Greek loanwords in Rabbinic literature: reflections on current research methodology

Timothy Michael LAW and Alison SALVESEN GREEK SCRIPTURE AND THE RABBIS LEUVEN – PEETERS – WALPOLE, MA 2012 PARIS TABLE OF CONTENTS Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII EMANUEL TOV Post-Modern Textual Criticism? . . . . . . . . . . . . . . . . . . 1 LORENZO CUPPI The Treatment of Personal Names in the Book of Proverbs from the Septuagint to the Masoretic Text . . . . . . . . . . . . . . . . 19 TIMOTHY MICHAEL LAW Kaige, Aquila, and Jewish Revision . . . . . . . . . . . . . . . . 39 MICHAEL GRAVES Midrash-Like Word Plays in Aquila’s Translation of Genesis . . 65 TIMOTHY EDWARDS Aquila in the Psalter: A Prolegomenon . . . . . . . . . . . . . . 87 ALISON SALVESEN Did Aquila and Symmachus Shelter under the Rabbinic Umbrella? 107 TESSA RAJAK Theological Polemic and Textual Revision in Justin Martyr’s Dialogue with Trypho the Jew . . . . . . . . . . . . . . . . . . 127 WILLEM F. SMELIK Justinian’s Novella 146 and Contemporary Judaism . . . . . . . 141 REINHART CEULEMANS Greek Christian Access to ‘The Three’, 250–600 . . . . . . . 165 CE JULIA G. KRIVORUCHKO Greek Loanwords in Rabbinic Literature: Reflections on Current Research Methodology . . . . . . . . . . . . . . . . . . . . . . 193 SHIFRA SZNOL Text and Glossary: Between Written Text and Oral Tradition . . 217 VI TABLE OF CONTENTS Index of Modern Authors . . . . . . . . . . . . . . . . . . . . . . 233 Index of Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . 239 Index of Biblical References, Patristic and Rabbinic Literature . . 243 Selective Index of Greek Renderings . . . . . . . . . . . . . . . . 255 Selective Index of Hebrew Words and Phrases . . . . . . . . . . . 257 GREEK LOANWORDS IN RABBINIC LITERATURE: REFLECTIONS ON CURRENT RESEARCH METHODOLOGY1 Julia G. Krivoruchko The paper starts with a short historical account of the research on Greek loanwords in rabbinic literature, beginning with the period of Wissenschaft des Judentums. Particular emphasis is placed on research practices of the last quarter of the twentieth–early twentyfirst century. The procedures of reconstructing Greek prototypes and etyma are analysed in the light of corpus linguistics and text linguistics data. Examples of published research on Greek loanwords are adduced to illustrate the methodological analysis. The paper seeks to improve the possibilities of correct identification of Greek vocabulary by warning against the unrealistic and ahistorical approaches current in the field. The study of Greek loanwords in post-biblical Hebrew and Aramaic texts is a field with an edifying history. Having accumulated a certain expertise in dealing with multiple problems of great complexity, it is now a lively and developing discipline that can profit from methodological generalisations. What follows is an attempt to see this scholarly domain through the eyes of a Greek linguist. 1. BRIEF HISTORICAL OUTLOOK OF THE GREEK LOANWORD STUDIES 1.1 Wissenschaft des Judentums: laying the foundations While etymologies of Greek loans have appeared in Talmudic lexicography since the Middle Ages, they did not attract much attention from Christian scholars until the Renaissance. Not until the mid-nineteenth century were Greek loanwords placed on the agenda of semitologists. 1 This article draws upon the ideas expressed earlier in Julia G. Krivoruchko, ‘O perspektivah izuweniq greweskih zaimstvovaniî v ravvinistiweskoî literature [O perspektivakh izucheniia grecheskikh zaimstvovanii v ravvinisticheskoi literature] (On the Prospects of Researching Greek Loanwords in Rabbinic Literature),’ Vestnik Evrejskogo Universiteta 2 (20) (1999), pp. 22–40. 194 J.G. KRIVORUCHKO Pioneers of this field came from Germany, traditionally strong in classical studies. Yet Greek and Latin loanwords would have remained an esoteric pursuit of few, if not for the political developments in post-1848 Europe. With the emancipation of Jews, Wissenschaft des Judentums embarked on the task of demonstrating that Jewish culture constitutes an integral part of the European cultural realm. And since the classical Greek heritage was perceived as the very core of Europeanness, there was no better way of proving the Jews’ allegiance to the ‘civilised world’ than showing that they partook in Greek culture. The traces of the ‘Hellenic spirit, the emancipator of nations’ were looked for – and found – in Talmud and Midrash.2 Moreover, it was claimed that Jewish studies could enrich Greek lexicography by discovering new lexemes and meanings.3 The fact that minor publications that emerged before the mid-1890s4 are now largely unknown, even to specialists, can be credited to Samuel Krauss (1866–1948). Born in the tiny train station of Ukk, he studied at the rabbinical seminary at Budapest and the local university (1884–1889), and further in Berlin. He received his doctorate at Giessen at the age of 27 and a rabbinical diploma from the Budapest seminary a year later. Towards the end of his doctoral studies on loanwords, Krauss accumulated substantial material for his magnum opus, ‘Griechische und Lateinische Lehnwörter im Talmud, Midrasch und Targum’, which generalised and systematised the data of his predecessors.5 The first volume of the work consisted of a grammatical sketch, and the second of a dictionary of loanwords, in which many new etymologies were suggested. Both volumes took only five years to complete, and they were proudly described on the cover page as the ‘Preisgekrönte Lösung der Lattes’schen Preisfrage’. A significant contribution to the dictionary has been made by Immanuel Löw (Szeged 1854–Budapest 1944), whose fingerprints can be dis2 3 4 5 Leopold Zunz, ‘The Geographical Literature of the Jews’, in Adolf Asher (ed.), The Itinerary of Rabbi Benjamin of Tudela, (London–Berlin, 1841), vol. 2, p. 303. Samuel Krauss, Griechische und Lateinische Lehnwörter im Talmud, Midrasch und Targum. Mit Bemerkungen von I. Löw, (Berlin, 1898–1899), vol. 1, pp. 200–205. Michael Sachs, Beiträge zur Sprach- und Altertumsforschung: aus Jüdischen Quellen, 2 Hefte, (Berlin, 1852–1854); Adolf Brüll, Fremdsprachliche Redensarten und ausdrücklich als fremdsprachlich bezeichnete Wörter in den Talmuden und Midraschim (Leipzig, 1869); Julius Fürst, Glossarium Graeco-Hebraeum, oder der griechische Wörterschatz der jüdischen Midraschwerke: ein Beitrag zur Kultur- und Altertumskunde (Strassburg, 1890); idem, ‘Quelques mots midrashiques empruntés au grec’, Revue des Études Juives 23 (1891), pp. 129–131. Krauss, Lehnwörter, vol. 1, pp. XXXIV–XLI. GREEK LOANWORDS IN RABBINIC LITERATURE 195 tinguished on most articles of the ‘Lehnwörter’. According to different calculations, the Wortlist of Krauss contained between 2260 and 2652 lexemes.6 Löw disagreed with Krauss on approximately eight hundred of them, that is, almost a third of cases,7 and greatly improved the final product. The dictionary should also have borne his name. The immediate response of the learned community to the ‘Lehnwörter’ ranged from appreciative but restrained to devastatingly critical. Many reviewers seem to have perceived it as a hasty publication of a promising but immature scholar. It has been pointed out that many readings suggested by Krauss originate from unreliable manuscripts, that certain Greek etyma cannot have the meanings that were ascribed to them, that phonetic reconstructions were not trustworthy, etc.8 In his classical ‘Introduction to Talmud’ (1920), H. Strack advised: ‘the new edition of the dictionary of loans in Talmud and Midrash is a necessity’.9 Contrary to what could be expected, such reception of Krauss’s work did not lead to its being quickly forgotten, nor promptly replaced by a better counterpart. Half a century later, G. Zuntz and H. Rosen continued to censure Krauss with the same zeal:10 ‘It would, however, be rash if the non-specialist, in consulting it, were to take his results for final and unquestionable, while in fact a highly alert scepticism is called for at every point.’11 ‘A great proportion of the proposed Greek 6 L. de Nobiscu, Review of Samuel Krauss, Griechische und Lateinische Lehnwörter im Talmud, Midrasch und Targum. Mit Bemerkungen von I. Löw. Bd. 1, (Berlin, 1899), Wiener Zeitschrift für die Kunde des Morgenlandes 17 (1903), p. 195, numbers 2260 words, while Daniel Sperber, ‘Prolegomena to a New Dictionary of Classical Words in Rabbinic Literature’, repr. in Essays on Greek and Latin in the Mishna, Talmud and Midrashic Literature (Jerusalem, 1982), p. 5, counted 2652. According to Günther Zuntz, ‘Greek Words in the Talmud’, JSS 1 (1956), pp. 132, the dictionary includes 2370 Greek and 240 Latin words. ‘Circa tremila lemmi’ in Marco Mancini, ‘Appunti sulla circolazione del latino nella Palestina del I secolo d.C.’, in R. Lazzeroni et al. (eds), Diachronica et Synchronica, Studi in onore di Anna Giacalone Ramat (Pisa, 2008), p. 294 is a rough approximation. 7 See preface to the index of Krauss, Lehnwörter, vol. 2, p. 622. 8 Among the important reviewers were Siegmund Fraenkel in Zeitschrift der Deutschen Morgenländischen Gesellschaft 52 (1898), pp. 290–300; 55 (1901), pp. 353–358; Felix Perles in Byzantinische Zeitschrift 8 (1899), pp. 539–546; 10 (1901), pp. 300–306; Leopold Cohn in Monatsschrift für Geschichte und Wissenschaft des Judentums 44 (1900), pp. 561–570; C. Levias, ‘Greek and Latin Loan-Words in Talmud, Midrash, and Targum’, The American Journal of Semitic Languages and Literatures 16 (1900), pp. 190–192. 9 Hermann L. Strack, Introduction to the Talmud and Midrash (New York, 1969), p. 6. 10 Zuntz, ‘Greek Words’, pp. 129–140; Haïm B.Rosén, ‘Palestinian koinß in Rabbinic Illustration’, JSS 8 (1963), pp. 56–72. 11 Zuntz, ‘Greek Words’, p. 129. 196 J.G. KRIVORUCHKO equivalents, whether new or traditional, is doubtful or even definitely wrong.’12 For the lack of a better reference tool, the ‘Lehnwörter’ was reprinted by Olms in 1964, and when in 1982 G. Stemberger revised Strack’s ‘Introduction’, it was still referred to as ‘a pioneering labour… badly in need of revision’.13 The need for revision being widely recognised,14 Krauss’ work became the most prominent milestone of the field and its universal point of reference. 1.2 Twentieth century: the epoch of minor notes and new apologetics Krauss and Löw did not abandon their studies of Talmudic realia upon completion of the ‘Lehnwörter’.15 However, they never returned to the linguistic part of the inquiry. Neither contributor was eager to bring their joint product to perfection, and although Krauss continued to publish on topics of Greek and Roman interest, he went on to research Jewish poetry, medieval philosophy, modern Jewish history, Mishnah, and Jewish-Christian contacts. In contradistinction to contemporaries, later generations were more lavish in their praise of the ‘Lehnwörter’ and less prone to ‘alert scepticism’. The impact of Krauss as a cultural historian overshadowed his flaws as a linguist. ‘Samuel Krauss’ insights in many areas of Jewish scholarship, expressed through his often-pioneering research, have truly passed the test of time’, wrote Steven Fine in the introduction to the volume dedicated to the fiftieth anniversary of Krauss’s death.16 12 13 14 15 16 Zuntz, ‘Greek Words’, p. 131. Hermann L. Strack and Günter Stemberger, Introduction to the Talmud and Midrash, transl. by Markus Bockmuehl (Cambridge, 1991). ‘A new dictionary of Greek (and Latin) loans and loan translations is an urgent need’: Eliezer Y. Kutscher, ‘The state of research of the Rabbinic Hebrew, in particular lexicography, and its functions’ (in Hebrew), in Archive of the New Dictionary of the Rabbinic Literature 1 (Ramat-Gan, 1972), p. IX; ‘… a dire necessity’: Saul Lieberman, ‘Defective Transliteration of Greek Words in Rabbinic Literature’, JQR 73 (1982), p. 62; cf. Sebastian Brock, ‘Greek Words in Syriac: Some General Features’, Scripta Classica Israelica 15 (1996), pp. 251–252. Among their most important publications were Samuel Krauss, Talmudische Archäologie, 3 vol. (Leipzig, 1910–12); idem, Synagogale Altertümer (Berlin, 1922); Immanuel Löw, Die Flora der Juden (Wien–Leipzig, 1926–1934); idem, Fauna und Mineralien der Juden (Hildesheim, 1969). For Krauss’ earlier works see Eli Strauss, Bibliographie der Schriften Prof. Dr. Samuel Krauss’ 1887–1937 (Wien, 1937). Steven Fine, ‘Preface’, in Steven Fine (ed.), Jews, Christians, and Polytheists in the Ancient Synagogue: Cultural Interaction during the Greco-Roman Period (Baltimore Studies in the History of Judaism; London, 1999), p. XVI. GREEK LOANWORDS IN RABBINIC LITERATURE 197 As time passed and ‘a new Krauss’ failed to appear, the only standard reference book on Greek loanwords not only retained its position, but grew in authority and came to be perceived as a ‘model of scholarly method and patient research’ – this time without caveats.17 Towards the third quarter of the century the existence of a list of over two thousand lexemes acquired symbolic significance – it was the evidence for the profound Hellenisation of Jews, with the quality of individual etymological conclusions no longer questioned. With the establishment of the state of Israel, the academic elite of the new nation was no less eager to project the image of the ‘European Israeli’ than their nineteenth-century predecessors that of the ‘European Jew’. The new round of emancipation resulted in equally apologetic scholarly motivation: the more Greek words can be found in Jewish sources, the more Hellenised and civilised ancient and medieval Israel was supposed to be. A race to come up with many novel Greek loanwords thus intensified. In 1972 the plan to publish a revised Krauss under the aegis of Bar-Ilan University was proclaimed, but never materialised, at least in its initially conceived form.18 Instead, the abundant research of legal, mercantile, architectural, naval, and other realia was published.19 Just as in the previous century, the discovery of loanwords was advertised as a valuable addendum for mainstream classical studies, and the classicists were urged to cooperate: Rabbinic lexical material has for the major part been totally neglected by classicists. And this despite the fact that there are many thousands of classical loan-words in Rabbinic literature, a literature which spans much of the Roman and Byzantine periods. A goodly part of this material is readily available in one form or another in European languages. A significant example in point is the second volume of S. Krauss’ Griechische und lateinische Lehnwörter im Talmud, Midrach und Targum … It may readily be used by a classicist with no knowledge of the Hebrew language, since all Hebrew characters are transliterated … This considerable body of material may be utilised by classicists for the further clarification of readings, semantic meanings, the dating of the appearance of the word, and for dialect research.20 17 18 19 20 Levias, ‘Greek and Latin Loan-Words’, p. 191. Daniel Sperber, ‘Prolegomena to a New Dictionary of Classical Words in Rabbinic Literature’, repr. in Essays, pp. 1–74. Daniel Sperber, A Dictionary of Greek and Latin Legal Terms in Rabbinic Literature (Ramat-Gan, 1984); idem, Nautica Talmudica (Ramat-Gan–Leiden, 1986); idem, Magic and Folklore in Rabbinic Literature (Ramat-Gan, 1994); idem, The City in Roman Palestine (Oxford, 1998). Daniel Sperber, ‘Studies in Greek and Latin Loan-Words in Rabbinic Literature’, Scripta Classica Israelica 2 (1975), pp. 163–164. This passage is indicative of numerous incongruous presumptions: e.g., that the loanwords borrowed during the Roman 198 J.G. KRIVORUCHKO The pleas to mobilise the classicists for Rabbinic studies fell by and large on deaf ears.21 Since most evident etymologies were already picked up, the post-war researchers were confined to commenting on new material, comparing the known texts and solving the thorny cruces interpretum left by the predecessors. Yet each advance required increasingly more effort, and the law of diminishing returns inevitably led to the marginalisation of the field. On the whole, research in Greek loanwords in the second half of the twentieth century was subservient to the agenda of historians who aimed at investigating realia rather then analysing the words as linguistic objects, or to that of philologists, who sought textological advances. And while the study of Hellenisation was flourishing, it was not matched by comparable advances in historical phonetics and morphology of loanwords.22 The only linguistic situation to be discussed in detail was that of Palestine at the time of Jesus, with later periods attracting only a fraction of the attention given to Hellenisation.23 The very format of loanword studies changed – they were now buried in ad hoc commentaries in text editions and minor notes in periodicals.24 With a few 21 22 23 24 and Byzantine periods can still be considered classical Greek words, that the passages containing loanwords can be efficiently studied in translation, etc. On the possibility of utilising the Rabbinic material for classical research see below sect. 5.1.–5.2. Among those responsive, few authors were extremely productive, see e.g. Howard M. Jacobson, ‘Greco-Roman Light on Rabbinic Texts’, Illinois Classical Studies 5 (1980), pp. 57–62; idem, ‘Ketiah Bar Shalom’, AJS Review 6 (1981), pp. 39–42 (on Latin material); idem, ‘More Roman Light on Rabbinic Texts’, Illinois Classical Studies 8 (1983), pp. 165–167; idem, ‘A Note on Joseph Qimkhi’s ‘Sefer Ha-Galuy’, The Jewish Quarterly Review, New Ser. 85 (1995), pp. 413–414; idem, ‘A Note on Isaiah 51:6’, Journal of Biblical Literature 114 (1995), p. 291; idem, ‘Two Greek Words in Genesis Rabbah’, Scripta Classica Israelica 16 (1997), pp. 212–214; idem, ‘‫’אנפלי‬, The Jewish Quarterly Review, New Ser. 90 (2000), p. 417; idem, ‘Pedagogical Advice in the Haggadah’, Zutot 6 (2009), p. 9; idem, ‘Two Greek Hero-Types in Ancient Jewish Texts’, Zutot 6 (2009), pp. 7–8, etc. For a short discussion of Fremdworten morphology see Gustaf H. Dalman, AramäischNeuhebräisches Handwörterbuch zu Targum, Talmud und Midrasch (Leipzig; 1905), pp. 182–187. The poverty of formal linguistic research in the domain has been recently deplored by Wilfred G.E. Watson, ‘Loanwords in Semitic’, Aula Orientalis 23 (2005), pp. 191–198. The bibliography on Hellenisation is too large to be adduced here. For a recent review see John C. Poirier, ‘The Linguistic Situation in Jewish Palestine in Late Antiquity’, Journal of Greco-Roman Christianity and Judaism 4 (2007), pp. 55–134. For different approaches to de-Hellenisation see Nicholas R.M. de Lange, ‘The Revival of the Hebrew Language in the Third Century CE’, Jewish Studies Quarterly 3 (1996), pp. 342–358, and Seth Schwartz, ‘Rabbinization in the Sixth Century’, in Peter Schaefer (ed.) The Talmud Yerushalmi and Graeco-Roman Culture (Tübingen, 2002), pp. 55–69. A particularly large number of short lexical studies has been authored by Daniel Sperber and published in Israeli periodicals, e.g., ‘‫ – פריפינון – כרתינון‬on the Nature of GREEK LOANWORDS IN RABBINIC LITERATURE 199 exceptions,25 solitary interpretations and guesses were not brought together, and linguistic generalisations were thus impeded. During the last decade of the twentieth century the scenery of Greek loanword research started to change, mostly as a result of gradual refinement of the neighbouring fields. First, significant progress has been achieved in Byzantine and Modern Greek lexicography. Lexikon zur Byzantinischen Gräzität with its emphasis on derivative morphology provided a valuable spurt for the etymologists and specialists in late ancient and early medieval languages.26 As its elder brother, Kriaras’s dictionary, grew in coverage, its usability dramatically increased.27 Almost simulta- 25 26 27 Ravia’s Manuscript of Yerushalmi’ (in Hebrew), Sidra 3 (1987), pp. 81–83; idem, ‘Etymological Studies in Rabbinic Hebrew’ (in Hebrew), Leshonenu 53 (1988), pp. 60–66; idem, ‘Greek Word-Plays in the Pattern ‘Katal’’, in Menachem-Zvi Kadari and Shimon Sharvit (eds.), Studies in Hebrew Language and Talmudic Literature Dedicated to the Memory of Dr. Menachem Moreshet (in Hebrew) (Ramat-Gan, 1990), pp. 221–224; idem, ‘‘Paggei Shevi’it’ and the Innocent Victim’ (in Hebrew), Sidra 7 (1991), pp. 158–157; idem, ‘Greek Words in Sefer Rushaina to Exodus’ (in Hebrew), Leshonenu 56 (1991–2), pp. 143–146; idem, ‘Difficult Greek Words in Later Rabbinic Literature’ (in Hebrew), Leshonenu 10 (1994), pp. 291–295; idem, ‘Clarification of Readings: Levdakos, Charkom, Lektica’ (in Hebrew), Sidra 14 (1998), pp. 165–169; idem, ‘Notes on Words and Their Variants’ (in Hebrew), Leshonenu 66 (2004), pp. 193–196, etc. Extensive lexical material is found in his books, in particular A Dictionary of Greek and Latin Legal Terms in Rabbinic Literature (in Hebrew) ([Tel Aviv], 1984), idem, Magic and Folklore in Rabbinic Literature (Ramat-Gan, 1994), idem, Nautica Talmudica (Ramat-Gan, Leiden, 1986), idem, Material Culture in Eretz-Israel During the Talmudic Period (in Hebrew) (Jerusalem, Ramat Gan, 1993–2006). Other characteristic examples of individual word studies are Haïm Rosén, ‘Etymological Notes’, in East and West: Selected Writings in Linguistics (Munich: 1982), pp. 444–449; Al Wolters, ‘Targumic ‫( כרובת‬Zechariah 14:20) = Greek Korufaía?’, JBL 115 (1996), pp. 710–713; Howard Jacobson, ‘‫’כלפי לייא‬, JQR 89 (1999), p. 387, and Giuseppe Veltri, ‘Greek Loanwords in Talmud Yerushalmi: Some New Suggestions’, JSS 47 (2002), pp. 237–240. Several collections of word studies have been published, see e.g., Saul Lieberman, Greek in Jewish Palestine: Studies in the Life and in the Manners of Jewish Palestine in the II–IV centuries C.E. (New York, 1942); idem, Hellenism in Jewish Palestine (New York, 1962); Daniel Sperber, Essays on Greek and Latin in the Mishna, Talmud and Midrashic Literature (Jerusalem, 1982). They included, however, no linguistic summary. On the other hand, theoretical papers were normally short, e.g. Daniel Sperber, ‘Rabbinic Knowledge of Greek’, in Shmuel Safrai (ed.), The Literature of the Sages. Second Part: Midrash and Targum, Liturgy, Poetry, Mysticism, Contracts, Inscriptions, Ancient Science and the Languages of Rabbinic Literature, (Assen, 2006), pp. 627–640. For an attempt to trace the impact of Greek in other fields see Abba Ben David, Biblical Hebrew and Mishnaic Hebrew (in Hebrew) (Tel-Aviv, 1967), pp. 135–152. Erich Trapp (ed.), Lexikon zur Byzantinischen Gräzität, besonders des 9–12 Jahrhunderts (Wien, 1994–). Emmanouil Kriaras (ed.), Lezikó tjv Mesaiwnikßv Elljnikßv DjmÉdouv Grammateíav, 1100–1669 [Leksiko tis Mesaionikis Ellinikis Dimodous Grammateias, 1100– 1669] (Thessaloniki, 1969–); 200 J.G. KRIVORUCHKO neously, two state-of-the-art lexica of Modern Greek appeared, and monolingual Greek lexicography established itself as a sine qua non for the research of earlier periods.28 Second, the inquiry in the textual history of Mishna and other rabbinic works became both increasingly precise and detailed, and more comprehensive. For many texts, it became possible to analyse the loanwords on the level of manuscripts and groups of manuscripts rather than that of editions.29 Yet interest in the linguistic side of the borrowing process was still slow to arise,30 and Krauss’s dictionary continued to be quoted as authoritative both inside and outside the field.31 2. WHAT MAKES GREEK LOANWORD STUDIES SO ARDUOUS? The fact that a century-long research effort failed to produce ‘a new Krauss’ does not look surprising in view of the sheer quantity and variety of the material covered by the original one. But the problematic nature of Greek loanword research lies not only in its scope. A considerable number of publications lament the difficulties awaiting the researcher of Greek loanwords:32 On considering the points here raised, or perusing Krauss’s book with the necessary critical attention, one may easily feel driven towards a 28 29 30 31 32 Instituto Neoellinikon Spoudon [Idryma Manoli Triantafyllidi], Lezikó tjv Koinßv Neoelljnikßv [Leksiko tis Koinis Neoellinikis] (Thessaloniki, 1998), further abbreviated to LKNE; Georgios Mpampiniotis [GeÉrgiov MpampiniÉtjv], Lezikó tßv Néav Elljnikßv GlÉssav (Athens, 1998; 3rd edition 2008). E.g., Gabriel Birnbaum, The Language of Mishnah in the Cairo Genizah: Phonology and Morphology (in Hebrew) (Mekorot U-Mehkarim. Sidrah Hadasha 10; Jerusalem, 2008), pp. 32–34, 47–48, 57–58, 68, 87–88, 108–112, 140–143, 169–171, 196–198, 227–228, 236, 258–264, 287–289, 326, etc. But see e.g., a recently announced presentation by Shai Heijmans, ‘Some Remarks on the Linguistic Tradition of Greek Loanwords in the Mishnah’, paper presented on the 15th World Congress of Jewish Studies, 2-6 August 2009. E.g., references to Krauss in Florentino García Martínez, ‘Greek Loanwords in the “Copper Scroll”’, in Florentino García Martínez and Gerard P. Luttikhuizen (eds.), Jerusalem, Alexandria, Rome: Studies in Ancient Cultural Interaction in Honour of A. Hilhorst (Leiden, 2003), pp. 119–145; Shim‘on Sharvit, Language and style of Tractate Avoth through the Ages (in Hebrew) (Beer Sheva, 2006), pp. 96–97, 177, 246–247, 251, etc.; Marco Mancini, ‘Appunti sulla circolazione del latino nella Palestina del I secolo d.C.’, in R. Lazzeroni et al. (eds.), Diachronica et synchronica: studi in onore di Anna Giacalone Ramat (Pisa, 2008), pp. 277–299. Ranon Katzoff, ‘Sperber’s Dictionary of Greek and Latin Terms in Rabbinic Literature: A Review-Essay’, JSJ 20 (1989), pp. 195–206. See recently Daniel Sperber, ‘Al ha-zoreÌ letaken et ha-“Lehnworter” shel Krauss: dugma me-ereÌ “pnkrysyn”’, JSJ (University of Bar-Ilan online publication) 3 (2004), pp. 13–17. GREEK LOANWORDS IN RABBINIC LITERATURE 201 completely negative attitude with regard to these studies. And indeed the labour and the uncertainty involved are great and the reward small. Many instances suggest that foreign words were liable to be transformed, or deformed, to such an extent and with such a lack of regularity as to make their recovery an unrewarding gamble.33 Long lists of etyma suggested for a single Hebrew/Aramaic word create the impression that there are few or no methodological criteria for establishing the correct Greek etymon. Zuntz was perplexed and baffled by Krauss’s attempts to interpret ‫( פרוטומי‬Ex. Rabbah ch. 15, 17, Lam. Rabbah II, 2, etc.)34: What, after all, is one to say when finding – to quote but one further example – one and the same word traced (p. 485) alternatively to protomß ‘bust’, pt¬ma ‘corpse’, próqema ‘public notice’, êpítimov ‘contraband’(?) – or even glans penis?’ 35 In fact, such – and longer – lists are not at all rare. For instance, ljÇstßrion ‘robber’s den’, staqeróv ‘starker’, stjqárion ‘Brustbild’, statßrion, Lat. statura ‘stature’, *listótaurov ‘a bull who is worshipped’ were all suggested to explain ‫ ליסטטרין‬in Shir Rabbah ad 6:4.36 Whether the ‘unrewarding gamble’ may ultimately become worthwhile, or at least less disappointing, depends on our chances of overcoming the obstacles to lexicohistorical research. Let us have a closer look on them one by one. 2.1 Inadequacy of writing system The fact that one Hebrew-script sequence can correspond to many different Greek words, is largely a consequence of the writing system. To get a taste of the inadequacy, let us assume for the sake of a very rough approximation that a combination of three Hebrew consonants records a Greek word (with a couple of vowels this would be a Greek word of an average length).37 Let us ignore for a moment the vocalisation of the last 33 34 35 36 37 Zuntz, ‘Greek Words’, p. 138. ‫ פרוטומו‬in ed. Buber (Vilna, 1878). Ibid. For references see Howard M. Jacobson, ‘Greco-Roman Light on Rabbinic Texts III’, Scripta Classica Israelica 7 (1983–1984), pp. 89–90. By an oversight the calculations adduced in Krivoruchko, ‘On the prospects’, pp. 28–29, were imprecise. Average word length in the corpus of Modern Greek is 5.45 letters, see Nick Hatzigeorgiu, George Mikros, and George Carayannis, ‘Word Length, Word Frequencies and Zipf’s Law in the Greek Language’, Journal of Quantitative Linguistics 8 (2001), p. 179; no average length calculation for the period corresponding to Rabbinic Hebrew is known to us. 202 J.G. KRIVORUCHKO syllable, as it depends on the adaptation of the word to Hebrew/Aramaic morphology and assume that the word starts with a consonant. Two positions are left to insert vowels. Greek script has seven letters for vowels (a, e, j, i, o, u, w), and ten popular combinations of vowel letters (ai, ei, oi, ui, au, eu, ju, ou).38 It is also possible that the word contains a consonant cluster, in which case nothing would be inserted, bringing the number of possible insertions to eighteen. The number of possible combinations of two from eighteen is 306. Certainly most of them are impossible for various linguistic reasons, but even one tenth of them would give the impressive thirty options. Let us now assume that one of the internal positions is occupied with a mater lectionis. Let us ignore hei as it mostly occurs at the end of the word. According to Krauss, alef in Greek words may represent a, e, i, o, u, w, au (seven options); yod: e, j, i, u, ai, ei, oi, eu (eight options); vav: o, u, w, ai, au, eu, ou (seven options).39 If Krauss’s data are correct, this would mean that with two consonants recording a cluster, there are at least seven options to interpret the remaining record, and in the worst case these options are seven times eight, that is, close to fifty. Again, some would be filtered out by the phonetic system, but the remaining are still too many. This primitive calculation shows that partially vocalised renderings of Greek are also hardly unambiguous. It has been always intuitively clear to scholars that the unvocalised Hebrew script, or Hebrew script with few and/or inconsistently placed vowel signs and matres lectionis, is capable of giving only a very poor and approximate picture of Greek phonetics, to such an extent that reading Greek in Hebrew garb is often called ‘deciphering’.40 Theoreticians of grammatology were right to use non-alphabetic renderings of Greek to illustrate inefficient writing systems.41 2.2 Inadequate compensation mechanisms for the inadequate writing system Historically, the lack of clear writing conventions for loanwords was compensated for by oral transmission. In order to facilitate the reading 38 39 40 41 The calculation reflects current perception of Hebrew-spelled Greek as resulting from the transcription process. Krauss, Lehnwörter, vol. 1, pp. 13–27, and the table on p. 64. Richard C. Steiner, ‘Textual and Exegetical Notes to Nicholas de Lange, Greek Jewish Texts from Cairo Genizah’, Jewish Quarterly Review 89 (1998–1999), p. 155. Florian Coulmas, Writing Systems: An Introduction to Their Linguistic Analysis (Cambridge Textbooks in Linguistics; Cambridge, 2003), pp. 82. GREEK LOANWORDS IN RABBINIC LITERATURE 203 of Greek, scribes and scholars often added vocalisation a posteriori. Most readers of primary manuscript material and/or scholars competent in Medieval Greek would not find any difficulty in admitting that vocalisation is secondary, frequently non-contemporaneous and occasionally incompatible with the initial record. As a result, the forms that the modern researcher encounters may vary from faithful representations of real lexemes to more or less lucky guesses or complete ghost-words. Evidently, in this latter case the non-synchronic evidence cannot be interpreted at face value. 2.3 Inadequacy of transmission Textological problems connected with Greek in Hebrew script are numerous, in so far as the compositions containing Greek loanwords were copied in the regions where the knowledge of Greek was either no longer current or altogether nonexistent. This resulted in notoriously faulty textual transmission42 with lots of mechanical mistakes leading to the corruption of initial renderings and confusion of word boundaries. Even diligent scribes misunderstood their predecessors when copying incomprehensible garbage. Yet while common palaeographical errors are in a sense predictable, folk etymological transformations through which the scribes tried to explain the confusing words to themselves, are much less so. This situation is generally typical for the transmission of Greek via a non-Greek alphabet in a non-Greek environment, cf. the witness of the researcher of commedia dialettale: The fact that the texts … were initially transmitted orally and were fixed in writing only on later stages is surely one of the reasons for the destruction of language material. One should add that the writers and the scribes were partially or completely ignorant of Greek language, which resulted in sequences of mistakes and distortions during the copying process, and finally in typographical errors.43 2.4 Lexico-semantic and text-linguistic inadequacy of the source texts A typical procedure of lexico-semantic analysis is based on collecting the contexts in which a word is used, that is, the syntactic units with this 42 43 Abraham Wasserstein, ‘Review of Daniel Sperber, “A Dictionary of Greek and Latin Legal Terms in Rabbinic Literature” (1984)’, Scripta Classica Israelica 7 (1983), pp. 147–149. A. Koumarianoú, ‘To enetograikikon idíwma tjv ‘Commedia Dialettale’: néev prooptikév (The Veneto-Greek Dialect of the ‘Commedia Dialettale’: new prospects)’, Thesaurismata. Bolletino dell’Instituto Ellenico di studi bizantini e postbyzantini, Venetia 3 (1964), pp. 140–145. 204 J.G. KRIVORUCHKO word. The semantic structures of the words, to which a given word is connected, are then analysed in order to define the potential structure of the word under analysis. To perform this procedure one should have, as a minimum, a sufficient number of syntactic structures with a given word, preferably full-fledged sentences. Elliptic (underdeveloped or incomplete) syntactic and/or semantic structures are not suitable for this procedure, as they do not exhibit clear syntactic and semantic connections between the words. Many so-called rabbinic texts lack clear structure both on the level of text and sentence. Mishnaic and Talmudic discourse as a matter of norm offers an abridged version of the argument and gnomic phrasing, so that a text results in a list of shorthands, to be interpreted by a student on the basis of his preexisting knowledge. Links between the parts of the text and even inside a sentence are not expressed, but implied and based on the information external to the text/sentence. Thus most Mishnaic and Talmudic texts, together with parts of Midrash and secondary exegesis, would be defined in text linguistics as texts of low cohesion but sufficient coherence. Their coherence (= the fact that the majority of these texts are still understandable) is essentially provided and secured by the oral environment in which these texts function, and/or by the corpus of traditional texts as a whole.44 As is clear from the above, rabbinic texts are the worst possible material for lexical semantics, since they do not provide syntactic and/or semantic units of sufficient completeness. The problem is particularly acute in case of hapaxes or quasi-hapaxes that constitute significant proportion of loanwords, when the problem of quality of contexts is exacerbated by their quantity: there are simply not enough collocations to deduce the semantics of the word. 2.5 Unclear degree of linguistic accommodation of the words under investigation In strict linguistic terms, not all the words dubbed ‘loanwords’ in this paper for the sake of simplicity, are indeed loans, that is, lexemes integrated into the lexicon and morphology of a host language. The items, with which Krauss and his successors dealt, represent varying stages of the borrowing process, from initial to very advanced, or do not partake in this process at all. 44 J. Renkema, Discourse Studies (Amsterdam–Philadelphia, 1993), pp. 35–40. GREEK LOANWORDS IN RABBINIC LITERATURE 205 It is expected that realia that entered the Jewish society with Hellenisation would preserve their Greek names, and that such nouns will be fully integrated into Hebrew or Aramaic for the lack of native terminology. Yet the great number of lexemes under inquiry are non-adapted borrowings or nonce borrowings, not Lehnwörter, but Fremdwörter, or just plain one-word code-switches of various types. Their appearance is due to the following main reasons: (a) A significant part of rabbinic literature was created by writers whose native language was neither Hebrew nor Aramaic: it is in fact second- (third-, etc.) language writing. The compositions written in Hebrew/Aramaic by Greek-speakers include Greek lexemes as a result of code-switching with their mother tongue. Speakers of other languages could have code-switched to Greek as well, particularly when living in the regions where Greek was the lingua franca. (b) Many Greek lexemes must have penetrated the rabbinic corpus from the manuscripts copied in the Byzantine realm, Greek glosses being integrated into the original compositions. Accidental code-switches or syntactically unconnected solitary terms, including the lexemes explicitly labelled in the texts as translations into Greek, are hardly ‘loanwords’. However, the status of the lexeme under investigation is not always obvious. This vagueness is not without consequences for etymologising, where one must decide whether the final consonant(s) of the word represent morphemes, acquired by the word during its integration into Hebrew/Aramaic morphology, or whether they are part of the Greek etymon. 2.6 Insufficient text-historical information about the source texts The Midrashim, Targumim and other types of texts in which the loanwords appear, are very often texts without a well-researched textual history. Therefore, the lexical material of Greek origin contained in them can neither be dated with certainty, nor ascribed to specific dialect or region. The cross-genre generalisations over this type of material are very risky, since it is easy to end in linguistically problematic comparison of data divided by centuries and even millennia.45 45 Johannes Niehoff-Panagiotides, ‘Byzantinische Lebenswelt und rabbinnische Hermeneutik: Die Griechischen Juden in der Kairoer Genizah’, Byzantion 74.1 (2004), pp. 51–109. 206 J.G. KRIVORUCHKO 2.7 Insufficient information about the reconstruenda Last but not least, the very targets of etymological reconstruction, the Greek etyma, may present a problem. Notwithstanding the spectacular advances that were made in the last decades both in Medieval Greek lexicography and dialectology, large strata of vocabulary and linguistic phenomena are still waiting to be described: neither dictionary of Medieval Greek has reached its omega, and its grammar is still unpublished. 3. LOANWORD RESEARCH: THEORETICAL PRINCIPLES 3.1 What conditions should a reliably reconstructed loan satisfy? As is clear from the above (sect. 2.1–2.7), an ideal reliably reconstructed loan is (0) uniquely defined: In case a supposed loanword seems to appear in different texts, each of them should be analysed independently, and no presumptions should be made about its being identical across the texts. (a) paleographically coherent: The researcher should read what s/he sees in the manuscript rather than presume an unknown number of corruptions that supposedly occurred during the transmission. Reconstructing the graphemic form may be acceptable if there is a firm evidence of confusion of given letters or signs in the specific hand through which the text was transmitted. Less strict would be consideration of the common confusions of the tradition, to which the scribe of the given work belongs. Still more compromising is the use of the editions, since due to the selective character of human visual perception the editors tend to read whatever they can understand rather than what actually appears in the manuscript. (b) grammatologically coherent: All the graphemes should be taken in the phonetic values ascertained for the reconstructed biblical pronunciation of the area,46 unless other values 46 For basic table, see Shelomo Morag, ‘Pronunciation of Hebrew’, in Jewish Encyclopedia, s.v. GREEK LOANWORDS IN RABBINIC LITERATURE 207 have been established by the orthographic research of a given period and region.47 (c) morphologically coherent: The phonetic sequence reconstructed through (a) and (b) should be either a valid Greek form or a Hebrew/Aramaic form which can result from the morphological adaptation of a Greek form, provided the rules of adaptation for the given period and region are known from elsewhere.48 A ‘valid Greek form’ here means a form that either occurs in the corpus of Greek texts of a given period and region, or a form that can be derived from the occurring forms through verifiable chronologically and geographically localised morphological paradigms. (d) semantically coherent on the level of lexeme: The phonetic sequence achieved through (a) and (b) should be a lexicosemantic variant which existed in the given period or region, that is, occurs in the corpus of contemporaneous Greek texts.49 Given the incompleteness and imperfections of the dictionaries and electronic corpora, as a last resort one may allow a single semantic transformation (metonymic, metaphoric or hypo-hyperonymic variation) to intercede between the registered LSV and the reconstructed one. (e) semantically coherent on higher levels: The reconstructed loan should form a part of syntactically and semantically coherent sentence, and the sentence – of larger textual entity. (f) text-linguistically coherent: The sentence with a reconstructed loan should not contradict the practice of code-switching or loanword usage observed in the given text. If the text is short, poorly preserved, and does not provide sufficient material for comparison, one is permitted to use the closest possible chronological and geographical parallels. 47 48 49 E.g., Abraham Wasserstein, ‘A Note on the Phonetic and Graphic Representation of Greek Vowels and of the spiritus asper in the Aramaic Transcription of Greek Loanwords’, Scripta Classica Israelica 12 (1993), pp. 200–208; Steven E. Fassberg, ‘The Orthography of the Relative Pronoun she- in the Second Temple and Mishnaic Periods’, Scripta Classica Israelica 15 (1996), pp. 240–250. This often will not be the case. For definitions of lexical-semantic variant and typology of semantic variation see Mikhail V. Nikitin, Leksiweskoe znawenie slova [Leksicheskoe znachenie slova] (The Lexical Meaning of the Word) (Moscow, 1983). 208 J.G. KRIVORUCHKO The requirements (c)–(e) are in fact applications of the general principles of etymology to the specific material. The requirement (e) presumes that the meaning of the passage is clear on the basis of itself, since using parallel contexts for its exegesis is limited by the criterion (0). 3.2 The principle of the best match In many cases it would not be possible to retrieve the solution answering all the above criteria because of the shortage of data or their poor systematisation. If several etyma close to the ideal are found, the one which complies with more criteria should be considered best. Good scholarly practice for suggesting a new etymon should include a proof that previously suggested etyma are worse matches (= that they violate more principles of loanword reconstruction). Like a jigsaw puzzle, loanword etymology demands fine matching, and if one aspect of the reconstruction is found unfit, it usually means that others were forced into their position too. 4. A SAMPLE OF LOANWORD RECONSTRUCTIONS ANALYSED ACCORDING TO 3.1–3.2 The following section aims to illustrate how to apply the criteria described above in assessing real etymological solutions suggested by various researchers. 4.1 BT Yoma 84a, Ketubbot 30b, Berachot 8a50 The author discusses the medical term serunke or sirvanke, a synonym to askara, which appears in a number of places in the Talmud, and argues that it derives from Gr. sunágxj ‘sore throat’. Since the Hebrew-script rendering does not match the consonant sequence of the Greek word – one would expect s-n-x or s-n-n-x – he suggests ‘une déformation paléographique, le i (‫ )י‬se transformant facilement en r (‫ )ר‬et le n (‫ )נ‬en v (‫’)ו‬.51 50 51 Discussed by Samuel S. Kottek, ‘Sur l’origine gréco-latine de certains termes médicaux utilisés dans le Talmud et le Midrash’, in Guy Sabbah (ed.) Le latin médical: la constitution d’un langage scientifique: réalités et langage de la médecine dans le monde romain (Saint-Étienne, 1991), pp. 41–52. Kottek, ‘Sur l’origine’, p. 45. GREEK LOANWORDS IN RABBINIC LITERATURE 209 While one cannot exclude that in a particular hand these consonants can be confused, it is hardly applicable to the totality of the transmission of the above Talmudic treatises. And while the explanation might in fact be correct, this type of reasoning does not answer the criteria of individual approach (0) and paleographic realism (a). 4.2 Midrash Samuel 24:1052 The passage in question is an interpretation of 1 Sam 19:13 ‫ותקח מיכל את‬ 53 ‫התרפים ותשב על המיטה‬, which is as follows: ‫ ניקורים‬:‫תני בשם רבי איבו‬ ‫של בדוקלי‬. As the scholar aims to understand the enigmatic two-word Rabbinic scholion, he starts by appreciating the transmission of the text. He quotes the editor of the Midrash Samuel Salomon Buber, who was aware of the complexity of the passage and pointed out that the above reading belongs to the editio princeps, while in the manuscript one finds ‫נקודים של בריקלי‬, and in Yalkut ‫ניקורים של ברוקלי‬. Various versions of the words have generated numerous etymological conjectures, such as Heb. ‫‘ נקד‬point’, nákov ‘fleece’, ∫gkov ‘Haaraufsatz’, Lat. veriloqui ‘speaking truth’ (pl.), barúlogov ‘slanderer’, bdeluktóv ‘ugly, unsightly’, eîdolikóv ‘phantasmal’, brákai (Lat. bracae) ‘trousers’, ‫‘ אמבורקלון‬shreds of horse cover’, bríkeloi ‘Larven’, etc. Following the short commentary on the opinions of predecessors, the scholar opts for the manuscript version of the second word and for the editio princeps version of the first, that is, for a consonant sequence identical with that of Yalkut, but with different vocalisation, as he assumes the first word reflects a form of Gr. nekróv. In his mind, the second word ‫ בדוקלי‬originates from brukólakav ‘a dead man that according to popular tradition gets out of his grave at night and goes among the living to drink their blood, and generally to hurt them’.54 The choice of the Greek etymon for the context prompts questions, since it is known in numerous variants. If one upholds the Slavic origin of the word,55 the form boulkólaz would have been the earliest, with 52 53 54 55 Discussed by Daniel Sperber, ‘Milim yavaniyot kashot be-sifrut ha-rabanit hameuheret,’ Leshonenu 10 (1994), pp. 291–295. KJV 1 Sam 19:13: And Michal took an image, and laid [it] in the bed, and put a pillow of goats’ [hair] for his bolster, and covered [it] with a cloth. Sperber, ‘Milim’, p. 288. The meaning of the Greek etymon after LKNE, s.v. brikólakav 1. See Kriaras s.v. boulkólaz with reference to Gustav Meyer, Neugriechische Studien (Wien, 1894), vol. 2, p. 20, and LBG s.v. bourkólakov. Sperber seems to approve both Ancient Greek and Slavic origin of the lexeme. 210 J.G. KRIVORUCHKO such variants as bourkólakav originating through dissimilation of laterals. Metathesis is a common phenomenon for the syllables of such structure, giving e.g. the verb broukolakiáhw. The first attestation of the phonetic form of the root that matches the Hebrew text comes from the vernacular satirical poem Spanos, and it reads as follows: kaì ™ ∫civ tou êgéneton ¿sper bourkolakiasménou nekroÕ kaì ≠frihen tò stóman tou (recensio A e cod. Vind. theol. gr. 244, line 269) ‘and his appearance became like that of a dead man that rose out of his grave, and his mouth was afoam’, cf. kaì ™ ∫civ aûtoÕ gégonen ¿sper ∫cin brikolakiasménou nekroÕ kaì ânépempen âfroùv tò stóma aûtoÕ (recensio B e cod. Vat. gr. 1139, line 91).56 The participles bourkolakiasménov and brikolakiasménov may by way of backward reconstruction give the required noun. While such backward reconstructions are legitimate, the absence of the noun itself from the early sources is warning, in particular if one acknowledges the possibility of the early dating of the Midrash.57 The comparison with Spanos, whose earliest acceptable dating is fifteenth century,58 appears anachronistic. Even if we presume, for the sake of alleviation, that the form in the Midrash is plural, it is unclear why this should be a plural in –oi rather than more modern-styled in –ev, in which case the rendering of [s] is also missing. It is also unclear why the Hebrew-script rendering does not reflect the final consonant [k] of the noun’s stem. Thus, the criterion (c) is violated. 4.3 ‘Aruk s.v. ‫פרוכימטוס‬59 ‘Aruk quotes Midrash Yelamdenu from the section beginning with the verse ‫( לכה ארה לי יעקוב‬Num 23:7)60: ‫למלך שמוציא כרוז למחר יוצא אני‬ ‫ודן את בני פרוכימטוס‬, ‘Like a king who issued a proclamation: Next day I am going to [hold court and] try my son, [and he tried him] “prokhimatos.”’ All the corrections and explanations of the foreign word in our dictionaries are senseless, and are not worth recording. Moreover, the whole passage in the quotation is not understandable. 56 57 58 59 60 Both versions are quoted after Hans Eideneier (ed.), Spanos: eine byzantinische Satire in der Form einer Parodie (Supplementa Byzantina 5. Berlin, 1977). Sperber, ‘Milim’, pp. 289–290. Eideneier, Spanos, pp. 24–26. Discussed by Saul Lieberman, ‘Defective Transliteration of Greek Words in Rabbinic Literature’, JQR 73 (1982), pp. 62–63. KJV Num 23:7: ‘And he took up his parable, and said, Balak the king of Moab hath brought me from Aram, out of the mountains of the east, [saying], Come, curse me Jacob, and come, defy Israel’. GREEK LOANWORDS IN RABBINIC LITERATURE 211 To solve the problem, the scholar addressed the section of the same Midrash quoted by ‘Aruk s.v. ‫אפקלין‬. Notably, the link between the passages is not suggested expressis verbis by the ‘Aruk itself, but at that stage the author obviously strives for methodological consistency by trying to find as close a parallel as possible. Unfortunately, further on the parallel passage brings to his mind yet another midrashic passage, this time from Lev. Rabbah 27.6, Pesikta de Rab Kahana 9, etc. As a result of the juxtaposition of all those, …the Yelamdenu passages quoted by ‘Aruk are quite understandable. It first compared the situation to that of a woman who complained to the authorities of the behaviour of her son, but when she realized that her charge may involve serious punishment, she twisted her accusation into an absurdity. Like-wise, when the Lord saw that the charge against Israel might carry severe punishment, he diverted the accusation to something else. And the Rabbis cited our parable: ‘Like a king who declared: Next day I am going to [hold court and] try my son, [and he tried him] ‫פרו]ס[כימטוס‬,’ prò sxßmatov, i.e., ‘pro forma only.’61 The insertion of samech is both prompted by and justified by external analogies. The problem with this interpretation is that while TLG lists 4585 cases with the form sxßmatov and other 55 with the form sxjmatov (without accent),62 not a single one of them has the meaning that the distinguished author wants them to have. The usage of the word prosxßmatov < prósxjma ‘anything used to disguise the real purpose of an action; a pretext’63 would be somewhat closer semantically, but does not match chronologically. In all probability, it was Lat. pro forma which made the researcher believe that the meaning ‘as a pure formality, to obey a formal demand’ is normal for the combination prò sxßmatov in Greek. The discussed reconstruction therefore fails the lexico-semantic criterion (d) and the most basic criterion of unique source (0). The question arises whether the transliteration of the ‘Aruk was indeed as ‘defective’ as the scholar imagined, and whether correcting the text, or correcting the text in this particular way, was at all necessary. Unprovoked emendations of the sources may result only in unfair matches between the imaginary text reality and the imaginary Greek. 61 62 63 Lieberman, ‘Defective Transliteration’, p. 63. Accessed on 23.03.2010. LKNE, s.v. prósxjma. 212 J.G. KRIVORUCHKO 4.4 The Mastaura Ketubba (T-S 16.374)64 The following example is taken from a documentary text, which does not constitute rabbinic literature strictu senso. Its analysis here is justified, since (a) the loans in documents are often analysed together with rabbinic loans,65 and (b) they are indeed contemporaneous with part of them, and ultimately originate from the same literate strata of Jewish society. In the line 19 recto of T-S 16.374, containing the list of dowry items, there appears a combination of letters ‫כילי‬. It has been interpreted as a Greek word xjlß ‘net, plait’ with reference to xjlß III in LSJ.66 The relevant section of the dictionary reads as follows: III. of various cloven or hooked implements: 1. in surgery, forked probe, Hp.Morb. 2.33. 2. notch of an arrow, Hero Bel.111.1, Hsch. s.v. glufídev; but also (pl.) the claws composing the hook (xeír), Hero ib.2; also the claws or arms of the skorpíov v, Vitr.10.10.4, 10.11.7. 3. rims of the eyelids, Ruf.Onom.20. 4. crack in the heels or other parts, Poll.4.198. 5. net, plait, Hsch. s.v. xjleutà kránj. It is clear that the last lexico-semantic variant (III.5) stands apart from the preceding four. Should LSJ be correct in its definition, it will be a semantic hapax, as such meaning does not seem to appear in other contexts. In fact, the gloss of Hesychius xjleu[ma]tà kránj· tà Åaptá, xjlàv ∂xonta poiáv may refer to Sophocles (fr. 486 Radt), with kránj being a part of the interpretation, pace the punctuation of the current edition. Notably, Hesychius also has a gloss s.v. kránea xjleutá· plektà êk sxoínou, Æ ãlljv tinòv Àljv malak±v ‘xjleutá helmets: woven of string or some other soft material’, where he evidently tries to shed light on the passage of Herodotus 7.89: 64 65 66 Discussed by Daniel Sperber, ‘Contributions to Byzantine Lexicography from Jewish Sources’, Byzantion 46 (1976), p. 60. Sperber used the edition of the ketubba by Jacob Mann, The Jews in Egypt and in Palestine under the Fatimid Caliphs; A Contribution to Their Political and Communal History, Based Chiefly on Genizah Material Hitherto Unpublished ([London], 1920–1922). It is now superseded by Nicholas R.M. de Lange, Greek Jewish Texts from Cairo Genizah (TSAJ 51; Tübingen, 1996), pp. 1–10. E. g., Jacob J. Rabinowitz, ‘Grecisms and Greek Terms in the Aramaic Papyri’, Biblica 39 (1958), 77–82; ‘More on Grecisms in Aramaic Documents’, Biblica 41 (1960), 72–74. LSJ = Henry G. Liddell, Robert Scott, and Henry S. Jones, Greek-English Lexicon with revised supplement, 9th ed. (Oxford, 1996), s.v. xjlß III. GREEK LOANWORDS IN RABBINIC LITERATURE 213 Aîgúptioi dè néav pareíxonto dijkosíav· oœtoi dè e˝xon perì mèn t±Çsi kefal±Çsi kránea xjleutá, âspídav dè koílav tàv ÷tuv megálav êxoúsav, kaì dóratá te naúmaxa kaì túxouv megálouv … ‘The Egyptians furnished two hundred ships. They wore woven helmets and carried hollow shields with broad rims, and spears for sea-warfare, and great battle-axes’.67 The same noun xjleúmata and adjective xjleutà are interpreted by the earlier lexicographer Julius Pollux: … ôpjtídia, ° kaì xjleúmata êkáloun oï poijtaí. málista dè oÀtwv Önómahon tà t¬n sxoínouv plekóntwn, Üv kaì kránj xjleutà tà plektà ¨Jródoton légein … (7.83) ‘ … ôpjtídia, which poets call also xjleúmata. Also, they called likewise everything that refers to the weavers of strings, such as Herodotus calling the woven helmets kránj xjleutà … ’. It is therefore very likely that Hesychius was trying to explain to himself the antiquated realia, and instead of deriving the verbal adjective from the verb xjleúei· Åáptei, plékei ‘sew, plait’ (s.v.) tried to link it to the noun xjlß, which in his own words meant öplß, ∫nuz boóv. gnáqov ‘hoof, [that is] ox’s nail, jaw’ (s.v.). Given that precisely these meanings are central for the semantic structure of xjlß in the rest of the Greek tradition,68 one may conclude that the usage of xjlß in the Hesychian gloss is but a metonymy ‘something shaped like öplß or gnáqov’. Yet the unfortunate ad hoc explanation invented by Hesychius was given a new life by the modern lexicographers. Considering the general context of the ketubba, one may note that ‘net’ is certainly not the same as the ‘plait’. Since both meanings rarely combine in one lexeme, the lexicographers would be better off translating it ‘a braid or woven object’. However, both ‘net’ and ‘plait’ are common realia. Still, no such terms for nets are known from elsewhere, and both plaits and nets would be unexpected in the given point of the marriage deed.69 As has been shown, the proposed etymology of ‫ כילי‬does not satisfy the criteria of lexico-semantic (d) and text-semantic (e) consistency. It comes as no surprise that ‫ כילי‬is, for the matter of fact, not a real word at 67 68 69 Herodotus, with an English translation by A. D. Godley (Cambridge, 1920). See LSJ, s.v. xjlß I A.2 and 3. For the editions, interpretation and significance of the ketubba see Julia G. Krivoruchko, ‘Notes on the Dowry Items in the Mastaura Marriage Deed (T-S 16.374)’, in María Ángeles Gallego and Juan Pedro Monferrer-Sala (eds.), The Semitic Languages of Jewish Intellectual Production: Memorial Volume for Dr. Friedrich Niessen (Cambridge Genizah Studies; Leiden, forthcoming). 214 J.G. KRIVORUCHKO all, but a paleographical mistake: ‘Mann read the seventh word as kyly… In fact the scribe began writing the next word but, after writing kyl, realised he had made a mistake, and crossed the letters out’.70 Thus, a forced attempt to retrieve a new Greek loan in the piece of Aramaic/Hebrew writing resulted in creating a vox nihili, a ghost-word.71 5. TOWARDS A REALISTIC APPROACH TO THE GREEK LOANWORD RESEARCH 5.1 Overcoming the Kinderkrankenheit of ‘extra data’ Since the very dawn of loanword studies, Krauss proudly claimed that it could enrich Greek philology with ‘newly discovered words’.72 Current lexicographical projects enable better assessment of the size of the corpora available for each language. The following table reflects the size of the Greek linguistic corpora: the corpus of literary and documentary texts Thesaurus Linguae Graecae, and the corpus of papyri and inscriptions Packard Humanities Institute, version 7. Corpus Scope Word Instances Word Forms TLG + PHI (VIII–XVI, +tech +christ +inscr/pap) 101,684,658 1,815,540 TLG (VIII–XVI) (VIII–XVI, +tech +christ -inscr/pap) 95,475,128 1,567,892 LSJ Corpus (VIII–VI) (VIII–VI, +tech –christ +inscr/pap) 34,746,312 1,147,454 Mostly Pagan (VIII–IV) (VIII–IV, -tech –christ -inscr/pap) 16,312,159 605,335 5,463,292 334,187 Strictly Ancient (VIII–IV) (VIII–IV, +tech -christ +inscr/pap) Abbreviations: tech(nical literature), Christ(ian literature), inscr(iptions) and pap(yri). The Roman numerals in the first and second columns refer to centuries.73 70 71 72 73 de Lange, Greek Jewish Texts, p. 5. The plates of the ketubba may be consulted on pp. 309–313, with line 19 recto visible on p. 310. In stricter terms, a ghost lexico-semantic variant. Krauss, Lehnwörter, vol. 1, p. 204. The table has been compiled by Nick Nicholas; its original full form published on 07.07.2009 in ¨Jlljnisteúkontov: an Occasional Blog on Greek Linguistics, http://hellenisteukontos.blogspot.com/2009/07/lerna-vic-correction-of-word-form.html, last accessed 24.09.2010. GREEK LOANWORDS IN RABBINIC LITERATURE 215 On the other side, the corpus which constitutes the basis of Historical Dictionary of Hebrew Language consists, according to the data published by the Academy of Hebrew Language, of 9,078,725 word instances and covers the chronological span of II BCE–first half of V CE.74 Aramaic material does not form part of this corpus. Even if we presume that it is of comparable size and make a generous allowance for the material after the first half of V CE, it is clear that combined Hebrew/Aramaic data does not reach the size of contemporaneous Greek corpus. Thus, statistically the probability that an item which does not occur in the Greek corpus will occur in the Hebrew/Aramaic one is low. Some may argue that a low probability is still not a zero probability. Yet unless an unregistered word is a derivative from the known root, its reconstruction borders on pure guesswork because of the inadequacy of the writing system. In fact, it is possible to reconstruct Greek loans without the mainstream (non-Hebrew script) data but impossible to validate them without such data, as the criteria (c) and (d) can only rely on the Greek corpus as a whole. Thus, it is high time to give up chasing new Greek words in Rabbinic materials, an honourable but outdated quest of the Wissenschaft des Judentums period. 5.2 On the brighter side: Nullus liber est tam malus ut non aliqua parte prosit From the viewpoint of the history of research it appears that analysing unfortunate etymologies allows us to pinpoint various lexicographical and philological shortcomings which would otherwise pass unnoticed, as well as to make interesting – and new – semantic observations. Thus, the ‘added value of Jewish data’ does exist, but it resides not where the nineteenth-century authorities expected it to be. 5.3 Greek loanword research: practical aspects We believe that the application of the principles outlined in 3.1–3.2 is useful for assessing already collected material and offering new etymological solutions. The critical re-assessment of the accumulated data is 74 See http://hebrew-treasures.huji.ac.il under ‫( נתונים‬http://hebrew-treasures.huji.ac.il/ PopUpWindow.asp?openOn=2), last accessed 24.09.2010. A version of database is also available on CD Taqlitor “Maagarim Alef” – min ha-mea ha-sheniyya lifne ha-sefira ha-nozerit ad ha-maÌazit ha-rishona shel ha-mea he-Ìamishit la-sefira ha-nozerit (Jerusalem, 1998). 216 J.G. KRIVORUCHKO certainly necessary for every etymology starting from Krauss’s Lehnwörter and including those discussed above. For the sake of methodological correctness, the more problematic the transmission of the loanword, the stricter should be the morphological and semantic criteria applied to its historical reconstruction. As to the new material, it is advisable that it should be researched in the appropriate format. Strict adherence to the procedures described above requires much effort and space, and the results are unlikely to fit into a short note, but rather into a lengthy article or Wortgeschichtestyled monograph. Importantly, these formats would allow the derivative benefits of loanword research (see 5.2) to be fully exploited. It remains a question whether such inquiries into individual word-histories will be produced in sufficient numbers given the publish-or-perish realities of modern academe.