Studies in Rabbinic Hebrew

This volume presents a collec� on of ar� cles centring on the language of the Mishnah and the Talmud — the most important Jewish texts (a� er the Bible), which were compiled in Pales� ne and Babylonia in the la� er centuries of Late An� quity. Despite the fact that Rabbinic Hebrew has been the subject of growing academic interest across the past century, very li� le scholarship has been wri� en on it in English.


1.
The discourse unit of the halakhic give-andtake conversation and its features 57

INTRODUCTION
The present volume contains eight articles on topics related to Rabbinic Hebrew. Seven out of the eight are revised versions of papers read at the Rabbinic Hebrew Workshop that was held at the University of Cambridge on the 5th and 6th of July, 2016. The eighth, my own article, is a translated and revised chapter from my doctoral dissertation.
Since the establishment of the Regius Chair of Hebrew by Henry VIII in 1540 the study of Hebrew has occupied a permanent place in the Cantabrigian curriculum. 1 As might be expected, Rabbinics and Rabbinic Hebrew were of lesser interest to the academic community in Cambridge than Biblical Hebrew, at least during the first five centuries of the University's existence. But the second half of the 19th century saw important developments which secured Cambridge's place on the world map of Rabbinic studies: in 1875 Schiller-Szinessy was appointed Reader in Talmudic and Rabbinic literature; in 1877 Charles Taylor, the Master of St. John's College, published the Hebrew text of Tractate Aboth from Codex Cambridge of the Mishnah with an English translation; in 1883 Wlliam Henry Lowe published the entire text of the Cambridge Mishnah codex; and in 1890 Solomon Schechter was appointed as Schiller-Szinessy's successor, in which capacity, a few years later, he examined the genizah of the Ben-Ezra synagogue in Cairo -a collection that after its transfer to Cambridge would have an unparalleled impact on the world of Rabbinic studies in general and Rabbinic Hebrew in particular. It is my hope that this volume will be an additional contribution to Cambridge's long and distinguished history of Hebrew research.
The modern academic study of Rabbinic Hebrew, which originated in the first half of the 20th century with Moses Hirsch Segal's seminal article on Mishnaic Hebrew and his subsequent Grammar of Mishnaic Hebrew, 2 shifted to the new-born state of Israel in the second half of that century. The ground-breaking works of Jacob Nahum Epstein, Hanoch Yalon and Eduard Yechezkel Kutscher, as well as the works that followed them, were and continue to be written almost exclusively in Modern Hebrew, 3 making the field quite inaccessible to those unfamiliar with the language. Fortunately, the situation seems to be changing, and works on Mishnaic Hebrew appear more often in English. Special mention should be made to the volume of collected articles in the 37th instalment of Scripta Hierosolymitana, edited by Bar-Asher and Fassberg, and to the proceedings volume of the Yale Symposium on Mishnaic Hebrew, edited by Bar-Asher Siegal and Koller. 4

Introduction
It is a pleasure to express my gratitude to Prof. Geoffrey Khan, for wholeheartedly supporting the idea of holding a Rabbinic Hebrew Workshop, and for making it financially possible to organise it. It is largely due to his encouragement that both the Workshop and the present volume came into being. I would also like to thank all invited lecturers for their contributions and for meeting various deadlines, rendering the editing process smooth and effective.
I am especially grateful to the administrative staff of the Faculty of Asian and Middle Eastern studies, as well as the staff of Gonville and Caius College, for their kind yet indispensable assistance, both before and during the Workshop. Special thanks go to Open Book Publishers, and especially to Alessandra Tosi, for her patience and guidance, and to Luca Baffa, for expertly typesetting this challenging volume. And finally, I would like to express my special thanks to Aaron Hornkohl, for correcting the English language of the articles, for preparing the index, and for making numerous suggestions that improved the manuscript considerably.

Introduction
In the Babylonian Talmud there frequently occur two similar proper names that differ in spelling as well as pronunciation: ‫רבה‬ Rabba and ‫רבא‬ Rava; the former ends with a heh and has a doubled bet, while the latter ends with an alef and has singleton bet. Since these similar names tended to be confused with each other, Rav Hai Gaon was sent a question in which he was asked to attribute each name to the proper Amora. In his response he divided all the bearers of one of these names into two lists according to the correct form. At the end he added an explanation for the difference between the names -it stems from a difference between the nouns from which they are derived: At the outset Rav Hai explains that the name ‫רבה‬ Rabba derives from the compound ‫אבה‬ ‫רב‬ Rav ʾAbba, while the name ‫רבא‬ Rava derives from the compound ‫אבא‬ ‫רב‬ Rav ʾAva. According to this explanation, the difference between the proper names results from a difference between the nouns ʾabba and ʾava. He goes on to explain the difference between these nouns, which is one not only of spelling and pronunciation, but also of meaning: the meaning of ʾabba is 'my father', and that of ʾava is 'a father'. He concludes by bringing examples from the Aramaic Targum: the Hebrew ‫י‬ ‫בִ‬ ‫אָ‬ 'my father' translated by the Aramaic ‫ה‬ ‫ּבָ‬ ‫אַ‬ ʾabba, while the Hebrew ‫ב‬ ‫אָ‬ 'a father' is translated by the Aramaic ‫א‬ ‫בָ‬ ‫אֲ‬ ʾava. 3 2 Shraga Abramson, Tractate ʿAbodah Zarah of the Babylonian Talmud (New York: Jewish Theological Seminary, 1957), p. 129. The vocalisation is copied from the source. Another version of this responsum was published by Benjamin M. Lewin, ʾIggeret Rav Sherira Gaʾon (in Hebrew; Haifa: Golda-Itskovski, 1921), appendices, pp. xiv-xv, according to MS Parma 327, but this version is missing and incomprehensible, and it is a wonder that Lewin did not comment on this. 1

. Rabba and Rava, ʾAbba and ʾAva
It is not clear whether this distinction existed in the living language or only in the copying and reading tradition of the Targum. The structure of the response seems to point to living language, since the distinction is introduced at the outset, while the Targum is only presented at the end in order to supply a proof or an example. In any case, we have here an important testimony of a distinction so far unknown from any other source. This distinction deserves an explanation: how did this threefold distinction evolved, according to which ‫אבה‬ ʾabba means 'my father' while ‫אבא‬ ʾava means 'a father'? I will first introduce the classical forms in Hebrew and Aramaic relevant to our discussion:

1: a father 2: the father 3: my father
Hebrew: ʾav ha-ʾav ʾavi have been unable to find any text that preserves this distinction; see the appendix below. Of course, the parallel distinction between ‫רבה‬ Rabba and ‫רבא‬ Rava does exist. In the case of proper names there is a recognisable tendency to use heh for a final a vowel even in the Babylonian Talmud; see Yechiel Kara, "Babylonian Aramaic in the Yemenite Manuscripts of the Babylonian Talmud" (in Hebrew; PhD dissertation, The Hebrew University of Jerusalem, 1982), p. 41; Shamma Yehuda Friedman, "Early Manuscripts of Tractate Bava Metzia" (in Hebrew), Alei Sefer 9 (1981), pp. 5-55, at pp. 14-16. It seems that this tendency, together with the influence of Rav Hai's response and the necessity to differentiate between personalities, combined to preserve this distinction specifically in these proper names. However, even in these names it is not preserved in all sources, and this has led some scholars to conclude that the very distinction is not original; see Shamma Yehuda Friedman, "Orthography of the Names Rabbah and Rava in the Babylonian Talmud

Studies in Rabbinic Hebrew
Babylonian Aramaic: ‫אבא‬ ‫איתת‬ ‫ליה‬ ‫הוה‬ ‫יוחנן‬ ‫דר׳‬ ‫קריביה‬ 'a relative of R. Yochanan had a father's wife' (b.Ketuboth 52b); ‫אלא‬ ‫אמרן‬ ‫ולא‬ ‫בה‬ ‫לן‬ ‫לית‬ ‫דאימא‬ ‫באחי‬ ‫אבל‬ ‫דאבא‬ ‫באחי‬ 'and what we said concern only the father's brothers, but concerning the mother's brothers this is not valid' (b.Baba Metzia 39b). 8 Meaning 3 -ʾav: Mandaic: ‫דאב‬ ‫שותא‬ ‫תיהויא‬ ‫כמא‬ 'how will be the conversation of my father'. 10 Babylonian Aramaic: ‫דאב‬ ‫אבוה‬ ‫ליה‬ ‫אמ׳‬ ‫שתליה‬ ‫מאן‬ ‫חרובא‬ ‫האיי‬ 'who planted this carob tree, so he said, my father's father' (b.Taanith 23a according to He). 11 8 The text of the quotations from Rabbinic Literature, unless otherwise specified, is according to the text that is presented in the Maagarim database of the Historical Dictionary Project of the Academy of the Hebrew Language, accessible at http://maagarim.hebrew-academy.org.il. 9 The final yod in Syriac is only an archaic spelling, and the pronunciation is ʾav.

Western Aramaic
In Western Aramaic the distinction between the emphatic and non-emphatic forms was preserved, so the difference between columns 1-2 was maintained. On the other hand, the meaning of the emphatic form ʾabba was expanded to include meaning 3 'my father' and it supplanted the original form ʾavi altogether. 12 The following examples demonstrate only meaning 3 (in meanings 1-2 the original forms were maintained):  17 Accordingly, in contrast with Eastern Aramaic, where columns 1-2 merged, in Western Aramaic it was columns 2-3 that merged:

Mishnaic Hebrew
The Aramaic form ʾabba was borrowed into Mishnaic Hebrew and is very common in Rabbinic Literature. However, it is used only in the (new) meaning 'my father'. 18 It is never used in the original Aramaic meaning 'the father', where the original Hebrew form ha-ʾav is maintained. 19 Here are some examples of the different forms: 1. Rabba and Rava, ʾAbba and ʾAva that it does not appear in the Mishna; it may have been reintroduced towards the end of the Tannaitic period. This is the system in Mishnaic Hebrew: How did the form ʾabba reach column 3 ('my father')? There are two possibilities: either it was an independent process, similar to what happened in Western Aramaic, 25 or it is a borrowing from Mishnaic Hebrew. 26 Here we should point once again to 25 This possibility also depends on the question of the extent to which this phenomenon occurs in Syriac. As noted above, the normal form for this meaning in Syriac is ʾav. I  The difference between the columns is now explained: in Mishnaic Hebrew, ʾabba only exists in column 3 and has a dagesh. This form was borrowed by Babylonian Aramaic, and this is why the dagesh appears only in column 3. In columns 1-2 it does not exist in Mishnaic Hebrew and could not affect Babylonian Aramaic, so the original Eastern Aramaic forms were maintained. 28 This explanation may also account for the difference in spelling. In the Babylonian Talmud a final a vowel is marked by alef in Aramaic words and by heh in Hebrew words, e.g., ‫מעולם‬ ‫תנא‬ ‫בה‬ ‫אדם‬ ‫שנה‬ ‫לא‬ 'it was taught, no one ever repeated it' (b.Yoma 26a); ‫יממא‬ ‫דהני‬ ‫צנעא‬ ‫ליה‬ ‫אמ׳‬ ‫ביתו‬ ‫לתוך‬ ‫בצנעה‬ ‫שיכניסם‬ ‫ובלבד‬ ‫אביי‬ ‫ליה‬ ‫אמ׳‬ ‫הוא‬ 'Abbaye said to him, [have we not learnt that] he should bring them into his house privately? He answered, the day is the disappeared, while in Eastern Aramaic the original ʾav appears in all three dialects, so ʾabba seems to be foreign. 27 To the best of my knowledge, there is no proof of direct influence of Galilean Aramaic on Babylonian Aramaic, so the only language which can be considered is Mishnaic Hebrew. 28 Even if we assume that the use of this form developed independently and only the dagesh is influenced by Mishnaic Hebrew, in columns 1-2 it does not exist in Mishnaic Hebrew, so the original eastern form was preserved.
[time of] privacy for these' (b.Moed Katan 12b). 29 According to my suggestion, the word in columns 1-2 is written with alef as an authentic Aramaic word, while in column 3 it is written with heh because it was borrowed from Hebrew. For this explanation we need not assume a tradition of exceptional conservative power. In Babylonian Aramaic the form ʾava was the ordinary form. Speakers of Babylonian Aramaic were exposed to Tannaitic texts, where they found only ʾabba and only in the meaning 'my father', so the form and the meaning seemed to them connected. Since these two phenomena are typical of Hebrew texts, they viewed it as Hebrew, different from their Aramaic form ʾava.

Mishnaic Hebrew -a bridge between Western and Eastern Aramaic
According to this suggestion, the form ʾabba 'my father' was created in Western Aramaic, borrowed into Mishnaic Hebrew, and then made its way into Babylonian Aramaic. Both phenomena -influence of Western Aramaic on Mishnaic Hebrew and influence of Mishnaic Hebrew on Babylonian Aramaic -are well attested. 30 Accordingly, Mishnaic Hebrew, which was studied 1. Rabba and Rava, ʾAbba and ʾAva by Jews in Palestine and Babylon alike, became a bridge between Western and Eastern Aramaic. I will adduce another example for this process. The word ‫כאן‬ kan 'here' was created in Western Aramaic. Its Aramaic origin is proven by the lack of the Canaanite Shift (in contrast with its Hebrew cognate ko), and its Palestinian origin is proven by the addition of final nun. 31 This word was borrowed into Mishnaic Hebrew and then again into Babylonian Aramaic. As a result, we have in Babylonian Aramaic a doublet: the original Babylonian Aramaic ‫הכא‬ haḵa alongside the Western Aramaic loan kan. 32

Appendix: did the distinction of spelling survive in the manuscripts?
In the second footnote of this article I mentioned Shraga Abramson's conclusion, that the distinction of spelling according to meaning has not been preserved in the texts that have reached us. I have rechecked a list of manuscripts and have been unable Aramaic see Yochanan Breuer, "The Hebrew Component in the Aramaic of the Babylonian Talmud" (in Hebrew), Leshonenu 62 (1999), pp. 23-80. 31 See, e.g., Harold L. Ginsberg, "Zu den Dialekten des Talmudisch to confirm this distinction. I do not claim that such a distinction never existed. There is no reason to doubt Rav Hai's clear testimony that he was familiar with texts that exhibited this distinction, but so far we have not been able to trace them. It is true that the spelling with heh is widespread in certain manuscripts, and one may conclude that this distinction does exist in them. 33 Therefore I would like to present the considerations for my claim that this distinction has not yet been found.
In my view, the distinction is proven only if the two spellings are distributed according to meaning, not according to language; i.e., if one spelling is typical of Hebrew and one of Aramaic, then the spelling is governed by language, not by meaning. Since within Hebrew ʾabba is used in only meaning 3 ('my father'), this distinction cannot be found in Hebrew. Therefore, the question is only if this distinction is to be found in Aramaic. In order to check it, I chose a group of texts where a spelling with heh was preserved, and separated the data between Hebrew and Aramaic. 34 I omitted proper names altogether, since according to the testimony of Rav Hai there are two distinct proper names, ʾabba and ʾava. In proper names it is impossible to know, whether by form or by context, the meaning of the name and, consequently, whether the spelling is dependent on the meaning. Spelling of names is thus useless for this investigation. Alef: ‫ברא‬ ‫אפום‬ ‫אבא‬ ‫ונכיסנא‬ 'I will slaughter father with son' (25b).

Meaning 2:
Alef: ‫דאבא‬ ‫שוקיתא‬ ‫מן‬ ‫דאימא‬ ‫קולפי‬ ‫טבא‬ 'the blows of the mother are better than the kisses of the father' (106a). According to these findings, the spelling with heh is widespread in Hebrew, but rare in Aramaic, as will be emphasised by two facts: (1) in Aramaic the spelling with heh occurs only four times, which is less than 4 percent of the occurrences of this word in Aramaic, and a little more than 6 percent of the occurrences of this word in meaning 3 in Aramaic. If we add to the total the Hebrew and the proper names, these four occurrences become such a small portion that no conclusion can be based on them.
(2) In the book of Halachot Pesuqot, there are twice as many occurrences of the spelling with heh in Hebrew as with alef, while in Aramaic there is no spelling with heh whatsoever. Accordingly, in these texts the spelling with heh is typical only of Hebrew, and if so, the spelling is dependent on language, not meaning.
This survey also explains the illusion that the distinction does exist in these texts: since the spelling with heh is widespread in Hebrew and is restricted to meaning 3 (which is the only meaning in Hebrew), while in Aramaic the normal spelling is with alef and is used in all meanings, it seems as if the spelling with heh is typical of meaning 3. However, separating the languages leads to the opposite conclusion: this distinction exists neither in Hebrew -where only meaning 3 exists, nor in Aramaic -where only the spelling with alef exists (with a few exceptions). 26

Studies in Rabbinic Hebrew
As to provenance, MSS Kaufmann and Parma A originated in Italy, 4 whereas MS Cambridge is a Byzantine manuscript, as evidenced by its codicological and palaeographical features. 5 Whereas Mishnaic Hebrew traditions in Italy are reflected in many sources -manuscripts, incunabula, maḥzorim, among others -and have merited substantial research, 6 the Byzantine tradition, in contrast, suffers from sparsity of sources and research. The study of Byzantine Jewry remained frozen for years until the turn of the twenty-first century, which saw the publication of texts from the Genizah by Nicolas de Lange and seminal studies by Israel Ta-Shma. 7 Although the precise nature of this community's tradition has yet to made clear, its ties to Eretz-Israel and its unique facets are beginning to emerge. 8  Nonetheless, the scribe-vocaliser of MS Cambridge has sporadically inserted partial vocalisation. 10 My use of the term 'scribe-vocaliser' here is deliberate: the manner of vocalisation, the ink, and its colour all attest that the text was penned and vocalised by the same person. 11 Most of the more than two hundred vocalised words in this manuscript were documented by William Henry Lowe, the editor of the version of the text known as The Mishnah of the Palestinian Talmud (Cambridge, 1883); others, however, escaped his notice or were misunderstood.

Studies in Rabbinic Hebrew
This raises the question of what led the scribe-vocaliser to vocalise these words in particular. In general, we can say that the vocalisations found in MS Cambridge serve to underscore or elucidate a textual variant or particular reading from this fifteenth-century Byzantine vocaliser's tradition, similar to the partial vocalisation found in manuscripts of other rabbinic texts, such as MS Erfurt of the Tosefta. 12 The sporadic vocalisations in MS Cambridge mirror a process whereby the vocaliser considered the different reading traditions of the Mishnah with which he was familiar, and decided either in favour of his own tradition or one that seemed worthy or correct. Thus, not only were specific, accurate, and unique reading traditions of the Mishnah preserved in fifteenth-century Byzantium, but it appears that its scribevocalisers were also familiar with alternative readings.
These partial vocalisations reveal both the uniqueness and the trustworthiness of the Byzantine tradition reflected in MS Cambridge. On the one hand, this tradition shares some of the features of the punctilious Italian tradition; on the other hand, as shown below, in some instances the Byzantine tradition also preserves earlier, more precise features than those found in the Italian tradition.
Nonetheless, MS Cambridge also indirectly reflects latefifteenth-century traditions. The vocalisations attest to the vocaliser's familiarity with these traditions, which were not necessarily of the highest accuracy. The purpose of his partial vocalisation of words was to highlight his ancient Palestinian tradition; in effect, through these partial vocalisations and superior textual traditions he preserved an early Byzantine tradition with parallels in MSS Kaufmann and Parma A, which predate Cambridge by several centuries.

Nusaḥ: textual variants
As noted, the presence of a vocalised word in a largely unvocalised text cannot be dismissed as a slip of the pen, but rather reflects particular interest on the vocaliser's part. Although unique textual variants are by no means rare in MS Cambridge, they are not systematically vocalised there. Evidently, the vocaliser generally thought one vocalised example per variant in the manuscript sufficient. It is the conjunction of a variant with additional factors that might interfere with the transmission of his tradition, which impelled the scribe-vocaliser to vocalise a word. The use of vocalisation confirms the vocaliser's familiarity with other reading traditions of the Mishnah that differ from the one he wished to transmit. Thus, vocalisation of the word can function to support a disputed reading.  17 Note that the orthography of MS Cambridge is usually defective. Thus, the word ‫חודש‬ is almost always spelled defectively there, 18 and the unknown phrase composed of two identical words ‫החדש(‬ ‫)החדש‬ would certainly lend itself to correction or erasure. As a means of stressing the correctness of his version, the scribe vocalised both words to indicate that this is not mistaken dittography.  19 Focused mainly on the omission of the initial alef and its implications for the provenance and dating of the texts, less attention has been paid to the influence of the silent alef on the realisation of the names and the status of the ayin.

‫ש‬
Did the name ‫לעזר‬ retain its biblical form ‫ר‬ ‫זָ‬ ‫עָ‬ ‫לְ‬ lʿazar even without the alef, or did additional changes take place when the alef was dropped, perhaps due to the weakness of the guttural ayin that followed it? 19  Two types of sources assist in clarifying how this abbreviated name was realised: transcriptions, on the one hand, and vocalisation traditions, on the other. The transcriptions into Greek in the Gospels and other literary sources attest to a pronunciation close to the biblical one, e.g., Ελαζάρον, Ελεαζάρον, λεαζάρος, 20 and to a new realisation, Λάζαρον, as the name of contemporary individuals. 21 On the other hand, the vocalisation traditions reflected in the various manuscripts of the Mishnah evidence only a pronunciation close to the biblical one: ‫ר‬ ‫זָ‬ ‫עָ‬ ‫]לְ‬ ‫. [אֶ‬ 22 The vocalisation ‫ר‬ ‫זַ‬ ‫עְ‬ ‫לַ‬ found in MS Cambridge, with a vowel under the first consonant, is supported by some of the transcriptions, but diverges from the general picture derived from manuscripts of the Mishnah. Although this might suggest that this vocalisation reflects the late Byzantine tradition of the scribe-vocaliser, this is not the case. Direct evidence for this vocalisation comes from a Genizah fragment of the Mishnah (T-S E1.57), 23 and a twelfth-century Oriental manuscript of tractates Aboth and Zebahim. 24 25 Thus, on the margins of the literary transmission that remained close to the biblical realisation there were also vernacular pronunciations that attest to metathesis. Perhaps the movement of the vowel to the consonant lamed was supported by the weak ayin, 26  i.e. ‫סלתים‬ ‫;של‬ 28 Hebrew/Aramaic words and phrases: ‫שליפרומבייה‬ (Kelim 11.5); 29 ‫שלמים‬ (Shekalim 6.3, Yoma 2.5, Sukkah 2.5, 4.9, Baba Bathra 4.6, Middoth 2.6); 30 and our current example, ‫.שלרופיים‬ The preservation of proximity in these instances is the result of a unique spelling that prevented subsequent separation.
Clearly, the preservation of ‫של‬ juxtaposed to ‫רופיים‬ shows that the spelling of ‫,שלרופיים‬ for which I have found no parallels, is not a corruption, but rather a form preserved because of its unusual spelling. The vocalisation of the entire word also witnesses the scribe-vocaliser's desire to indicate that this form is neither a mistake nor a corruption.
This word displays another unique feature, which is the alef > yod shift. Much has been written on this exchange. 31 However, in his comprehensive treatment Breuer has shown that a distinction must be made between yod > alef and alef > yodshifts and that the alef > yod shift is the result not of a phonological process, but of a morphological exchange. He demonstrates that in MH the alef > yod exchange is not free, but takes place in the III-alef pattern, which became identical with the III-yod pattern. 32 This explanation, however, does not fit ‫ים‬ ‫יִ‬ ‫,רֹופְ‬ the word under discussion here, because the expected result of such 28 The spelling with samekh hid the ‫של‬ from the separators. 29 The plene spelling apparently kept the ‫של‬ from being separated. With respect to the first of the two bets, this hapax has two vocalisation traditions in manuscripts of the Mishnah: 36 one (Parma A) has dagesh lene; the other Cambridge (and Paris) indicates a fricative after the resh. 37 In MS Kaufmann, we find signs of hesitation: the consonant bet has a faded dagesh, but closer examination of the word suggests that the dagesh was blotted close to its writing. 38 On the other hand, MS Kaufmann does not mark rafeh over the bet. Perhaps the vocaliser of MS Kaufmann debated the matter and decided to take no steps, whereas the vocaliser of MS Cambridge used vocalisation to underscore the fricative bet in his tradition against the backdrop of another, opposing tradition that stresses the plosive bet, here represented by Parma A.

‫ין‬ ‫יִ‬ ַ ‫ד‬ ‫אֱ‬
The Mishnah in Nedarim 11.10 states: ‫המשיא‬ ‫אף‬ ‫אומ׳‬ ‫יהודה‬ ‫רבי‬ ‫נערה‬ ‫היא‬ ‫ין‬ ‫יִ‬ ַ ‫ד‬ ‫אֱ‬ ‫אצלו‬ ‫וחזרו‬ ‫ניתגרשה‬ ‫או‬ ‫ניתאלמנה‬ ‫פי‬ ‫על‬ ‫אף‬ ‫קטנה‬ ‫בתו‬ ‫את‬ 'R. Judah says: also if one gave in marriage his daughter who was a minor, and she became a widow, or she was divorced and returned to him, and she was still a maiden'  Kutscher's analysis, that the adverb ‫עדיין‬ is composed of ‫עד‬ + another element -the plural pronominal suffix ‫ינּו(‬ ֵ ‫ד‬ ‫)עָ‬ or ‫ן/אן‬ ‫יִ‬ ‫אַ‬ -has been accepted in scholarship. 39 As for the different forms, Kutscher proposed that the Hebrew word was borrowed from Akkadian adīni and that in Biblical Hebrew the initial alef became ayin, i.e., ‫,עדן‬ ‫,עדנה‬ due to mistaken affinity, renewed by biblical scribes and MH, to Hebrew ‫.עד‬ This suggested circular process, in which ‫עדיין‬ returns to its original source through a 'mistaken' folk etymology, seems somewhat convoluted. It is perhaps simpler to assume that what we have here is the known alef/ayin alternation in MH. 40 The textual witnesses are divided as to the first consonant of ‫:עדיין‬ alef or ayin. 41 The Genizah fragments analysed by Birnbaum attest exclusively to alef. 42 MS Kaufmann and the Babylonian tradition tend toward alef, although forms with ayin are found there, 43 whereas MS Parma B has both forms in equal distribution. 44 MSS Parma A and Cambridge of the Mishnah represent an opposite direction: the usual spelling there is ‫,עדין/עדיין‬ with a single exception that reads ‫.אדיין‬ 45 In other sources of MH the form with ayin is the dominant one, as shown by Yeivin,Sharvit,and Breuer. 46 It appears that the uniqueness of the form with initial alef in MH sources in general, and in MS Cambridge in particular, led to its vocalisation as a means of its preservation.

‫יו‬ ‫הֱ‬
The vocalisations in MS Cambridge are also found in verbal forms. Here I address only one instance. Sanhedrin 4.5 describes the process of questioning witnesses in capital cases:  that the latter give truthful testimony. The quote begins with 'Perhaps you will state' and concludes with a prooftext from the Bible and a halakhic midrash on the verse cited. As is characteristic of direct speech, it addresses the audience in the second person plural -‫,תאמרו‬ ‫,אתם‬ ‫אתכם‬ -and the speakers refer to themselves in first person plural -‫.שסופינו‬ This makes it certain that the verb ‫,היו‬ which is inserted in the direct speech, refers to the witnesses and functions as an imperative. 47 The root ‫הי״ה‬ is conjugated in two ways in MH: as II-yod form and as a II-waw form. 48  imperative. Given the consistent testimony of all the manuscript witnesses, I differ from Haneman, who contends that the original conjugation of the second person plural in the qal stem was only with waw, and that our example is an anomaly, perhaps even a graphic exchange of waw and yod. 51 Examination of the distribution of the roots ‫הו״י/הי״י‬ in this pattern in MSS Cambridge and Kaufmann elicits an opposite picture from that found in Parma A. ‫היו‬ appears three times with yod (in our mishnah, in Aboth 1.1, and in Aboth 1.3), and ‫הוו‬ only once (in Aboth 2.3). In MS Kaufmann it appears three times with yod (once in our mishnah and twice in Aboth). 52 A similar picture also emerges from other sources. 53 This contrasts with the second person singular that is usually found in the root ‫.הו״י‬ In essence, not only did the vocaliser of MS Cambridge vocalise the word correctly, he was aware of both the problematic nature of this form and the alternative tradition ‫יו‬ ‫.הָ‬ This is another example of how he underscores his tradition. 54 Hillel's statement and appended explanation that a person must employ his teacher's style of expression have sparked much debate and varied interpretations in the relevant scholarship. 55 The phrase ‫הין‬ ‫מלא‬ presents the main difficulty, and the different traditions diverge in their understanding and realisation of this phrase, as seen from the variant readings cited above. Nonetheless, additional sources support the tradition represented in MS Cambridge, which reads the vowel a in the second radical. 56 Eliezer Shimshon Rosenthal treats this expression at length and has shown that we must follow the version found in Maimonides and an ancient interpretation from geonic responsa, which indicate that this is the active participle of an Aramaic form of the root ‫:מל״א‬ ‫ין‬ ‫אִ‬ ‫מלָ‬ meaning 'to fill', and is therefore connected neither to ‫מלוא‬ nor to ‫.הין‬ 57 The vocalisation ‫הין‬ ‫א‬ ‫מלָ‬ is found in other sources, as Rosenthal notes. However, among the manuscripts of the Mishnah, MS Cambridge is the sole manuscript that has retained this reading.

‫ה‬ ‫הסֹוכַ‬
In where the word appears for the first time, MSS Cambridge and Parma A vocalise it ‫ה‬ ‫.סֹוכַ‬ Note that in Parma A this word appears in a long continuous section of unvocalised text; nevertheless, the vocaliser of Parma A chose to vocalise this word alone, affirming its unique tradition. 59 In MS Kaufmann, on the other hand, the entire line from ‫מה‬ ‫שהיזיק‬ to ‫סוכה‬ is unvocalised. In the facsimile edition there is a dagesh in the kaf of ‫;סוכה‬ in the scanned MS, however, there is no dagesh. The Arukh (s.v. ‫)סך‬ also attests to the version without dagesh in Baba Kamma and connects it to biblical ‫.שוכה‬ As Bar-Asher notes, Parma B always reads ‫סֹוכה‬ and Paris ‫;סוּכה‬ ‫סוּכה‬ is also attested by the vocaliser of MS Kaufmann (in Makhshirin) and K 2 (i.e., the second vocaliser, 'Kaufmann 2', in Zabim). 60 These are, in effect, two nouns that appear in MSS Cambridge, Parma A, and Parma B, where a distinction is made between ‫ה‬ ‫סֹוכָ‬ 'branch' and ‫ה‬ ‫סּוּכָ‬ 'shelter', 61 whereas MSS Kaufmann (once), K 2 , and Paris unite the two nouns in the common ‫ה‬ ‫עָ‬ ‫ּפֻ‬ pattern. What  emerges from this consideration is that the sole witness to ‫סוּכה‬ in this meaning of 'branch' is found once in the vocalised version in MS Kaufmann; all the other witnesses are from second-rate manuscripts.
Bar-Asher thinks that this is not an indication of a mistake on the part of the vocalisers, but rather root or pattern alternations ‫;סוך-סכך(‬ pattern alternation: ‫ה‬ ‫עָ‬ ‫.)ּפולה-ּפֻ‬ 62 But given the quality and number of witnesses to ‫ה‬ ‫,סֹוכַ‬ this suggests that the testimony of the manuscripts that distinguish between ‫ה‬ ‫סֹוכַ‬ and ‫סוּכה‬ represents an original, reliable tradition, whereas the unifiers blurred (in a natural, early or late process) the distinction between two close but different meanings. In any event, MS Cambridge highlights the fricative version.

‫ף‬ ‫הסֵ‬
Another noun for which the traditions of Mishnaic Hebrew reflect different patterns is ‫.הסף‬ 63 Its vocalisation twice in MS Cambridge witnesses its vocaliser's adherence to his task of elucidating his tradition.
One This noun appears seven times in the Mishnah: in five of these occurrences MS Cambridge's version is plene with a single yod; it is written defectively twice. The manuscripts of the Mishnah attest to two patterns for this noun: the segholate pattern with the extended diphthong ‫ל‬ ‫יִ‬ ‫,קַ‬ and its contracted diphthong ‫ל‬ ‫,קֵ‬ similar to the nouns ‫ל‬ ‫יִ‬ ‫יל-לַ‬ ‫,לֵ‬ ‫ל‬ ‫יִ‬ ‫יל-חַ‬ ‫.חֵ‬ 64 Since the material has already been analysed by Bar-Asher, I restrict my discussion to mapping the distribution of the forms in the various manuscripts vis-à-vis MS Cambridge. 65 One tradition (the scribe of MS Kaufmann 66 and MS Paris) attests only the pattern ‫ל‬ ‫יִ‬ ‫קַ‬ and is familiar mainly with the doubleyod spelling. 67 A second tradition (Parma B, and MS Kaufmann in Kelim 14.5, where, it seems, an original ‫ייף‬ ‫ּסַ‬ ‫הַ‬ was later corrected to ‫ף‬ ‫ּסֵ‬ ‫)הַ‬ attests the contracted form ‫יף‬ ‫.סֵ‬ The third (Parma A) knows both alternatives and the three spellings.
It is difficult to identify the tradition reflected in MS Cambridge. On the one hand, it underscores the defective spellings by vocalising them with ṣere, and the plene always has one, not two, yods. On the other hand, because of this manuscript's preference for defective spelling, a single yod could be understood as an extended diphthong. Perhaps the double vocalisation in this 64

Orthography: Homographs
Another sphere that invites vocalisation is that of orthography. As noted above, MS Cambridge is largely unvocalised. Moreover, it consistently adheres to defective spelling, not only in closed but also in open syllables. 74 Defective spelling inevitably creates homographs; we therefore find the use of vocalisation to distinguish between them. Vocalisation can also serve to refine a discussion or a textual reading. 75 A significant example comes from Abodah Zarah, in which three words in the same mishnah are vocalised.

‫דודיך‬
The Mishnah in Abodah Zarah 2.  The feminine form ‫ה‬ ‫לָ‬ ‫טָ‬ is a hapax in the Mishnah. In MSS Cambridge, Kaufmann, and Parma A it appears in the ‫ה‬ ‫עָ‬ ‫ּפָ‬ pattern, like ‫ה‬ ‫אָ‬ ‫.נָ‬ MS Paris has shewa in the first radical, whereas the Yemenite tradition and the printed editions, both early and late, have a noun that differs consonantally: ‫.טליה‬ 77 Examination of the manuscripts of the Mishnah and of various traditions suggests we are dealing with two separate patterns, which resulted in suppletion: on one hand, ‫ה‬ ‫לֶ‬ ‫טָ‬ (ms), ‫ה‬ ‫לָ‬ ‫טָ‬ (fs), ‫ים‬ ‫לִ‬ ‫טָ‬ (pl), based on the pattern of ‫ה‬ ‫פֶ‬ ‫יָ‬ (ms), ‫ה‬ ‫פֶ‬ ‫יָ‬ (fs), ‫ים‬ ‫פִ‬ ‫יָ‬ (pl), and on the other hand, ‫ה‬ ‫לֵ‬ ‫*טְ‬ )> ‫י‬ ‫לִ‬ ‫,*טְ‬ ms), ‫טליה‬ (fs), ‫ים‬ ‫יִ‬ ‫לָ‬ ‫טְ‬ (pl), based on the pattern of ‫י‬ ִ ‫ד‬ ‫ּגְ‬ (ms), ‫גדיה‬ (fs), ‫ים‬ ‫יִ‬ ָ ‫ד‬ ‫ּגְ‬ (pl). 78 The first pattern is seen in the BH and MH masculine form ‫ה‬ ‫לֶ‬ ‫,טָ‬ and the feminine form ‫ה‬ ‫טלָ‬ is attested in reliable manuscripts of the Mishnah, as presented above. The plural form ‫ים‬ ‫לִ‬ ‫טָ‬ is found three times in MS Parma A (in Tamid 3.3), but is also attested by the scribe of MS Kaufmann. Although this scribe generally uses the plene form with consonantal yod, 79 in this case he almost uniformly writes ‫טלים‬ defectively (five of six occurrences). 80 The defective form ‫טלים‬ is also found at Qumran, in both biblical and non-biblical texts, and even in MS Leiden of the Palestinian Talmud  Additional evidence for this pattern comes from the plural declension found once in the Mishnah. The phrase ‫קרבן‬ ‫טלאי‬ ‫לשכת‬ (Middoth 1.6), with the biblical plural, is found in the printed editions; in the manuscripts, however, it is declined according to the first pattern: MS Parma A reads ‫קרבן‬ ‫ה‬ ‫לֶ‬ ‫,טְ‬ which can be interpreted as an orthographic alternation between the ‫-י‬ and ‫-ה‬ suffixes. 82 Note that Parma A vocalises this word, even though it appears in an unvocalised section of the manuscript. This isolated instance of vocalisation highlights the rare form. In MSS Kaufmann and Paris a similar version was preserved, but with a lamed/resh alternation: ‫קרבן‬ ‫.טרי‬ 83 The second pattern is represented mainly by the biblical plural form ‫טלאים‬ and the Mishnaic Hebrew form ‫.טליים‬ The latter is the tradition adhered to consistently by the vocaliser of MS Kaufmann (see above). This form appears four times in MS Cambridge 84 and in Parma A as well. 85  We therefore have here two pattern systems that have already undergone suppletion in the Bible: ‫.טלה-טלאים‬ In the Mishnah, however, the conjugation of ‫לה‬ ‫טָ‬ expanded and is found in the feminine and in the plural forms. In Palestinian Aramaic we find ‫.טלי-טלייה-טליין‬ 87 This reveals the struggle between the two patterns. Although ‫טלאים‬ and ‫טליים‬ are supported by the Bible and by Aramaic, the forms ‫טלה-טלה-טלים‬ continued to exist. With respect to the forms ‫ה‬ ‫יָ‬ ‫לְ‬ ‫טַ‬ and ‫ה‬ ‫ּיָ‬ ‫לִ‬ ‫,טְ‬ found in the Yemenite tradition and the printed editions, respectively, it is difficult to determine if they were created by analogy to the second, dominant pattern or reflect an early tradition.

Ketiv and qere
Another characteristic of MS Cambridge is the small number of corrections. The manuscript was penned by one or two scribes with an eye to penmanship and design; it appears, however, that, following its completion, the manuscript was set aside and not studied. 88 The few corrections made during the writing process are attested here and there in delicate signs of erasure, 89 or superlinear dots that mark incorrect word order. 90

The Vocalisation of MS Cambridge of the Mishnah
Marginal notes mentioning variants 91 and additions of words or letters above the line by the scribe 92 are also found sporadically in the manuscript. For the most part, the scribe took care not to make corrections or erase textual variants. I argue that the scribe used vocalisation to resolve the conflict between his desire to adhere closely to a particular nusaḥ, on the one hand, and the need to correct it, on the other hand. Indeed, there are instances of ketiv and qere in MS Cambridge.

‫ראשית.‬ ‫בו‬ ‫שיש‬ ‫לשני‬ ‫ראשון‬ ‫ומעשר‬ ‫ראשית‬
And whence that first-fruits come before priest's-due? after all, the one is called priest's-due and the first, and the other is called priest's-due and the first. But first-fruits come first because they are the first-fruits ‫ים[‬ ִ ‫ר‬ ‫יכֵ‬ ‫]בְ‬ of all produce; and priest's-due precedes first tithe since it is termed first; and first tithe before second because it includes the first. as cited in Melekhet Shlomo ad loc.). MS Kaufmann, on the other hand, presents the spelling and vocalisation ‫ים‬ ִ ‫ּכּור‬ ‫,ּבִ‬ ostensibly an expansion of its meaning of 'the result of an action'.
The version in MS Cambridge, with yod in the first syllable, may represent a vocal shewa spelled plene, but this seems unlikely. 93 It may also reflect indecision as to the correct version: that of MS Kaufmann (vocalising the initial syllable with yod) or the versions that appear reasonable based on the context and other manuscripts (defective spelling in the second syllable). Here the vocaliser settled matters without intervening in the consonantal text. This mishnah deals with the purity or impurity of liquids, and sets the Halakhah -pure or impure -for various situations. In this instance, we have two casks, one of which is pure; the other is impure. The continuation 'kneaded dough from one of them' refers to the casks mentioned in the previous sentence. The second phrase concerning the doubt as to whether the water came from the pure cask also leads to this conclusion. The expected version ‫טמאה‬ does appear in MSS Parma A, Parma B, and Paris, but MSS Cambridge and Kaufmann have an identical example of ketiv and qere: the ketiv is ‫הטומאה‬ and the qere is ‫.הטמאה‬ Ketivim of ‫טמאה‬ as ‫טומאה‬ appear in six other places in MS Kaufmann (Kelim 10.8; Negaim 6.2, 13.8; Tohoroth 4.10, 6.3, 6.4), 94 and also in MS Vatican 60 of Sifra we find ‫ה‬ ָ ‫טהֹור‬ ‫לִ‬ ‫ה‬ ‫אָ‬ ‫טמֵ‬ ‫מִ‬ ‫ה‬ ‫אָ‬ ‫ומֵ‬ ‫טְ‬ ‫לִ‬ ‫ה‬ ָ ‫הֹור‬ ‫טְ‬ ‫,מִ‬ with the waw in the last word crossed out. The many occurrences in MS Kaufmann, whose version is supported by MSS Cambridge of the Mishnah and Vatican of Sifra, clearly testify to a stable tradition of ‫טומאה‬ in the sense of ‫טמאה‬ and negate the argument that this is a mistake or simply a copyist's error. This is another example of a common phonological phenomenon in Mishnaic Hebrew: variation before a labial consonant and the realisation ṭəmeʾa as ṭumeʾa. This variation often takes place in Mishnaic Hebrew between vowels, usually in closed syllables. 95 This word, however, provides evidence of the variation of an ultra-short vowel (vocal shewa) before a labial consonant. But additional sources from this period attest to vowel variation in this position: the Isaiah Scroll from Qumran, Palestinian Aramaic dialects, and Greek transcriptions, as Kutscher has shown. 96 Thus, in Mishnaic Hebrew the influence of labial consonants extended to ultra-short vowels. 97

Foreign Words
Any discussion of the vocalisation in MS Cambridge must address the scribe-vocaliser's treatment of foreign words. Some 10 percent of the vocalised words belong to this category and they are mainly Greek words. This phenomenon is important, as is the vocalisation of these words, because it may assist identification of the precise region in Byzantium where the scribe-vocaliser resided. To date, however, it has proven impossible to identify the specific locale.
This differs from what we find in other manuscripts of the Mishnah: in MS Paris, for example, most of the unvocalised words are foreign, which suggests 'that he did not know how to read them'. 98 In contrast, the vocaliser of MS Cambridge chose to vocalise these words specifically; moreover, his vocalisation represents a tradition that can at times differ in terms of spelling and vocalisation from the tradition of other manuscripts of the Mishnah. Two examples follow.  The origin of this noun is the Greek βῆμα. 100 Most of the rabbinic sources that vocalise this word attest to ḥireq in the first syllable, 101 with the exception of its rare vocalisation with an e-vowel in MS Cambridge and a Genizah fragment.

‫ה‬ ‫ימַ‬ ‫בֵ‬
In his discussion of loanwords, Heijmans describes the realisation of the Greek vowel η over time and determines that it was pronounced [e] in the Hellenistic-Roman period, but that a shift from [e] to [i] took place in Byzantine times. He sees the pronunciation with ḥireq as reflecting a late realisation of the Greek η. 102 Thus MS Cambridge reflects an earlier form as compared to those found in other manuscripts.

Studies in Rabbinic Hebrew
Cambridge has i. The realisation a for Greek eta is strange, and apparently represents a development later than the realisation with i. 104 Heijmans argues that the person who vocalised with i knew the Greek word as pronounced after the Greek [e]>[i] shift. In any event, the ḥireq found in MS Cambridge has a basis in a known process that took place in Greek and seems to reflect knowledge of this form.

Conclusion
I have presented here only a fraction of the vocalised words scattered throughout MS Cambridge of the Mishnah. I have attempted to demonstrate that these select examples reflect deliberate choices on the vocaliser's part. MS Cambridge shares some superior traditions -as reflected in the words ‫טמאה-טומאה‬ ‫,אדיין‬ ‫,הסוכה‬ ‫,הסף‬ ‫,בכרים‬ ‫,טלה‬ -with Italian manuscripts; others, such as ‫,היו‬ ‫,לעזר‬ ‫,מלאין‬ ‫,רופיים‬ ‫בימה‬ and ‫,מילפפון‬ are uniquely Byzantine. In addition, we have seen that, despite its relatively late date, MS Cambridge reflects a superior, Byzantine tradition of MH, which is supported by the witnesses of the Italian tradition, MSS Kaufmann, and Parma A. On the other hand, we have also seen that the Byzantine tradition has unique features that are undoubtedly early and accurate. This enables us to add to our knowledge a hidden, ancient Palestinian tradition that circulated in Byzantium. This independent tradition evidences affinity to the other extant, superior sources of Mishnaic Hebrew.

CONVERSATIONS IN THE MISHNAH
Rivka Shemesh-Raiskin

The discourse unit of the halakhic giveand-take conversation and its features
Two types of halakhic texts form the core of Tannaitic literature, in general, and of the Mishnah, in particular: the formulation of law and halakhic give-and-take. The formulation of law is an abstract presentation of the laws, whereas halakhic give-and-take is a presentation of the Sages' views on halakhic subjects in order to determine the laws. For example, citation [1] presents a formulation of law concerning the onset of a fast undertaken because of a drought:  The context of halakhic give-and-take may include not only the presentation of the views of the debating parties in succession, but also the actual debate between them regarding their views. In such cases, a halakhic give-and-take conversation takes place.
For example, citation [3] begins with a presentation of the views of R. Eliezer and R. Joshua regarding when one should start praying for rain in the Amida prayer. This is followed by a halakhic give-and-take conversation between the two sages, including two exchanges between them: 3 [3]  Said R. Joshua: Since rain during the holiday is but a sign of a curse, why should one make mention of it?
R. Eliezer said to him: He, too, does not ask [for rain], but only mentions 'who causes the wind to blow and the rain to fall' in its due season.
He said to him: if so, one should mention it at all times.
A halakhic give-and-take conversation must contain at least one exchange between the discussants, that is, an expression of the comments spoken by an addressor and an addressee or an expression of the comments spoken only by an addressor. The first exchange in the conversation, which is often the only one, begins at the place where a real conversation between the debating parties begins. Occasionally, the exchange appears after the presentation of the views of one or both of the parties, but the presentation of the views is not included in the halakhic give-and-take conversation itself. 4 In other words, the halakhic give-and-take conversation begins at the stage of the exchanges rather than at the stage of the 3. The Halakhic Give-and-Take Conversations in the Mishnah presentation of views. The presentation of views and the give-andtake conversation are separate discourse units.
For example, citation [3] begins with a presentation of the views of R. Eliezer and R. Joshua regarding when one should start praying for rain in the Amida prayer. The halakhic give-andtake conversation after the presentation of these views begins with R. Joshua's question, because it is only from this point that the other party's response begins. This conversation contains two exchanges. The first exchange is made up of two parts and includes R. Joshua's question and R. Eliezer's response. The second exchange contains R. Joshua's assertion, which raises an additional difficulty regarding R. Eliezer's view; this is a partial exchange since it does not contain the other party's response.
Halakhic give-and-take conversation is a part of argumentative discourse. Muntigl and Turnbull employ the term 'conversational arguing' for this type of discourse, and present other terms for it that are used in the research, such as 'disputing', 'conflict talk', and 'oppositional argument'. 5 In their view, conversational arguing involves the conversational interactivity of making claims, disagreeing with claims, countering disagreements, along with the processes by which such disagreements arise, are dealt with, and are resolved. Arguing has been studied in numerous disciplines, including philosophy, rhetoric, anthropology, sociology, psychology, and linguistics. 6 Halakhic give-and-take conversation functioning as argumentative discourse therefore has three prominent characteristics: (a) it is dialogic in nature; (b) it represents a controversy between the discussants; (c) and it has a suasive goal. a) Dialogic nature: This characteristic is reflected in the fact that halakhic give-and-take conversation expresses an actual spoken dialogue held between discussants, whether conversation held in the Tannaitic and Amoraic literature is viewed as reflecting an actual discussion between sages or as the product of redaction that presents these dialogues as conversations of this kind. Various scholars have discussed these two approaches as they apply to the nature of conversations in Tannaitic literature. Albeck describes the discussions between Tannaim as generally being face to face, and occurring in the Sanhedrin, the seat of the president, in private study halls, as well as while the Tannaim were strolling along. 7 Sharvit explains that some Talmud researchers and language scholars have interpreted the saying ‫שאדם‬ ‫רבו‬ ‫כלשון‬ ‫לומר‬ ‫חייב‬ 'because a man must employ the style of expression of his teacher' (Eduyoth 1.3) to mean that R. Judah the Prince, the redactor of the Mishnah, did not edit the words of the Tannaitic rabbis, and instead quoted them verbatim, since, as he notes in this statement, the Tannaitic scholars themselves were careful to cite the laws in the actual words of their rabbis. 8 De Vries believes that Albeck's claim that R. Judah the Prince only collated and arranged the actual wording of the Mishnah, without making any changes therein, arises from a literary-historic point of departure from within the Mishnah, rather than a historic one; according to De Vries, R. Judah the Prince not only collated and redacted the Mishnah, but also formulated and adapted it. 9 A similar view was expressed by Epstein. 10 Bendavid describes the Oral Torah learning method and the way it was transmitted from one generation to the next, 11 and maintains that the documentation of the discussions and arguments contained in the Talmud, the questions and answers and various kinds of give-and-take, is quite precise in its representation of what the speakers said -'if not word for word, the actual style of what was said' 12 -and reflects contemporary spoken Hebrew, and is 'a true reflection of how people living in the Hebrew language negotiated, how they asked and responded, laughed and vociferated, recounted events and joked, in the study hall and the marketplace, when discussing matters of Torah and holding mundane conversations'. 13 In contrast to this approach, which views the conversations as a reflection of the actual discussions held among the sages, is the one that considers these conversations to be the outcome of editing. Neusner believes that the language of the Mishnah is in fact a revision of the natural language of Middle Hebrew. 14 According to Blondheim, Blum-Kulka, and Hacohen, the successive editors of the Talmud tried to make the conversations in the Talmudic text appear as transcripts of oral debates taking place in a study hall. 15 This is also the basis of Blondheim and Blum-Kulka's analysis of a Talmudic text from the perspectives of conversation analysis and historical pragmatics. 16 According to Raveh, direct speech might have reflected one characteristic of the art of the oral story, the medium used by the narrator to imitate speech in the represented world. 17 Kahana examines the construction of three controversies in the Mishnah, and claims that these controversies are not to be viewed as complete protocols of the discussions by the rabbis, or as a neutral and unbiased documentation of the main lines of disagreement. 18 Simon-Shoshan in his book about the narrative discourse in the Mishnah, includes the dialogues within the type of texts that he terms 'speech acts'. 19 In his view, the Mishnah occasionally presents dialogues between two rabbis in order to expound on the underlying logic of opposing halakhic positions. He relates to the dialogues as a feature of the narrative, 3. The Halakhic Give-and-Take Conversations in the Mishnah but views them at most as marginal stories because no significant change occurs as a result of the conversation, and each of the rabbis leaves the encounter holding the same opinion as before. He argues that the debates between the schools of Hillel and Shammai, which conclude with the narrator stating that in response to the House of Shammai's arguments the School of Hillel changed their view, can be considered stories.

b) Representation of controversy between discussants:
This characteristic is reflected in the fact that the main motivation behind halakhic give-and-take conversation is the existing controversy between the discussants. 20 Blondheim and Blum-Kulka maintain that intensive interpersonal argument was indeed the trope of the study process engaged in by the Tannaim  According to Belberg, the culture of the sages can be described 'as a "culture of controversy", in which discussion and argument were the building blocks of creativity'; see Mira Belberg, Gateway to Rabbinic The study by Schiffrin, 22 along with those of Blum-Kulka, Blondheim, and Hacohen, show that controversy in rabbinic literature also impacted the shaping of the tradition of controversy in Jewish and Israeli society.
c) Suasive goal: This characteristic is reflected in the fact that the main intention of the addressor in expressing his halakhic position in give-and-take conversation is to persuade the addressee of the correctness of his assertion.

A description of two aspects drawn from conversation analysis
A study that I am conducting on halakhic give-and-take conversations in the Mishnah includes all halakhic give-and-take conversations found in the Mishnah -190 conversations, which include 240 exchanges between addressor and addressee. 23  debating parties in halakhic give-and-take conversations can be divided into three types: 24 In most of the conversations (117 conversations = 62 percent) one party is an individual and the other party is a group; in fewer than a third of the conversations in the corpus (56 conversations = 29 percent) 25 both parties contain two exchanges. A small proportion of the conversations in the corpus (seven conversations = 3.5 percent) contain a larger number of exchanges -with three, four, or five exchanges. Similar to the findings from the study of the corpus undertaken by Meir, "Questions or Answers", pp. 163-164, which includes 145 controversies, she found that the most frequent structure for controversies contained one stage; furthermore, 16 controversies (11 percent) contained a two-staged dialogue, and 11 had unique structures. 24 In Meir, "Questions or Answers", p. 161, the author similarly categorises the controversies into three groups, according to the participants in the controversy: 1) controversies between two collective figures; 2) direct controversies between two Tannaim; 3) direct controversies between a Tanna and an anonymous collective figure. Although the controversies discussed in her article are not identical to the give-and-take conversations in this study, the disparity involving group size is similar to the disparity described here between types of conversation. Meir characterises the controversies from the third group as being more uniform in terms of the structure of the controversy and as smaller in scope, and the controversies from the second group as having developed models that are exceptional in terms of the structure and course of the text. 25 In most of the conversations of this kind, the individual is a sage and the group is a group of sages (other conversations: a sage and a group of students [seven conversations], a sage with other groups -an unknown group [three conversations], Sadducees [one conversation]), and one conversation between a Galilean heretic and Pharisees. The group with whom the sage is holding the discussion (a group of sages, a group of students, or an unknown group) is generally presented in the pattern of ‫לו‬ ‫אמרו‬ 'they said to him'. In two out of 105 conversations in which a sage holds a discussion with other sages, the sages are presented using the term ‫חכמים‬ 'sages'; in other conversations, the sages are presented in the pattern of ‫(לו/לפניו)‬ ‫אמרו‬ 'they said (to him/before him)'. Meir, "Questions or Answers", pp. 164-165, maintains that the expression ‫אמרו‬ 'they said' marks an opinion held by more than one sage or the opinion of an individual sage that became accepted by many. are individuals; and in a small number of the conversations (17 conversations = 9 percent) both parties are groups (in most of these conversations -14 conversations -the parties are the House of Hillel and the House of Shammai).
In this study on halakhic give-and-take conversations in the Mishnah, the conversations are studied from aspects that belong to different linguistic areas: discourse analysis, pragmatics, conversation analysis, and rhetoric. This article will describe two aspects of conversation analysis that were investigated: adjacency pairs in conversations (in section 2.1) and argumentative steps in conversations (in section 2.2).

Adjacency pairs in the halakhic give-and-take conversations in the Mishnah
'Adjacency pair' is a term used in the theoretical approach known as conversation analysis. 26 This term relates to a pair of turn types in a conversation that come together, i.e., a turn of one type on the part of the addressor leads to a turn of a different type on the part of the addressee, for example question and answer, complaint and apology, a greeting answered by another greeting. 27 This investigation of halakhic give-and-take conversations in the Mishnah examined adjacency pairs that appear in both parts of the exchange. The examination included all the exchanges comprising two parts (151), excluding partial exchanges (88), which contain only the words of the addressor, thus making it impossible to examine the adjacency pairs in them. Table 1 presents five adjacency pairs in order of their frequency in conversations -based on the first part of the pair: asking, asserting, telling a story, explaining, and reprimanding. The first column of the table presents the pairs, and the second column shows the prevalent and rare options for each pair (alongside each, the number of its occurrences is noted, and for frequent options, their proportion as a percentage is shown in relation to the overall occurrence of the pair; the final column shows the overall number for each pair). 28 Identities. 117), there are typical preferred second pair parts which are common in conversation, but occasionally a turn that appears with a non-typical dispreferred second part, for example (in the following pairs the preferred second part will be presented after the dash compared to the dispreferred part: request -acceptance versus refusal, offer/invite -acceptance versus refusal, assessment -agreement versus disagreement, questionexpected answer versus unexpected answer or non-answer, blame -denial versus admission. And see a different approach in Amy Tsui, "Beyond the Adjacency Pair", Language in Society 8 (1989), pp. 545-564, according to which conversation is not arranged in adjacency pairs, but rather as a three-part exchange. 28 The prevalent options in each pair were determined in consideration of their proportion compared to the overall number of the occurrences of each adjacency pair. In the last two adjacency pairs -4 and 5 -no prevalent options have been presented due to the overall sparse number of occurrences of each of them. The table shows that there are two prevalent adjacency pairs in halakhic give-and-take conversations in the Mishnah -the pairs in which the first part involves asking (including qal va-chomer, i.e., a fortiori, questions) or asserting (including gezerah shavah, i.e., analogy, and a fortiori assertions). These pairs were found in 85 percent of the exchanges that were examined (128 exchanges: 81 with asking and 47 with asserting). From this it follows that when the discussant presents his position, he prefers to do so by asking or asserting, whereas presenting by telling a story, explaining, or reprimanding is very rare in halakhic give-and-take conversations. 29 In addition, we see the most common combinations in these two prevalent adjacency pairs. In pairs in which the first part is asking, the prevalent combinations are with a second part that is answering, asserting, or asking; 30 and in pairs in which the first part is asserting, the only prevalent combination is with a second part that is asserting (in 74 percent of the occurrences of this pair = 35 exchanges). 31 In more than half of the exchanges which are made up of two parts -in 58 percent of them (87 occurrences) -asking+answering pairs were found (52 occurrences) as were asking+asserting pairs (35 occurrences). In other words, the first party chooses to express his position 29 There are three adjacency pairs that are not prevalent in the corpus, and their first parts involve telling a story, explaining, or reprimanding. When the first part is telling a story, the prevalent combination is with a second part that is asserting. To these should be added four adjacency pairs represented by just one or two occurrences, which have not been presented in this table: requesting+giving permission or ordering; and one occurrence for each of these adjacency pairs: answering+answering, vowing+declaring, ordering+asserting. 30 Rare combinations of asking are followed by a second part determining of law, praising, or reprimanding. In one exchange, the question is followed by a nonverbal response ‫אחר‬ ‫לדבר‬ ‫והשיאו‬ 'and led him to another subject'. 31 Rare combinations include asserting with ordering, asking, reprimanding, and declaring. by asking and the other party chooses to respond by answering, or the first party opens by asserting and the other party also responds by asserting. An asking+answering pair can be found, for example, in citation [3] of R. Eliezer and R. Joshua, which discusses when one should begin to mention rain in the prayers. R. Joshua asks a question: ‫מזכיר?‬ ‫הוא‬ ‫למה‬ ‫בחג‬ ‫ברכה‬ ‫סימן‬ ‫גשמים‬ ‫ואין‬ ‫הואיל‬ 'since rain during the holiday is but a sign of a curse, why should one make mention of it?', and R. Eliezer responds: ‫״משיב‬ ‫אלא‬ ‫אומ׳‬ ‫אינו‬ ‫הוא‬ ‫אף‬ ‫בעונתו‬ ‫הגשם״‬ ‫ומוריד‬ ‫הרוח‬ 'he too does not ask [for rain] but only mentions "who causes the wind to blow and the rain to fall"in its due season'.
The asserting+asserting pair can be found, for example, in citation [4], in the conversation between R. Tarfon and R. Elazar ben Azariah about tithes taken from the fruits of the seventh year outside the land of Israel in the lands of Ammon and Moab: R. Tarfon argues, based on an analogy (gezerah shavah) that infers from the law regarding the giving of tithes in Egypt, that the obligation to give the poor-man's tithe applies in the lands of Ammon and Moab as well, and R. Elazar ben Azariah responds making a parallel claim, inferring from the law regarding the giving of a second tithe, that one is obligated to give a second tithe in Ammon and Moab as well.
The examination of adjacency pairs described here is aimed at examining the most prevalent adjacency pairs in conversations and the most prevalent combinations among them. The two adjacency pairs found most prevalent in this examination -the asking+answering pair and the asserting+asserting pair -are familiar pairs in the theoretical context of conversation analysis, 32 32 Jose, "Sequentiality of Speech Acts", examined speech acts sequentially in conversations between female adults and preschool children, employing a quantitative method of analysis. As opposed to the separate description of speech acts and of adjacency pairs in this research on halakhic give-andtake conversations in the Mishnah, in Jose's research there is a combination of the two, since he examined, as mentioned, speech act sequentiality in conversational discourse. Jose found in the conversations sequential patterns, whose initiating acts are questions, statements, and directives and whose responses are answers, agreements, interjections, and repetitions. The most common sequential patterns which Jose found in the conversations that he examined are question-answer and statement-acknowledgment, the most common speech acts being statements and directives (which also include questions). Although the examination of speech act sequentiality in Jose's research is different in many aspects from the examinations which were undertaken in this study on halakhic give-and-take conversations in the Mishnah -e.g., from such aspects as the nature of the conversations and research method -both studies arrive at similar conclusions as to the frequency of speech acts and adjacency pairs in the relevant conversations. And see in Jose, "Sequentiality of Speech Acts", pp. 67-69, a review of several sequential models of speech act production, one of them is the adjacency pairs. Jose maintains that some of those models lack empirical basis in real discourse, while those which had empirical basis examined a particular type of discourse or a limited discourse. and are also suitable for the common speech acts found in halakhic give-and-take conversations in the Mishnah -asserting, asking, and answering -and these are described in this study in the context of the pragmatic description of speech acts.

Muntigl and Turnbull's Model
Exchanges in halakhic give-and-take conversations in the Mishnah were analysed in this study based on a model presented by Muntigl and Turnbull (hereinafter: M&T), 33 which is described in this section. M&T examined arguments in naturally occurring conversations between university students and family members. 34   2. Challenge -disagreement by means of which a speaker questions an addressee's prior claim and demands that the addressee provide evidence for his or her claim, while suggesting that the addressee cannot do so, e.g., 'why do you say that?'; 37 3. Contradiction -disagreement by means of which a speaker presents a proposition that directly refutes the previous claim, e.g., 'no, that's just wrong'; 38 4. Counterclaim -proposing a claim as an alternative to the former one, without directly contradicting or challenging that claim, e.g., the utterance 'bananas are the most popular fruit' in response to the utterance 'apples are the most popular fruit'. 39 Also found were frequent combinations of contradiction + counterclaim and other act combinations.
the orderliness of the T2-T3 sequence is a consequence of interactants' concerns about face/identity: the more speaker B's T2 act damages speaker A's face, the more likely A is to respond with a T3 act that directly supports A's T1 claim; T3 acts that support T1 reflect A's attempt to repair damage to their own face occasioned by the face-aggravating T2 act. 36 M&T, p. 229, characterise these acts as meta-dispute-acts, because they comment on the conversational interaction. 37 According to M&T, pp. 229-230, the typical syntactic form of challenges is interrogative, appearing with question particles. 38 According to M&T, p. 231, the contradicting proposition negates the previous claim, so that if the previous claim is positive the contradiction contains negative markers, and if the previous is negative the contradiction contains positive markers. 39 According to M&T, p. 231, counterclaims tend to be preceded by pauses, prefaces, and mitigating devices.
M&T's study was done in the context of an approach that views argument as a face-threatening activity. In the wake of the examination of the distribution of these acts in argument, 40 M&T rank the degree of aggressiveness of the acts, i.e., in terms of the extent to which they damage another's face, from most to least face aggravating: irrelevancy claim, challenge, contradiction, combination contradiction+counterclaim, and counterclaim. The most aggravating act is an irrelevancy claim, because it limits any further discussion and attacks the most fundamental social skill of a conversationalist; next in aggressiveness is the challenge, since it directly attacks the competency of the other to back up his or her claim; contradiction is less face-aggravating, since it does not directly attack the other speaker; the combination act contradiction+counterclaim is less aggravating, since it contains a contradiction that repudiates other's claim, which is somewhat mitigated by a counterclaim that offers more information on the basis of which to negotiate the disagreement; and the counterclaim is the least face-aggravating, because it does not overtly mark opposition, but provides an alternative claim by opening up the topic for discussion.
In accordance with this ranking, M&T classified the acts into three categories: the highly aggressive category -irrelevancy claim and challenge; the moderately aggressive categorycontradiction and contradiction+counterclaim; and the less aggressive category -counterclaim.

Examining argumentative steps in halakhic give-andtake conversations in the Mishnah
In this study on halakhic give-and-take conversations in the Mishnah, an effort has been made to describe exchanges in conversations according to M&T's model and to compare findings with those of their study as well as of another study conducted according to this model, namely that of Blondheim and Blum-Kulka (hereafter B&BK), 41 which will be described in section 2.2.3 below.
The examination undertaken in this study is called an examination of argumentative steps and comprises two parts. The first part of the examination analysed the 116 two-part exchanges that contain the most prevalent speech acts: asserting, asking, and answering (i.e., 77 percent of the 151 two-part exchanges). Each of the exchanges was examined individually, 42 even when the exchange was part of a conversation containing multiple exchanges. In each exchange, the second part of the exchange was examined in relation to the previous part, i.e., the second part spoken by the addressee that comes in response to the first part spoken by the addressor. In this way, it was possible to assess the degree of the addressee's response in relation to the previous remarks by the addressor. The words of the addressor, i.e., the first part of the exchange, cannot be similarly assessed, because they do not always relate to something said previously, and consequently, the speech acts in the first part of the exchanges in the corpus were not included in this examination.
The second part of the examination included 40 two-part exchanges in conversations including multiple exchanges also contain the most prevalent speech acts of asserting, asking, and answering. In these conversational exchanges the second and (if appropriate) following exchanges were examined in order to find the argumentative step between the exchange that was examined and the exchange that preceded it in the conversation. In each exchange, the first part of the exchange was examined in order to find its relation to the second part of the exchange that preceded.
It should be noted that in the classification of exchanges in the corpus of the conversations in this study, dilemmas of classification often arose regarding the attribution of a particular exchange to one of the four types of steps. For example, is a particular argument a contradiction, i.e., does it expresses direct opposition to the previous claim, or is it merely an alternative counterclaim that does not directly contradict the claim; is a particular argument a contradiction to the previous claim or does it also contain a challenge, i.e., does it also expresses disagreement and demands that the addressee provide evidence for his or her claim, while suggesting that he or she cannot do so. It appears that this type of dilemma is typical of many classificatory studies, and M&T also report several cases that posed a challenge to them in their study. 43 Further to this, it is possible that dilemmas are due to the fact that the classification categories are themselves somewhat ambiguous, which often makes it difficult to distinguish among them. M&T note in some of the categories the different definitions that were provided for 43 M&T, p. 240. it by previous researchers, as well as terminological variety in the case of certain categories, which is especially relevant in the categories of challenge (M&T, p. 229-230) and contradiction (M&T, p. 231). It is also possible that dilemmas arose due to the different nature of the conversations under examination here -halakhic give-and-take conversations that appear in a text written during the classical period, as opposed to the nature of the naturally occurring oral conversations in modern English that formed the basis for M&T's classification. M&T explain at the beginning of their classification that former classification systems have been based on children's arguments, compared to their system of classification, which has been based on arguments between adults and adolescents. They comment that, because of this difference, there may be a need to modify the classification scheme in order to adapt it to these kinds of arguments. Despite these dilemmas in examining the corpus of halakhic give-and-take conversations in the Mishnah, each of the exchanges was classified into one of four types of argumentative steps, without creating combinations between steps or removing cases that aroused doubt. The working assumption was that, despite the dilemmas, the findings can be examined and compared in general terms to the findings of the studies of M&T and of B&BK.
In this section, findings regarding the four types of argumentative steps that emerged from the two parts of the examination of the exchanges in halakhic give-and-take conversations in the Mishnah will be presented first, followed by a sampling of each of the steps in the conversations in the corpus. Table 2 presents the findings regarding the four types of argumentative steps found in the 116 two-part exchanges (the types of argumentative step are presented in the first line; the second line notes the number of exchanges of each type of step, and alongside the number is its proportion in terms of a percentage of the overall number of exchanges examined in this part of the examination). Table 3, which follows, presents the findings for the different types of argumentative steps that were found in the 40 exchanges that are part of conversations with multiple exchanges.   Table 2 shows that the frequency of argumentative steps in ordinary two-part exchanges is -in descending order -counterclaim, contradiction, challenge, and irrelevancy claim. Table 3 shows that in exchanges that are part of conversations with multiple exchanges no irrelevancy claims were found at all, and that from among the three remaining types of argumentative steps, challenge was the most frequent, followed by contradiction and then counterclaim.
A comparison between the findings of the two types of exchanges from the two parts of the examination enables us to draw a number of conclusions. First, in both types of exchanges an irrelevancy claim is a rare step. Second, contradiction is in the mid-range in terms of frequency in both types of exchanges. Third, there is a marked difference between the two types of exchanges in terms of the argumentative step that is most prevalent in them: in exchanges of the first part of the examination, the counterclaim is most prevalent -which for M&T is the act of the lowest grade of aggressiveness in the ranking; on the other hand, in the exchanges taken from the second part of the examination, the most prevalent is challenge, which is the act of the highest grade of aggressiveness according to this ranking. And fourth, there is a further difference between the two types of exchanges in terms of the degree of aggressiveness of the acts: in the ordinary exchanges, the common acts are of the intermediate and the low aggression levels -contradiction and counterclaim -which represent 77 percent of the argumentative steps in these exchanges, whereas the acts of the high aggression level -irrelevancy claim and challenge -can be found in only about a quarter of the exchanges (23 percent); on the other hand, in the exchanges from the second part, which are part of conversations having multiple exchanges, there is similarity between the proportion of the act of the highest aggression level -challenge (52.5 percent) -and the proportion of the acts of the intermediate and low aggressive levels (47.5 percent).
These conclusions are indicative of the more aggressive nature of the exchanges of the second type as compared to those of the first type. It would appear that in ordinary twopart exchanges, the nature of the discussion in halakhic giveand-take conversations in the Mishnah is not aggressive -the discussant is much more likely to prefer the use of a counterclaim or contradiction than challenge or irrelevancy claim. The nature of the discussion emerges as more aggressive, on the other hand, when multiple exchanges appear in the conversation; in the situation of a conversation, in an exchange that comes in the wake of a previous exchange, the speaker chooses to relate more aggressively to the previous turn -he is much more likely to make use of challenge and contradiction, while keeping the use of counterclaim to a minimum. In both types of exchanges we find that steps with intermediate and low aggression levels are more common than steps at the high aggression level; however, whereas in exchanges of the first type the disparity is more evident (intermediate and low aggression levels cover 77 percent of all the argumentative steps), in exchanges of the second type, which are part of conversation, the disparity between the high level and the intermediate and low levels is far smaller (52.5 percent compared to 47.5 percent).
The four types of argumentative steps that appear in halakhic give-and-take conversations in the Mishnah will be described and demonstrated with examples below:

(a) Irrelevancy claim
Irrelevancy claims are rare in ordinary exchanges (4 exchanges = 3 percent) and are completely absent from exchanges that are part of conversations. For example, in citation [5], R. Akiba presents his position that it is possible to purify a zav (one who is afflicted with gonorrhoea) after an examination has shown that the ziva (the affliction) was caused by a type of food or drink. This is followed by a conversation between him and anonymous sages: They said to him: [then] there would henceforth be no zavim! He said to them: the responsibility [for the existence] of zavim is no concern of yours! The anonymous sages ‫לו(‬ ‫)אמרו‬ maintain that this position of R. Akiba could lead to a situation where there would be no more zavim, because they will able to attribute their condition to some food or drink, and R. Akiba admonishes them, arguing that they are not responsible for the existence of zavim.
The irrelevancy claim emphatically clashes with the previous claim presented in the first part of the exchange, with an explanation of its implications, and it contains an explicit admonishment of another, placing him on the side that opposing that of which the speaker considers himself part.

(b) Challenge
Challenges are found in the two types of exchanges and are the most prevalent argumentative step in exchanges that are part of conversations (in the first type 23 = 20 percent; in the second type 21 = 52.5 percent).
For example, citation [6] starts with a presentation of the views of the School of Shammai and the School of Hillel over the question of whether it is permitted to bring the priest's share of the dough and gifts set aside for him on a holiday -the hallah ‫)חלה(‬ is separated from the dough and the gifts are part of an animal sacrifice. This is followed by a conversation between the School of Shammai and the School of Hillel: The School of Shammai say: They may not take to the priest the priest's share of the dough or priests' dues to the priest on a holiday whether they were separated on the preceding day or were separated on the same day; but the School of Hillel permit it.
The School of Shammai replied to the School of Hillel with a logical analogy: a priest's share of the dough and priests' dues are a gift to the priest and the Heave-offering is a gift to the priest; just as they may not bring Heave-offering so they may not bring the priests' dues.
The School of Hillel replied to them: not so! Would you maintain the argument in the case of Heave-offering which one may not separate and also the same argument in the case of priests' dues which one has the right to separate?
The School of Shammai presents a claim based on an analogy between this case and that of a Heave-offering (donation), which is also a gift to the priest and is not given on a holiday, and the House of Hillel rejects that argument with an a fortiori question, which raises a difficulty regarding inference from the law about a Heave-offering regarding what may be done with hallah and gifts on a holiday: ‫במתנות‬ ‫תאמרו‬ ‫בהרמתה‬ ‫זכיי‬ ‫שאינּו‬ ‫בתרומה‬ ‫אמרתם‬ ‫אם‬ ‫לא,‬ ‫בהרמתם?‬ ‫זכיי‬ ‫שהוא‬ 'Not so! Would you maintain the argument in the case of Heave-offering which one may not separate and also the same argument in the case of priests' dues which one has the right to separate?' -They maintain that in Heave-offering there is a reason that it is forbidden to bring it on a holiday, but that this reason does not apply to hallah and gifts.
This form of challenge is a prevalent one (in the first type of the exchanges 16 occurrences = 70 percent; in the second type 8 occurrences = 38 percent). It is made up of two components: the first component -rejection of a previous question or claim using the negation word ‫לא‬ 'no', and the second element -an a fortiori question, the pattern of which is usually ‫אמרת/אמרתם‬ ‫אם‬ ‫ש…?‬ ‫ב…‬ ‫תאמר/תאמרו‬ ‫ש…‬ ‫ב…‬ 'if you said for… that…, would you say for… that…?'. In a challenge of this and other kinds that have not been demonstrated here, 44 the speaker expresses both disagreement with the previous claim along with a demand to present evidence to strengthen the claim.

(c) Contradiction
Contradiction is an argumentative step of intermediate frequency in both types of exchanges (in 34 percent of the exchanges in the first part of the examination and in 35 percent in the exchanges in the second part). Contradictions of various and sundry types were found in the corpus, and in all of them the discussant's argument presents direct opposition to the previous argument. 45 Three types of contradictions found in the corpus will be instanced here.
Some contradictions come in response to an a fortiori question and present evidence from a different case. For example, citation [7] discusses the question of whether it is permitted on the Sabbath to carry out labours related to a Passover offering to which apply a rabbinical rest restriction ‫שבות(‬ ‫,)איסור‬ i.e., which are forbidden by the rabbis: R. Eliezer said: is it not self-evident, seeing that slaughtering, which is an act of work, overrides the Sabbath, should not these, which are under only a rabbinical rest restriction override the Sabbath? R. Joshua replied to him: A festival-day will prove against this, for on it they permitted functions that come within the category of rabbinical rest restriction.
R. Eliezer answered him: how so, Joshua? What proof can you deduce from a voluntary act for an obligatory act? […] not possible to base classification on negation words alone, as M&T found, but it was also necessary to understand the nature of the argumentative step in order to characterise what was said in it by the speaker as a contradiction of the previous speaker's words.
R. Eliezer is of the view that acts that are forbidden on the Sabbath because of rabbinical rest restriction are permitted for a Passover offering on the Sabbath, and bases himself on an a fortiori inference from the act of slaughtering, which although a form of labour forbidden on Sabbath by the Torah, is permitted on the Sabbath for a Passover offering by the Torah, which is much more authoritative than a rabbinical restriction: ‫אם‬ ‫מה‬ ‫את‬ ‫ידחו‬ ‫לא‬ ‫שבות‬ ‫משום‬ ‫שהן‬ ‫אלו‬ ‫השבת,‬ ‫את‬ ‫דוחה‬ ‫מלאכה‬ ‫משם‬ ‫שהיא‬ ‫שחיטה‬ ‫השבת?‬ 'is it not self-evident, seeing that slaughtering, which is an act of work, overrides the Sabbath, should not these, which are under only a rabbinical rest restriction override the Sabbath?', and R. Joshua contradicts the a fortiori argument with evidence from a festival, when it is permitted to carry out labour to prepare food, though rabbinical restrictions on labour still apply: ‫טוב‬ ‫יום‬ ‫שבות‬ ‫משום‬ ‫בו‬ ‫ואסר‬ ‫מלאכה‬ ‫משום‬ ‫בו‬ ‫שהיתיר‬ ‫יוכיח,‬ 'a festival-day will prove against this, for on it they permitted functions that come within the category of rabbinical rest restriction'.
Contradictions of another type come in response to a question and offer an explanation. For example, citation [3] above presents the view of R. Eliezer that one should begin reciting ‫הרוח‬ ‫משיב‬ ‫הגשם‬ ‫ומוריד‬ 'who causes the wind to blow and the rain to fall', in the silent prayer of Shmoneh Esreh from the first day of Sukkot, in contrast to R. Joshua's view that the time to begin reciting it is on Shemini Atzeret, at the end of Sukkot. R. Joshua asks a question that challenges R. Eliezer's point of view: ‫ברכה‬ ‫סימן‬ ‫גשמים‬ ‫ואין‬ ‫הואיל‬ ‫מזכיר?‬ ‫הוא‬ ‫למה‬ ‫בחג‬ 'since rain during the holiday is but a sign of a curse, why should one make mention of it?', that is to say, why should one make mention of rain during Sukkot if rain could prevent people from sitting in the Sukkah. In response, R. Eliezer presents an explanation of his opinion, offering a more precise reading of the matter at hand: ‫ומוריד‬ ‫הרוח‬ ‫משיב‬ ‫אלא‬ ‫אומ׳‬ ‫אינו‬ ‫הוא‬ ‫אף‬ ‫בעונתו‬ ‫הגשם,‬ 'he too does not ask [for rain] but only mentions "who causes the wind to blow and the rain to fall"in its due season'. In his view, this statement does not represent a request for rain, but merely notes the might of the Lord, who brings down the rain when it is needed.
Contradictions of a further type are those in which the opposing claim has a parallel construction to the previous claim. For example, in citation [8], in the second conversation in the second exchange, Rabban Jochanan ben Zakkai makes an claim that contradicts the words of the Sadducees in the previous exchange and is formulated as a parallel construction:

. The Halakhic Give-and-Take Conversations in the Mishnah
The Sadducees say: we protest against you, O Pharisees, for you say: the Sacred Scriptures render the hands unclean and the books of the sectarians do not render the hands unclean.
Rabban Jochanan ben Zakkai said: have we not against the Pharisees save only this? Behold they say: the bones of an ass are clean and the bones of Jochanan the High Priest are unclean! They said to him: because of our love for human beings, we declare their bones unclean, so that man does not fashion the bones of his father or his mother into spoons. He said to them: even so the Sacred Scriptures, in proportion to the love for them so is their uncleanness, and the books of the Sectarians which are not beloved of us do not render the hands unclean.
In the first exchange, Rabban Jochanan questions the fact that the bones of an animal carcass are pure, whereas the human bones make one unclean; and the Sadducees claim that human bones are unclean because of their importance: ‫טומאתן…‬ ‫היא‬ ‫חיבתן‬ ‫לפי‬ 'because of our love for human beings, we declare their bones unclean…'. In the second exchange, he responds with a claim having a parallel construction: ‫טומאתן…‬ ‫היא‬ ‫חיבתן‬ ‫לפי‬ ‫הקודש‬ ‫כתבי‬ ‫אף‬ 'even so the Sacred Scriptures, in proportion to the love for them so is their uncleanness…'.

(d) Counterclaim
Counterclaims are the most prevalent argumentative step in ordinary exchanges (43 percent), but are not prevalent in exchanges that are part of conversations (12.5 percent). A counterclaim presents a response to the previous claim, but does not pose a challenge or present a contradiction in regard to it. A prevalent type (80 percent of ordinary exchanges) is when a question appears and the counterclaim presents an explanation of that question. For example, citation [9] tells of R. Nechonia ben Hakkanah, who composed two prayers for those entering the study hall:

Comparing the findings from this examination of argumentative steps to the findings of previous studies
Following the examination of the argumentative steps in halakhic give-and-take conversations in the Mishnah based on the model of M&T described above in section 2.2.2, the findings were compared to those of M&T's studies on naturally occurring conversations, as described in section 2.2.1 above, as well as to those of B&BK's study, which will be described in this section below. The frequency of the four steps found in the two studies is presented in Table 4 below. B&BK examined a single talmudic text (b.Baba Kamma 56b-57b) according to M&T's model. They found that, in contrast to the expectations of M&T, the Talmudic debate shows a pattern which is the opposite of the facework expected: throughout the Talmudic debate, the response to challenge is not a face-saving defence, but a counter attack, tit-for-tat style, and it would even appear that the more aggressive the challenge, the more animated the counter attack. 46 According to B&BK's evaluation, the Talmudic debate is considered aggressive, since its highly aggressive turns outnumber its mildly aggressive turns. B&BK present several results about the frequency of the four type of arguments: the frequency of the most mild, mitigated form of disagreement was by far the lowest; there are almost two and a half of the most aggressive turns for every one of the least aggressive turns; and overall, the frequency of the high-aggression pair is only slightly lower than that of the low-aggression pair (47.3 percent and 52.8 percent, respectively). B&BK propose a possible explanation for the results, which is that in Talmudic debate, challenges are based on authoritative Tannaitic texts, and that the response to challenges of this kind is T2-rather than T1-oriented.
It should be noted that examination in this study of halakhic give-and-take conversations in the Mishnah differs from the two other previous studies in two respects. First, each exchange was examined on its own, even when it was part of a conversation that includes multiple exchanges. And second, the arguments in the analysed corpus are not necessarily made up of three turns, unlike the three-turn exchange for arguing in M&T's study. 47 Consequently, only the first and central subject in M&T's study -characterizing the acts of disagreement and their level of aggressiveness -was examined, and the second issue of regularities in the sequences, i.e., the influence of the second turn on the third turn, was not, because the structure of the arguments in the corpus did not allow for examination of this in a similar way. Further, it should be noted that the number of exchanges that were examined in the corpus under examination, as described in section 2.2.2 above, is similar to the number of segments examined in M&T's study, which included 164 threeturn argument exchanges. It is, however, different in its scope from the corpus examined in the study by B&BK, which included one Talmudic text (b. Baba Kamma 56b-57b), and which, due to considerations of scope, treated only the first eight turns of its 23 turn-sequences. 48

The Halakhic Give-and-Take Conversations in the Mishnah
The findings of the two previous studies and of the current one on halakhic give-and-take conversations in the Mishnah are presented in Table 4. The table notes for each step its proportion as a percentage of the overall number of exchanges or turns examined in each study, without noting the actual number of occurrences in each study. The data regarding the combination of contradiction+counterclaim were not noted in the findings of the study by M&T, since this combination was not examined in the two other studies. The findings in the first row of this study on conversations in the Mishnah are divided into two internal rows according to the types of exchanges from both parts of the examination, and the findings in the second row of M&T's study are divided into two internal rows according to the two types of turns examined in it -T2 (the turn of the second speaker) and T3 (the turn of the first speaker).
of the Talmud (Berakhoth), for example: irrelevancy claim -midi, shani hatam , hacha bemai askinan; challenge -iy hachi , maytivey, matkif; contradiction -kashya , mibeʿey ley; counterclaim -ela mai, ela meʿata. In fact, the numerical data that they present that appear above as well as in Table 4 below relate to the frequency of the formulae in the four categories in tractate Berakhoth, and not in the Talmudic text analysed in their article, from which only the first 8 turns of its 23 turn-sequences were analysed.
As already indicated, in their study of naturally occurring conversations, M&T found the following frequency of the acts: counterclaim, contradiction, challenge, and irrelevancy claim; hence the acts of low and intermediate levels of aggressiveness -counterclaim and contradiction -are much more frequent than acts of high levels of aggressiveness -irrelevancy claim and challenge.
B&BK found in their study of a Talmudic text a different order of frequency of the acts: contradiction, challenge, irrelevancy claim, and counterclaim. This order shows that the frequency of the high-aggression pair is only slightly lower than that of the low-aggression pair (47.3 percent and 52.8 percent, respectively). Therefore, they concluded that the examined Talmudic debate could be more aggressive than the conversations that were examined by M&T.
In the present study of halakhic give-and-take conversations in the Mishnah a distinct difference was found between the exchanges examined in the two parts of the study: in ordinary two-part exchanges, the findings were similar to those of the study by M&T; the order of the frequency of the acts is identical to the order found in their study, and similarly, it was found that the acts of low and intermediate levels of aggressiveness are much more frequent than acts of high levels of aggressiveness. On the other hand, in the exchanges in the second part of the examination, which are part of conversations with multiple exchanges, the findings were more similar to those of the study by B&BK: the order of the frequency of acts is similar to the order found in their study, and similarly, it was found that the frequency of acts with a high level of aggressiveness is similar to the frequency of acts with low and intermediate levels of aggressiveness. As noted, in ordinary exchanges, the nature of the argumentative steps is not aggressive, but in exchanges that are parts of conversations with multiple exchanges, when the exchange comes in response to a previous exchange, the nature of the steps is more aggressive. this sketch will be preliminary and restricted to three main points: 1) delineating the corpus in terms of time, place, and genres; 2) positioning Tannaitic Aramaic in the wider context of Aramaic dialects; 3) spelling out methodological difficulties (and possibilities) inherent to the Tannaitic Aramaic manuscript evidence. In addition, we shall exemplify how some of these more theoretical considerations affect the interpretation of a test case. While Tannaitic literature is generally written in Hebrew, the Mishna, Tosefta, Sifra, and Sifre do occasionally contain Aramaic phrases, sentences, or even short texts. They represent instances of code-switching in a Hebrew text or -in the case of longer pieces -may constitute self-contained Aramaic compositions, original-language quotations of sorts, that were integrated into the wider Hebrew context. There is, of course, much more Aramaic on every page of rabbinic literature, but it stands to reason that the countless instances of isolated Aramaic words in Tannaitic Hebrew texts were mainly loanwords that had been incorporated into Hebrew to varying degrees and become part of that language. 4 They will therefore not be considered Tannaitic Aramaic in this sketch.

METHODOLOGICAL REMARKS AND A TEST CASE
Thus defined, the corpus of Tannaitic Aramaic comprises some 350 words, with the biggest chunk (200+ words) coming not from the rabbinical works enumerated above, but from Megillat Taanit, which dates from the same period and is traditionally associated with rabbinic circles (b.Shabbath 13b). 5

Tannaitic Aramaic
Gustaf Dalman referenced most of the Tannaitic Aramaic pieces (including doubtful ones from the Babylonian Talmud), but a complete list remains a desideratum. 6 The same holds for the grammar: no systematic description of Tannaitic Aramaic has ever been prepared. 7 Klaus Beyer edited most of the texts and provided a classification of their dialects, 8 but he did not utilise reliable rabbinic manuscripts and his editions do not always provide the best accessible text. David Talshir, in a two-page abstract of a lecture, was the first to point out the importance of the manuscript evidence and to call attention to some of the methodological problems associated with it. 9 Michael Sokoloff included most of the lexical material in his Dictionary of Judean Aramaic, 10  Any scholar wishing to provide a comprehensive description of Tannaitic Aramaic is faced with difficulties on three levels. Firstly, one has to test the homogeneity of the language of the corpus at the time of composition: are there indications of diachronic changes, dialectal variation, and different registers? Secondly, one has to consider the possibility of editorial changes once the original sources were incorporated into the extant literary texts. And thirdly, one has to account for possible effects of the transmission process on the language, and adopt a corresponding assessment of the manuscripts' textual reliability.
What signs are there, then, for variation in Tannaitic Aramaic? Diachronic change is not traceable in the corpus, even though the different Aramaic pieces were probably not produced at a single point in time. The Tannaitic Aramaic material has, by definition, a firm terminus ante quem: the final composition of the Tannaitic literary sources in the second century CE. However, these sources contain much older material, and the explicit attribution of some of the Aramaic texts to known rabbinic figures suggests that the material spans three centuries: Yose ben Yoezer, quoted in m.Eduyoth 8.4, lived in the second half of the second century BCE, Hillel, quoted inter alia in Aboth 1.13, lived approximately one hundred years later, and Rabban Gamaliel I, whose missives are preserved in t.Sanhedrin 2.5, was a leading authority in the Sanhedrin in the first half of the first century CE. Be that as it may, since attributions are not usually unanimous, 12  cannot be taken at face value, the general hypothesis of the chronological variety of the material should be retained. Geographical variance, i.e., possible dialectal differences in the material, is also difficult to assess. Beyer and Sokoloff assume a Judaean origin for Tannaitic Aramaic, 13 and it is indeed plausible (in light of both the rabbinical figures mentioned and the wider historical context) that the texts were produced in Jerusalem or its vicinity. However, Hillel the Elder, who was mentioned in the previous paragraph, is traditionally associated with Babylonia (e.g., t.Negaim 1.16), and if he was indeed born and brought up in the east, that could have affected his idiolect. 14 Different textual genres often correspond to different linguistic registers and are thus another source of linguistic variation in Tannaitic Aramaic. Indeed, the extant texts attest to diverse genres that can be assumed to correspond to a range of problems of identification of the rabbis in question and of divergent textual evidence in different rabbinic writings exist for other pieces as well. If at all, these can only be resolved by case studies that combine philology as well as textual and literary criticism. 13  registers, from the strictly formal to the more casual. One group of texts that stands out in the corpus are legal documents and formulas. 15 Their language, form, and style are rooted in the Imperial Aramaic legal tradition, which continued into post-Achaemenid times throughout the Middle East. 16 The scribal tradition had a conservative influence on the language, which contains less innovative and dialectal features than other texts. 17 The chronicle accounts of Megillat Taanit and the letters of Rabban Gamaliel I were written in an official or semi-official language, definitely not in legalese.  19 Proverbs travel easily between different communities and places and might preserve language features not original to the context in which they have come down to us.
The discussion in the preceding paragraph has moved to the fore the dichotomy of spoken vs. written language. The two are never exactly the same, and in written texts of different registers one can expect literary language with various degrees of influence from the vernacular. However, to determine, which feature of Tannaitic Aramaic represents literary Aramaic (and which kind of literary Aramaic), and which the vernacular, is a tricky task, not least so because of the very limited corpus. Essentially, it can only be achieved through comparison with other, roughly contemporaneous Aramaic dialects from the area. In other words, in order to determine the nature of Tannaitic Aramaic, one has to establish its place on the dialectal map of the Aramaic dialects from Palestine. Natural reference points and comparanda would be Biblical Aramaic, and the more innovative Aramaic of Daniel in particular, 20  of the literary texts from the Qumran caves, 21 i.e., the literary language of the Hasmonean period, and the language of the sparse contemporaneous epigraphic material from Judaea. 22 The Aramaic of Targum Onqelos and Jonathan represents another possible candidate for a literary language from Roman Palestine, even though it is now usually assumed that in its present form the language also contains (secondary?) eastern features. 23 The later Jewish Palestinian Aramaic is also important, since it represents a Jewish dialect that was promoted to a literary language in Byzantine times. 24 Precursors of this dialect were certainly spoken (but not written) in Roman Palestine, and similarities with Jewish Palestinian Aramaic in the Tannaitic corpus could thus be interpreted as vernacular features. In theory, the identification of lexical and morphological isoglosses with the aforementioned dialects should allow us to establish their relation to Tannaitic Aramaic. In practice, however, determining the nature of Tannaitic Aramaic is not that simple. The secondary processes of composing the Tannaitic texts and subsequently copying them several times over a period of 800 years or more surely affected the language that is preserved in the best manuscripts. The effects that composition and transmission may have had on the language in the medieval manuscripts are secondary, and thus differ in nature from the internal variation discussed above. In fact, these processes are possible sources of contamination that might mask to a certain extent the 'original' Tannaitic Aramaic, with its internal variation. It is not always feasible to tell original language features from later contamination, especially since many of the comparable dialects that could be used for establishing the nature of Tannaitic Aramaic are also possible sources of secondary contamination. In the following, we shall discuss (in roughly chronological order) these sources of contamination and point to the methodological problems associated with each one of them. For the most part, there is no reason to differentiate between contamination at the time of composition or during transmission.
As said above, similarities between Tannaitic Aramaic, on the one hand, and Biblical Aramaic, Qumran Aramaic, or the Aramaic of Targum Onqelos, on the other hand, may be interpreted as features of two related (post-Achaemenid Aramaic) literary languages, respectively, and would then help to place Tannaitic Aramaic on the dialectal map. However, since the books of Daniel and Ezra became part of the Jewish canon, and since Targum Onqelos subsequently garnered quasicanonical status in Judaism as well, the languages of these works acquired prestige, and later Jewish authors and copyists imitated them. 25 Any feature shared by these dialects might thus also be the result of imitation on the part of the copyists of the Tannaitic Aramaic texts. 26 Thus, ‫עבידתה‬ ‫בטילת‬ 'the cult ended/ was stopped' (Megillat Taanit 28 = t.Sotah 13.6) was probably influenced by the similar wording in Ezra 4.24, 27 and the choice of lexemes in ‫רגלוהי‬ ‫מעל‬ ‫סיניה‬ ‫ושרת‬ 'and she pulled his sandal from his feet' (t.Yebamoth 12.15, MS Erfurt) is clearly based on Targum Onqelos to Deuteronomy 25.9. However, such influence is not necessarily restricted to specific textual correspondences, but can also be of a more general nature. Perfect forms of the internal passive of the G-stem, such as ‫אחידת‬ 'she was taken' (Megillat Taanit 9 and 20), are possible candidates for linguistic influence, 28 especially in light of common passive t-stem forms, e.g., ‫אתנטילו‬ 'they were taken' (Megillat Taanit 11). Tannaitic orthography, too, was influenced by Biblical Aramaic, e.g., in retaining the <h> in the C-stem participle ‫אנחנא‬ ‫מהודעין‬ 'we 25 For a discussion of the prestige and influence of Targum Onqelos cf., e.g., Abraham Tal 30 could well be an original Tannaitic language trait. 31 Since Qumran Aramaic texts and contemporaneous epigraphic material did not become canonical, they can serve as a test case: a linguistic feature found in Qumran Aramaic, but not in Biblical Aramaic and Targum Onqelos, is in all likelihood ancient and does not result from secondary influence. However, due to the similarity between the dialects and the restricted corpora, such features are very rare. A case in point might be the syntagm of the negated infinitive to express a prohibition, e.g., ‫למספד‬ ‫דלא‬ … ‫להתענאה‬ ‫דלא‬ 'one must not fast … one must not eulogise' (Megillat Taanith 1 = m.Taanith 2.8). It is well attested in epigraphic Aramaic from the late Second Temple period, e.g., ‫למפתח‬ ‫ולא‬ 'and one must not open' on funerary inscriptions from Jerusalem. 32 Even though this syntagm is also found in Biblical Aramaic, its prominence in the epigraphic corpus and the fact that a corresponding construction appears in contemporaneous Hebrew point to an authentic language feature. 33 29  The case of Jewish Palestinian Aramaic is even more complex. Predecessors of this literary language were probably spoken in Palestine in Tannaitic times, and linguistic characteristics of the dialect in Tannaitic texts could thus be traces of the vernacular of the time. 34 On the other hand, once this dialect was promoted to a literary language in Amoraic times, it also acquired prestige and might have served as a model for changes in the transmission of the Tannaitic Aramaic corpus. Presumably, Tannaitic Aramaic attests to both original vernacular-like traits that resemble Jewish Palestinian Aramaic and secondary influences. A possible example of the former would be the use of ‫טליא‬ 'the youths' (t.Sotah 13.5) instead of ‫.עולימיא‬ 35 The lexeme ‫טלי‬ is not employed in the literary Aramaic dialects of Tannaitic times, even though it existed in the spoken idiom (Mark 5.41). On the other hand, the 3pl Perfect ending ‫-ון‬ in the same context ‫דאזלון(‬ ‫טליא‬ ‫נצחון‬ 'the youths who went were victorious', t.Sotah 13.5, MS Vienna) could be a secondary change introduced by a copyist. MS Erfurt has forms without n, and such 'regular' Perfect forms are also found elsewhere in the corpus (e.g., m.Sotah 9.15, Megillat Taanit 7, 36). 36 And in contradistinction to the previous example, the ending ‫-ון‬ is not unequivocally attested in Aramaic texts from Tannaitic times. 37 Once the Babylonian Talmud became authoritative, Jewish Babylonian Aramaic, too, served as a prestigious literary language and exerted influence on Jewish copyists and scribes. Apart from possible authentic (but certainly very rare) traces in the idiolect of Tannaitic figures from the east (discussed above), all 34 Cp., e.g., the extraordinary Qumran Aramaic spelling ‫-וי‬ for the 3ms suffix pronoun, Muraoka Jewish Babylonian Aramaic traits in the Tannaitic material can be dismissed as late corruptions. A number of such Babylonian forms are easily recognizable in the Tosefta MS Erfurt, e.g., the participle with clitic pronoun ‫מהודענא‬ 'we declare' and the C-stem infinitive ‫לאפוקי‬ 'to bring out' in t.Sanhedrin 2.5. 38 In the preceding paragraphs, we have pointed to numerous possible examples of linguistic forms in Tannaitic Aramaic texts that could be secondary: results of linguistic updating and alignment to the norms of prestigious literary languages that affected the text in the manuscripts up to the Middle Ages. However, apart from Jewish Babylonian Aramaic forms, which can confidently be assigned to the transmission process, the interpretation of other language traits remains equivocal, and we cannot tell original from secondary forms with certainty. But while the interpretation of the data might sometimes be contestable, the validity of the methodological assumption of linguistic interference during the copying of the manuscripts can be ascertained. For in the Aramaic Levi Document we possess one Aramaic text from late Second-Temple period Palestine for which we can compare the language in the contemporaneous Dead Sea Scrolls with a medieval copy from the Cairo Genizah. 39 There is not much overlap between the surviving fragments, but even this very restricted corpus evinces linguistic updating of the kind we have assumed for the Tannaitic Aramaic texts, e.g., in the spelling of C-stem participles and infinitives with <h>. 40 Thus far we have tried to disentangle the different layers of the consonantal texts in Tannaitic Aramaic that we encounter in the medieval manuscripts. When taking into account all possible uncertainties of the original language situation and every possible source of interference during the transmission process, even the consonantal skeleton sometimes remains elusive. Additionally, in some of the manuscripts some words of the Tannaitic Aramaic corpus are also pointed with vowel signs. This further increases the variability and variegation of the material. As with the Hebrew parts, the consonantal and vocalisation traditions of each manuscript are to be judged separately. 41 Due to the sparsity of the material, it is doubtful whether one can reach definite conclusions about the reliability and the independence of the vocalisation traditions. We shall only exemplify the divergence 40 44 and the latter would seem to be a translation, given that the predication is in Hebrew. 45 The version in m.Eduyoth 8.4 is thus primary, and it stands to reason that its Aramaic is the original language of these rulings. 46 However, the very fact that the Aramaic material was reworked confirms our methodological caveat above that the texts might have been affected at the time of their composition: other texts, too, could be the result of partial translation, though this is impossible to prove.
Turning to the consonantal text in the manuscripts, one notes minor differences in the Aramaic: 47 43 Either close to Bet-El or farther to the north-west (cp. 1 Kgs 11.26); the exact identification is uncertain.  In addition to these erroneous forms, two Jewish Palestinian Aramaic orthographic conventions are also clearly secondary (for this dialect was not a written language when the rulings were produced): one is the spelling of the final -ā of the definite article with <h>, not <ʾ>, in MSS Kaufmann and Cambridge, and with the noun ‫מיתה‬ also in MS Parma A. Interestingly, the 48 The form could perhaps be interpreted as a plene spelling of ‫דכן‬ from MS Parma A. But <ʾ> for short a would be exceptional. 49 But according to Sokoloff, Dictionary of Judean Aramaic, p. 64 s.v., this is a singular construct. The incongruence would then remain unexplained.
epithet ‫שרייא‬ is consistently spelled with <ʾ>. The other one is the spelling <yy> for consonantal y, especially in the definite plural ending -ayyā, in MSS Kaufmann and Parma A, and once in MS Cambridge. In addition, the plene spelling <yh> of the 3msg suffix, though common in Targum manuscripts, is also unattested until the end of the Second-Temple period, and therefore probably secondary in our piece.
The adjusted text of the Mishnah -with emendations and non-Jewish Palestinian Aramaic orthography -would thus run like this: ‫ודי*‬ ‫דכין/דכן‬ ‫מטבחיא‬ ‫בית‬ ‫משקי‬ ‫ועל‬ ‫דכי‬ ‫קמציא‬ ‫איל‬ ‫על‬ ‫שריא‬ ‫יוסה‬ ‫לה‬ ‫וקרון‬ ‫מסאב‬ ‫למיתא‬ ‫.יקרב‬ This short text evinces some potentially diagnostic language traits that merit discussion. One orthographic-phonological trait is the spelling <dy> of the nominalizing particle. This spelling as a separate word is typical for older strata of Aramaic, including Biblical Aramaic. Qumran Aramaic has both this spelling and the proclitic <d->, as in later dialects, 50 and prima facie a similar picture emerges for Tannaitic Aramaic. However, the orthography of the particle in the manuscripts oscillates, as in the parallel ‫אמר‬ ‫די‬ (t.Sotah 13.6, MS Vienna), ‫דיאמר‬ (MS Erfurt), and ‫דאמיר‬ (Megillat Taanith 28). The spelling <dy> is thus hardly diagnostic and could well be secondarily influenced by Biblical Aramaic orthography.
Two morphological traits are also of interest. The mpl passive participle 'clean' is spelled ‫דכיין‬ in MS Kaufmann, and ‫דכן‬ in MS Parma A. The former spelling presumably represents dakayin, as in Biblical Aramaic (and later western dialects), the latter dakan, as in Targum Onqelos. 51 Since the sound change underlying the Targumic form is typical for Babylonia, 52 one may assume that the Tannaitic form was dakayin, and that ‫דכן‬ in MS Parma A is secondary. The second morphological feature has already been mentioned in our methodological remarks: the Jewish Palestinian Aramaic 3pl Perfect ending ‫.-ון‬ The fact that all good manuscripts have the reading ‫,קרון‬ not ‫,קרו‬ could be marshalled in support of the authenticity of the form, which would then be a vernacular feature. But such forms with -n are not otherwise attested until well into the Common Era, which would make this an extreme outlier. However, the interpretation as an original language feature becomes a little more probable if one takes into account that the form is not part of Yoseʼs rulings and could thus be later than these. A date sometime in the first two centuries CE is more easily reconcilable with the vernacular interpretation, but it is hypothetical. Ultimately, we cannot decide which of the interpretations of the form is more probable: it could be an original vernacular feature or a secondary scribal imitation of Jewish Palestinian Aramaic. Individual syntactical and lexical traits from Yoseʼs third ruling are best discussed together. In the relative clause ‫יקרב‬ ‫די‬ ‫למיתא‬ 'one who touches a corpse/dead body/the dead', the noun ‫מיתא‬ appears with the definite article, even though the referent is indefinite. This usage is typical of eastern Aramaic, where the article had lost its function of marking definiteness, and the syntactic peculiarity is thus best interpreted as secondary influence from Targumic Aramaic. Presumably, ‫במיתא‬ ‫דיקרב‬ in Targum Onqelos to Numbers 19.11 (for Hebrew ‫ת‬ ‫מֵ‬ ‫ּבְ‬ ַ ‫ע‬ ‫ֹגֵ‬ ‫נ‬ ‫,הַ‬ without the definite article) is the source of the determined form, for Yoseʼs halakhic ruling seemingly recapitulates the command from this verse. 53 This rather surprising fact did not escape the rabbis, who -assuming that Yose was not simply reiterating the plain meaning of the biblical verse -offered explanations on which specific situations Yose could have been referring to (b.Abodah Zarah 37b). The reason behind the talmudic discussion also bears on the lexical peculiarity of the Tannaitic piece. The G-stem verb ‫קרב‬ with different verbal arguments conveys different meanings: with the prepositions ‫על‬ (of humans) or ‫ל-‬ it expresses the notion 'to come near someone/something', while the notion 'to touch someone/something' usually requires an argument with the preposition ‫.ב-‬ 54 Only in the later Jewish Palestinian Aramaic does this strict distinction unravel and the notion 'to touch something' also comes to be expressed by an argument with ‫.ל-‬ 55 This leaves us with two possible interpretations for the Tannaitic text: either Yose meant to say 'one who comes near a dead body', i.e., he wanted to convey a notion different from the biblical verse, or the unusual verbal argument with ‫ל-‬ is a Jewish Palestinian Aramaic vernacular feature. 56 The former is difficult in terms of content. And the latter would be all the more noteworthy in light of the proposition ‫ב-‬ in Targum Onqelos, as well as in the Palestinian Targumim to Numbers 19.11, which were undoubtedly known to the copyists. Although Tannaitic literature was composed mainly in Hebrew, it also incorporates a number of brief texts in Aramaic. The language of these short pieces (and of the related Megillat Taanit) can be called 'Tannaitic Aramaic'. Due to the very small corpus, and since it is preserved only in medieval manuscripts, this language is very difficult to characterise and describe with precision. In this sketch we have tried to list and discuss the methodological problems that face every student of Tannaitic Aramaic. We have then applied these to a test case. It turned out that it is indeed possible to go beyond the manuscript evidence and excavate a more original form of the Tannaitic Aramaic dialect, e.g., by identifying and eliminating secondary traits. However, other linguistic features remain ambiguous. We can tell why this is the case, and we can point to the possible interpretations of the data, but we cannot reach a definite conclusion.
was due mainly to the early Hebrew translation of the book only seventy years after it had been written. 4 Originally, there were two separate translations of the book. One was Judah Ibn-Tibbon's translation, under the title Sefer Ḥovot ha-Levavot, which was more widely known and consequently is available today in many manuscripts and printed editions. The other was by Joseph Qimḥi. His translation was not as popular as Ibn-Tibbon's, and perhaps that is why we have only a small remnant of it today. 5 Judah Ibn-Tibbon was born in Granada, probably in 1120. 6 He was a physician, a translator, a merchant, and a book collector. 7 Around 1150 he moved to southern France and became a prominent figure in the Jewish community of Lunel. Ibn-Tibbon was a fountain of knowledge; people consulted with him and he would lend books from his private library. Bahye Ibn-Paquda's Ḥovot ha-Levavot was the first book Ibn-Tibbon translated. After  8 Like many medieval authors and translators, 9 Ibn-Tibbon complained that Hebrew was inadequate in comparison with other languages (especially Arabic); some called this deficiency ‫קוצר‬ ‫הלשון‬ 'language insufficiency'. 10 It was clear to these authors and translators that the Hebrew of previous ages had been sufficient for all the needs of the people at the time. Since the ancient texts (the Bible, rabbinic literature, and early liturgy) dealt with limited subjects, the Hebrew reflected in them was limited as well. As they knew Hebrew mostly from these sources, it was insufficient for composing original works and for translating works from different languages that dealt with different and wider issues that did not appear in earlier Hebrew writings.
None of the previous periods of Hebrew was sufficient on its own to be used as a source for structures and lexicon to create a whole translation. Therefore, Ibn-Tibbon decided to combine Biblical Hebrew, Rabbinic Hebrew, liturgy, and previous medieval Hebrew works -both syntactically and lexically. 11 On different occasions, he derived new lexemes from roots and other lexical stems taken from classical literature, and occasionally he shifted the meanings of biblical and rabbinic lexemes. In the prefaces to two of his translations, Ibn-Tibbon reveals to the reader the changes he had to make in the lexicon, and he is apologetic for these actions.
Considering the arguments and efforts of these authors and translators, one might expect that the lion's share of the lexicon in their writings would consist of neologisms of different kinds (both morphological and semantic neologisms). The analysis of the nominal lexicon used by Ibn-Tibbon in his translation of Duties of the Hearts serves as a useful source of confirmation or refutation. I believe that the analysis presented below indeed refutes this assumption, or at least suggests a different perspective on this impression.

The nominal lexicon in Ibn-Tibbon's translation of Duties of the Hearts
In Ibn-Tibbon's translation of Duties of the Hearts, I have found 2,102 nominal entries (1,878 lexemes and 224 phrases). 12 As is shown in Table 1, almost 50 percent of the entries are taken from the Bible, approximately 26 percent from rabbinic literature, a small portion from the liturgy, and around 8 percent from medieval writings composed prior to the era during which Ibn-Tibbon engaged in his translation work. Just under 15 percent are neologisms coined by Ibn-Tibbon. Although not all entries were taken 'as is' from classical Hebrew writings, these findings shed a different light on the perception of medieval Hebrew as presented by authors and translators of that era. In other words, if Hebrew could not provide sufficient words and phrases to express deep ideas and nuances, neologisms should have constituted the main portion of the lexicon and classical Hebrew entries should have been in the minority. The fact that most of the vocabulary in Ibn-Tibbon's translation was taken from classical Hebrew suggests that reservations and complaints regarding the state of Medieval Hebrew might be due less to the actual state of Hebrew and more to a perceived need to defend against claims of medieval authors and philosophers (e.g., Abraham Ibn-Ezra) critical of the way other authors tried to make changes in the Hebrew language.
The following is a description of representative entries used by Ibn-Tibbon to translate Duties of the Hearts. 16 The aim of this description is to present and examine the nature of the Rabbinic Hebrew lexicon in the nominal lexicon of Ibn-Tibbon. It will hopefully shed light on the rich semantic and morphological variety of Medieval Hebrew, both from the perspective of Rabbinic 13 New meanings for lexemes which occur in Classical Hebrew. 14 New lexemes which were created by using existing morphological elements. 15 Compound noun which did not occur in Classical Hebrew but were based on Classical Hebrew lexemes. 16 In this paper I will not discuss phrases of any kind.
Hebrew and from the perspective of Hebrew morphological and semantic mechanisms.
In each example the Hebrew entry, as it appears in Ibn-Tibbon's translation, will be followed by the Arabic equivalents in Ibn-Paquda's original. For each equivalent I will cite one example, which will include the Arabic original, 17 the Hebrew translation of Ibn-Tibbon, and the English translation of Hyamson. 18 In a footnote I will present the treatise and the chapter the example is cited from. Overall, Duties of the Hearts consists of an introduction and ten treatises:

Rabbinic entries in Ibn-Tibbon's translation of Duties of the Hearts
The rabbinic nominal entries can be divided into six categories:

Biblical lexemes with rabbinic meanings
In total, 33 lexemes from this category were found in Ibn-Tibbon's translation. Although not all the examples presented here reflect new or unknown meanings, they certainly comprise the largest part of Rabbinic Hebrew in Ibn-Tibbon's nominal lexicon.
(1) ‫ר‬ ‫בֶ‬ ‫אֵ‬ The biblical meaning of this lexeme is 'pinion (i.e., wing)', while its rabbinic meaning is 'limb, organ'. 19 These original and later meanings reflect a simple metonymy, in which the original meaning represents a specific example and the later meaning a more simplified and general meaning that is based on the biblical meaning. This lexeme is used by Ibn-Tibbon to translate four different Arabic equivalents: As is clearly evident in these citations, equivalents (a) to (c) correspond to the rabbinic meanings. Apparently, equivalent (d) is the result of a mistake in the translation, probably made by Ibn-Tibbon himself, who mistakenly translated with this lexeme the word ‫אצחאב‬ which in Hebrew means ‫ים‬ ִ ‫ר‬ ‫בֵ‬ ‫חֲ‬ 'friends'. 24 (2) ‫ה‬ ‫ּפָ‬ ‫נֻ‬ ‫חֲ‬ While the biblical meaning of this entry is 'profaneness, pollution', the rabbinic meaning is 'fawning and praising in order to please someone'. 25 As in the previous example, the rabbinic meaning, which is employed in the Mekhilta de-Rabbi Ishmael, reflects a metonymy in comparison with the original biblical meaning: the rabbinic meaning represents the method of realising the concept that appears in the biblical meaning. In Ibn-Tibbon's translation, there is one equivalent for this Hebrew entry: (3) ‫יס‬ ‫ּכִ‬ The biblical meaning of this word is 'bag, purse', 27 and its rabbinic meaning is 'skin pocket in which glands are placed'. 28 This entry has two Arabic equivalents in Ibn-Tibbon's translation: The semantic shift from the original biblical meaning to the rabbinic meaning is expressed by a metaphor based on the resemblance of shape and designation between the two. 26 The service of God, chapter 3. 27 Brown-Driver-Briggs, Lexicon, p. 476 28 Ben-Yehuda, Milon ha-Lashon ha-Ivrit, pp. 2346-2347. 29 Examination of creation, chapter 5. 30 Ibid.

(4) ‫ב‬ ‫ׁשָ‬
Biblical meaning: 'returning, coming back'; rabbinic meaning: 'penitent'. 31 Mishnaic Hebrew reflects a meaning that is more metaphorical in comparison with the biblical meaning. This metaphorical shift represents the movement of meaning from the physical field to the spiritual-cognitive field. In Ibn-Tibbon's translation, this entry has one Arabic equivalent: In the dictionaries of Even-Shoshan and Ben-Yehuda the lexeme ‫ה‬ ָ ‫ר‬ ‫הָ‬ ‫זְ‬ ‫הַ‬ is claimed to be a neologism of Medieval Hebrew. As revealed by the Historical Dictionary Project of the Academy of the Hebrew Language, it is found already in the Babylonian Talmud, Shebuoth 47b (MS Vatican 140). This lexeme also appears in the liturgy of Yannai and Ha-Kalir, in different manuscripts and in Genizah segments. 39 However, it is doubtful whether Ibn-Tibbon was familiar with these specific writings and witnesses, and it is possible, even probable, that he created this neologism on his own.
The following examples (7 and 8) reflect another phenomenon that is characteristic of Rabbinic Hebrew; the assimilation of 37

Rabbinic lexemes with new meanings
As is common in many developing languages, semantic shifts are an elementary method for enriching an existing vocabulary and for bringing back into use lexemes that were once part of the lexicon. Like many others before him, Ibn-Tibbon used metaphors and metonymies for this purpose. On rare occasions, he used ellipsis, folk etymology, and loan shifts. All these rare cases involve biblical lexemes or other medieval neologisms, and therefore I will not present them here. 62  The roots ‫,גב״ר‬ ‫,נא״י‬ ‫שד״ל‬ and ‫שת״פ‬ all exist in the Hitpael stem in Rabbinic Hebrew, but do not occur as verbal nouns. The roots ‫זמ״ן‬ and ‫סל״ק‬ also exist in Rabbinic Hebrew, but the relevant meanings of these roots are semantic neologisms coined by Ibn-Tibbon. here presents this method and its basic components. Nevertheless, we should take into consideration the fact that this method was not employed exclusively with Rabbinic Hebrew by Ibn-Tibbon, for he used the same approach and principles when enriching the lexicon with lexemes from all periods of Hebrew. It seems that methodologically, Ibn-Tibbon was familiar with semantic processes and with the grammatical characteristics of Hebrew and Arabic and that he unquestionably knew how to use them in order to enrich the Hebrew lexicon.

Linear word-formation: Rabbinic stems
As can be seen in the above examples, the same Hebrew lexeme is frequently used to translate several Arabic equivalents. This obviously reflects the condition of medieval Hebrew, and especially the richness of, and variety in, the Arabic lexicon, in comparison with the insufficiency of Hebrew. Although this is the case with most of the Hebrew entries, one should take into consideration that, at times, the opposite occurred, when the same Arabic lexeme had several Hebrew equivalents. Frequently Ibn-Tibbon created neologisms by adding suffixes to an existing Hebrew lexeme (a lexeme from an earlier stage of Hebrew or a neologism of his own). This suggests a moderately automatic way for creating neologisms and enriching the Hebrew lexicon. Similarly, for Ibn-Tibbon the creation of verbal nouns and nouns from existing Hebrew roots has become a productive method for new lexemes.
Semantically, the lexicon of Rabbinic Hebrew in this translation is varied. An analysis of all the rabbinic entries suggests that the semantic fields from which they were taken were rich and broad, and they correlate with all the subjects Ibn-Paquda deals with in his book: Halakhah (Jewish law), nature, proficiency, economics, time, the human body, faith, knowledge, society, and culture.
In light of all that has been stated above, I have some reservations regarding the declarations of Jewish authors and

Yehonatan Wormser 1
The distinction between the two branches of Rabbinic Hebrew -the Palestinian branch and the Babylonian branch -has been well accepted from the very beginning of the modern study of Rabbinic Hebrew. Zacharias Frankel was probably the first to comment on this distinction, in 1859. 2 More than fifty years later, in 1912, Jacob Nahum Epstein briefly mentioned this distinction as a known fact. 3 In 1933, Harold Louis Ginsberg published a comprehensive study about it, 4 and five years later Epstein introduced a detailed description of this subject in his monumental introduction to the text of the Mishnah. 5 Later scholars, such as Kutscher, 6 Bendavid, 7 Rosenthal, 8 Bar-Asher, 9 and Breuer,10 continued in this course, expanding and detailing the basic distinction. However, the latest developments in this domain, in which numerous details of this distinction have been questioned or proven wrong (that is to say, linguistic features which were considered characteristic only of one branch were also found in texts of the other branch), have blurred this distinction. The two most important scholars who have dealt with such cases are Friedman 11 and Breuer. 12 One of the features that has remained a fairly stable distinguishing feature up to present is the spelling of the conjunction ‫אלא‬ 'but (rather)': in Babylonian texts it is frequently (but not always) written with yod, ‫,אילא‬ while in Palestinian texts it is written with the standard defective spelling. The different spelling methods reflect different pronunciations: in the Land of Israel the vowel of the initial alef was probably the segol, but in Babylonia, according to the testimony of manuscripts with Babylonian vocalisation, 13 along with Yemeni oral traditions, 14 it was ṣere or ḥireq. The first to indicate this difference in spelling was probably Sokoloff, in a short comment in his doctoral dissertation. 15 But the issue became widely known only a few years later, after Yeivin published a thorough study in which he examined the spelling of ‫אלא‬ and ‫אילא‬ in a wide range of different manuscripts. 16 He introduced his conclusions very carefully, emphasising that they were liable to necessitate revision on the basis of future manuscript research. Nevertheless, this distinction has been well accepted, even though, as we shall see, it has not always enjoyed complete confirmation in further findings. This acceptance was also strengthened by the parallel Aramaic dialects of the period: the form ‫אילא‬ is very common in Jewish Babylonian Aramaic texts, 17 but in Palestinian Aramaic it occurs very rarely. 18 In this paper I would like to examine what can be learnt about this matter from texts of the well-known and widespread genre of the late Midrash, the Tanḥuma-Yelammedenu (TY) genre. TY literature, according to most studies, was created in the Land of Israel after the Amoraic period. Initially it included written summaries of oral sermons (derashot), which were compiled into unified collections. 19 A few of those collections are known nowadays as the two editions of Tanḥuma (the 'standard' edition 6. The Distinction between Branches of Rabbinic Hebrew and Buber edition), Shemot Rabbah, Bemidbar Rabbah and Devarim Rabbah (two different editions). But it is clear that there were more TY editions, from which we have only remnants preserved in Cairo and European Genizah fragments, and in short quotations in yalkutim (medieval collections of Midrashim), while their full texts have been lost. As to its linguistic character, the Hebrew of TY literature reflects its Palestinian sources very clearly. 20 Indeed, the Palestinian linguistic features were not equally preserved in all TY editions, and in at least a few of them, some of these features were considerably blurred. 21 From the perspective of the Palestinian linguistic features we can single out a group of Cairo Genizah fragments of lost TY editions, 22  is consistently employed in those texts (e.g., ‫עליי‬ 'on me', ‫בניי‬ 'my sons' etc.), 24 final nun frequently substitutes radical final mem (e.g., ‫אדן‬ instead of ‫אדם‬ 'man, person', ‫כשן‬ meaning ‫כשם‬ 'like'), 25 and consonantal alef is always omitted in certain words (e.g., in the name ‫,אלעזר‬ which is written ‫,לעזר‬ or in the construct ‫כאילו‬ 'as if', which appears as ‫.)כילו‬ 26 From this group, our main interest here is in one TY edition, which is represented in four Genizah fragments. 27 The Palestinian linguistic character of this edition is obvious: except for the abovementioned features, which all appear in those texts, we find here the extraordinary form ‫כיויכול‬ instead of ‫כביכול‬ 'seemingly'. That is, a waw had substituted the bet, a well-known Palestinian spelling phenomenon. 28 Other striking forms in these texts are the 6. The Distinction between Branches of Rabbinic Hebrew constructs ‫שו‬ and ‫,שי‬ meaning ‫שההוא‬ 'that he', ‫שההיא‬ 'that she'. The elision of h is witnessed also in the equivalent form in Jewish Palestinian Aramaic of the Byzantine period ‫,דו‬ shortened form of ‫,דההוא‬ which frequently occurs in this dialect. 29 Considering all these features, it seems beyond doubt that this text represents an original early Palestinian linguistic tradition.
There is only one feature in this text that seemingly contradicts this assumption -the spelling of ‫,אלא‬ which occurs twentysix times in the text, all of them in the 'Babylonian' form ‫.אילא‬ Given the frequency, it cannot be explained as a scribe's spelling mistake. It also cannot be assumed that yod was used as a vowel letter representing the vowel of segol in the initial alef -because yod is employed frequently in this text to represent ṣere, but it never comes with segol.
Rather we should raise the question, how did it come about that a typical Babylonian form appears in an otherwise Palestinian text? We are not able to provide a certain explanation, but there are three reasonable options: it could be an independent development in the Hebrew of the Land of Israel; it may be due to the influence of a foreign linguistic tradition; or the explanation might involve a combination of the two aforementioned options. According to the first alternative, it may be that the gemination of the lamed was simplified for some reason. The loss of gemination might then have brought about the lengthening of the preceding vowel, the segol. This lengthening could then have been realised as substitution of the segol by a ṣere: ‫א‬ ‫ּלָ‬ ‫אֶ‬ < ‫א‬ ֿ ‫לָ‬ ‫*אֶ‬ < ‫א‬ ֿ ‫לָ‬ ‫,*אֵ‬ a common process in the Tiberian vocalization system. 30 As for the second option, since TY literature is considered a relatively late stratum of Rabbinic Hebrew, i.e., from after the Amoraic period, it is possible that when this text was written, the Babylonian Talmud and even Geonic literature had already reached an exclusive and authoritative position in the Jewish literary canon. In such a situation, the Babylonian linguistic tradition could have had an impact even in regions where the Palestinian traditions were practiced.
Whatever the reason behind this form, if we consider a few findings from Tannaitic Hebrew, its absolute attribution to the Babylonian branch seems quite dubious: Eldar 31 and Yeivin 32 have found a few occurrences of the form ‫א‬ ‫לָ‬ ‫,אֵ‬ vocalised with ṣere and without dagesh in Tiberian manuscripts; Eldar also commented on the occurrence of the spelling ‫אילא‬ in MS Cambridge, Add.470.1 (widely known due the edition published by Lowe); 33  branches. We actually find ourselves in line with the attitude advocated by Bendavid more than fifty years ago: 40 Now, after detailing hundreds of tiny differences between the Palestinian version and the Babylonian version, it is advisable to qualify our words and resist an overly schematic division. In reality, there is no clear Palestinian or Babylonian type. The literature of the sages of the Land of Israel abounded in Babylon for generations, and the formulation of their sayings was sometimes precisely and sometimes less precisely preserved.
[…] There is but a difference of proportions between the two types -Palestinian and Babylonian -(linguistic) features occurring frequently (in one branch), rarely (in the other branch).
It seems that this view has not gained sufficient attention among researchers of Rabbinic Hebrew, who, in many cases, have tended to attribute linguistic features only to one branch, ignoring or objecting to the possibility of their presence in the other branch. 41 In my opinion, the distinction between Palestinian Hebrew and Babylonian Hebrew should most often be regarded as a relative rather than absolute distinction. Bendavid pointed to the influence of the sages of the Land of Israel on Babylonian Jews, but, as a matter of fact, the influence was mutual. There was continual interaction between the two communities during the Tannaitic and Amoraic periods and thereafter, with scholars travelling or migrating from one country to the other. By this 40 43 It is likely that, in many cases, even the written texts moved from one place to another, and continued to be edited in their new location. The result of such cases is a kind of combination of the different traditions, as may have happened, according to Epstein's assumption, 44 in a few manuscripts that were written in the Land of Israel, but vocalised in Babylon. 45 Therefore, we should rarely if ever expect to find a criterion on the basis of which it is possible absolutely to distinguish between the branches. Whenever an apparently distinctive feature is 42 45 In most cases it is probably impossible to determine whether the fusion of traditions represents testimony authentic of living Hebrew, i.e., the language of an author of a rabbinic text as an actual representation of a Palestinian or Babylonian tradition, or just late corruptions introduced by a copyist. The reason for the importance of the findings presented here is that the main text discussed is clearly an original text of the Palestinian tradition, so the assumption that the appearance of ‫אילא‬ here is an original feature seems very reasonable.
identified, it should be remembered that any characteristic of the Hebrew of one branch may appear, to one degree or another, in the other branch. 46 Recognition of this fact does not entail rejecting the fundamental concept of the linguistic distinction between the two branches. It just puts it in its right perspective. If this view is accepted, we should abandon any attempt to find a single criterion to determine the type of a particular text, as Yeivin proposed regarding the form ‫:אילא‬ 47 A manuscript in which this word is written only in defective spelling is probably a Palestinian manuscript. Indeed, it is not absolute evidence, because there are also a few Babylonian manuscripts in which this word is written only defectively, and therefore, depsite this spelling, it is possible that this is a Babylonian manuscript. On the other hand, a manuscript in which the plene spelling is found, constantly or occasionally, is certainly a Babylonian manuscript.
In conclusion, we have pointed out the fact that the form ‫,אילא‬ which is considered a characteristic of the Babylonian branch of Rabbinic Hebrew, is also found in texts that belong to the Palestinian tradition. It seems that this tendency intensified after the Amoraic period, in the Hebrew of TY. There are two possible reasons for this situation: it may be an independent development in Palestinian Hebrew or, alternatively, a result of Babylonian influence on the Palestinian branch. Whatever the reason, the fact is that a characteristically Babylonian form has come to be found, however rarely, in the Palestinian tradition. But according to our 46

Shlomi Efrati
The Psiqata of the Ten Commandments (henceforth: PsTC) is a relatively unknown rabbinic composition. 1 It has an unusual transmission history and relations between its textual witnesses are intriguing. In what follows I will briefly describe PsTC and stylistic differences make it clear that PsTC is an independent work, which at a certain point was incorporated into PsR. Like most of rabbinic literature, PsTC is not a continuous, uniform composition, but rather a compilation of fragments. It incorporates various sayings, homilies, and stories, and shows little effort, if any, to integrate or harmonise these into a coherent and continuous text. In general, it seems that PsTC faithfully preserves sayings of Palestinian Amoraim, the rabbis of the third to fifth centuries CE. 7 Therefore, any attempt to study PsTC should take into account at least two levels of development: the traditions cited in PsTC (which themselves may have undergone a long process of development before they were integrated into PsTC), and the redaction and composition of PsTC itself. Of course, it is not always easy to distinguish the different components of PsTC. For instance, differences in wording between PsTC and parallel composition(s) may, on the one hand, stem from alterations of the text made by the editor/redactor of PsTC (or its parallel(s)); or, on the other hand, represent earlier variations in the form of textual traditions that were faithfully preserved in each of the parallels.
The study of PsTC, or in fact any ancient composition, becomes even more complicated when we consider the textual transmission of such a composition and its implications for evaluating its text. It goes without saying that the text of PsTC as 7. Two Textual Versions of Psiqata of the Ten Commandments we have it, preserved mainly in late, medieval manuscripts and anthologies (see below), is somewhat removed from its original form, due to copyist errors or secondary interventions in the text. True, by collating textual witnesses and carefully examining the text it is possible to discern secondary readings and reconstruct a more reliable text. However, not every textual variant can be accounted for, and, more important, not every variation reflects a corruption of an original text. Sometimes such variants represent a degree of fluidity in the 'original' text itself.
I would like to demonstrate such 'original' variants through the intriguing textual situation of PsTC. This composition is known through two main channels of transmission: medieval European manuscripts, on the one hand, and citations in eastern anthologies, on the other. Let us briefly examine these channels.
As mentioned above, the complete text of PsTC is preserved only as part of Psiqata Rabbati, and came down to us in the textual witnesses of this latter composition. These include only four independent (direct) witnesses: 8 Three medieval manuscripts, the earliest of which dating to 1270, 9 and the first printed edition of PsR. 10 These direct witnesses represent one branch of transmission, which is made evident by many secondary readings shared by all of them. To give only two examples: 2 (= PsR 22, 111b)  The mention of R. Eleazar b. Yose at the beginning of section 2 is awkward: the following sentence states clearly that two people were walking ‫תרויהון(‬ ‫מהלכין‬ ‫)הוון‬ and goes on to tell only of R. Yeremiah and R. Zeʻera. It seems that the name of R. Eleazar b. Yose was mistakenly copied from the beginning of section 3, due to the repetition of the phrase ‫לשלשל‬ ‫צריך‬ at the end of sections 1-2.

PsTC
These examples exhibit simple and common copying mistakes. In both of them, however, the corrupted text appears consistently throughout all of the direct textual witnesses. As it is rather unlikely that several scribes made exactly the same mistakes independently, it is quite probable that all of the direct witnesses stem from a certain older copy of PsR that contained these -and many others -corrupt readings.
Besides the textual branch of the direct witnesses there is another line of transmission of PsTC, preserved mainly as citations in two medieval anthologies, or Yalqutim: Midrash HaGadol (= MG), a fourteenth-century Yemenite anthology, 15  HaMaʻasiot (= SM), an anthology of tales whose date and provenance are not quite clear. 16 In general, these two Yalqutim tend to agree almost verbatim (when they overlap). Obviously, there is some close relationship between them, though the exact nature of that relationship is not entirely clear. 17 Each of these Yalqutim cites passages from PsTC that are absent in the other, and I will regard them as (independent) witnesses of a certain version of PsTC, a version clearly distinct from the one preserved in the direct witnesses. One important feature of these Yalqutim is that they seem to be completely unfamiliar with other parts of PsR, apart from chapters 21-24 (= PsTC). This is a strong indication that PsTC was circulating independently of PsR. 18 Even more important than the evidence of independent circulation are the numerous variations between the version of PsTC preserved in the Yalqutim a human king, if he is a philanthropos, benevolent and kind, 24 would allow his servants one day (out of seven) to handle their private affairs; but God allows his people six days for their own work and demands only one day -the Sabbath -for himself. The loanword ‫*פילנתרפוס‬ is not attested, as far as I am aware, anywhere else in rabbinic literature. It is probably due to its rarity that it was replaced with the similar looking and better-known word ‫.פילוסופוס‬ In this case the version preserved in MG not only enabled us to reconstruct the original form and meaning of the parable, but also enriched our knowledge of Rabbinic Hebrew. 25 However, variant readings in the two versions do not always reflect an error or secondary reading in one (and sometimes both) of the versions. Not infrequently the two versions exhibit what seem to be good, genuine, reliable, yet different texts. This is especially evident when examining the relatively long Aramaic tales that are included in PsTC. Many of these are presented in rather different forms in each of the versions of PsTC, yet both forms are in good Jewish Palestinian Aramaic. Now, the use of JPA declined and eventually ceased during the first centuries after the Arab conquest. Medieval scribes and authors had little (if any) familiarity with this dialect, being much more familiar with Jewish Babylonian Aramaic, due to the immense influence of the Babylonian Talmud. 26 Therefore, the use of JPA in both 7. Two Textual Versions of Psiqata of the Ten Commandments versions, in a seemingly free and lively manner, testifies to their antiquity and, presumably, authenticity.
In order to demonstrate the character and significance of such variant tales I would like to present and discuss one lengthy example. 27 The following story exemplifies a problematic aspect of a common Jewish ritual item tefillin (phylacteries). Tefillin are considered a marker of piety and righteousness. However, not everyone who wears them is indeed pious and trustworthy, as can be seen from the following incident. A certain man reached his destination, presumably far away from home, just before Shabbat. When he saw someone wearing tefillin, standing in prayer, he decided to leave his money with him (carrying money during Shabbat is forbidden according to Jewish law). After Shabbat the man came back to ask for his money, at which point the other person denied having received any money from him. The first man, angry yet helpless, cried out: 'It is not you that I believed, but that holy name that was on your head' -that is, the tefillin. But the story does not end here. Elijah the prophet appeared to the man and told him how to retrieve his money: he should go to the hypocrite's wife and tell her that her husband asks her to give him back the deposit. In order for the wife to believe him, he should tell her that she and her husband ate leaven on Passover and pork on the Day of Atonement (a day of fasting) -an incriminating secret that served as an agreed sign between them. The man did so, and the wife innocently gave him back his money. When her husband returned and found out, he began beating her. But when his wife told him all that had happened, and that their transgressions were . Full text and translation of both versions of the story are given in the appendix to this paper. exposed, they decided 'to return to how they used to be', implying that the couple, presumably proselytes, would now return to live as gentiles. 28 This tale appears nowhere else in rabbinic literature, except in PsTC. However, the two versions of PsTC exhibit two rather different forms of the same story. Let us examine a few of the more interesting differences between the two. It happened that a certain man came to a certain place on a Friday toward sunset and had with him money to deposit.

Direct witnesses Yalqutim
He entered a synagogue and found a certain man praying with tefillin on his head.
He entered a synagogue and found a certain man standing and praying with his tefillin on him. Some say he was a proselyte.
This section serves as an exposition, presenting the two main characters of the story. The direct witnesses describe them only in terms of their actions: the one carrying money, the other He wrapped himself in his cloak and stood and prayed in that place, and said in front of Him: "Master of the world, It was not him that I believed, but Your holy name that was on his head." What did that man do? He went and stood praying in front of the ark, and said: "Master of the world, It was not him that I believed, but Your holy name that was on his head I believed." After the hypocrite denied he was given any money, the poor man who gave him the money, furious and helpless, rebuked him and then cried out to God. The Yalqutim report that the man 'went and stood praying in front of the ark ‫ארונא(‬ ‫',)קומי‬ that is the chest dedicated to holding the scroll(s) of the holy scripture(s), a physical and conceptual focal point of Jewish synagogues. This specific use of the common Aramaic word ‫ארונא‬ is unique to JPA. 32 The direct witnesses, on the other hand, present this episode in Hebrew rather than Aramaic, and in a more elaborate way: 'he wrapped himself in his cloak and stood and prayed in that place', etc. The Hebrew appears somewhat unexpectedly in the middle of an Aramaic passage, but it must be noted that such shifts of language are not uncommon in 'classic' rabbinic compositions. 33 In fact, the conclusion of this very tale is in Hebrew, according to both versions. The change in language and content reflects different literary choices made in each of the versions.  37 Give me this object (= the deposit)." These people (= we) eat leaven on the night of Passover and of that thing on the day of the Great Fast.ʼ And (then) she shall give (it) to you." This passage relates the contents of Elijah's revelation, and reveals the hypocrite's and his wife's hidden sins. The Yalqutim version uses the relatively rare self-referential clause ‫עמא‬ ‫דאיליך‬ 'these people', which is unique to JPA. 38 Note that these words, ‫עמא‬ ‫,דאיליך‬ are graphically and phonetically similar to the clause ‫הּ‬ ‫מַ‬ ‫עִ‬ ‫ליה‬ ‫דהוה‬ 'that he had with her', which is found in the direct witnesses exactly at the same place. 39 It seems that the Yalqutim preserve a genuine Aramaic phrase that was omitted or replaced in the version of the direct witnesses. But this is not to say that the version of the direct witnesses is secondary or less reliable. Just at the end of Elijah's words, the direct witnesses use the phrase ‫פלן‬ ‫ת‬ ‫מַ‬ ‫קְ‬ ‫מִ‬ 'this object', a fine specimen of JPA. 40 By now it should be clear that the differences between the two versions of PsTC, at least as far as this tale is concerned, are much more than mere scribal errors. Each version presents, in general, a good text, from both a literary and a linguistic perspective. The fact that both versions use good, authentic Palestinian Aramaic is extremely important, as it shows that these versions are not Psiqata DeRav Kahanah, VaYehi 7 (pp. 128-129): ‫עמא‬ ‫הלין‬ ‫על‬ ‫דאמ׳‬ ‫מה‬ ‫כל‬ ‫מייתין‬ ‫עמא‬ ‫הלין‬ ‫לא‬ ‫ואי‬ ‫מביניכון,‬ ‫עיבריא‬ ‫אילן‬ ‫נפיק‬ ‫אתון‬ ‫עליהון.‬ ‫אתא‬ 'all that he (i.e., Moses) has said concerning these people (i.e., us, the Egyptians) has happened to them (i.e., us). Come, let us expel these Hebrews from among you, or else these people (i.e., we, the firstborn) will die'. Note that in each of these instances the term ‫עמא‬ ‫אילין‬ refers to 'others', i.e., proselytes or gentiles. the product of medieval scribes or redactors, who were no longer able to use JPA to such an extent. 41 Rather, both versions were given their final form in a historical and geographical context in which this dialect was, if not actually spoken, at least in common literary use. It would seem, therefore, that the differences between these versions, rather than representing corruptions or reworking of an original fixed text, reflect some fluidity in the text itself.
The nature and meaning of this fluidity can be explained in several ways. It is possible that they represent a kind of 'creative transmission', that is, the active and intentional interventions of later transmitters in an original text. 42 Indeed, it is sometimes possible to discern a secondary reading in one version or the other. 43 But in most instances both versions preserve equally reliable readings. Moreover, the scope and frequency of the textual variants examined here, which are by no means exceptional, 44 may suggest that there was something in the text itself that made it especially susceptible to such alterations. In other words, it is the text itself that was -to a certain degree -changeable and fluid. According to this model, the differences between the two versions reflect a relatively early stage of transmission, when the redaction -that is, the process of choosing, arranging and ordering the segments of PsTC -was completed, and also the text of these segments was more or less fixed -but not entirely. A certain degree of freedom was allowed, or perhaps inevitable, during this early, possibly oral, 45 stage of transmission. 46 Whether the curious textual situation of PsTC represents the inherent fluidity and openness of the text itself or the creative reshaping of a (hypothetical) original text by its transmitters is not easy to decide. Perhaps more important, however, is the recognition that both versions are equally important for the study and understanding of this composition. This is especially true in regard to the subject of this volume, that is, the study of Rabbinic Hebrew and Aramaic. As has been demonstrated above, both versions of PsTC represent authentic and common use of Palestinian Aramaic, and offer rich, invaluable materials for the study of this dialect. I hope that future researchers will make use of these treasuries, thus enriching our knowledge of this most important, yet somewhat neglected branch of Aramaic.
it cannot be denied that loanwords were especially vulnerable to scribal errors. But when a comparison is made between the form of loanwords in excellent manuscripts of the Mishnah, on the one hand, and the form of their etymons according to Greek pronunciation in the Roman and Byzantine periods, on the other hand, many incongruities vanish, and a more consistent picture emerges.
A case in point is the word ‫קנוניא‬ 'conspiracy'. The word is documented 35 times in rabbinic literature, most notably in two well-known tractates of the Babylonian Talmud, Baba Metziah and Baba Bathra. 3 But in the three best manuscripts of the Mishnah -MSS Kaufmann, Parma, and Cambridge -the word appears in a slightly different form, with a yod after the qof: ‫,קינונייא‬ ‫,קינוניה‬ ‫קינונייא‬ etc. 4 This loanword is derived from Greek κοινωνία. 5 In Classical Greek the letter-combination οι represented the diphthong [oi], making the Hebrew form less than an exact equivalent. However, in the relevant era (i.e., at the end of the Hellenistic period and the beginning of the Byzantine period), the combination οι represented the rounded front vowel /y/; 6 consequently we may assume that the pronunciation of the word 8. Vowel Reduction in Greek Loanwords in the Mishnah in those times was /kynonia/, which is the pronunciation reflected in the spelling ‫.קינוניא‬ We see therefore, that that an overlap exists between the form of the loanword in good manuscripts and its pronunciation in Koine Greek. Systematic research confirms that most incongruities between Hebrew forms and Greek forms can be explained in this way, and that both the vocalisation and the spelling of the loanwords in good manuscripts reflect a reliable tradition of pronunciation of these words. 7

Vowel reduction
A systematic comparison between the vowels in loanwords and their equivalents in Greek shows that in a considerable number of words we find a shewa in Hebrew against a vowel in Greek. In most cases this reduction is evidenced in an open pretonic syllable, i.e., the syllable before the final syllable. 8 Thus we find, for instance, that the Greek πάρδαλις was loaned as ‫ס‬ ‫לֵ‬ ‫ּדְ‬ ְ ‫ר‬ ‫ּפַ‬ 'leopard' (Baba Kamma 1.3), and ἀτελής was loaned as ‫ס‬ ‫לֵ‬ ‫טְ‬ ‫אַ‬ 'market' (Bekhoroth 5.1). It ought to be emphasised, that the term 'pretonic syllable' refers here to the syllable structure of the word after it was loaned, because during their passage from Greek to Hebrew (and Aramaic), words often changed their syllable structure, especially due to loss of endings. For example, in the word χαράκωμα > ‫קֹום‬ ְ ‫ר‬ ‫ּכַ‬ 'palisade' (Ketuboth 2.9), the vowel reduction occurred in the pretonic syllable after it had been loaned into Hebrew. The material also shows that the vowel /a/ was more prone to reduction than other vowels. In the following subsections data from MS Kaufmann is presented, occasionally with examples from other manuscripts of the Mishnah. 9