A New Attempt at Reconstructing Proto-Aramaic Part II Sergey Loesov Russian State University for the Humanities, Moscow For my pupils Evgeny Barsky and Maksim Kalinin The paper is the second part of the author’s essay in the Proto-Aramaic reconstruction. In the first chapter of the present contribution, the writer tries to make plausible the idea according to which the Eastern Aramaic plural nominal ending -ē made its appearance in Proto-Eastern Aramaic and had been borrowed by this language from the Assyrian variety of Akkadian. The borrowing hypothesis has been forwarded for want of a better solution. In the second chapter, the author attempts to prove that the Middle Eastern Aramaic resultative constructions qtīl- l- and qattīl- may have started their career already in Proto-Eastern Aramaic. 1.1. The -ē Suffix of the Noun and the Decline of Morphological Definiteness1 To my knowledge, “Proto-Eastern-Aramaic” (unlike “Proto-Aramaic”) is not ein Begriff in Semitic philology.2 This is probably due to two reasons: 1) The postProto-Aramaic development of Aramaic has been never consistently thought through in its entirety. 2) At least in Semitic studies, the “continuum” (i.e., “languages-in-contact/Wellentheorie”) reasoning is generally less demanding than the Stammbaum one: the former does not seem to require stringent pieces of proof (say, of extralinguistic historical nature) to show a contact origin of a grammatical isogloss. If a common feature cannot be readily interpreted as an exclusively shared innovation, people tend to look for a contact explanation. A third reason is that since Huehnergard (1995) the theory according to which “Imperial Aramaic … is the approximate ancestor of the subsequent dialects” (Kim 2008, 509 n.15) has become almost the gospel truth. A much more plausible interpretation has been developed by Klaus Beyer (e.g., Beyer 1966, 1984, 2004). For him, Old and Imperial Aramaic are means of written communication that conceal (“verdecken”) the development of spoken varieties, and this approach is ultimately a weaker claim and therefore more appealing than the majority view that has all the spoken Middle Aramaic languages descend from Imperial Aramaic. 1 Part I of this study is Loesov (2012). In the present Part II, the writer continues his essay in Proto-Eastern-Aramaic reconstruction. The writer acknowledges the support of RFU/РГНФ (project 1101-00142а). I am grateful to Ilya Arkhipov, Evgeny Barsky, Maria Bulakh, Maksim Kalinin, Stephen Kaufman, Leonid Kogan, Aleksey Lyavdansky, and Vera Mostovaya who helped me in various ways with the research done for this paper. 2 Kim (2008, 525) does mention (with disapproval) “the notion of a monolithic ancestral ProtoEastern-Aramaic,” but what he means is a putative common ancestor of all Eastern Neo-Aramaic languages, allegedly located around 1000 AD. 92 Sergey Loesov According to my working hypothesis (probably to be falsified in the course of further research), various salient features of what is traditionally called “(Middle) Eastern Aramaic” are shared innovations that have to be attributed to a postulated “Proto-Eastern-Aramaic.” In Loesov 2012, I have hopefully shown on hand of evidence that the Proto-Eastern-Aramaic verbal prefix l- antedates the emergence of Imperial Aramaic. Same is true of the nominal suffix -ē that we are now going to consider. 1.1.1. The evidence and its explanation I have to say from the outset that for me the -ē suffix is another feature of ProtoEastern-Aramaic. It is the standard masculine-plural-definite nominal ending in the three Middle Eastern Aramaic literary idioms, i.e., Classical Syriac, Classical Mandaic, and Jewish Babylonian Aramaic. Reflexes of this -ē are the plural nominal markers in the whole of Eastern Neo-Aramaic, while they do not show up in Western Neo-Aramaic (Spitaler 1938, 104 ff.). Outside Middle Eastern Aramaic, the -ē suffix appears in Hatran (the dated documents of the corpus were written in 44 BC – 238 AD, see Beyer 1998), in the cuneiform “Uruk” incantation (of unknown date),3 in “Old Syriac” (dated texts were written between the years 6 and 243 AD, Drijvers and Healey 1999). Note that in terms of genealogical and “phasal” subgrouping, these three corpora belong to the Eastern Aramaic of the “pre-Middle” type, and the -ayyā suffix is hardly attested in them at all. Outside Eastern Aramaic proper, the -ē suffix is sporadically attested in various recensions of Imperial Aramaic, starting from the Achaemenid epoch. It is widely believed to show up in mm ‘the nations/peoples’ occurring twice in Ahiqar (TADAE III, 1.1:98, 189, the available copy was written in the 5th century BC). In the Palmyran corpus (dated texts were written in 44 BC – 274 AD), the -ē suffix is much rarer than -ayyā and is used in free variation with it (Cantineau 1935, 123f.; Rosenthal 1936, 76). In Onkelos-Jonathan, -ē appears as a rare allomorph of -ayyā with no self-evident distribution vis-à-vis the latter (Tal 1971, 90ff.; Kaufman 1974, 127 n. 56; Garr 2008; Barsky et al. 2012, 554 f.). Finally, according to Beyer (2004, 59), the -ē suffix was attested in the 7th century BC “in keilschriftlichen aramäischen Ortsnamen.” The simplest explanation of the evidence is as follows: wherever the -ē suffix appears as a less frequent allomorph of -ayyā, the former must be due to an adstrate Eastern Aramaic influence in a Western Aramaic (most probably, Imperial Aramaic) corpus. This is what Cantineau proposed regarding Palmyran: “ - étant la forme traditionelle et - la forme de la langue parlée” (Cantineau 1935, 124; italics added). Onkelos-Jonathan was written in Imperial Aramaic and influenced by a Babylonian adstrate (cf. Nöldeke 1904, 50; Beyer 1984, 35ff.). The consensus 3 “Although the tablet is likely to be Seleucid in date, there is at present no way of accurately dating the script, and the text of this incantation may have been copied from an older Vorlage, e.g. from the Achaemenid period. The language, however, is not Reichsaramäisch” (Geller 1997–2000, 127f.). A New Attempt at Reconstructing Proto-Aramaic 93 opinion has the mm of Ahiqar as a terminus ante quem for the emergence of the ē suffix in a variety of spoken Aramaic, which means that the -ē suffix must anyway be older than creation of the Imperial Aramaic, i.e its inauguration as the official language of the Achaemenid empire.4 As for the cuneiform “Uruk” incantation, Hatran and Old Syriac, they are pure manifestations of Eastern Aramaic (with Geller 1997-2000, Beyer 1984 and 1998). For obvious reasons, the evidence of the allegedly Aramaic proper names in cuneiform Akkadian texts is the least important, yet it enhances the plausibility of the picture. 1.1.2. The etymology of the -ē suffix How does this evidence square with our ideas about the origin of the -ē suffix? There is no consensus about its etymology, except that it was not the masculineplural-definite nominal ending in Proto-Aramaic (save perhaps for the nisba nouns). Three theories have been enjoying support since the late 19th century: 1) *-ayyā > -ē; 2) generalization of the -ē that since prehistoric times had been used to the right of the nisba āy- in the whole of Aramaic; 3) borrowing of the Assyrian masculine plural ending -ē. Theory (1), being the weakest claim, is the most appealing one, but it has no phonological justification. A shift ayyā > -ē is attested nowhere in historical Aramaic, and the last-syllable stress makes it improbable in prehistoric times as well (Rosenthal 1936, 76 n.6; pace Nöldeke 1904 and Cantineau 1931).5 Theory (2) is based on the assumption that kaŝdāyē < *kaŝdāyayyā should be “a natural Aramaic development, a simplification of the overly cumbersome *-ayayyâ” (Kaufman 1974, 128 n. 58). Thus this theory presupposes two unexplained (and to my mind improbable) developments: the ad hoc contraction -ayyā > 4 Beyer (2004, 50) believes that Ahiqar was written in Eastern Aramaic and then rewritten in Imperial Aramaic, but his suggestion is too daring. 5 Since the point is important and I have not found a more detailed argumentation in subsequent literature, I will cite Franz Rosenthal’s words (1936, 76 n.6) in their entirety: Die älteste Ansicht, daß -ê eine jüngere Entwicklung von -ai̭ i̭ â darstelle, ist wegen ihrer sprachlichen Unmöglichkeit aufgegeben. Bei einer Akzentverschiebung auf die Pänultima, wie sie N ö l d e k e , Beiträge 1904, p. 50s aushilfsweise angenommen hat, und Eintritt eines Langvokalabfalls wäre der Übergang von -ai̭ i̭ â zu -ê allerdings zu verstehen; doch sind diese Voraussetzungen eben in alter Zeit nirgends gegeben. Gleichfalls aus einem Akzentwechsel, allerdings in “protoaramäischer” Zeit, läßt die neueste Theorie, die C a n t i n e a u (De la place de l’accent de mot en Hébreu et en Araméen biblique, Bulletin d’études orientales I, 1931 p.81s) vorbringt, die beiden Endungen -ai̭ i̭ â und -ê hervorgehen. Doch gründet sich diese Theorie auf der These von der durchgängigen Pänultima-Betonung des “Protoaramäischen” und erscheint in Verbindung mit einer der schwächsten Stellen dieser These, nämlich dem Faktum der ständigen Ultima-Betonung des st. emph. im Aram., die sich nicht auf einfache Weise in eine Pänultima-Betonung im “Protoaramäischen” venwandeln läßt. Man kann darum für ihre Richtigkeit wie auch für die Existenz einer alten Pänultima-Betonung den Beweis nicht als erbracht ansehen. 94 Sergey Loesov -ē in this particular surrounding (i.e., in nisba nouns) and the subsequent generalization of -ē to combine with all relevant nominal bases.6 I suggest, on the contrary, that Western Aramaic swapped the too heavy -āy-ayyā for -āy-ē by way of a non-phonological replacement, i.e., the -ē suffix was borrowed from Eastern Aramaic at an early date, when an interdialectal borrowing was still an easy thing to do.7 Theory (3), shared by the present writer, is a strong claim,8 therefore it requires typological and historical justifications. The borrowing hypothesis will look more plausible if we relate it to the fact that the postpositive article of Proto-Aramaic was doomed to forfeit its pristine discourse function in the whole of Middle Eastern Aramaic. It is natural to ask whether this shared loss had had its beginnings in the immediate common ancestor of the Eastern Aramaic languages. Turning to typological studies of definiteness: Joseph Greenberg proposes that, cross-linguistically, the definite article can live through three epochs, to wit: the article expresses genuine definiteness (Stage I); the marker in question shifts to the “non-generic article” that does not anymore have to do with definiteness/referentiality but still cannot be used in all and any syntactic surroundings of the noun either (Stage II); the erstwhile article becomes a gender morpheme or a meaningless nominal element (Stage III).9 For Greenberg, one of the best documented examples of this evolution is Aramaic: “A particularly interesting case is Aramaic which, in a recorded history of almost 3000 years, has gone through the three stages mentioned here. An -ā (< *hā) suffixed to nouns in the masculine and feminine, singular and plural, which were already provided with sex/number markers based on the inherited Semitic system, functions as a definite article in the earliest inscriptional language (ninth century BC) and as late as the Aramaic portions of Daniel and Nehemiah (generally dated second century BC). From the early Christian era onward, we have to reckon with two dialect groups, Western and Eastern, in which the former is more conservative in this matter. The Eastern literary dialects (e.g. Syriac, Babylonian Talmudic) have a non-generic article while the Western literary dialects still have a definite article (Christian Palestinian Syriac, Targum Onkelos, etc.). The contemporary dialects of Aramaic still are distinct in this respect, but each group has advanced one further stage. West Aramaic now has a non-generic article while in Eastern Aramaic, it is a general noun marker. One Eastern dialect, that of Tūr-Abdīn, has developed a new prefixed article, this time distinguished for gender and number, thus 6 Kaufman (1974, 128) opts for theory (2) and adds that “the preservation of this morpheme as a characteristic only of Eastern Aramaic might be partly due to Akkadian.” – In this case, the addition of Akkadian into the play does not help understand anything. 7 The present writer carries the onus probandi, i.e. he will have to produce examples of nonlexical interdialectal borrowings in the history of Aramaic. 8 It was first put forward in Bauer-Leander (1927, 204) and most recently supported in Beyer (2004, 50). 9 See Lyons (1999, 337ff.) for a general evaluation of Greenberg’s theory, under the heading “The life cycle of definite articles.” A New Attempt at Reconstructing Proto-Aramaic 95 renewing the process which started in prehistoric times at least 3000 years earlier” (Greenberg 1978, 59f.; reprinted in Greenberg 1990, 250f.). Lyons (1999, 339) remarks regarding Greenberg’s semantic evolution of the exponents of definiteness: “It is far from obvious why a formative with an important discourse function should lose it, and in many cases cease to have any grammatical or semantic function.” The reader of Loesov (2012) will also remember that we have to explain why the speed of the common Aramaic drift has been faster in the Eastern than in the Western branch. I hope that different contact histories of the two branches will help answer both “whys” of the last lines. Given the loss of the article’s expressivity in Eastern Aramaic, what can we say in defence of theory (3)? It is self-evident that there is a ‘hierarchy of borrowability,’ with words for cultural items being among the elements that are most easily borrowed, and bound inflectional morphemes underlying the most stringent constraints” (Pakendorf 2009, 85; with references to earlier literature). A recent textbook on language contact discusses the problem in a way that is important for our study: Inflectional morphology is applied at the sentence level, not at the word level, and so within the framework of the utterance of the recipient language; it does not, by default, accompany individual words, since it is not an inseparable component of the meaning of those words, and hence not directly relevant to the goal for which the word is being borrowed in the first place (namely specificity of reference). This is <…> the main reason why the borrowing of inflectional morphology is rare compared to that of derivational morphology. The one exception is inflectional morphology that accompanies the single word, notably the expression of plurality (Matras 2009, 212; italics added).10 I guess that plurality markers on substantives are borrowable with relative ease11 because their meaning has to do with extralinguistic reality and is therefore “reified”: this meaning is not syntactic (unlike, e.g., the gender value of agreement targets), pragmatic/discursive (nominal definiteness) or otherwise “abstract” (as e.g. modal, temporal, aspectual markers on verbs). Simply put, dog is not the same “fact of life” as dogs. Arabic broken plurals probably provide an inner-Semitic typological parallel to the development I posit for Proto-Eastern Aramaic. If we accept the Central Semitic (= CS) hypothesis, and I believe there are good reasons to do so,12 then Arabic 10 “Gardani (in press) provides a literature-based survey of borrowed inflectional morphology, in which the overwhelming majority of cases involve the marking of plurality on nouns.” This is a fn. by Matras ad loc. The study he refers to is now Gardani 2008. 11 I.e., in comparison to other types of inflectional affixes. 12 See now Kogan (forthcoming, chapter 3), with a research history of the CS hypothesis. Cf. the following cautiously worded judgement in this chapter of the book: “All in all, one is forced to acknowledge that the interpretation of the “South Semitic” broken plurals as innovations as advanced by Diem (1980, 82) and Ratcliffe (1998a, 211–12; 1998b, 119–22) is the easiest and the most appealing way of understanding this phenomenon. Accordingly, it is rather the second possibility outlined above — infiltration into Arabic of innovative “South Semitic” features — that may look a more attractive way of defending the CS affiliation of the Arabic language.” In this lexical study, L. Kogan strongly supports the CS affiliation of Arabic: he endorses, with much caution, the “exclusivelyshared-morphological-innovations” approach of Hetzron, Huehnergard et al. (i.e., the New Imperfec- 96 Sergey Loesov broken plurals are a borrowing from a non-CS Semitic language that used to have broken plurals vom Haus aus (like e.g. Geez). In the case of Arabic, this morphological borrowing can be construed only as a side effect of massive borrowing of nouns, their singular and plural forms finding their way to Arabic from a source “South Semitic” language, so to speak, separately. This is comparable to the way Latin and Greek nominal forms such as datum and data, corpus and corpora, apocryphon and apocrypha entered English. The difference is that English borrowed such words essentially from books to books, while in the case of Arabic the borrowing must have happened “mouth to mouth,” i.e., in the course of a real-time contact of prehistoric Arabs with an unidentified speech community. Of course, our inability to trace down a sizable layer of “South Semitic” nominal loans in Classical Arabic is disappointing, yet one hopes that (if the CS hypothesis is viable) such basic terms with obscure etymology as qalb- (rather than the PS *libb-) ‘heart,’ qamar- rather than *warx-‘moon’ etc. had been borrowed from an otherwise unknown language with a variety of productive broken plural shapes.13 The above case of Arabic broken plurals may sound speculative because theorydependent. Some kinds of broken plurals formation in Tigre produce a more palpable inner-Semitic parallel. Bulakh and Kogan (2011) showed, on hand of a detailed analysis of evidence, two things: (1) certain types of singular-plural correspondences in Tigre are not attested in the rest of Ethio-Semitic yet constitute an isogloss with (a variety of) Arabic; (2) these singular-plural correspondences appear not only in loanwords from Arabic but in the indigenous lexicon of Tigre as well. This is a clear-cut case of morphological borrowing, i.e. (depending on one’s theoretical stance) borrowing of inflectional morphemes. As I have just mentioned, all Eastern Aramaic varieties display the decline of status, i.e. collapse of the Proto-Aramaic nominal definiteness paradigm, while it is fully alive in the whole of Middle Western Aramaic and even nowadays its vestiges hold on in Western Neo-Aramaic. I propose to relate this double-edged grammatical innovation of the Eastern Aramaic noun (vis-à-vis the Proto-Aramaic one)14 to the contact of Eastern Aramaic with the Neo-Assyrian variety of Akkadian. It follows from our historical records that Assyrian may have been a substrate language for Proto-Eastern-Aramaic. Assyrian (as the rest of Akkadian) did not have morphological means to encode definiteness, so the Eastern Aramaic loss of definiteness as a value of the nominal inflection could be attributed to the substrate influence. As I suggest, the Assyrian masc. pl. suffix *-ē replaced (save for certain monosyllabic bases) the Aramaic masc. pl. definite ending *-ayyā: this may have been partly due to a putative Imālah pronunciation of *-ayyā (i.e., somewhat akin to *-ē), but mostly because *-ayyā is the only Proto-Aramaic inflectional marker of the noun that looks completely alien (in terms of its phonological shape) to the Astive isogloss) and then introduces his own findings related to innovations in the basic lexicon exclusively shared by NWS and Arabic. 13 See now Kogan (forthcoming, ch. 3) for various issues related to innovations in the basic lexicon of Arabic. 14 I.e., -ē rather than - ayyā and the decay of morphological definiteness. A New Attempt at Reconstructing Proto-Aramaic 97 syrian repertoire of nominal inflection in the plural (compare the table in Loesov 2012, 439 with Hämeen-Anttila 2000, 77 ff.), therefore it may have been unusual and difficult for the Akkadian-Aramaic bilinguals. This replacement, coupled with the insufficient sensitivity of Assyrian-Akkadian bilinguals to the morphological encoding of definiteness, will have triggered promotion of other “full” (i.e., definite) endings into the indiscriminate use and the consequent decline of morphological definiteness.15 That is to say, accepting Greenberg’s commonsense diachronic typology of definiteness markers, we can explain the speedier bleaching (i.e., desemanticization) of the Eastern Aramaic definite endings by the contact of ProtoEastern-Aramaic with Akkadian. If the borrowing pedigree of the -ē suffix is accepted as plausible, it has to be juxtaposed (or even confronted) with the probable suggestion that Akkadian more or less stopped being spoken as a mother tongue during the early first millennium BC, and this happened due to infiltration of Arameans rather than conquest. This means that within a few generations speakers of Akkadian had been progressively loosing this language. Schematically: children used to speak Akkadian worse than their parents, grandchildren had only a passive command of Akkadian, the grandgrand-<…>children would be unable to follow the performance of a native speaker of Akkadian. Sociolinguistically, this is what “Aramaization of Assyria” means. We know next to nothing about the actual progress of this Aramaization,16 but we have to posit that this was a process wherewith Akkadophones were losing Akkadian and acquiring Aramaic. Speculatively, one can reckon with alternative scenarios, those of a catastrophic development. There could have occurred a mass extermination of Akkadophones (around the fall of the Neo-Assyrian Empire, 614-609 BC), one could also imagine Akkadophones being confined to isolated language 15 Let me adduce one of the best-known examples of suffixed markers of definiteness losing their force. Proto-Slavic cliticized a declinable demonstrative pronoun (with the base *jь) to declinable attributive adjectives in order to express the definiteness of their heads: *mǫžь dobrъ ‘a good man’ vs. *mǫžь dobrъ-jь ‘the good man’ (nom. ms); *žena slĕpa ‘a blind woman’ vs. žena slĕpa-ja ‘the blind woman’ (nom. fs). This method of coding definiteness gave rise to two types of adjectival inflections, “long” (i.e., definite) and “short” (i.e., indefinite). In the long paradigms, the postpositive article fused in due course with nominal endings to its left. In most of individual Slavic languages, the long paradigms ousted the short ones, and this brought an end to the morphological expression of definiteness. Thus, in Russian short forms of adjectives lost (except for lexicalized rests) inflection for case and are now used as predicates or adverbials. 16 Simo Parpola’s explanation is perhaps too rich in detail, I am not sure we do know that much: “By the end of the 8th century BC, the provincial system covered the entire Levant from Palestine to central Iran, and it was further expanded in the seventh century. At this time Aramaic was already spoken all over the Empire and Assyrian imperial culture had been dominant everywhere for centuries. The Aramaization of Assyria was calculated policy aimed at creating national unity and identity of a kind that could never have been achieved, had the Empire remained a loose conglomeration of a plethora of different nations and languages. And it did pay off. Even though Akkadian retained its position as the language of the ruling elite and cuneiform script continued to be used for prestige purposes, Aramaic soon also became part and parcel of the imperial administration. It was by no means the only language of the subjected peoples but equal in status with Akkadian, and eventually it became the language of the ruling class as well” (http://www.betnahrain.net/1History/Parpola1.htm), cf. also Parpola (2007) to the same effect. 98 Sergey Loesov pockets and gradually dying out (Hackl forthcoming), yet all this does not square well with our historical data, among other things because Neo-Assyrian political elite included speakers of Aramaic (cf. the Assurbrief KAI 233). It is a commonplace (mentioned above in connection with Arabic broken plurals) that plural markers are normally borrowed from the lexicon. Now Kaufman (1974) counts some one hundred fifty Akkadian loanwords in the whole of Aramaic, and I believe not much would have changed had they been a few more hundreds. The absence from our Syriac dictionaries of a sizeable layer of Akkadian loans for trivial notions proves definitely that the spoken Aramaic of the early 1st millennium BC did not possess a lot of Akkadisms, or at least they had failed to penetrate deeply into the Wortschatz of the language. Yet, if Assyrian Akkadian was indeed a substrate of Proto-Eastern-Aramaic, the low number of lexical Akkadisms in Aramaic can be easily explained: the Akkadian speech community was melting down, the winning Aramaic did not feel a need to renew its basic vocabulary by borrowings from the weaker (although culturally superior) partner, yet massive bilingualism may have impaired a fine-tuned domain of grammaticalized pragmatics in Aramaic, i.e. the obligatory expression of definiteness / referentiality.17 Curiously, the number of Aramaisms in our Akkadian corpus is close to that of Akkadisms in Aramaic (von Soden 1966-77; Abraham and Sokoloff 2011). Hackl (forthcoming, n. 59) observes: In von Soden 1977 a total of 282 Aramaic loanwords was proposed (collected from NeoAssyrian, Neo-Babylonian and Late Babylonian texts). This number has now been reduced to 85 certain/possible loanwords by Abraham/Sokoloff 2011. Be this as it may, one thing is certain: once upon a time, Akkadian was replaced by Aramaic in the whole of Mesopotamia, and the scenario of a “peaceful” death of a language (as was the case of Akkadian, according to the consensus opinion) implies of necessity heavy lexical borrowing into the dying language, and usually grammatical borrowings as well (see countless examples in typologically oriented studies of the 20th century AD endangered spoken languages). In so far as the written Akkadian corpus of the first millennium BC does not display a massive and manifold influence of Aramaic, this will be because the writers of this corpus did their best to stick to what they considered to be a standard variety of Akkadian.18 Yet it looks like the grammatical pressure of Aramaic as the mother tongue of the writers of Akkadian was sometimes irresistible. An example is the marking of direct object. Akkadian had probably lost most of its erstwhile distinctions between the nominative and the accusative case of the noun soon after 1500 BC, thus leaving the difference between the subject and the direct object morphologically un17 Compare the modest amount of substrate words in Ibero-Romance languages, or of Gallic (Celtic) words in French. In the meantime, it is likely that the indigenous populations of Iberia and Gallia that would ultimately abandon their languages in favour of a form of Latin outnumbered the conquerors who had come from the Apennine Peninsula. 18 The problem has been successfully obscured in the recent research (Hackl forthcoming). It merits a close study by a scholar equally devoted to Aramaic and Akkadian philology. A New Attempt at Reconstructing Proto-Aramaic 99 marked, and this is a state of affairs intolerable for most of Aramaic throughout its history, at least in so far as the pragmatically definite nominal direct object goes. As is well known, in Neo-Assyrian and Neo-Babylonian the preposition ana ‘to, for’ sometimes appears as a marker of direct object (nota accusativi). In this function, ana is a loan translation of the Aramaic preposition *la19 with the basic motion meaning ‘to(wards).’ In Aramaic, as in various other West Semitic languages, *la developed into the regular marker of indirect object (semantically speaking, this is addressee/recipient). As we will learn from a forthcoming part of this study, in the whole of Eastern Aramaic (and probably already in Proto-Eastern-Aramaic), *la further became a marker of direct object,20 having completely ousted from this slot the Proto-Aramaic nota accusativi * iyyāt, still holding on, in one way or another, through the whole of Middle Western Aramaic. At a certain point of Aramaic-Akkadian contact (no doubt when there were already a large number of bilinguals) this accusative *la was loan-translated into Akkadian as ana. The earliest instance in the core Akkadian I know of obtains in the Middle Assyrial Laws:21 mussa … ana šuāša u līdānīša-ma ilaqqe-šunu ‘her husband … shall take her and her children’ (A V 12-14). Driver and Miles (1935, 476) observe that in this sentence ‘ana introduces the direct object which is resumed by the pron. suffix’, i.e. by a bound “accusative” pronoun: ilaqqe-šunu ‘he shall take-them’.22 Since ana as nota accusativi can hardly be a result of inner-Akkadian development, it is likely that the wording of this clause is due to Aramaic-Akkadian interference. In the peripheral Akkadian, even earlier examples of this kind are known: a-na an-né-tim be-lí lu-ú i-de ‘May my lord know this’ (ARM XXVI 312: 35’).23 The expected wording is an-ni-tam/an-né-tim be-lí lu-ú i-de (cf. e.g. ARM 2, 22:32). ki-i-me a-na-ku a-na LUGAL be-lí-i[a] a-ra-aḫ-am ‘Just as I love the king, my lord’ (EA 53:41, a letter from Qatna) Returning to the Akkadian influence on Aramaic: despite the low number of lexical borrowings from Akkadian, Aramaic (both Old and Middle) has two productive derivational morphemes almost certainly borrowed from Akkadian: the nominal abstract suffix -ū(t) and the causative verbal prefix š-/s-. The -ū(t) suffix is highly expansive, to the degree of becoming “parasitisch” (Barth 1894, 415), while š-/s- is hardly attested with more than a dozen Aramaic roots (Loesov 2009, 490f.; a review of data gleaned from reference tools). 19 GAG § 114e, AHw 48b. It is likely that in this function *la was initially used with personal and definite/referential nominal direct objects. 21 Our main extant texts were copied in the 11th century BC, the original is believed to be created in the 14th century (Roth 1995, 154). 22 I am grateful to Leonid Kogan for alerting me about this example. 23 I am grateful to Ilya Arkhipov for alerting me about this example. 20 100 Sergey Loesov Given this evidence and the above typological considerations, the borrowing of the plural nominal ending -ē from Akkadian into Proto-Eastern-Aramaic does not look as improbable as it would seem on a first sight. 1.2. *Qatīl- l- Is No Longer Passive. Both *qatīl- l- and *qattīl- Are Resultative The non-active adjective *qatīl- may have existed already in Proto-Semitic, although it is only parsimoniously represented in Akkadian. Proto-Aramaic promoted *qatīl- to the passive participle of the G-stem, thus drastically enhancing its grammatical relevance. This was the first step along the path of grammaticalization that ended up by *qatīl- becoming the base of the Eastern Neo-Aramaic Preterit in the G-stem. As is well-known, the pragmatically non-passive qtīl- l- predicate is attested in all three Middle Eastern Aramaic literary corpora (see the reference grammars). The active-diathesis reading of qtīl- l- in Syriac is proven beyond doubt by the fact that the NT Peshitta is able to use qtīl- l- to translate past-time active transitive sentences of the Greek original: Peshitta was not happy with the Syriac Preterit qtal alone, and this was because Eastern Aramaic had already developed qtīl- l- as a periphrastic conjugation encoding resultative for transitives. Thus there is no reason to believe (pace Bar-Asher Siegal 2011) that qtīl- l- was invariably passive in Middle Eastern Aramaic (be it Syriac, JBA or Mandaic).24 As I tried to show in Loesov 2011, 424 ff., non-passive qtīl- l- may have appeared in Aramaic simultaneously with the resultative construction for intransitives, *qattīl-, which means that properly speaking the shared innovation of Eastern Aramaic is the birth of resultative/perfect as an independent aspectual category. Naturally enough, the newborn member of the Eastern Aramaic tense-aspect repertoire had two guises, *qatīl- l- and *qattīl-, depending on the transitivity value. Consider a few NT examples for the ergative conjugation of the verbs bd ‘to do’ and psq ‘to decide’; 25 as we will see, the original Greek sentences are active transitive: (1) hānnā dēn meddem da-snē lā ḇīḏ lęh (Lk 23:41) ὐ ἄ π ἔπ α ε (aor. act.) ‘(For we receive the due reward of our deeds): but this man hath done nothing amiss’ (KJV). (2) w- en saḵlūṯā ḇīḏā lī (Acts 25:11) ἀ ῶ α ἄ α υ πέπ α ά (perf. act.) ‘(For if I be an offender,) or have committed any thing worthy of death’ (KJV). (3) psīq-hwā lęh gēr l-pawlos d-ne brīh l- ep̄ esos (Act 20:16) 24 Yet I share wholeheartedly Bar-Asher Siegal’s feeling that the ergative conjugation of Eastern Aramaic is not (contra Kutscher 1969 and much of the subsequent literature) a loan translation from Iranian, see Kogan and Loesov (2012, 579–84). 25 I thank my doctoral student E. Barsky for permission to use examples of predicative adjectives from his dissertation in progress. Kutscher (1969) has six additional NT examples, see for their analysis Ciancaglini (2008, 35). E. Barsky works among other things on a corpus-based description of qtīl- l- in Syriac. 101 A New Attempt at Reconstructing Proto-Aramaic ε ί ε ὁ Παῦ πα απ ῦ α Ἔφ (pluperf.) ‘For Paul had determined to sail by Ephesus’ (KJV). The reverse equivalence is known as well: in Greek translations of Syriac Vorlagen, qtīl- l- can be rendered by the transitive aorist. Consider an example from the opera of Isaac of Nineveh, a seventh-century ascetic writer.26 (4) sb d ktybn fem. pl. hwy fem. pl. lh l s klhyn d-qylyth ktybt fem. pl. mš lpt fem. pl. (Bedjan 1909, 407) Γέ ἔγ α ε υ ῦ υ αὐ ῦ υ … π υ π υ (Pirard 2012, 647; aor. act.) ‘A certain Father had written on all the walls of his cell various things’ (tr. from Syr. Wensinck 1969, 273). ktybn hwy lh … ktybt mšḥlpt = ἔ α υ π υ π υ , that is to say the Greek philologist who translated the Syriac text (probably in the 10th century) understood the construction as an active-transitive one, with lh as the exponent of the agent. We will have to juxtapose these examples of the resultative qtīl- l- with morphosyntactically (= in terms of morphological semantics) comparable tokens of predicative qattīl- of certain basic motion verbs: (5) talmīḏawhy azzīlīn-hwaw la-mḏittā (Jn 4:8) α α αὐ ῦ ἀπε ύ ε σα π (pluperf.) ‘For his disciples were gone away unto the city’ (KJV). (6) w-lā attī-hwā lwāṯhon yēšu (Jn 6:17) α ὔπ ἐ ύ ε π αὐ ὁἸ ῦ (pluperf.) ‘And Jesus was not come to them’ (KJV). (7) w-saggīyē men ihūḏāyē attī īn-hwaw lwāṯ mārtā w-maryam (Jn 11:19) π ἐ ῶ Ἰ υ α ἐ ύ ε σα π Μ α α Μα (pluperf.) ‘And many of the Jews came to Martha and Mary’ (KJV). (8) nāšā gēr menhon men ru qā attī īn (Mk 8:3) α αὐ ῶ ἀπ α ἥ ασ (pres./perf. ἥ ‘For divers of them came from far’ (KJV). ) (9) meṭṭol d-ḏaywē saggīyē allīlīn-hwaw bęh (Lk 8. 30) ὅ εἰσῆ ε α απ αὐ (aor.) ‘Because many devils were entered into him’ (KJV). (10) w- ezzlaṯ l-ḇaytāh w- ešk aṯ barṯāh kaḏ ramyā b- arsā w-nappīq mennāh šē ḏāh (Mk 7:30) α ἀπ ῦ α ἶ αὐ ῆ πα ἐπ α α ἐ ε υ ό (perf. part.) ‘And when she was come to her house, she found the devil gone out, and her daughter laid upon the bed’ (KJV) In the Syriac text of Mk 7:30, unlike in the Greek original with its accusativuscum-participio clause ( … α ἐ υ ), the predicative adjective nappīq stands in an independent clause (as in the preceding examples), i.e., it does not depend syntactically on eškḥaṯ .. kaḏ and functions as a finite verb form. 26 The example is courtesy Maksim Kalinin, my doctoral student. 102 Sergey Loesov It is worth noting that the Greek Vorlagen of examples (2) – (3), (5) – (8) and (10) display verb forms of the morphological Perfect aspect (including a participle), while examples (1) and (9) have aorists. The Perfect base of the Greek verb (even in the Hellenistic period) is known to have a pronounced resultative clang that is valid for the pluperfect form as well. Zerwick (1963, 98) remarks about the pluperfect in the NT Greek that “it indicates a past state of affairs constituted by an action still further in the past … the pluperfect is not used simply because the action denoted was prior to the past time of the main verb or the narration in general (as Latin or English would use the pluperfect).” Yet, as we have seen, NT Peshitta shows no one-to-one correspondence between qattīl of intransitive motion verbs and the perfect-aspect shape of intransitives in the Greek Vorlage, which entails that the appearance of qattīl in the above examples is meaning-oriented, it is no blind “translation technique” in the spirit of the kaige-revision or Aquila. The resultative *qatīl- l- is well-known in both JBA and Mandaic (Nöldeke 1875, 381 ff.). If my suggestion about the new Resultative making its first steps already in Proto-Eastern-Aramaic is correct, one expects the intransitive resultative *qattīl- to appear in these corpora as well. No morphosyntactic analysis of deverbal adjectives in Classical Mandaic is available (the predicative qattīl- had not been described in the syntactic part of Nöldeke’s Syriac grammar either, unlike *qatīl- l-), yet the paragraph on qattīl- in Nöldeke’s nominal morphology of Mandaic sounds promising: Eine in den verwandten Sprachen seltene, im Aram. stark um sich greifende Adjectivbildung, die freilich im Mand. noch nicht so zahlreiche Vertretung hat wie im Syr., wo sie besonders dient, um zu vielen intransitiven Verben Verbaladjectiva zu bilden. Aus der grossen Zahl führen wir an <examples follow>’ (Nöldeke 1875, 123 f.). As for JBA, reference tools do not help, since they do not offer a discussion of *qattīl- in this corpus, yet there seems to be a certain number of resultative/pasttime *qattīl- tokens of basic motion verbs. It may well be that in this respect the situation in JBA was not radically different from that in Syriac. Consider the following examples for three basic motion verbs, corroborating the above Syriac evidence:27 ‫נ‬ ‫נ‬ ‫נ‬ ‫נ‬ ‘So he said to them: Wait for me, because I have forgotten a golden dove. He went back and said grace and found a golden dove’ (BT Berakoth 53b, tr. Talmud Soncino). ‫נ‬ ‫נ‬ ‘Perhaps some other man after his decease went and told them’ <i.e., he made something known to the inhabitants of the netherworld>28 (BT Berakoth 18b, tr. Talmud Soncino).29 ‫נפ‬ 27 ' ‫נ‬ ‫נע‬ ‫נ‬ ' They are courtesy Maksim Kalinin as well, BT is cited according to the Vilna Edition. Cf. Jacob Newsner’s translation: ‘“Perhaps someone else died and went and told them what had happened.” 29 may turn out to be a *qattīl- predicate as well, lit. ‘he died and went,’ but this is a moot point since in JBA the verb škb has the e-vowel in the suffixing conjugation (DJBA 1142). 28 A New Attempt at Reconstructing Proto-Aramaic 103 ‘For R. iyya b. Abba said: It is unseemly for a scholar to go abroad in patched sandals. — Is that so? Did not R. iyya b. Abba go out in such?’ (BT Berakoth 43b, tr. Talmud Soncino). At the time of the discussion, R. iyya b. Abba was already an authority of a remote epoch. ‫ע נפ‬ ‫ע‬ ‘And the reason why he does not do so is because Scripture said: ‘And he shall go out unto the altar,’ i.e., until he has gone outside the whole altar’ (BT Yoma 58b, tr. Talmud Soncino). ‫נ ע‬ ‫ע‬ ‫פ‬ ‫נ‬ ‘You say: Why has not Messiah come? Now to-day is the Day of Atonement and yet how many virgins were embraced in Nehardea!’ (Yoma 19b, tr. Talmud Soncino).30 Orthographically, in the JBA corpus can of course stand for both active participle and *qattīl-. Yet in the text the meaning is doubtless that of resultative/perfect, and in the previous four examples the forms are unlikely to be active participles, so the cumulative evidence makes it plausible that this is another token of *qattīl-. As I suggested in the first part of this study, the productive *qattīl- for adjectival concepts is likely a Proto-Aramaic innovation (Loesov 2012, 429). To illustrate: according to my hypothesis, in Proto-Aramaic, a notion like ‘needy, deficient’ was probably *xassīr rather than *xasir (as against Biblical Hebrew). Now then, when one looks at the material gathered in Nöldeke’s grammars of Syriac and Mandaic and at the examples cited in the present paragraph, one’s impression is that both *qatīl- l- and *qattīl- may have started their activity of resultative predications already in Proto-Eastern-Aramaic, while the outcomes of this innovation are different in individual daughter-languages of Proto-Eastern-Aramaic, as the evidence of the Eastern Aramaic idioms spoken in the 19th – 20th centuries AD shows. In this respect, the case of Mandaic may be particularly instructive. Nöldeke’s list (1875, 124) of *qattīl- forms has only stative-adjectival concepts, such as e.g. z dyq ‘gerecht,’ h kym ‘weise,’ q lyl ‘leicht.’ If the picture reflects faithfully the state of affairs in Mandaic spoken around the year 500 AD, this would mean that the evolution of verb in Mandaic was slower than in other Eastern Aramaic idioms, and as a consequence the new resultative forms has never taken root in this variety. Spoken neo-Mandaic of the 20th century still had the reflex of *qatal- as the basic exponent of the past time. Abbreviations ARM = Archives royals de Mari. Paris BT = Babylonian Talmud CS = Central Semitic DJBA = M. Sokoloff. A Dictionary of Jewish Babylonian Aramaic of the Talmudic and Geonic Periods. Ramat-Gan, 2002 EA = J. A. Knudtzon. Die El-Amarna Tafeln. Leipzig, 1915 JBA = Jewish Babylonian Aramaic 30 Except that the Soncino translation has “You have said,” which perhaps does not do full justice to the conjugated participle of the original. 104 Sergey Loesov KAI = H. Donner, W. Röllig. Kanaanäische und aramäische Inschriften. 3 Bde. Wiesbaden, 1962– 1964 TADAE = B. Porten; A Yardeni. Textbook of Aramaic Documents from Ancient Egypt. 4 vols. Jerusalem, 1986-1999. References Abraham, Kathleen, and Michael Sokoloff. 2011. “Aramaic Loanwords in Akkadian: A Reassessment of the Proposals.” Archiv für Orientforschung 52:22–76. Bar-Asher Siegal, Elitzur A. 2011. “On the Passiveness of One Pattern in Jewish Babylonian Aramaic—A Linguistic and Philological Discussion.” Journal of Semitic Studies 56:111–43. Barsky, Evgeny, and Maria Bulakh, Sergey Loesov, Aleksey Lyavdansky, Nikolay Seleznyov. 2011. Review of Aramaic in Its Historical and Linguistic Setting, by Gzella Holger and Margaretha L. Folmer, eds. Babel und Bibel 6:545–564. Bulakh, Maria, and Leonid Kogan. 2011. “Arabic Influences on Tigre: A Preliminary Evaluation.” Bulletin of the School of Oriental and African Studies 74:1–39. Barth, Jakob. 1894. Die Nominalbildung in den semitischen Sprachen. Leipzig: Hinrichs. Bauer, Hans, and Pontus Leander. 1927. Grammatik des Biblisch-Aramäischen. Halle, Saale: Niemeyer. Bedjan, Paul. 1909. Mar Isaacus Ninevita de perfectione religiosa. Parisus, Lipsiae: Harrassowitz. Beyer, Klaus. 1966. “Der reichsaramäische Einschlag in der ältesten syrischen Literatur.” Zeitschrift der Deutschen Morgenländischen Gesellschaft 116:242–54. Beyer, Klaus. 1984. Die aramäischen Texte vom Toten Meer samt den Inschriften aus Palästina, dem Testament Levis aus der Kairoer Genisa, der Fastenrolle und den alten talmudischen Zitaten. Göttingen: Vandenhoek und Ruprecht. Beyer, Klaus. 1998. Die aramäischen Inschriften aus Assur, Hatra und dem übrigen Ostmesopotamien (datiert 44 v. Chr. bis 238 n. Chr.). Göttingen: Vandenhoek und Ruprecht. Beyer, Klaus. 2004. Die aramäischen Texte vom Toten Meer samt den Inschriften aus Palästina, dem Testament Levis aus der Kairoer Genisa, der Fastenrolle und den alten talmudischen Zitaten. Bd. II. Göttingen: Vandenhoek und Ruprecht. Cantineau, Jean. 1931. “De la place de l’accent de mot en Hébreu et en Araméen biblique.” Bulletin d’études orientales I:81–98. Cantineau, Jean. 1935. Grammaire du palmyrénien épigraphique. Paris: Le Caïre. Ciancaglini, Claudia. 2008. Iranian Loanwords in Syriac. Wiesbaden: Harrassowitz. Diem, Werner. 1980. “Die genealogische Stellung des Arabischen in den semitischen Sprachen. Ein ungelöstes Problem der Semitistik.” In Studien aus Arabistik und Semitistik, A. Spitaler zum 70. Geburtstag, edited by Diem Werner and Stefan Wild, 65–85. Wiesbaden: Harrassowitz. Drijvers, Han, and John Healey. 1999. Old Syriac Inscriptions of Edessa and Osrhoene: Texts, Translations and Commentary. Leiden and Boston: EJ Brill. Driver, Godfrey R., and John Miles. 1935. The Assyrian Laws. Oxford: Clarendon Press. Garr, W. Randall. 2008. “The Determined Plural Ending -ē in Targum Onkelos.” In Aramaic in Its Historical and Linguistic Setting, edited by Gzella Holger and Margaretha L. Folmer, 173–206. Wiesbaden: Harrassowitz. Gardani, Francesco. 2008. Borrowing of Inflectional Morphemes in Language Contact. Frankfurt am Main etc.: Peter Lang. Geller, Mark. 1997–2000. “The Aramaic Incantation in Cuneiform Script (AO 6489 = TCL 6, 58).” Jaarbericht ex Oriente Lux 35–36:127–146. A New Attempt at Reconstructing Proto-Aramaic 105 Greenberg, Joseph H. 1978. “How Does a Language Acquire Gender Markers?” In Universals of Human Language, edited by Joseph H. Greenberg, Charles A. Ferguson and Edith A. Moravcsik. Vol. 3, 47–82. Stanford: Stanford University Press. Greenberg, Joseph H. 1990. On language: selected writings of Joseph H. Greenberg. Edited by Keith Denning, Suzanne Kemmer. Stanford: Stanford University Press. Gzella, Holger, and Margaretha L. Folmer, eds. 2008. Aramaic in Its Historical and Linguistic Setting. Wiesbaden: Harrassowitz Verlag. Hackl, Johannes. Forthcoming. “Language Death and Dying Reconsidered: the Role of Late Babylonian as a Vernacular Language.” In Proceedings of the St. Petersburg Session of the 53rd Rencontre Assyriologique Internationale Held in Moscow and St Petersburg, edited by Cornelia Wunsch. http://iowp.univie.ac.at/sites/default/files/IOWP_RAI_Hackl.pdf/. Hämeen-Anttila, Jaakko 2000. A Sketch of Neo-Assyrian Grammar. Helsinki: State Archives of Assyria, Studies XIII. Huehnergard, John. 1995. “What is Aramaic?” Aram 7:261-82. Pirard, Marcel. 2012. Abbā Isaak tū Syrū Logoi askētikoi. Hagios Oros: Hiera Monē Ibērōn. Kaufman, Stephen. 1974. The Akkadian Influences on Aramaic. Chicago: University of Chicago Press. Kim, Ronald. 2008. “Stammbaum or Continuum? The Subgrouping of Modern Aramaic Dialects Reconsidered.” Journal of the American Oriental Society 128:505–31. Kogan, Leonid. Forthcoming. Genealogical Classification of Semitic. The Lexical Isoglosses. Kogan, Leonid, and Sergey Loesov. 2012. Review of Iranian Loanwords in Syriac, by Claudia Ciancaglini. Babel und Bibel 6:575–86. Kutscher, Eduard Y. 1969. “Two ‘Passive’ Constructions in Aramaic in the Light of Persian.” In Proceedings of the International Conference on Semitic Studies, Jerusalem, 19–23 July 1965, edited by Ze‘ev Ben- ayym, Aharon Dotan and Gad Sarfatti, 132–151. Jerusalem: Israel Academy of Sciences and Humanities. Loesov, Sergey. 2009. “The Aramaic Languages.” In Languages of the World: The Semitic Languages. Akkadian. Northwest Semitic, edited by Anna Belova, Sergey Loesov, Leonid Kogan, Olga Romanova, 414–495. Moscow: Academia. (In Russian.) Loesov, Sergey. 2012. “A New Attempt at Reconstructing Proto-Aramaic. I.” Babel und Bibel 6:421– 56. Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. Matras, Yaron. 2009. Language Contact. Cambridge: Cambridge University Press. Nöldeke, Theodor. 1904. Beiträge zur semitischen Sprachwissenschaft. Strassburg: K.J. Trübner. Nöldeke, Theodor. 1875. Mandäische Grammatik. Halle: Buchhandlung des Waisenhauses. Pakendorf, Brigitte. 2009. “Intensive contact and the copying of paradigms – an Ėven dialect in contact with Sakha (Yakut).” Journal of Language Contact 2: 85–110. Parpola, Simo. 2007. “The Neo-Assyrian Ruling Class. Beihefte zur Zeitschrift für die Alttestamentiche Wissensch No 374.” In Studien zu Ritual und Sozialgeschichte im Alten Orient = Studies in Ritual and Society in the Ancient Near East: Tartuer Symposien, 1998–2004, edited by Thomas R. Kämmerer, 257–274. Berlin and New York: Walter De Gruyter. Ratcliffe, Robert. 1998a. The “Broken” Plural Problem in Arabic and Comparative Semitic: Allomorphy and Analogy in Non-Concatenative Morphology. Amsterdam, Philadelphia: John Benjamins. Ratcliffe, Robert. 1998b. “Defining Morphological Isoglosses: The ‘Broken’ Plural and Semitic Subclassification.” Journal of the Near Eastern Studies 57:81–123. Rosenthal, Franz. 1936. Die Sprache der palmyrenischen Inschriften und ihre Stellung innerhalb des Aramäischen. Leipzig: Hinrichs. 106 Sergey Loesov Roth, Martha. 1995. Law Collections from Mesopotamia and Asia Minor. Atlanta: SBL. Soden, Wolfram von. 1966–1977. “Aramäische Wörter in neuassyrischen und neu- und spätbabylonischen Texten: Ein Vorbericht. Teil I.” Orientalia 35:1–20 [1966]; “Teil II.” Orientalia 37:261–271 [1968]; “Teil III.” Orientalia 46:183–197 [1977]. Spitaler, Anton. 1938. Grammatik des neuaramäischen. Dialekt von Ma lūla (Antilibanon). Leipzig: F. A. Brockhaus. Tal, Abraham. 1971. The Language of the Targum of the Former Prophets and Its Position within the Aramaic Dialects. Diss. Hebrew University, Jerusalem. Wensinck, Arent. 1969. Mystic Treateses by Isaak of Nineveh, translated from Bedjan’s Syriac Text with an Introduction and Registers. Wiesbaden: Harrassowitz. Zerwick, Max. 1963. Biblical Greek. Rome: Editrice Pontificio Istituto Biblico.