articles: Semitic Studies !yt is attested at least eleven times. It appears in the two extant Northwest Syrian documents of any appreciable length, KAI 202 (ZKR) and 222– 224 (Sefire), and in a peripheral inscription (KAI 320, Bukan).13 It is missing from the contemporaneous peripheral inscriptions written in Old Western Aramaic and found in Zincirli (KAI 216–218).14 Consider all eleven extant examples: ZKR (1) w-hwsp[t …] !yt kl mµgt […] ‘And I added <to a previously mentioned city> a whole cir- cle of […]’ (KAI 202 B 4f.).15 (2–3) w-bnyt !yt […] !yt !pš ‘I (re)built [a GN and] Afis’ (KAI 202 B 10f.). (4–5) w-k[tbt b]h !yt !šr ydy [w-kl m]n yhg" !yt !š[r ydy] Zkr mlk Ñm[t] ‘And [I wrote on] it the story of my achievements.16 [Now, whoever] destroys17 the story [of achievements] of ZKR king of Hamath …’ (KAI 202 B 14–17). (6) […] !š! w-!yt […..] š[r]šh ‘[Execute?] the man and […] his folk’ (KAI 202 B 27f.). Sefire (7) l-t!th b-µylk w-![tm l-t!]twn b-µylkm l-šgb b[y]ty [w-hn "q]r[k l-]y!th l- šgb !yt "qr[y] ‘<If> you do not come with your army and [you do not] sg. come with your armies to strengthen my house, [and if pl. your offspring does not] come to strengthen (!yt) [my] off- spring’ (KAI 222 B 31ff.). (8–9) !hbd !yt ktk w-!yt mlkh 13 We do not take into account assumed tokens of !yt restored in partly broken portions of texts, e. g., KAI 310:4, 10 (Tell Dan). 14 The Zincirli vernacular was not Aramaic (cf. KAI 214–215), while the site of KAI 320 was at the time not a Semitic-speaking area at all. 15 The translation follows Gibson 1975:11. The exact meaning of mµgt is un- known, cf. DNWSI 611, with ref. to previous literature. 16 The translation follows Gibson ibid. The exact meaning of !šr is unknown, cf. DNWSI 128 (!šr), with ref. to previous literature. 5 17 DNWSI 715.