<<

CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC: CAN WE DISPENSE WITH ANACHRONISM?

Shlomo IZRE'EL*

Introduction In paper delivered at the meeting of the team of the forthcoming new edition of The Dictionary of Northwest Semitic Inscriptions (DNWSI) in Leiden at the end of 2001 (Izre'el 2003), tried to set up a methodology for marking out West Semitic words in cuneiform texts from the Canaanite region. In that paper I dwelt on primary identification of Canaanite glosses, basically discussing the writing system and scribal traditions. The following are the main points to be considered: • Cuneiform writing does not separate between words, and thus signs can be attributed to strings on either side. • Cuneiform signs can be interpreted either as syllabic signs or as logograms. • Cuneiform signs are polyphonic and can bear different, sometimes quite remote values. Also, cuneiform signs have different values in different areas and periods, and must be interpreted according to their assigned scribal traditions. • When a sign is not inscribed carefully or the surface upon which it was inscribed has been damaged, there are many more options for reading and restoration than in alphabetic or consonantal, linear or pictographic script. • Non-Akkadian phonemes that form part of the Semitic stock and could have been part of the phonological system of the languages of the Levant in the second millennium BC (d, t, z, g, d, g,, h, h, o) are not represented faithfully by the cuneiform writing system. Laryngeal and pharyngeal consonants can be indicated in spelling only indirectly; in many cases such consonants are in effect ignored. In order to represent o, Canaanite scribes used the cuneiform u series.1 • Consonantal doubling is not always manifested in cuneiform writing, and it is less common in Amarna texts than others. However, when it is manifested, it

• Professor , Department of Hebrew and , Tel Aviv University

66 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

must be interpreted as reflecting genuine phonological or morphophonological doubling. • Vocalic length is not always manifest in writing. However, when plene writing does occur, one must attend to its meaning, which in some cases represents

genuine phonological length. Apart from these issues of decipherment and interpretation, one must also pay attention to the following primary methodological concerns: • The language in which the are written was thought by the local scribes to be Akkadian (Izre'el 1987b). However, one might rightfully say that the language was a mixed system, and in some scribal traditions it was based more on Canaanite than on Akkadian and therefore included more Canaanite features than Akkadian. The wisdom needed to distinguish disparate elements is not trivial, since Canaano-Akkadian is a single, unified system. • The inherent variation represented in the writings of the Amarna Canaanite scribes and their respective scribal traditions needs close scrutiny before one can determine non-Akkadian components. Still, much of the existing variation is concealed by the writing system, which was not only a foreign language to the scribes, but was in itself traditional and constrained by learning habits.

In this paper I would like first to suggest some further methodological concerns, in order to deal with what can definitely be seen as Canaanite- mainly in the domain of glosses-and then see to what extent the data at hand can be used to show variation in the Canaanite dialect continuum of the 14th century BC. I will cover only the area of The Land of the Bible (to use Aharoni's term, 1967), which fits the interest of this workshop, which is dealing with dialectal variation in Hebrew. I should further emphasize that what will be discussed here are only those words that can be unequivocally isolated from the bulk of the Canaano-Akkadian flow as genuine Canaanite, either by their being marked by a Glossenkeil or otherwise. As will be shown later, this is not always clear. I have not discussed individual grammatical features integrated in the Canaano-Akkadian linguistic system, neither have I included in this study Canaanite roots that were integrated in the system through derivation. On the other hand, I have discussed, in some cases, Canaanite nominal stems that have adapted Akkadian inflectional affixes (e.g.,/ibu/, #9). I have not included proper names in the study. Although proper names may reveal some tendencies in local dialects, they tend to preserve older features and are therefore unreliable for synchronic studies. Lastly, one should notice that the data presented here is not complete, as the aim of this paper is to discuss variation rather than to

Vol. XXXVIII 2003 67 present a list of all Canaanite lexemes attested. This will be given in the forthcoming edition of DNSWI (a preliminary list appears in Izre'el 1998b).

Methodological Concerns Let me first discuss some methodological concerns which must be dealt with when trying to determine genuine variation in the Canaanite data.

Akkadian, Canaanite, and Canaano-Akkadian First, one must always remember that even when scribes used glosses and marked them by a Glossenkeil, this is not an indication that a word is Canaanite. Glossenkeil-marked words can be Canaanite, Canaano-Akkadian, Akkadian, Hurrian, Egyptian, or even Sumerian. Since the two main components from which Canaano-Akkadian is built, Canaanite and Akkadian, are two languages belonging to the same language family, some forms can be identical in both languages, at least in writing. This is true not only for some words within the flow of the text, but also for some glosses. Note the following two examples:

(1) LUMES•_ma-sa-ar-/massarta/ "garrison" (EA 263: 24, central Palestine)2

(2) GAZ•_di12-[k]a-t[i] /dikat/ "I was killed" (EA 287: 73, Jerusalem) In the first example, the Akkadian word "garrison" is explanatory to the Sumerogram, which means just "men", thus indicating what type of men are needed. In the second example, the Canaano-Akkadian passive verb interprets a Sumerogram which indeed means "kill", but is usually found in the Amarna letters as a component of the logogram indicating the Papiru groups (Rainey 1978a, 72). Some glosses can be interpreted as either Canaanite or Akkadian and therefore cannot help in determining actual Canaanite lexemes, neither can they be used to determine variation in lexical usage in 14th century . I mentioned some examples (pi "mouth", maska "leather", murra "myrrh" and mu-tu-mi "death") in Izre'el 2003, 28. A case in point to our discussion here is found in the following prostration formula: (3) a- sepe(GIRMES) sarri(LUGAL) beliya(EN-)7 samas(dUTU) is -tu sa-me-e 8 7-su 7-ta-a-an 9 us-he-hi-in 10 i-na pa-an-te-e •_

ba-at-- 11 u se-ru-ma •_su-uh-ru-ma

At the feet of the king, my lord, the sun from heaven, I prostrate

7 times and 7 times on the belly and on the back. (EA 232: 6-11,

Acre)

68 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC Obviously, the glosses here reflect Canaanite /batnu/ (BH baetaen) and

/Suhru/3 respectively. Similar prostration formulae can be found in many

Amarna letters from all over the region. However, the lexeme used for "belly" or "front (of the body)" is not pantu or batnu , but-mostly-kabattu, which means "liver" in Akkadian. Because of this shift in meaning ("liver" •„ "front"), and because two of the occurrences of kabattu are preceded by a Glossenkeil, it has been regarded by some as Canaanite (in its adverbial form) (CAD K, 14;

Sivan 1984, 131, 235). If so, there is room for the conclusion that there is dialectal variation between two lexemes with the same meaning. However, there is no reason to assume a Canaanite loan in the case of kabattu (cf. AHw, 416;

Rainey 1978a, 75), except, perhaps, for the meaning ("front"), all the more so since the two occurrences where Glossenkeils are found are not clear of problems. One must also note that Glossenkeils serve not only as gloss markers, but also in other functions (Artzi 1963). Still, there are two occurrences that may be of interest here:

(4) se-ru-ma •_ u -ib-du-ma "on the back and on the belly" (EA 316: 9, Yursa)

(5) a-na muh-hi ka-bi-di-ia muh-hi •_ su-r[i-]ia "on my belly, on my back" (EA 147: 39, Tyre) The first example attests to a rather peculiar spelling of the word, which I cannot explain. It is possible to render it ka-ib-tu-ma to agree with the consonantal skeleton of the Akkadian lexeme. However, stepping away from the face value of the sign DU is unwarranted in Canaano-Akkadian.4 The second example attests to a rather rare form in Akkadian, also meaning "liver", which is found-besides in this Amarna occurrence-almost exclusively in lexical texts (in the form gabidu or kabidu, CAD G, 6; AHw, 272). Since it is only the second word, Suhriya, that is marked by a Glossenkeil, we may perhaps conclude that the word is not to be regarded as a borrowing from Canaanite but as a scholarly word, acquired though the curriculum of this scribe. The above rendering of the syllabic string, ka-bi-di-ia, which at face value would be read ga-bi-ti-ia, points towards a possible Canaanite-like reading of this string as kabidiya/, which may have been similar or identical to the cognate Akkadian/ form acquired at school. Still, one may further ask whether the meaning "belly" or "front (of the body)", which is not attested in Akkadian, can be attributed to Canaanite *kabidu. Given the scanty data at hand, any suggestion brought forward would be conjectural, although the attestation of the two examples above, one from northern and one from southern Canaan, may lead towards such an assumption.5

Vol. XXXVIII 2003 69 What do we do with batnu, then? Can we say anything on variation in the Canaanite domain as regards the geographical distribution of these two lexemes? In this case, the answer is clear, I believe, and is negative. Variation in the Canaano-Akkadian corpus, as real and prominent as it may be, does not necessarily reflect variation in the substratal dialects. In our case, while a lexical isogloss that separates Acre from both Tyre and Yursa is, of course, a possibility, the scantiness of data, the problems in interpretation, and especially the immediate textual context, prevent us from drawing such a conclusion. It is possible, even plausible, that the two lexemes were both part of the lexicons of the respective scribes, and the individual variant used in the specific formula stemmed from reasons other than dialectal variation. The gloss batnumma may have been attracted by the Akkadian lexeme pante "chest", which it translates (CAD B, 78-9 s.v. bamtu), due to their similarity in form. Another variant of this formula is found in the following examples:

(6) •_ ka-ba-tum-ma u sa-sa-lu-ma (EA 215: 4-5, unknown provenance) (7) it UZUsa-sa-lu-ma u kabattumma(UZUHAR) (EA 211: 5-6. unknown provenance) The use of sasallu here is interesting. sasallu is an academic word attested in Akkadian with the meaning "back", mostly in medical contexts (CAD S2, 169-70).6 The use of the Sumerogram in the second example (#7) may point to an academic-scholarly inclination of the scribe. Perhaps not only the use of sasallu instead of the common seru (or Canaanite Suhru) but also the common use of kabattumma stem from a similar origin, viz., an academic-scholarly tendency. The actual source and road followed by kabattu "liver" to arrive at its meaning "belly", "chest" or "body-front" in this formula still remains an enigma.

Linguistic change and geographic variation The Canaano-Akkadian materials make a unique corpus from which data on the vocalization of Canaanite dialects can be extracted. Nevertheless, cuneiform writing is highly conventionalized and conceals many subtleties that might be significant for variation studies. Discerned variation in texts from different periods must be meticulously investigated in order to determine whether it indicates variation also in the Canaanite substratal dialects from the synchronic point of view. After all, the data we are working with precedes by many centuries data from the Northwest Semitic dialects we have at our disposal from other sources. Of course, even without population migration and change in

70 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

political and social structure, languages do change in time, and therefore one must be extremely careful not to rely on historical reconstruction when dealing with 14th century data, but draw conclusions only from the actual data at hand. Take, for example, the Jerusalem Amarna letters. As I have stated in Izre'el 2003, 20, while the Hebrew of the Bible (at least the bulk of it) may reflect the dialect of Jerusalem, still (1) it is a first millennium BC dialect, whereas the Amarna evidence is from the second millennium BC, quite a few centuries earlier; and (2) it reflects a dialect of Hebrew, the roots of which may have been different from the local Canaanite dialect attested by the scribe of Jerusalem of the Amarna period. Apart from these two concerns, one must always remember that BH as it is known to us today reflects only in a very limited way the language of its original scribes. Its vowel system is anachronistic by at least a thousand years and reflects the preservation of a reading tradition rather than a living language. Information on the contemporary vocalic system and some of the consonants is drawn completely from secondary data like transcriptions, and very rarely from contemporary materials. Let us look at just one example extracted from these letters: (8) I sent as gifts to the king, my lord, x prisoners, 5000 ..., and 318 porters for the caravans of the king, my lord, (but)

la ki[-h]u i-n[a]•uu-ga-•v•_sa-de4-e[ ] 57uRuia-lu-naKI they have been taken in the countryside (of?) Yalona. (EA 287: 56-57)

The form la-ki[-h]u /lakihu/ is a 3 pl m qal passive verb in the suffix conjugation, a pattern unknown to us from the Hebrew Bible. It stands against the active form [la]-ka-hu I /lakahu/ in this very same letter (1. 36; see below,

#26). In BH, qal passive forms are attested only sporadically and indirectly, and their pattern is always reflected as •Eu•E•Ea•E (Bauer-Leander 1922, 316). In

Canaano-Akkadian,•Ea•Ei•Eis the commonly attested qal passive pattern, and there is no attestation of a •Eu•Ea•E pattern7 (Rainey 1996, II, 303-6). On the contemporary-synchronous level, there is no variation to be noticed. The question of whether *•Eu•Ea•Eis a later innovation within Canaanite or a borrowing of a different system cannot be answered (Bauer-Leander 1922, •˜38a' suggest a

Proto-Semitic origin for this pattern; see further below, on the discussion of

#30). One point of interest to which we shall return later (p. 80) is the retention of the middle short vowel in both forms, which enables the passive-active distinction to be overt. BH as it has been transmitted to us had lost this vowel in the 3 sg f and in the 3 pl forms before its stage of vocalization by the Massoretes.

Vol. XXXVIII 2003 71 Uncertainty about provenance While in many cases the provenance of a letter is known to us, in many others we lack any data about the origin of the letter. In some cases the name of the sender is unknown and his city is not mentioned; in other cases the beginning of the letter with its address is missing; in yet other cases the name of the city is mentioned, but archaeologists and historians either disagree about its exact location or do not have a clue for determining it. Chemical and petrographic work on the Amarna letters done by Yuval Goren and his colleagues from Tel Aviv University will enable a more exact siting by studying the components of the clay of the respective tablets. However, even that endeavor will not always determine the actual provenances of city rulers, since in some cases scribes were writing tablets on behalf of city rulers while actually being in a location quite remote from their cities. According to Goren's finds, about 10% of the letters from Canaanite rulers to Egypt were written in the Egyptian administrative centers in Canaan, notably Gaza, Beth Shean, Sumur and Kumidi (Goren, Finkelstein and Na'aman, forthcoming).8

Scribal education and the movement of scribes Another consideration to be born in mind is the possibility that scribes who had received their education in one place did their service in a different location, sometimes quite remote from their school. This is the case with the Jerusalem scribe, who had probably received his education somewhere in Syria (Moran 1975). This may also be the case with one scribe of Ribhaddi, prince of Byblos (Izre'el 1987b). At this juncture, one must note that the last letters of Ribhaddi were written during the time he was out of him home town and located in Beirut (EA 136-8). Scribes sometimes wrote letters for more than one Canaanite ruler, as is shown by the almost identical letters EA 201-6 (cf. also Izre'el 1977, 159- 63). Since the Amarna letters themselves do not include any data on their respective scribes, the only way of revealing such discrepancies is by philological and linguistic analyses. Therefore, conclusions in this regard can never be on the positive side, and elimination of data must be used in order to reach any reasonable conclusions.

Scanty data and lack of comparative contemporary data from the region The data at our disposal is so scanty that it is hardly necessary to say that we usually lack comparative materials from the region. That is to say, most of the forms isolated from the corpus are unique, and comparable forms from the same corpus that might reveal similar linguistic features are usually nonexistent or very rare.

72 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC Take, for example, the form i-bi (ibi or ibi) "enemies" in a letter from Shechem: (9) i-bi ti-sur-ru-na I shall protect the enemies. (EA 252: 28) As there is no other occurrence of this lexeme in Canaano-Akkadian, one must go to cognate Canaanite and other NWS dialects for comparative materials.9 From the Canaanite region the only attested cognate is BH 2oyeb (Baumgartner 1967, 37). However, Ugaritic attests to a form that seems to be closer to the one attested in the Shechem Amarna letter: alphabetic ib, syllabic e-bu (Del Olmo Lete and Sanmarin 2003, I, 4). Spelling requires that we transcribe the stem as having an i vowel, which may or may not be long, viz., ibu/ or /ibu/. Whether or not the contemporary Jerusalem dialect or any other / Canaanite dialects had a form similar to the BH one cannot be determined. Another example is the following gloss attested in a letter from the Egyptian pharaoh to the prince of Gezer:

(10) GADAMES •_ ma-al-ba-si "clothing" (EA 369: 9) I have tried to show elsewhere that the scribe of this text, although writing in Egypt, was one from Gezer (Izre'el 1995, 109-18). Whether this form was shared by other contemporary Canaanite dialects is impossible to determine for lack of data. While BH attests to a different pattern (ma•E•Eu•E; Baumgartner 1974,

555), comparative material from other Semitic languages suggests that this pattern is Proto-Semitic (op. cit., 117-18 n11). The origin of the Hebrew pattern

(ma•E•Ea•E) for the cognate noun is still to be sought. '° Our last example is the notorious hifil form hi-ih-be-e, attested in a letter from the Transjordanian city Pihilu:

(11) Imu-ut-balu(dISKUR)-mi 6 in--bi-itIa-ia- 7 •_ hi-ih-be-e Mutba'lu fled away; he has hidden Ayyab. (EA 256: 5-7) The normalization of this form suggests a long e, i.e., /hilibe/ (Sivan 1984, 174). This formation may be an outcome of a reconstructed form *hihbia, yet if so, we do not know the intermediate steps. We have no other suffix-conjugation hifil form to compare it with. We do have one other suffix-conjugation verb of a prima verb: na-as-sa-a "he has become elevated" (EA 366: 13; #36). This form, which is attested in a letter by Suwardata of Gath, is 3 sg m suffix- conjugation nifal (Rainey 1978a, 85; Izre'el 1995, 104 n5), which is ambiguous as well with regard to its ending. It can be interpreted as either /nassa'a/ or as I nassa/ (or perhaps even, although less likely, as I nassa2 /).11 I myself tend to prefer the first alternative. In any case, this nifal form cannot offer any insight as regards the hifil form under discussion. Furthermore, any assumption regarding

Vol. XXXVIII 2003 73 the contemporary Jerusalem dialect or any other Canaanite dialect can rely only on historical reconstruction rather than on actual data. However, as I have reiterated time and again already, we have no means for doing this in any reliable manner.

The Search for Linguistic Variation in Canaan Having laid down some methodological problems and difficulties, let us try nevertheless to search for variation in the dialect continuum of Canaan in the 14th century BC. I shall start with Tyre. Tyre is known as a Phoenician city, but it does relate to our discussion here not only because it marks the northernmost site of the dialect continuum we are to deal with, but also because Phoenician along with Aramaic -is usually taken as an anchoring point for determining the relationship among Hebrew dialects in the biblical period (Garr 1985, 229-35).

Tyre The scribe of Tyre has been the subject of quite an intensive research effort ever since Albright tried to demonstrate the intimate knowledge this scribe had of Egyptian literature and language (Albright 1937). While Albright suggested that the Tyrian scribe was Egyptian, the existence of Canaanite glosses and other linguistic features in his writing may also suggest that he was a Canaanite who had acquired his knowledge of Egyptian language and culture during his education (cf. Finkel 1977, 114-6). In any case, the Canaanite glosses attested in the Tyre letters may be taken as evidence of the Canaanite dialect of Tyre. In the Amarna letters from Tyre there are eight unambiguous words that can be regarded as purely Canaanite. Seven of them are marked as glosses. Among these words are four verbs and four nouns. Let us look at the verbal forms first. (12) ,sa id-din ri-ig-ma-su i-na sa10-me 14 ki-ma baP/i(dISKUR) ta[r]-gu5-ub gab-bi 15 mati(KUR-ti) a-tu ri-ig-mi-su Who cries in the sky like Baal and the entire land fears his cry. (EA 147: 13-15)

(13) •_ ku-na 37 a-na pa-ni sdbi(ERINMES) rabi(GAL) "Prepare for a big army ." (EA 147: 36-37) (14) it as-sum hapsi(ZAG) .sarri(LUGAL) 55 be-li-ia da-na-ti 56 \ nu-uh-ti \ ba-ti-i-ti And due to the strong arm of the king, my lord, \ I am relaxed, \ I am confident. (EA 147: 54-56) None of the verbal forms correspond to cognate forms attested from texts

74 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

originating further south in Canaan, be it contemporary or later. Some have

northern parallels. The root of ta[r]-gu5-ub/targub/"feared" is attested only in

Ugaritic (and ; De Moor 1969, 188; Moran 1992, 234 n4). The ("long")

imperative ku-na/kuna/"prepare" (intransitive) has a parallel only in nifal in

Hebrew, viz., hikkon (hikkon likrat elohaeka "prepare for your god", Amos 4:

12; Moran 1992, 234 n8). nu-uh-ti/nuhti/"I am relaxed" is paralleled in

Hebrew by nahti. ba-ti-i-ti/batiliti/"I am confident" is, again, paralleled by

Hebrew batahti (pausal batahti). Byblos may attest a counterpart of this form,

ba?•v-•uti4?•v-[t]a?/batihta/ (Huehnergard 1996, 101 11. 20-21; commentary •uon p.

109) "you are confident".

Some of these forms may have existed also in contemporary Canaanite

dialects, but are not attested. Some forms may reflect earlier stages of forms that

may well have existed in Proto-Hebrew or even in Hebrew or other Canaanite

dialects of the biblical period, yet, again, have remained in obscurity due to the

accident of recording. Still, ,•ãrgb may have been lost in the Southern Canaanite

dialect, and especially in Hebrew, where its possible semantic parallel •ãyr is

most common. The stem difference between qal and nifal forms of •ãkwn may be

significant as a dialectal marker as wel1.12 The mediae verb nuhti is

paralleled by the Arabic corresponding inflection (Wright 1896, •˜ 155). Another mediae vocalis verb (mediae yodh, in this case) attested in a letter found at

Kamid el-Loz, located further inland and to the north of Tyre, has an a vowel

even in closed syllables: -al-ta/ halta /,ha-al-ti/ halti /, with the possible

meaning "you/I fear" (Arnaud 1991, 10, 11. 8, 12); cf. BH halti (Isaiah 23. 4;

Baumgartner 1967, 298). Two other mediae vocalis verbs that should be

mentioned here are si-ir-ti / sirti / "I am besieged" (EA 127: 34, Byblos) and si-i r-ti /sirti/"I am denounced"(EA 252:14, Shechem;see below,#30). These

are passive qal forms, an inflection which we have already met above (#8). As

for nuhti, while it may well reflect dialect variation in 14th century Canaan, it is

possible to restore a similar form for Proto-Hebrew or Proto-Phoenician (cf. Friedrich-Rollig 1970, 78). Still, pausal forms like nahti (Job 3: 26) may perhaps

suggest a different historical origin (Bauer-Leander 1922, •˜56v).

The fourth attested verb, batihti, may have well been widespread

throughout the entire Canaanite continuum and beyond, as the change from i to

a in the second syllable may have been the result of a later change (Blau 1981;

Blau 1986; Rendsburg 1997, §5.7.1). Still, the existence of BH pausal forms

with a games: (e.g., batahti, Psalms 31: 7, 143: 8, and especially bataha "she has

not trusted", Zephaniah 3: 2, with a games: in an open syllable) may attest to an

early •Ea•Ea• pattern.

Vol. XXXVIII 2003 75 The nominal lexemes are the following:

(15) u-na muhhi(UGU-hi) muhhi(UGU-hi) •_ su-r[i-]ia 40 u-bal a-ma-ta5 sarri(LUGAL) be-li-ia On my belly, on my back I carry the words of the king, my lord. (EA 147: 39-40)

(16) u it li-id-din URUu-suKI 12 a-na ardisu(IR-su) DUG •_q a-ku-ni•_ me-ma 13 a-na si-ti7-su And may he give Usu to his servant to drink a jug of water. (EA 148: 11-13; also 1. 31, spelled me-e-ma; further EA 146: 20; EA 150: 21; EA 155: 10)

(17) u id-d[in] 56 [pa-n]i-su a-na •_ u-bu- sar[ri](LUG[AL]) [be-li-ia] And he has given attention to serve the king, my lord. (EA 152: 55-56; also fi-bu-di, EA 151: 20)

(18) a-ma-ta5,.sarri(LUGAL) •_ pa-ni-mu [i-]la-ak By? the order of the king I will go before him. (EA 155: 46)

We have already dealt above with the possibility that ka-bi-di-ia / kabidiyal "my belly/front" is a Canaanite lexeme and mentioned that sa -r[i-]ia / Suhriya / "my back" is one . The lexeme Suhru is attested from various sites in Canaan, notably Gath (EA 64: 7; EA 65: 5; EA 281: 7; EA 282: 7; EA 284: 5, also EA 232: 11 from Acre and EA 306: 11 from an unknown provenance in southern(?) Palestine). Thus, there is no attested variation for the Canaanite lexeme for "back" , at least in a context like the one in which this lexeme occurs. It is to be noted, however, that we lack data from many contemporary Canaanite sites, including Jerusalem, as no similar prostration formula is attested in texts from them. The Akkadian cognate of this lexeme is seru, attested also in similar prostration formulae in Canaano-Akkadian, two examples of which have already been cited above (#3, #5; see further Ebeling in Knudtzon 1915, II, 1514; Rainey 1978a, 91). Since laryngeal and pharyngeal consonants were lost in Akkadian, the attested form conceals its history in this respect.13 Interestingly, Ugaritic, a language much closer to Canaanite, attests to a cognate lexeme without h: zr (Del Olmo Lete and Sanmartin 2003, II, 1005). Other Semitic languages do attest to cognates with h (see Baumgartner 1983, 946). As for the vowel, Semitic languages attest to forms with either u or a (loc. cit.). As regards Hebrew, both sohar "skylight" (Genesis 6: 16) and saharayim "noon" have been suggested as cognates (for both see Baumgartner 1983, 946), though considerable semantic changes must be assumed in both cases (BOhl [1909, 83]

76 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC has relied on the Amarna occurrences to render the Genesis occurrence as "Dach") .

The lexeme for "water", me(-e)-ma •^ memal•^ , shows a long e, recalling the

extended or contracted diphthong in the Hebrew independent (mayim) and

construct (me-) forms respectively (Baumgartner 1974, 546). The final vowel a

may well be reconstructed also for Proto-Hebrew within the pl m morpheme (cf.

Brockelmann 1908, 453; Bauer-Leander 1922, 514g; Blau 1976, 31; for Ugaritic

see Huehnergard 1987a, 296-7). While Tyre thus attests to a form that can be similarly reconstructed for BH, an early variant nevertheless exists in the cuneiform data. In Aphek, a small fragment of a vocabulary has been discovered. The vocabulary has Sumerian and Akkadian glosses, and these are followed by what seem to be West-Semitic ones. One of the lexemes listed in that vocabulary is the word for "water":

(19) [A]MES •^ma-wu •^ mu-mi We have here a roughly contemporary variant to Canaanite mewl. Still, one should note that although the fragment was discovered in Aphek, its provenance may have been remote from that site. There are several ways in which this string can be rendered, and space does not permit us to delve into this discussion here (for suggested, although by no means conclusive or comprehensive, explanations, see Rainey 1976, 138; Sivan 1984, 14 n4; Edzard 1985, 251). Still, I should note that since the form is attested in a vocabulary, its case is probably the one used in dictionary forms, viz., nominative. Therefore, the vocalic difference between e in the Tyre forms, all of which represent the oblique case, and a (or 6), cannot be compared and used as evidence for any geographic variation.

The forms u-bu-ud •^ubud•^ and u-bu-di •^ubudi•^ are interpreted as qal

infinitive (construct) forms of •ãbd "serve" (Grave 1982, 166 30; Moran 1992,

239 n1). Both seem to occur in a similar context: they are preceded by a preposition (ana "to, towards") and followed by a nomen regens (sarri "king"). If so, there is variation in the ending of the form within the Tyre letters

themselves. This may need further research (for work done on case marking in the Amarna letters see especially Kossmann 1987-8; Rainey 1996, I, 172-7).

The last cited form, pa-ni-mu •^panimul•^ , is interesting in several ways. In prepositional use its cognates are usually preceded by another preposition (lipne "before" etc .; Blau 1976, §51). It is also rarely attested independently, in a prepositional usage with the meaning "before" in Phoenician and in Ugaritic

(DNSWI II, 919; Del Olmo Lete and Sanmartin 2003, II, 675), and perhaps also in BH (BDB 816a s.v. pnh 6). The ending is usually interpreted as a shortened

Vol. XXXVIII 2003 77 form of the pl m pronominal suffix *-humu (Friedrich-Rollig 1970, •˜235; Sivan

1984, 133). However, the context hardly allows for a plural interpretation

(Moran 1992, 241; cf. Albright 1937, 197 n2). Now, there are rare forms in BH

of the pl m pronominal suffix with final -o, i.e., -amo or -mo (which may well

reflect an earlier form of the same pronoun; cf. Brockelmann 1908, 256 •˜94 n2).

One of these forms is panemo "his face" (Psalms 11: 7). Interestingly, this latter

occurrence also refers to a singular, as do some other similar occurrences of this

rare form with certain prepositions (Jouon-Muraoka 1993, •˜94i, •˜103f,m).

Recalling that later Punic attests to a sg m suffix pronominal with m (Friedrich-

Rollig 1970, 112; Krahmalkov 2001, 54-5), these occurrences, both in Tyre and

in BH, may require further research, although it is possible that the co-

occurrence is accidental. There are two other occurrences of a similar pronominal suffix in the Canaano-Akkadian corpus: (20) nu-pu-ul-mi 26 ta-ah-ta-mu "Fall beneath them ." (EA 252: 25-27, Shechem)

(21) u mi-im-ma •_ ma-ah-si-ra-mu and all they need. (EA 287: 16, Jerusalem) In both cases the forms are for the 3 pl m and can be compared to standard BH -am and especially to its rarer variant -amo (Jouon-Muraoka 1993, 288-9), as well as to Phoenician -om (Krahmalkov 2001, 54-5). Therefore, as regards form, variation is not attested.

Jerusalem Let us leave Tyre now and travel south to Jerusalem. The Canaanite lexemes of the Jerusalem letters are attested as follows:14

(22) i-ka-lu ka-ar-si-ya •_ u-sa-a-ru 7 i-na pa-ni sarri(LUGAL) beliya(EN-ri(sic)) They denounce me before the king, my lord. (EA 286: 6-7) (23) a-mur mat(KUR) [UR]U[g]az-riKI mat(KUR) URUas-ka-lu-naKI 15 uRU U l[a-ki-s]iKI i-din-nu a-na sa-su-nu 16 akala(NINDAHA)

s amna(IHA) u mi-im-ma •_ ma-ah-si-ra-mu Look, the land of Gezer, the land of Ashkelon and Lachish gave them food, oil and whatever they need. (EA 287: 14-16) (24) [a-]mur mat(KUR) URUu-ru-sa10-lim an-n[i-]ta 26 [l]a-a abiya(LUAD.DA.A.NI) la-a -mi-i[a] 27 [n]a-ad-na-an-ni •_

U •_ zu-ru-uh[ sarri(LUGAL-ri) dan]nu([KALA]G.GA) S 28 [n]a-ad-na-an-ni a-na ia-a-si

78 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

Look, this land of Jerusalem, neither my father nor my mother gave it to me. The strong arm of the king gave it to me. (EA 287: 25-28; also EA 288: 34) (25) a-mur sarru(LUGAL) beliya(EN-ia) sa-du-uk a-na ia-a-si Look! O king, my lord! I have a just case. (EA 287: 32)

(26) [la]-ka-hu u-nu-ta5-su-nu u as-•-u•v 37 [e•v-mid u-re-e •_ ga-ag- gi-m[i] They took their tools, and I had to take shelter by a support from the roof (EA 287: 36-37) (27) mu-se-er-ti a-na sarri(LUGAL-ri) be[liya](E[N-ia]) 54 [x LU].MEa-si-ru 5 li-im[ ] 55 [3 M]E[ 1]8 LU.MESu-bi-li-mi harrani(KASKALHA) sar[ri](LUGA[L-ri]) 56 la-ki[-h]u i-n[a]

•uu -ga-ri•v •_ sa-de4-e [ ] 57 URU ia-lu-naKI I sent as gifts to the king, my lord, x prisoners, 5000 ..., and 318 porters for the caravans of the king, (but) they have been taken in the countryside (of?) Yakina. (EA 287: 53-57) (28) a-mur a-na-ku LUru-i sarri(LUGAL-ri) 12 u u-bi-il GUN s arri(LUGAL-ri) a-na-ku Look, I am a friend(?) of the king; and a tribute-bearer of the king, I am. (EA 288: 11-12) (29) If there are no troops this year,

h al-ka-at a-ba-da-at 53 •_ gab-bi mat(KURHA) sarri(LUGAL-ri) beliya(EN-ia) all the land of the king, my lord, will be lost. (EA 288: 52-53) We have already noted above the qal passive in Canaano-Akkadian, and discussed the Jerusalem form la-ki[-h]u /lakihu/ "they have been taken" (#8 = #27), as well as the form si-ir-ti /sirti/ "I am denounced", which is attested from Shechem as follows:

(30) ka-bi 14 ka-ar-si-ia •_ si-ir-ti 15 i-na pa-ni sarri(ILUGAL)-ma be-li-ia I am denounced before the king, my lord. (EA 252: 13-15) Both /lakihu/ and /sirti/ are suffix conjugation forms. In #22 we have a prefix-conjugation form of the latter verb, u-sa-a-ru /usaru/. While the ¥a¥i pattern of the suffix conjugation is unknown to us from BH, the yu¥¥a¥ pattern of the prefix conjugation, which conforms to the Amarna evidence, is attested sporadically; e.g., yukkah "it will be taken" (Jouon-Muraoka 1993, 166-7). While the origin of the ¥u¥a formation for the suffix-conjugation passive of qal in the remains of this pattern in BH is unknown, one may perhaps speculate that it is a result of paradigm analogy, to conform with both the prefix conjugation of

Vol. XXXVIII 2003 79 qal as with both the suffix- and the prefix conjugation of pual and hufal. The

attested •Ea•Ei•E pattern reminds us of Aramaic •E•Ei•E (Bauer-Leander 1927, 93;

Leander 1928, 44). I also mentioned earlier that the passive-active contrast is possible in the Canaanite dialect of Jerusalem in the 14th century BC because short vowels in open syllables did not elide. Other forms that attest to this phonological feature beside the already mentioned lakihu and lakahu - are abadat (#29;- cf. Canaano-Akkadian halkat which it translates) and - with a different set of syllables - also obilimi (#27). Forms that attest to an analogical syllable structure occur also in other Canaano-Akkadian texts, e.g., sa-pa-tu-ni apatuni/ "they judged me" in a text discovered at Hazor /Swhich possibly originates from Enishasi in the Bashan area (Goren 2000, 37). As we shall see below, this is not the only possible syllable structure attested in the Canaanite dialect continuum of the period. Both these forms conform to a historical stage previous to the forms attested in BH, where we have 'add (e.g., Exodus 10: 7; pausal abada, e.g., Proverbs 11: 7) and *spatuni (cf. spatunu "they ruled (lit. judged) us", Daniel 9: 12) respectively.15 We have already discussed the pronominal suffix of the form ma-ah-si-ra-mu mahsiramu/, which does conform to later evidence. As for the /noun itself, it seems to represent a cognate to BH mahsor "lack, want" (Baumgartner 1974, 501), which fits the context well. If so, the Amarna occurrence represents a different pattern than the Hebrew one. Some scholars, precisely because of this difference, suggested the rendering of this gloss as a cognate to Hebrew maaser "tenth" (Baumgartner 1974 , 583-4). However, recalling that 14th century Canaanite is not a direct ancestor of BH, this forced interpretation is uncalled- for. The spelling of the gloss zu-ru-uh /Zoro/ represents a case of vowel harmony, either as a result of the assimilation of the original vowel i (cf. Arabic dira; Hava 1970, 227) or of an epenthetic vowel (following the deletion of that vowel; cf. Hebrew aezroa, Baumgartner 1967, 269). The difference between the BH forms and 14th century Jerusalem Canaanite Zoro is, thus, either phonetic or diachronic. Still, one notes the difference in gender, which is masculine here and feminine in BH (rarely masculine; loc. cit.). Both occurrences of this gloss have a zero ending, and it seems that nouns in the construct state did not have case endings, at least in the nominative. One other form that concurs with this conclusion is obil (#28; for obilimi see below). As I mentioned above in the discussion of the forms ubud and ubudi in Tyre (#17), the question of case endings in the construct state is yet to be investigated

80 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

thoroughly. In this case, one may notice a -0 ending in the singular non-genitive noun.

Both the syntactic structure of the sentence sa-du-uk a-na ia-a-si and the

form sa-du-uk are open to various interpretations (Moran 1992, 329 n8).

Although it seems clear that the form sa-du-uk is the predicate, its subject could

be the king, or it could be impersonal, as in the interpretation offered above. As

for the form itself, it can be interpreted as an infinitive absolute /sadok/, as a

passive participle /saduk/, as a suffix conjugation of a •Ea•Eu•E pattern /saduk/

(Rainey 1996, II, 306; the absence of a final -a may be questioned, cf. Rainey 1996, II, 287-8), or as an infinitive construct /saduk/.16 Our inability to make a

clear disambiguation of this form and context is a direct outcome of the lack of

an anchor to such a structure in BH or in the other NWS languages, and

therefore any attempt to draw further conclusions regarding history or

contemporary variation will be futile. The gloss ga-ag-gi-m[i] /gaggimi/ is related to BH gag and to Ugaritic gg (Baumgartner 1967, 169). It further demonstrates the original doubling of the second g (cf. BH gaggot), a case ending, and the enclitic particle -mi. The enclitic -mi also seems to have been used throughout the entire Canaanite- speaking continuum (Rainey 1996, III, 234-48; Izre'el 1998a, §2.5.5.5, §3.8.1).

The forms u-bi-li-mi /obilimi/ (#27) and u-bi-il /obil/ (#28) are interpreted

as representing qal participles of •ãy/wbl (Bohl 1909, 25), conforming in

meaning to BH hifil forms. The transcriptions given to the two forms bear the

underlying assumption that the initial /y/ had been dropped (Sivan 1984, 56),

since initial y is usually represented in script in the Amarna letters. If so, these

two forms are markedly different from the potential BH cognate forms not only

in morphology, but also in phonology. The form is a pl m construct

state form and can be rendered either with a long i or with a long e following its

stem. The spelling is incapable of determining the quality of the vowel in this

case (Izre'el 1998a, 8-9, 16-7). The enclitic -mi, already accounted for above, is

found here attached to a noun in the construct state, a feature attested from

various sites in the Canaanite area (Rainey 1996, III, 236; see another example

in #32 below), as is the case with the enclitic -m in Ugaritic (Tropper 2000, 827-

8) and in some remains in BH (Waltke-O'Connor 1990, §9.8). The last Jerusalem form to be discussed here is the famous sa-de4-e ade/17 "field, countryside". This form represents a contracted /sfinal triphthong conforming well to that found in BH (Birkeland 1940, 43-4; Friedrich-Rollig 1970, 24). One difference between the Amarna form and its BH cognate sadae is the length of the final e (the precise quality of which we are unable to tell). As

Vol. XXXVIII 2003 81 for Hebrew, (Tiberian) Massoretic Hebrew does not make a quantitative distinction between vowels (Khan 1997), so that between the two systems must rely on historical reconstruction of the latter (cf. Rendsburg 1997). As for other second millennium BC NWS languages, the data regarding (contracted) triphthongs is rather scanty. Still, Ugaritic is known to have preserved case distinction in nouns with final contracted triphthongs (Huehnergard 1987a, 288-92). A letter by Ribhaddi of Byblos, sent from Beirut, attests one form that suggests a similar monophthongization:

(31) u [i]a-pu •_ ha-mu-du 127 sa-a sa-pi-ir i[s]-tu 128 sarri(LUGAL) be-li-

And it is beautiful •_ nice what has been sent on behalf of the

king, my lord. (EA 138: 128) Here [i]a-pu stands for /yapa/ ( 30 i-li su-su-mi a-bi-ia 31 u u-sur-ru-su-nu I shall protect the enemies. The men who captured the city and my god are the despoilers of my father, but I will protect them. (EA 252: 28-31; Shechem) For su-su-mi /Sosumi/ cf. BH sosim (Baumgartner 1990, 1485). We shall see later a different type of triphthong attested in the Megiddo texts (#60).

Gath

From Jerusalem let us move further south to Gath.

(33) a-nu-ma 21 10 sinnisati(MUNUSMES) 22 •_ me-ki-tu 23 •_ ia-pa-ak-ti I send herewith 10 women (and) eye paint. (EA 64: 20-23) (34) yu-us-si-ra sarri(ILUGAL-ri) 11 beliya(EN-ia) sabi(ERINMES)

pi-ta-ti 12 ma-a-da ma-gal 13 u yi-ki-im-ni 14 •_ ia-si-ni May the king, my lord, send very many troops and may he take

me away \ take me out. (EA 282: 10-14)

(35) [l]i-[i15-ma-ad sarr]i(LUGAL-ri) beliya(EN-ia) 8 ki-ma G[AZ?ME]S •_ mi-hi-sa 9 P[N1 u] PN2 May the king, my lord, be informed that PN1 and PN2 have been killed. (EA 335: 7-9)

(36) apiru(LUSA.GAZ) s•ua•v13 yi-na-as-si •_ na-as-sa-a 14 i-na mati

(KURKI.HA) na-da-an 15 ilu(DINGIR-lu4) sa sarri(LUGAL-ri)

beliya(EN-ia)•ua•v-na i•ua•v-si

82 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC The Apiru who had become elevated in the lands, the god of the king, my lord, gave (him) over to me. (EA 366: 12-15) (37) And Surata, Man of Acre, 23 and Indaruta, 24 Man of Achshaf, 25 su- ni-ma in4-ne-ri-ru \ na-az-a-ku 26 i-na 50 narkabati

(GISGIGIRHA) 27 a-na mu-hi-ia it was they two who rushed to help me with 50 chariots. (EA 366: 22-27)

(38) u lu-u tu-te-er 34 matu(「K」URKI.HA) sa sarri(LUGAL-ri) beliya(EN-ia) 35 a-na ZAG!H-si\up-si-hi ... and the land of the king, my lord, be restored to its borders. (EA 366: 33-35) The context in which the verb ia-pa-ak-ti occurs is interpreted here as

containing two different dispatches: women and eye paint (cf . Moran 1992, 135-

6 and n2 with further bibliography). The root of the verb is clearly •ãypk, being a

cognate to Aramaic •ãnpk (e.g., DNWSI, 741-3; Sokoloff 1990, 356-7) and BH

pwk (Baumgartner 1983, 869; for Phoenician see DNWSI, 903 s.v. pwq1). •ã The correspondence between Aramaic npk and Gath-Canaanite •ãypk

reminds us, interestingly, of the known Phoenician-Hebrew pair •ãytn : •ãntn

(Harris 1939, 37). While most occurrences of the verb "give" in the Amarna correspondence are Canaano-Akkadian, using the Akkadian root •ãndn (Ebeling

in Knudtzon 1915, II, 1476-9; Rainey 1978a, 83), the Byblos letters attests one

occurrence of the local form ia-ti-na "let him give" (EA 83: 31; Sivan 1984 ,

156). The corresponding •ãntn is curiously attested with a non-assimilated n (cf .

below, #42) in a letter sent to Byblos by a high-ranking military official:(18)

(39)la-a-mi an-ti-in4-nu 8 e-re-eb 9 awili(LUM[E]S)sa URUsu-mu-riKI 10[a -n]a aliya(URUKI-ia) I do not let people from Sumur enter into my city. (EA 96: 7-10)

The form ia-si-ni seems to require a transitive meaning, and therefbre most

scholars interpreted the gloss as a Canaanite hifil form (Ebeling 1910 , 64; Rainey 1996, II, 192). However, hifil would most probably be spelled with a PI

sign to designate /yo/ (="yu"; AS #223; Jucquois 1966, •˜28A),19 and the value

iu/ for the cuneiform sign IA is virtually impossible for Amarna (AS #104; /

Jucquois 1966, 66 #142), Transitive qal forms with a pronominal suffix are , however, attested in both Amarna and BH. The Amarna occurrence, in Canaano-

Akkadian, can hardly be interpreted as a Canaanite hifil (cf. Rainey 1996 , II, 192-3). It goes as follows:

(40) us-si-ir-mi eleppa(GISMA) a-na 29 mat(KUR) ia-ri-mu-ta u u-sa- ka 30 kaspi(KU.BABBARMES) lu-bu-si is-tu sa-su-nu

Vol. XXXVIII 2003 83 Send a boat to the land of Yarimuta and I will send you silver (and) clothing from them. (EA 82: 28-30, Byblos) The BH example, although not transitive, shows a pronominal suffix attached to a qal verb: (41) banay ysauni wenam My children left me and they are gone. (Jeremiah 10: 20)

While one may further pursue the question of transitivity in the Amarna

Gath form, I believe the form must be interpreted as qal, and hence it is a cognate to BH yese. The latter's historical relationship to the form yasi remains to be explored. A similar pattern in the prefix conjugation of primae yodh or primae waw verbs is attested elsewhere in the Canaanite continuum, notably in forms of •ãytn "give" (see above).

The verb mi-hi-sa /mihiSa/ "they (dual) have been killed" presents the change of a•¨i _h(i)S in a rather specific phonological environment (Izre'el 1998a, 11). Similar forms that exhibit i (or e) in the first syllable are further attested in EA 220: 24 (unknown provenance), EA 264: 8, 12 (Gath Carmel), EA

273: 23 (central Palestine) and EA 313: 4, 19 (unknown provenance). As regards the dual as a morphological marker, the data is too scanty for any further comment (cf. Izre'el 1998a, 26). A comprehensive research of Canaano-

Akkadian dual forms may come up with some insights regarding the substratal dialects. The forms na-as-sa-a /nassaa/ and na-az-a-ku /nazaku/ represent suffix conjugation forms of the nifal stem, which conform to the historical reconstruction of the cognate BH patterning before attenuation took place (nissa, e.g., Daniel 11: 12; nizaku, Judges 18: 22). The first form may further attest to a final -a morpheme of the 3 sg m (see above in the comments to #11). The last form, up-si-hi, is tentatively rendered as /upsihi/, with reference to physical boundaries as compared with Ugaritic ps (van Soldt 1997). The final syllabic sign, HI, has been interpreted as a 3 sg f pronominal suffix, which contradicts the form found in Semitic in general, where the 3 sg f genitive suffix is s/ha(:)(Lipinski 1997, 306-7; cf. Rainey 1996, I, 78).

Shechem We have already discussed some forms from the Shechem letters above: the nouns ibu "enemy" (#9) and /Sosu/ "despoilers" (#32), the passive qal verb "I am denounced" (#30) and the prepositional phrase /t ahtamu/ "beneath/sirti/ them" (#20). EA 252 from Shechem includes many Canaanisms. It cites a local saying that has become rather famous in early Amarna studies:

84 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

(42) ki-i na-am-lu 17 tu-um-ha-su la-a 18 ti-ka-bi-lu u ta-an-su-ku 19 ka-ti awili(LU-li) sa yi-ma-ha-as-si When an ant is stricken, wouldn't it fight back and bite the hands of the man who strikes it? (EA 252: 16-19, Shechem)

The Akkadian dictionaries list a cognate to namlu in the form of namalu

(CAD) or namalu (AHw, CDA) from Akkadian sources (CAD N1, 208; AHw,

725; CDA, 235). A close scrutiny of the data given, however, leads us to

conclude that the Akkadian data is doubtful: there is a single occurrence in a

medical text in Standard Babylonian of which the interpretation is uncertain

(thus CAD s.v. namalu) and there are a few proper names that include this form, which, of course, need not mean "ant". While words from •ãnml are attested in

some Semitic languages (cf. Baumgartner 1983, 662),20 the common word for "ant" in Akkadian is kulbabu (CAD K , 501-2; AHw, 501). Therefore, what is

listed in the Akkadian dictionaries under these respective entries should not be

taken as supporting evidence against the conclusion that the form namlu in our

text, which in any case differs in pattern from the Akkadian word, is not

Akkadian, and hence to be regarded Canaanite. Hebrew attests the cognate

(

distinction suggests that the Shechem form may also be a collective noun, or, perhaps better, a name of species (cf. Jouon-Muraoka 1993, § 135; for the 3 sg f agreement see Rainey 1996, I, 132-5). Since the Shechem occurrence is the only attestation of this word from 14th century Canaan, it is impossible to suggest whether Canaanite dialects had the form namal(a)tu for the individual, neither can it serve to draw any synchronic variation. The verb ta-an-su-ku /tanSuku/ seems to be strictly Canaanite, given the fact that cognate roots are attested in Hebrew and Ugaritic (Baumgartner 1983, 688-9; Del Olmo Lete and Sanmartin 2003, II, 653), and noting the Canaanite personal prefix ta- and indicative marker -u. However, the root is common to both Akkadian and Hebrew, as well as is the inflectional pattern yaqtul/iprus (AHw, 758; Baumgartner 1983, 688; Hebrew also attests yissak). Since Canaano- Akkadian shares the personal prefixes and the Tense-Mood-Aspect markers with Canaanite rather than Akkadian (for a concise exposition see Izre'el 1998a, §§2.5.2, 2.5.5.2), forms where root and pattern are shared by Akkadian and Hebrew would be identical in Canaanite and Canaano-Akkadian. This specific word, however, shows yet another feature which is attested in neither Akkadian nor Hebrew for this verb, viz., the non-assimilation of n. How do we account for this feature?

Vol. XXXVIII 2003 85 Prima facie, one would tend to conclude that since this feature is unattested in Akkadian, it must be the result of substratal influence. It is only one more step to the logical conclusion that the Shechem dialect of Canaanite did not assimilate n in this environment. However, there is one other possibility which might explain the attestation of a non-assimilated n here, viz., that it is a morphophonemic spelling, preserving the n-radical of the Akkadian root and thus noting the contrast between the Canaanite mother tongue of the scribe and the language he was writing in, which he regarded as Akkadian. Still, the comparative evidence lends support to first assumption, i.e., that the attested vocable reflects actual pronunciation. First, non-assimilation of n to the following consonant is attested in Canaano-Akkadian forms elsewhere also

(Izre'el 1998a, 9-10), including, perhaps significantly, in some occurrences of place names (Harris 1936, 28 n9). BH has some primae-nun verbal forms of the a•Eu• pattern where the first radical has been preserved, mostly in pause (Bauer-

Leander 1922, § 151; Jouon-Muraoka 1993, §72). In Ugaritic, which is, grosso modo, contemporary to the Amarna evidence, forms with non-assimilated n are also attested (Tropper 2000, 144-7). Further support comes from Aramaic, where both etymological and non-etymological n is reflected in writing in some dialects and forms (Spitaler 1952-54; Kutscher 1970, 374-5).21 In Phoenician and Punic, the data is too scanty. Still, Krahmalkov (2001, 26) explains the non- assimilated forms as reflecting a dialectal feature. While further research is still needed, at this stage I would tend to view the Shechem form as yet another piece of evidence that there existed a variant to forms with assimilated n, although not necessarily a geographical variant but perhaps a structural one. We have already seen one such form from a northern(?) site in #39 above. One other prima nun form in the same text is the imperative nupul "fall": (43) nu-pu-ul-mi 26 ta-ah-ta-mu u27 ti-ma-ha-su-ka "Fall beneath them so that they strike you ." (EA 252: 25-27, Shechem)

BH does not attest a singular imperative form of •ãnpl, but the plural does include all three root consonants: niplu (Hosea 10: 8, Jeremiah 25: 27). Besides the sporadic attestations in BH of non-assimilated n in the prefix conjugation of primae nun verbs with the u pattern mentioned above, imperative forms from the same class (and in some cases also in other patterns) do tend to preserve the n more generally (Gesenius-Kautsch-Cowley 1910, §66c; Jouon-Muraoka 1993,

§72 c (2)). The timbre -—and the very existence - of the first vowel in the

Shechem form cannot be regarded as a local variant to the attested BH cognates, because the history of the related BH forms may be similar.

86 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

One other interesting form is the interrogative pronoun "who", which will be discussed further below (p. 90).

Other sites Before we move to what I would regard as the most interesting locale from our point of departure here, Megiddo, I would like to comment on some Canaanite words attested from other places in central and southern Canaan. I will not list all attested Canaanite words, but only those that have some significance to our discussion on variation.

(44) at-tu-nu tu-sa-ab-li-tu-na-nu 32 u at-tu-nu 33\ti-mi-tu-na-nu It is you who make us live and it is you who put us to death. (EA 238: 31-33, unknown provenance) The gloss ti-mi-tu-na-nu /timitunanu/ is one of the few Canaanite forms which are interpreted as hifil verbs. We have already discussed the suffix conjugation hifil form *hahbe (#11), as well as the two alleged Canaanite hifil forms ia-si-ni (#34) and u-sa-ka (#40), which are better explained otherwise.

One other is the gloss usually transliterated) yv-ki-il-li-ni "he despised me" (EA 245: 38, #62 below). The first vowel of the form is of intriguing interest here. In fact, only timitananu is instructive as regards the quality of the vowel, since the cuneiform sign which stands for y-syllables is polyphonic, and can represent y+any vowel (AS #223).22 The quality of the first vowel of either the piel or the hifil pattern in 14th century Canaanite is impossible to determine. Canaano-Akkadian over- whelmingly exhibits a u in this position (Rainey 1996, 136-7), and it has even been suggested that for Proto-Hebrew even i cannot be excluded (Steiner 1980). While this issue still needs further research, the case in point here as regards the form ti-mi-tu-na-nu is that the i vowel is epenthetic, whether pronounced as i or otherwise. While u must be surmised as the Proto-Semitic vowel and a may reflect, in the case of hifil, a contraction of the string /uha/ (yuha¥¥i¥•„ya¥¥i¥), there is no track one can posit for an original i (cf. also Steiner 1980, 515). The data and the structure of Canaano-Akkadian cannot give us a reliable route to reach any solid conclusions regarding the initial vowel of the piel or hifil stems in the Canaanite substrata without further research.

(45) u ti-na-mu-su SIG4 18)\la-bi-tu is-tu 19 [s]u-pal tap-pa-ti-si 20 u a-na-ku la-a i-na-mu-su 21 is-tu su-pal sepe(GIRMES)22 sarri(LUGAL-ri)be-li-ia A brick may move beneath its companion, but I will never move from beneath the feet of the king, my lord. (EA 296: 16- 22, central Palestine; also EA 266: 19-25, Gath Carmel)

Vol. XXXVIII 2003 87 The gloss la-bi-tu /labittu/ "brick" represents a form with -t for this noun, as against the BH datum, lbena (Baumgartner 1974, 492-323), which represents an original -at morph (•ƒlabinatu). Unfortunately, we have no parallel datum from the Canaanite dialect of Jerusalem.

(46) su-si-ir-mi g IGI.KARMES ma-[a]t-ni-a 10 a[-n]a pa-ni sa bi(ERINMES) r[a]bi(G[A]LMES) 11 \ pi-[t]a-[t]i 12 [s]a sa[r]ri(LU[G]AL) be[liy]a(E[N-i]a) "Send supplies to the large army •_ archers of the king , my lord."

(EA 337: 8-12, unknown provenance; similarly 11. 19-22, with a

Glossenkeil) While BH attests to the noun taenae "basket" (Baumgartner 1974, 361), Phoenician attests also to a noun with a prefixed m-, mtn "offering" (DNWSI, 618, cf. also tn2, p. 426; Moran 1983). The data in both languages are too scanty to determine the semantics and distribution of the respective nouns. As regards its phonological form, the Amarna vocable suggests /matnia/. Besides these and other forms already discussed above, there are still other forms that present differences in either semantics or form from their respective BH cognates; e.g.,

(47) mu-ul-ka /mulka/ "royalty, authority"(EA 333:20, unknown provenance); cf. BH mluka "royalty, kingship", mamlaka "kingdom , reign", malkut "reign, kingdom", mamlkut (construct state) "kingdom, dominion, reign", as well as, for the pattern, molaek "Molech" (BDB, 574-5); Phoenician mlk "kingdom" (Tomback 1978, 181); Ugaritic mlk "kingship, sovereignty" (Del Olmo Lete and Sanmartin 2003, II, 555). (48) mil-ka /milka/ "king" (EA 333: 21, unknown provenance); cf. BH maelaek "king", malki "my king"; Ugaritic malku (Huehnergard 1987a, 147).

(49) su-lu-uk-ta /sul(l)uhta/ "shipment" (EA 265: 8, Gath Carmel); cf. BH misloah "sending"; cf. also silluhim "sending away, parting gift" (BDB, 1019-20). (50) ha-at-nu-ta5 /hatnata/ "marriage (agreement?)" (TT 2: 24 = Rainey 1999, 157*); cf. BH hatan "bridegroom, daughter's husband"; *hatunna "marriage, wedding" (BDB, 368). When I discussed the word for "water", I mentioned the gloss mu-mi attested in the Aphek fragment. This vocabulary fragment attests one other item which is of interest here:

(51) [GESTIN]MES •_ ka-ra-nu •_ ye-nu "wine"

88 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC This gloss, to be transcribed as /yenu /, attests to diphthong contraction, a marked feature of northern, or rather non-Judahite Hebrew (Harris 1939, 29-32; Garr 1985, 35-40). Of course, the commonly accepted transliteration of the PI sign with e is made on the basis of comparative evidence rather than on any support from the text itself. We have already seen an example of contracted diphthong in the Tyre dialect in the forms for "water" (mema). The Megiddo letters attest the gloss l[e-l]a / lela / "night" (EA 243: 13; Rainey 1996, I, 167; III, 1, 134).24 Byblos Canaanite attests the form ke-e-Si/keSi/ "summer fruit" (EA 131: 15), which also is a proof of the quality and length of the middle vowel. Both lexemes, yenu and kesu, are used frequently to illustrate other dialects of Hebrew, like the one reflected in the Gezer calendar (ks, KAI #182: 7; Garr 1985, 38), the one reflected in the Samaria ostraca (yn, KAI #185, 3; Garr loc. cit.), or the word-play in the book of Amos on kayis "summer fruit" and kes "end" (Amos 8: 1 -2; Rabin 1981, 122). As it stands now, contracted ay is the norm in the Canaanite continuum.25 Unfortunately, 14th century BC Jerusalem Canaanite lacks any data by which we can learn of the cognate words in that dialect.

As has been repeatedly mentioned above, the data have hitherto been too scanty to reach any solid conclusions regarding variation of individual features or forms in 14th century Canaan. Here is another example. Two texts, one from

Megiddo and one from an unknown provenance, attest to a a`•E•Ei.• pattern in the verb for "plough", viz., /ahrisu/ "I plough":

(52) a-nu-um-ma a-na-ku-ma 11 ir-ri-su•_ ah-ri-su 12 i-na URU s u-na-ma Now, it is I who plough in Shunam. (EA 365: 10-12, Megiddo)

(53) a-nu-ma11 i-ri-su•_ ah-ri[-su] 12u i-ba-ka-•umu?•v Now, I plough and pluck. (EA 226: 10-12, unknown provenance) While these are the only attestations of this verb in the Canaanite material, we can still point to the fact that the cognate verb in BH is attested with the pattern a•.•Eu.• rather than a•.•Ei•E, e.g., taharos "you shall (not) plough"

(Deuteronomy 22: 10). It may be assumed that the a• • i• pattern of the prefix- conjugation qal in Hebrew had been replaced by other patterns (or changed into hifil, Blau 1976, 50). Therefore, lack of data does not permit us to determine whether BH represents a later stage in the development of the Jerusalem dialect of Canaanite or preserves an older variant form. On the other hand, real, i.e., contemporary variation is represented by

Vol. XXXVIII 2003 89 forms of the interrogative pronoun "who", which is notably well attested (mostly in similar syntactic structures) by different scribes and from different locations: (54) mi-ia-mi awilu(LU) kalbu(UR.GI12) 17 u la-a yi-is-te-mu 18 a-wa-ti7 sarri(LUGAL) beliya(EN-ia) 19 mar(DUMU) samsi(dUTU) Who is the man, the dog, that would not listen to the speech of the king, my lord, the son of the Sun? (EA 324: 16-19, Ashkelon) (55) mi-ia-ti a-na-ku u la-a 13 u-wa-se-ru harrani(KASKAL-ra-niHA) 14 sarri(LUGAL -ri) beliya(EN-ia) Who am I that I will not send the caravans of the king, my lord? (EA 255: 12-14, Pihilu) The variant forms are /miya/, which in the area covered by this study is always attached to the enclitic -mi, and /miyedi/ 26 The variant miya(mi) is attested in letters from Byblos in the north to the southernmost sites, and can be regarded as virtually identical to Phoenician and Ugaritic my (DNWSI 619; Del Olmo Lete and Sanmartin 2003, II, 607) and as a virtual ancestor to Hebrew mi. The form miyati is attested in letters from Kamid el-Loz in the Lebanese valley, from the Bashan region in the north, and from Shechem and Pihilu in central Palestine. This form attests to an additional morph with t, which is known from other pronominal systems, yet is unique in the Hamito-Semitic continuum attached to the human interrogative particle (Brockelmann 1908, 326-7; Lipinski 1997, 328-9).

Megiddo The most prominent cluster of Canaanite words that point toward contemporary variation in 14th century Canaanite comes from the Megiddo letters. We have already discussed (#52) the verb ah-ri-su, which shows the a i pattern for the qal. The same letter attests also the following gloss:

(56) a-na-ku-ma •_yv-hu-du-un-ni 25 ub-ba-lu LUMESma-as-saMES 26 is-tu URUia_puK[I] I alone bring corvee-workers from Yapu. (EA 365: 24-26)

The unique form yv-hu-du-un-ni, although obviously derived from •ãyhd, is ambiguous as regards its underlying form (for suggested explanations see

Rainey 1978a, 73; 1996, I, 72; Sivan 1984, 286; DNWSI, 454 with further bibliography). (The noun ma-as-sa /massa/ I [also 11. 14, 23], probably West

Semitic as well, is attested already in the Old Babylonian Alalah texts [CAD Ml,

327; Moran 1961, 57].)

90 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC We have also mentioned the Megiddo gloss l[e-l]a I Lela I (p. 89 above), showing the contraction of the diphthong layl. Two other glosses are attested in the following example:

(57) u la-a-mi ni-le-u 14 ZU.SI.GA ba-ka-ni •_ ka-•usi•v-ra 15 u la-a-mi 16 a -Asi KA a-bu-u1-11\ sa-ah-ri 17 is-tu pa-niIla-ab-a-ya

And we cannot do the plucking, and we cannot go out of the

gate because of Lab'aya. (EA 244: 13-17) The scribe of this letter interestingly uses tripartite gloss sets.27 It reminds us of the trilingual vocabularies used in cuneiform schools, where each lexeme is represented by a Sumerian logogram, its Akkadian counterpart, and finally the local word, the latter two in syllabic writing. We have mentioned already the

Aphek fragment, which is a fragment of a vocabulary like the one just described

(Rainey 1976). Megiddo was a political and probably also a scribal center during the Amarna period.28 The employment of such tripartite gloss sets is understandable in the context of cuneiform scribal education, where students learned Akkadian by using Sumero-Akkadian lexical lists and expanding them to include local glosses (Artzi 1990). As for the Canaanite glosses, both have cognates in BH: ka-•usi•v-ra / kaSira / is to be compared with kasir "harvest"

(Baumgartner 1983, 1048-9); sa-ah-ri /Slagri / is to be compared with saar "gate" (Baumgartner 1990 , 1490-4).

All other Canaanite words occur in a single text:

(58) balpnumma(TIL.LA-nu-um-ma)•_ ha-ia-ma 7 nu-ub-ba-lu-us-su a-na sarri(LUGAL-ri) belinu(EN-nu)

Alive we shall bring him to the king, our lord. (EA 245: 6-7)

(59) u tu-sa-ah-mi •_ tu-ia 9 sisitiya(MIANSE.KUR.RA-ia) u iz-zi-iz-mi 10 EGIR-su •_ ah -ru-un-u 11 u ir-ka-ab-mi 12 it-ti Iya-as-da-ta 13

it a-di ka-sa-di-ia 14 u da-ku-su•_ ma-ah-su-u

And my mare was removed \ shot, so I stayed behind him, and I

rode with Yasdata, and before I arrived, they killed him. (EA

245: 8-14)

(60) i-na-mi libbi(SA) GISMA•_ a-na-yi29 u-ta-as-sa-ru-us-su 30 a-na s arri(LUGAL-ri) "In a boat I will send him to the king ." (EA 245: 28-30)

(61) u Isa-ra-ta 34 la-kj-mi kasap(KU.BABBARHA) ip-ti-it-ri-su 35 i-na katisu(SU-ti-su)•_ ba-di-u "And Surata has taken his ransom from his hand ." (EA 245: 33-

35)

(62) sa-ni-tam mi-na-am-mi ep-sa-ku-mi 37 a-na sarri(LUGAL-ri)

Vol. XXXVIII 2003 91 beliya(EN-ia) 38 i-nu-ma SIG-ia •_ yv-ki-il-li-ni 39 u DUGUD •_

yv-ka-bi-id 40 ahheya(SESHA-ia) se-eh-ru-ta5 Furthermore, what have I done to the king, my lord, that he despised me and honored my lesser brothers? (EA 245: 36-40) Out of the eight glosses in this letter, two (yv-ki-il-li-ni,yv-ka-bi-id) resemble what can be directly reconstructed as Proto-Hebrew in its standard form; two (ha-is-ma, ah-ru-un-u) have non-standard or partial counterparts in the Hebrew Bible; the other four (ma-ah-su-u, a-na-yi, ba-di-u and perhaps tu-ra) may be regarded as representing a different dialect than the one from which BH emerged. Furthermore, one of these glosses (ma-ah-su-u) represents variation also from the synchronic point of view. The verbal form yv-ki-il-li-ni / yvkillini / "he despised me" has its counterpart in BH *ykilleni(sg yakel, 1 Samuel 6: 6), which goes back to exactly the same form as the Megiddo one, including, notably, also the epenthetic vowel i between the verbal stem and the pronominal suffix (see Bauer-Leander, 337). As regards the first vowel, see the discussion above for #44. The form yv-ka-bi-id yvkabbid/ "he honored" has its counterpart in BH ykabbed (Baumgartner/ 1974, 434-5), which, again, goes back to exactly the same form as that represented by the Megiddo spelling (for the first vowel in both forms see the discussion above, p. 87). The adverbial phrase ha-ia-ma / hayyama / "(while he is) alive" is construed by the adjective hayy + adverbial accusative a + enclitic -ma (cf. the occurrence of the enclitic also within the adverbial phrases batnumma and Suhrumma above, #3). BH attests similar occurrences of the form hay (e.g., waet maelaek haay tapsu hay "The king of Ai was taken alive", Joshua 8: 23), and enclitic -m on adverbs is attested in rare retentions from an older period in Hebrew (Jouon- Muraoka 1993, 329-30). The prepositional phrase ah-ru-un-u /ahronhu/ reminds us of the Hebrew adjective aharon. While commonly connoting "last", this adjective is also used in the sense of "(the one) behind" (BDB, 30), e.g., (63) wayyasaem aet hasspahot waet yaldehaen risona waet le a wiladaeha aharonim waet rahel waet yosep aharonim And he placed the maids and their children first, and Leah and her children behind, and Rachel and Joseph behind. (Genesis 33: 2) BH does not attest, however, the form as a prepositional phrase, neither does this form ever occur with a pronominal suffix.29 Instead, Hebrew usually employs ahare- (BDB, 29-30; see also Sivan 1984, 133 n14).

92 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

The verb ma-alt-su-u / mahSuhu / "they killed him" represents a vowel

syncope (

which also goes back to *mahasuhu. In the BH form, the antepenultimate

syllable (i.e., the penultimate of the verb without the suffix) is lengthened and

the syllable before it is elided. The Megiddo form, in contrast, shows the

deletion of the vowel in the antepenultimate. This feature reminds us of a similar

vocalic reduction in Aramaic (Bauer-Leander 1927, 66). As for the meaning of the verb, the glossed Akkadian verb means "kill". In Canaano-Akkadian, as well

as occasionally in Akkadian, mahasu implies killing rather than striking or

smashing, which is the more commonly attested use of this verb in Akkadian

(CAD Ml, 71-77; for "kill" see p. 75, s.v. mahaasu lc). BH, which attests the cognate verb less frequently, also seems to prefer the meaning "strike, smash" for •ãmhS (Baumgartner 1974, 541).

The gloss a-na-yi /andyi/ has a cognate in BH aniyya (Baumgartner 1967,

69), which differs in pattern and in gender marking from the Megiddo form (cf.

Ugaritic any, UT •˜ 19.248; cf. Del Olmo Lete and Sanmartin 2003, I, 85-6). The collective ani "fleet" (Baumgartner, loc. cit.) goes back to *uny-, a form that would not include a triphthong. Both the triphthong preservation and the vowel syncope manifested by the form / mahSuhu/ require that we render the form a- na-yi with a long vowel in the penultimate. There is no basis for comparison with the contracted triphthongs we have seen in Jerusalem, Shechem and

Byblos, because this word differs in its morphological and phonological structure from the forms discussed. Non-contracted final triphthongs are nevertheless attested also in BH, as, e.g., in the by-form saday "field" (Barth

1894, 375-9).

The prepositional phrase ba-di-u / badihu / (

Friedrich-Rollig 1970, 23) recalls Phoenician bd, as well as Punic PNs with the component bod (op. cit., p. 31). BH has byado. As for the meaning, both the

Akkadian and the context demand that we translate this phrase as an ablative "from his hand" . BH seems not to attest an ablative meaning for the cognate phrase byad, and uses miyyad instead (BDB, 390-1). The verb tu-ra / tura / is a prefix-conjugation passive qal form of

wry(•`yry), which conforms to the forms found in other Canaanite sites of the •ã period (see discussion for #22 above). However, BH attests to a nifal form instead (yiyyar "he will be shot", Ex 19: 13). Of course, the replacement of passive qal forms by nifal in Hebrew (Jouon-Muraoka 1993, •˜58d; Blau 1976, 51) means it is possible that Hebrew too used a similar form in older periods.

To sum up, the Canaanite words from Megiddo seem to reflect a dialect

Vol. XXXVIII 2003 93 different from the one we can reconstruct as the forerunner of standard BH. As mentioned, the verb mahSuhu further attests to contemporary variation. This is manifest by the vowel syncope it reflects. We have discussed above (p. 80) non- syncopated forms attested in the Jerusalem Amarna letters (and elsewhere). While the forms attested from Jerusalem do not match precisely the form mahSuhu in syllabic structure, we have also seen one form which does match it, viz., sapatuni, from a northern dialect. Therefore, we can safely conclude that the Megiddo dialect was different at least in this respect. The cluster of other forms that show differences from the dialect represented in standard BH may be significant in that they allow us to draw the tentative conclusion that contemporary dialects did exist and that the Megiddo dialect was distinct by several features, not only by this one individual feature. As we have seen some other divergencies from what can be reconstructed as a forerunner of standard BH also in texts from other 14th century Canaanite sites, we wish to suggest as a tentative hypothesis that there existed a Megiddo dialect, which was distinct from the attested Canaanite continuum. Having visited Megiddo, I cannot refrain from recalling the northern dialect of Hebrew as reflected in the Song of Deborah (Judges 5). Commentators differ in the rendering and interpretation of various forms and structures of this text, and some would even challenge the usually accepted early date of that poem. While I will not delve into discussing the language of the Song of Deborah here, let me say just a few words. Most recently, reviewing the work of Waltisberg (1999), Rendsburg (2003)30 writes as follows: The features which Waltisberg considered to be Aramaisms are instead to be understood as Aramaic-like features, lexical and grammatical traits shared by I(sraelian) H(ebrew) and Aramaic. There are, in fact, many other IH features in the poem, including items with parallels in Aramaic and items with parallels in Ugaritic-Phoenician. After listing these items, Rendsburg concludes: This long list of items ... together demonstrate that Judges 5 is to be seen as an Israelian composition. Recalling that the geographical setting of the Sisera war near Megiddo, I am tempted to ask whether we cannot look for signs for an historical line connecting the Canaanite dialect of Megiddo of the 14th century and the language of the Song of Deborah. Any conclusive evidence as regards the historical continuity of the 14th century Megiddo dialect and the one suggested for the Song of Deborah is yet to be presented. Still, we may note at this point

94 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC that one "Aramaism" has been revealed in this study, viz., vowel syncope in the form mahSuhu. Admittedly, vowel syncope is not a rare feature or one that is unlikely to occur independently, so I would like to look further at what insights can such a comparison bring forth. With Rendsburg, who followed old traditions in biblical studies, I do see the Song of Deborah as an archaic poem, so that even the chronological distance between the composition of these two bodies of text is not too large.

Synopsis In order to enable a more general look at my findings, I have listed below the relevant forms that give any insight on linguistic variation as discussed in this paper. The variant forms are here arranged according to grammatical features. No reference is given where no contemporary data is available. Still, in the section on morphology, some comparative data from later NWS dialects, mostly BH, are listed sporadically. The list is to be regarded as an index of forms discussed in the paper rather than as a comprehensive list of attested grammatical features.

Phonology and phonetics Diphthong contraction (ay): Aphek (#51), Tyre (#16), Megiddo (#21), Byblos (p. 89); Jerusalem: no data Final triphthong contraction: Jerusalem sade (#27) -Shechem Sosu (#32); Byblos-Beirut yapu (#31)-cf. Megiddo anayi (#60) Contraction of vowels in tertiae aleph verbs: Pihilu hihbe (#11) — ambiguous in Gath nassaa 1 nassa (#36 and comments to #11)

Contraction of iya•„ a (or a) in badihu: Megiddo (#61); Phoenician-Punic bd, bod -BH byad ?Dropping of initial lyl: Jerusalem obil (#27); Obilimi (#28) Non-assimilation of In!: Shechem tanSuku (#27); unknown (northern?) provenance antin(n)u (#39) Vowel syncope: Megiddo magiThu (#59)-not in Jerusalem abadat (#29); not in Enishasi ,.Sapatuni (p. 80) Vowel harmony (specific): Jerusalem Zoro (#24); Gath, Gath Carmel, etc. mihiSa(#35)

Morphology and the lexicon ?3 sg f genitive pronominal suffix: Gath upsihi (#38)-Semitic s/ha(:) -(i)mu in the sg: Tyre panimu (#18) -Shechem, Jerusalem: -amu pl. (#20, #21)

Vol. XXXVIII 2003 95 Human interrogative pronoun: miya(mi)•`miyati (Kamid el-Loz, Bashan, Shechem, Pihilu) (p. 90)

Gender difference: Jerusalem Zoro (m) (#24) -BH zroa (f)

Feminine marking: Gath Carmel etc. labittu (-t) (#45) -BH lbenda(•ƒ-at)

Case ending in sg construct nouns: Jerusalem -0 (Zoro , #24; obil, #27)

- Tyre -0•`-i (ubud•`ubudi, #17) Qal patterns: Tyre nuhti (#14) - BH nahti -Arabic nuhti -Kamid el- Loz halti (p. 75) -BH halti Tyre batihiti (#14); Byblos batihta (p. 75) BH Megiddo etc. 'ahriSu (#52)-BH taharos Jerusalem saduk(?) (#25)

Passive qal (suffix conjugation): Canaanite •Ea•Ei• (p. 71) -BH *•Eu•Ea• -—

cf. Aramaic•E•Ei•E Qal vs. nifal: Tyre: kuna (qal) (#13) -BH hikkon (nifal) Passive qal vs. nifal: Megiddo: turd (#59) -BH yiyyarae Qal vs. hifil: Gath: yasi (#34) -BH yasi2eni; cf. ysd2uni Jerusalem Obil (#27); Obilimi (#28) -BH *mobil Root variants: Gath •ãypk (#33) -BH •ãpwk -Aramaic ,•ãnpk

(Byblos 1ytn (p. 83); Pheonician •ãytn -EA 96 4lntn (#39); BH 4•ãntn) Nominal patterns and formations: Tyre mema (#16)•`Aphek mu-mi (#19), mulka (#47), milka (#48), sul(l)uhta (#49), hatnuta (#50), matni'a (#46), malbasi (#10), anayi (#60), mahsiramu (#23), ibi (#9), yvhvdvnni (#56), saduk(?) (#25) A collective noun (or name of species): namlu (#42) Particles: Megiddo badihu ablative (#61) BH miyyad Megiddo 2ahronhu (#59) -BH (aharon)

Lexical semantics: Tyre •ãrgb "fear" (#12), Gath, Megiddo •ãmhS "kill"

(#35, #59), Jerusalem, Shechem •ãSyr "denounce" (#22, #30)

What Now? The title of this paper suggests a discussion of Canaanite varieties in the second millennium BC. Its subtitle raised the question whether we can dispense with anachronism in our exploration of synchronous variation. I believe we can answer this question now. The evidence for synchronous variation is extremely scanty. We have a single feature in Megiddo (vowel syncope) which stands against evidence of the retention of vowels in a similar environment from other dialects; the variant forms of the human interrogative pronoun; and an ambiguous difference in the final vowel in the word for "water" between Aphek

96 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC and Tyre. Along with these features, I am intrigued by the variant forms of the mediae vocalis verbs as reflected by nuhti and halti, and I wonder whether these too cannot be regarded as reflecting dialectal variation rather than variant forms dependent on the respective roots. Apart from these instances, I have not been able to show any synchronous variation. Therefore I must conclude that in order to deal with linguistic variation in 14th century Canaanite we cannot dispense with anachronism, at least not at this stage of research. However, for studying both standard BH and, especially, non-standard features in the Hebrew Bible in their Canaanite background, research into Canaanite varieties can be proved to be useful indeed. Further research may improve our ability to show synchronous variation. As stated at the beginning of this paper, my discussion has focused on prominent and genuine Canaanite words. At this point, I would like to point out some of the interesting issues that can be instructive for the study of dialectology in the Levant and the development of the Canaanite dialects, as well as other issues in Semitic linguistics, by studying not only of genuine Canaanite words, but also features attested within the Canaano-Akkadian structure. Some of these features are, the use of the infinitive absolute as a finite verb, which is taken to be northern (Rendsburg 2002, 76-9) and indeed seems to be attested (only?) in northern Canaano-Akkadian texts; the Tense-Mood-Aspect system, in particular the use of the suffix conjugation and the jussive forms for the past (cf. Moran 1950, 24-52, Rainey 1996, II, 221-7, 348-50); the reflection of the Barth law in prefix-conjugation qal forms (Rainey 1978b; Rainey 1996, II, 61-75); the vowel following the personal prefix in piel and safel prefix conjugation forms (cf. the remarks above, p. 87); impersonal expressions (Moran 1950, 61; Izre'el 1998a, 63); gender agreement for the collective (Rainey 1996, I, 132-5; cf. above, p. 85); case endings (Kossmann 1987-8; cf. Blau 1970, 36-7); assimilation and non-assimilation of n before a following consonant (cf. above, p. 85); possible reflections in spelling of Canaanite morphs and proper names that may reveal phonological variation in spite of their anachronistic values (Sivan 1984 in passing; cf. also my remarks on the sibilants and on the difference between spellings of // and /g/ in n. 1); and many others. I have eliminated from my discussion texts originating in places outside of Biblical Canaan (apart from a few instances of using them as comparative evidence). Of course, a comprehensive study including the larger Canaanite continuum is desired. A start was made more than 60 years ago by Harris (1939), whose work is chronologically deeper, as it includes also later materials. Sivan's work (1984), with all its limitations (see Huehnergard 1987b; van Soldt 1989),

Vol. XXXVIII 2003 97 can be used as a start towards a comprehensive, large-scope study to replace Harris' 1939 work. The study of dialects in the first millennium BC, either as reflected in the Bible (an enterprise mostly taken by Rendsburg; see his state-of- the-art paper in this volume) or as reflected in external sources (e.g., Garr 1985), may also take advantage of the diachronic aspect, which can be disclosed by meticulous, punctilious scrutiny of the Amarna textual materials and other second millennium sources (inscriptions in consonantal script, e.g., Sass 1988 and Egyptian materials; e.g., Hoch 1994; Sivan and Cochavi-Rainey 1992). For such an endeavor to take place, some theoretical premises must stand in the background, including extralinguistic issues. Among the latter, one may mention immigration, entailing the introduction of new dialects or languages into the area under scrutiny, which in itself may entail either absorbing sub- or adstratal features into the local dialects, or - vice versa - absorbing substratal features coming from outside when newcomers integrate into the local society and take over their language. Of course, new insights may thus lead to far-reaching conclusions regarding the origin of some sections of the population in that area (cf. Izre'el 1987a, 90). Will the linguistic study help in resolving these intriguing issues? Let the future decide.

Notes 1 There is no way to determine whether any of the shifts realized in either Phoenician or Hebrew had occurred prior to the Amarna period in that region (cf. Harris 1939, 40; Segert 1997, 59; Rendsburg 1997). Where Ivam unable to reconstruct the status of the phoneme, I use capital letters as follows: Z for */dl, S for */t/, S for either */z/ or */d/. This way I acknowledge our inability to tell whether the attested form has preserved the original phoneme or already manifests its respective later counterpart. The Amarna scribes did differentiate in writing between /g/ and / /. /g/ is always written with h-syllabograms; for words with / /, the scribes vacillated between using the h-syllabograms and not indicating this sound at all (Izre'el 2003, 23). For /h/ and/h/ in secondmillennium "Phoenician" see Harris 1939, #13, #39; Krahmalkov 2001, 19. For the o vowel as a result of the Canaanite shift see Bohl 1909, 25-6; Harris 1939, §17;Sivan 1984, §5.2. 2 A Glossenkeil is marked here by a backslash (•_) . A phonemic string (transcription, normalization) is given between slashes (/---/) 3 The standard spelling of s-syllables for etymological /z/ is not unequivocal (pace Sivan 1984, 38), as analogy with early Aramaic inscriptions teaches us (Gan 1985, 27), all the more so in the cuneiform syllabary used in Canaano-Akkadian, which did not distinguish between /su/ and /zu/ (AS, #5; Jucquois 1966, 233). 4 In contrast to what the notation of Amarna in Jucquois 1966 , 66 #206 may imply, the value to for DU is attested exclusively in letters from sites outside the borders of Canaan, as, e.g., Amurru (Izre'el 1991, II, 118). (The single occurrence from Byblos, EA 132: 50, requires explanation.) 5 One other possibility is that the idiom goes back to Egyptian , since Egyptian tradition may stand behind the two-way prostration (cf. Moran 1992, xxx n85). This will need further exploration; a preliminary survey of Egyptian sources does not support such a hypothesis.

98 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC 6 CAD marks this meaning as questionable; sasallu mostly bears the meaning "tendon of the hoof or heel" (CAD S2, 168-9). 7 The doubling of the second radical is explained as secondary (Bauer-Leander 1922, 219, 316; Jouon-Muraoka 1993, I, 166-7; Blau 1976, 32). 8 I thank Yuval Goren for sharing this information with me . For now, see . 9 The form as it appears in the text is adapted to the Canaano-Akkadian structure . This is evident since the plural marker lacks the final -ma (cf. Sivan 1984, §2.1). For this form see Albright 1943, 32 n26 ("my foe"); cf. Rainey 1978a, 73; Rabiner 1981, 16, 18 n23, who insisted on a plural interpretation given the appositions; differently Moran 1992. 306. 10 Arabic attests to both patterns (pace Baumgartner; cf. Hava 1970, 676). 11 For 3 sg m forms with final -a see Ebeling 1910, 56-7; Rainey 1996, II, 287-8. 12 For a possible BH occurrence of Awn in qal cf. waykunnnu (Job 31: 15), usually emended to be read as a polel form (e.g., Baumgartner 1974, 442). 13 The vowel e need not be a reflection of the lost consonant but may be the result of e's

approximation to r (and .s; GAG, •˜9b). 14 One may perhaps add to the following list the gloss \ ta-ZA-ka[ ..] (EA 287: 41-42), which has resisted any solid interpretation so far (cf. Izre'el 1987a, 90; Moran 1992, 330 n11). 15 As for 'abadat , it attests the morpheme -at of the suffix-conjugation 3 sg f, preserved in BH only in non-final position (e.g., ktalatni; Bauer-Leander 1922, 309, 342). Other NWS languages have preserved the t in final position also in either the noun (Phoenician; Garr 1985,

3:6c) or the verb (Aramaic; op. cit., •˜3:17c). In any case, the difference between the•˜ 1st millennuim NWS dialects and the Amarna evidence is a chronological one. Ugaritic has -at (or -t) preserved in all environments (Tropper 2000 , §§52.2,73.331.2). 16 Although BH suggests an imperative with an a thematic vowel (cf . the prefix conjugation yisdak; Baumgartner 1983, 941), I wonder whether this vocable could not also be taken as an imperative with the same form /saduk/ , perhaps attested in the so-called Amorite personal names (Gelb 1980, 365). If so, the translation would be: "Look: 0 king, my lord! Justify me!" 17 For the sibilants in Jerusalem and their spelling see Izre'el 2003 , 19-20. 18 Although it comes in a context where words of Ribhaddi of Byblos are quoted , this form seems not to attest Byblos Canaanite but the local dialect of the scribe of this letter. 19 Mari PNs , which, according to Moran (1961, 62) suggest an a for possible causative forms, attest to diphthongs in causative froms of •ãws (loc. cit.). 20 Akkadian further attests to lamattu (•ãlmn) in a lexical list . Both CAD, 67 and AHw, 533 suggest that it is a West Semitic word. 21 I thank Gary Rendsburg , who prodded me into looking again at the possibility that this form reflects an actual pronunciation, and Gideon Goldenberg, who discussed with me the problem of similar forms in Aramaic. 22 One other form that has been explained as hifil is ya-ah-li-ku "lose" (EA 254: 9, Shechem; Rainey 1996, II, 190). This form still needs further research (cf. Rainey 1978a, 72, where this form is interpreted as a transitive qal form). Note that both the first and the second signs can be rendered with any possible vowel, and this form can therefore be transliterated yi-ih-li-ku also (Knudtzon 1915, 810). 23 Where for EA Canaanite the erronous labinat is cited . 24 The reading of the second sign was suggested by Rainey 1975, 405 n85. Knudtzon's reading was l[i-e]l (1915, I, 788 with note d); Schroeder's drawing (1915, 181) seems to allow for [l]a. 25 The diphthong aw (or contracted vowels resulting from this diphthong) still has to be sought. Cf. Izre'el 2003, 28 for a possible attestation of the word for "death" (mu-tu-mi) in Byblos. 26 In similar formulae one also finds Akkadian mannu "who" (pace Izre'el 1978, 26 with n60).

Vol. XXXVIII 2003 99

27 Cf . also the following example: s um-ma-mi ga-am-ra-at-mi 31 d/u(URUKI) i-na BA.UG7 32 i-na mu-ta-a-an 33 i-na x-ri Look, the city is consumed by death, by plague, by...(EA 244: 30-33) The context requires a word related to death or illness, but the traces on the tablets have so far resisted any conclusive interpretation. Moran's dab-ri (cf. Hebrew daebaer "pestilence"; Moran 1992, 299 n5) is a desperate solution, and seems not to agree with the traces of the sign as drawn by Moran himself (who kindly placed at my disposal his collations of the Amarna letters). It is possible that the context goes from the general to the particular (BA.UG7, Akkadian mutu, means "death"; mutan(u) means "plague, epidemic"), and hence the repetition of the preposition ina before each noun, which explains the absence of a Glossenkeil. In any case, if the last word was indeed Canaanite, the language usage here looks like a play on the tripartite gloss set. 28 One may recall in this connection the finding of a Gilgamesh fragment in Megiddo (Goren, Finkelstein and Na'aman 1999; cf. Edzard 1985, 251; Artzi 1985, 270; van der Toorn 2000, 105). The assumption that Megiddo was a scribal center may further help us understand several features of its Canaano-Akkadian grammatical system, such as the use of the Akkadian stative 1st person marker -aku instead of the more common This, however, needs further research. 29 An alternative interpretation of this phrase as "afterwards" has also been suggested (cf. discussion in Campbell 1965, 199). However, the narrative seems to have a conjunction before each new sentence, while there is none before EGIR-su. 30 I thank Gary Rendsburg for making available to me his unpublished work.

References Aharoni, Y. 1967: The Land of the Bible: A Historical Geography, translated by A. F. Rainey, London: Burns & Oates. AHw: W. von Soden, Akkadisches Handworterbuch, I-III, Wiesbaden: Harrassowitz, 1965-1981. Albright, W. F. 1937: "The Egyptian Correspondence of Abimilki, Prince of Tyre," Journal of Egyptian Archaeology 23, 190-203. Albright, W. F. 1943: "An Archaic Hebrew Proverb in an Amarna Letter from Central Palestine," Bulletin of the American Schools of Oriental Research 89, 29-32. Arnaud, D. 1991: "Une lettre de Kamid-el-Loz," Semitica 40, 7-16. Artzi, P. 1963: "The Glosses in the El-Amarna Documents," Bar Ilan 1, 57-84. (Hebrew) Artzi, P. 1985: "Response to Edzard," in Biblical Archaeology Today: Proceedings of the International Congress on Biblical Archaeology, Jerusalem, April 1984, Jerusalem: Israel Exploration Society, 269-273. Artzi, P. 1990: "Studies in the Library of the. Amarna Archive," in J. Klein and A. Skaist (eds.), Bar-Ilan Studies in Assyriology dedicated to Pinhas Artzi, Bar-Ilan Studies in Near Eastern Languages and Culture; Publications of the Bar-Ilan University Institute of Assyriology, Ramat-Gan: Bar-Ilan University Press, 139-156; pl. III. AS: W. von Soden and Wolfgang Ro11ig, Das akkadische Syllabar, 4., durchgesehene und erweitete Auflage, Analecta Orientalia, 42, Roma: Pontificio Istituto Biblico, 1991. Barth, J. 1894: Die Nominalbildung in den semitischen Sprachen, 2. Auflage, Leipzig 1894. (Reprinted: Hildesheim: Olms 1967.) Bauer, H., and P. Leander. 1922: Historische Grammatik der hebrdischen Sprache des Alten

Testamentes, I: Einleitung, Schriftlehre, Laut- und Formenlehre, Mit einem Beitrag (•˜•˜6-9) von Paul Kahle und einem Anhang: Verbparadigmen, Halle. (Reprinted: Hildesheim: Olms, 1965.)

100 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC

Bauer, H., and P. Leander. 1927: Grammatik des Biblisch-Aramaischen, Halle. (Reprinted: Hildesheim: Holms 1966.) Baumgartner, W. 1967-1990: Hebraisches und aramaisches Lexikon zum Alten Testament, Dritte Auflage, I-IV, Leiden: Brill. BDB: F. Brown, S. R. Driver and C. A. Briggs, A Hebrew and English Lexicon of the Old Testament, Oxford: Clarendon Press, 1907. Birkeland, H. 1940: Akzent und Vokalismus im Althebraischen, mit Beitragen zur vergleichenden semitischen Sprachwissenschaft, Oslo: Dybwad. Blau, J. 1970: On Pseudo-Corrections in Some Semitic Languages, Jerusalem: The Israel Academy of Sciences and Humanities. Blau, J. 1976: A Grammar of , Porta linguarum Orientalium N.S., 12, Wiesbaden: Harrassowitz. Blau, J. 1981: "On Pausal Lengthening, Pausal Stress Shift, Philippi's Law and Rule Oredering in Biblical Hebrew," Hebrew Annual Review 5, 1-13. Blau, J. 1986: "Remarks on the Chronology of Philippi's Law," in Proceedings of the Ninth World Congress of Jewish Studies, Jerusalem, August 4-12, 1985; Division D, Volume I: Hebrew and Jewish Languages, Other Languages, Jerusalem: World Union of Jewish Studies, 1-4. (Hebrew) Bohl, F. M. Th. 1909: Die Sprache der Amarnabriefe mit besonderer Berucksichtigung der Kanaanismen, Leipziger Semitische Studien, Band 5, Heft 2. (Reprinted: Leipzig: Zentralantiquariat der DDR, 1977.) Brockelmann, C. 1908; 1913: Grundriss der vergleichenden Grammatik der semitischen Sprachen, I-II, Berlin: Reuthner & Reichard. CAD: The Assyrian Dictionary of the Oriental Institute of Chicago, Vols. 1-11, 13, 15-17, 21 (letters A-N, Q, R, S, S, S, Z), Chicago: The Oriental Institute, 1956-1999. Campbell, E. F. 1965: "Shechem in the Amarna Archive," Appendix 2 in G. E. Wright. Shechem: The Biography of a Biblical City, New York: McGrow-Hill, 191-207. CDA: J. Black, A. George and N. Postgate, A Concise Akkadian Dictionary, SANTAG, 5, Wiesbaden: Harrassowitz, 1999. De Moor, J. C. 1969: "Studies in the New Alphabetic Texts from Ras Shamra I," Ugarit Forschungen 1, 167-188. Del Olmo Lete, Gregorio and Joaquin Sanmartin 2003: A Dictionary of the Ugaritic Language in the Alphabetic Tradition, translated by Wilfred G.E. Watson, Handbook of Oriental Studies; Section 1: The Near and Middle East, 67, I-II, Leiden: Brill. DNSWI: J. Hoftijzer and K. Jongeling, Dictionary of the North-West Semitic Inscriptions, With appendices by R. C. Steiner, A. Mosak-Moshavi and B. Porten, I-II, Handbuch der Orienatlistik, Erste Abteilung: Der Nahe und Mittlere Osten, 21, Leiden: Brill, 1995. Ebeling, E. 1910: "Das Verbum der El-Amarna-Briefe," Beitrage zur Assyriologie und semitischen Sprachwissenschaft 8, 39-79. Edzard, D. O. 1985: "Amarna und die Archive seiner Korrespondenten zwischen Ugarit und Gaza," in Biblical Archaeology Today: Proceedings of the International Congress on Biblical Archaeology, Jerusalem, April 1984, Jerusalem: Israel Exploration Society, 248-259. Finkel, M. 1977: The Tyre Letters of the El-Amarna Archives: A Study of Selected Linguistic Aspects, M.A. Thesis, Tel Aviv University, unpublished. Garr, W. R. 1985: Dialect Geography of Syria-Palestine, 1000-586 B.C.E . Philadelphia: The University of Pennsylvania Press. Gesenius-Kautsch-Cowley, 1910: Gesenius' Hebrew Grammar as edited and enlarged by late E. Kautzsch, second English edition revised in accordance with the twenty-eighth German edition (1909) by A. E. Cowley, Oxford: Clarendon Press. Vol. XXXVIII 2003 101 Goren, Y. 2000: "Provenance Study of the Cuneiform Texts from Hazor," Israel Exploration Journal 50, 29-42. Goren, Y., I. Finkelstein and N, Na'aman. 1999: "El-Amarna Tablets from Megiddo," Revelations from Megiddo: The Newsletter of The Megiddo Expedition 1999/4, Goren, Y., I. Finkelstein and N, Na'aman, forthcoming: Inscribed In Clay, I: Provenance Study of the Amarna Letters and Other Ancient Near Eastern Texts. Grave, C. 1982: "Northwest Semitic sapernu in a Break-up of an Egyptian Stereotype Phrase in EA 147," Orientalia 51, 161-182. Harris, Z. S. 1936: A Grammar of the , American Oriental Series, 8, New Haven: American Oriental Society. Harris, Z. S. 1939: Development of the Canaanite Dialects, American Oriental Series, 16, New Haven: American Oriental Society. Hava, J. G. 1970: Al-Fara 'id: Arabic-Anglish Dictionary, Beirut: Dar el-Mashreq. Hoch, J. E. 1994: Semitic Words in Egyptian Texts of the New Kingdom and Third Intermediate Period, Princeton, New Jersey: Princeton University Press. Huehnergard, J. 1987a: Ugaritic Vocabulary in Syllabic Transcription, Harvard Semitic Studies, 32, Atlanta, Georgia: Scholars Press. Huehnergard, J. 1987b: "Northwest Semitic Vocabulary in Akkadian Texts," Journal of the American Oriental Society 107, 713-725. Huehnergard, J. 1996: "A Byblos Letter, Probably from Kamid el-Loz," Zeitschrift ,fur Assyriologie 86, 97-113. Izre'el, Sh. 1977: "Two Notes on the Gezer-Amarna Tablets," Tel Aviv 4, 159-167; pl. 18(5). Izre'el, Sh. 1978: "The Gezer Letters of the el-Amarna Archive-Linguistic Analysis," Israel Oriental Studies 8, 13-90. Izre'el, Sh. 1987a: "Early Northwest Semitic 3rd Pl. M. Prefix: The Evidence of the Amarna Letters," Ugarit Forschungen 19, 79-90. Izre'el, Sh. 1987b: "Some Methodological Requisites for the Study of the Amarna Jargon: Notes on the Essence of That Language," in B. J. Beitzel and G. D. Young (eds.), Amarna in Retrospect: A Centennial Celebration, Winona Lake, Indiana: Eisenbrauns. = Izre'el, Sh. 1991: Amurru Akkadian: A Linguistic Study, with an Appendix on the History of Amurru by Itamar Singer, Volume I, Harvard Semitic Studies, 40; Volume II, Harvard Semitic Studies, 41, Atlanta, Georgia: Scholars Press. Izre'el, Sh. 1995: "The Amarna Glosses: Who Wrote What for Whom? Some Sociolinguistic Considerations," Israel Oriental Studies 15, 101-122. Izre'el, Sh. 1998a: Canaano-Akkadian, Languages of the World/Materials, 82, Munchen: LINCOM Europa. Izre'el, Sh. 1998b: "A New Dictionary of Northwest Semitic and the Amarna Glosses," Israel Oriental Studies 18, 421-430. [On: J. Hoftijzer and K. Jongeling, Dictionary of the North- West Semitic Inscriptions, with appendices by R. C. Steiner, A. Mosak-Moshavi and B. Porten, I-II, Handbuch der Orienatlistik, Erste Abteilung: Der Nahe and Mittlere Osten, 21, Leiden: E. J. Brill, 1995.] Izre'el, Sh. 2003: "Vocalized Canaanite: Cuneiform-Written Canaanite Words in the Amarna Letters -Some Methodological Remarks," DS-NELL (Dutch Studies - Society of Near Eastern Languages and Literatures) 5/1-2, 13-34.) Jouon, P. 1993: A Grammar of Biblical Hebrew, translated and Revised by T. Muraoka, (Volume I: Part One: Orthography and Phonetics; Part Two: Morphology; Volume II: Part Three: Syntax; Paradigms and Indices), Subsidia Biblica, 14/I-II, Roma: Pontificio istituto Biblico. 102 ORIENT CANAANITE VARIETIES IN THE SECOND MILLENNIUM BC Jucquois, G. 1966: Phonetique comparee des dialectes moyen-babyloniens du nord et de l'ouest, Bibliotheque du Museon, 53, Louvain: Institut Orientaliste. KAI: H. Donner and W. Rollig, Kanaanaische und aramaische Inschriften, Zweite durchgesehene und erweiterte Auflage, 3 vols, Wiesbaden: Harrassowitz, 1966-1969. Khan, G. 1997: " Phonology," in A. S. Kaye (ed.), Phonologies of Asia and Africa (Including the Caucasus), Winona Lake, Indiana: Eisenbrauns, Vol. 1, 85-102. Knudtzon, J. A. 1915: Die El-Amarna Tafeln, Anmerkungen und Register bearbeitet von C. Weber und E. Ebeling, I-II, Vorderasiatische Bibliothek, 2, Leipzig. (Reprinted: Aalen: Otto Zeller, 1964.) Krahmalkov, C. R. 2001: A Phoenician-Punic Grammar, Handbook of Oriental Studies; The Near and Middle East, 54, Leiden: Brill. Kutscher, E. Y. 1970: "Aramaic," in Th. A. Sebeok (ed.). Current Trends in Linguistics, Vol. 6: Linguistics in South WestAsia and North Africa, The Hague: Mouton, 347-412. Leander, P. 1928: Laut- und Formenlehre des Agyptisch-Aramaischen, Halle. (Reprinted: Hildesheim: Georg Holms 1966.) Lipinski, E. 1997: Semitic Languages: Outline of a Comparative Grammar, Orientalia Lovaniensia Analecta, 80, Leuven: Peeters. Moran, W. L. 1961: "The in its Northwest Semitic Background," in G. E. Wright (ed.) The Bible and the Ancient Near East: Essays in Honor of William Foxwell Albright (Reprinted: Winona Lake, Indiabna: Eisenbrauns, 1979), 54-72. = W. L. Moran, Amarna Studies: Collected Writings, edited by J. Huehnergard and Sh. Izre'el, Harvard Semitic Studies, 54, Winona Lake, Indiana: Eisenbrauns, 2003, 197-218. Moran, W. L. 1975: "The Syrian Scribe of the Jerusalem Amarna Letters," in H. Goedicke and J. J. M. Roberts (eds.), Unity and Diversity: Essays in History, Literature and Religion of the Ancient Near East, Baltimore, 146-166. =W. L. Moran, Amarna Studies: Collected Writings, edited by J. Huehnergard and Sh. Izre'el, Harvard Semitic Studies, 54, Winona Lake, Indiana: Eisenbrauns, 2003, 249-274. Moran, W. L. 1983: "A Note on igi-kar, 'provisions, supplies,'" Acta Sumerologica 5, 175-177. = W. L. Moran, Amarna Studies: Collected Writings, edited by J. Huehnergard and Sh. Izre'el, Harvard Semitic Studies, 54, Winona Lake, Indiana: Eisenbrauns, 2003, 297-300. Moran, W. L. 1992: The Amarna Letters, Baltimore: Johns Hopkins University Press. Rabin, C. 1981: "Leshonam shel Amos veHoshea" (The Language of Amos and Hosea), in B.-Z. Luria (ed.), Iyunim beSefer Tre-Asar, Divrei Hug haIyun beTanach beVet Nesi haMdina, Jerusalem, 115-136. (Hebrew) Rabiner, S. 1981: Linguistic Features of the El-Amarna Tabletsfrom Akko,Megiddo and Shechem, M.A. Thesis, Tel Aviv University. (Hebrew with English summary) Rainey, Anson F. 1975: "Morphology and the Prefix Tenses of West-Semitized el-'Amarna Tablets," Ugarit Forschungen 7, 395-426. Rainey, A. F. 1976: "A Tri-lingual Cuneiform Fragment from Tell-Aphek," Tel-Aviv3, 137-140. Rainey, A. F. 1978a: El-Amarna Tablets 359-379, Alter Orient und Altes Testament, 8, Second edition, revised, Kevelaer: Butzon und Bercker. Rainey, A. F. 1978b: "The Barth-Ginsberg Law in the Amarna Tablets," Eretz Israel 14, 8*-13*. Rainey, A. F. 1996: Canaanite in the Amarna Tablets: A Linguistic Analysis of the Mixed Dialect Used by Scribes from Canaan, Handbuch der Orientalistik, Erste Abteilung: Der Nahe und Mittlere Osten, 25, 4 volumes, Leiden: Brill. Rainey, A. F. 1999: "Taanach Letters," Eretz-Israel 26, 153*-162*. Rendsburg, G. A. 1997: "Ancient Hebrew Phonology," in A. S. Kaye (ed.). Phonologies of Asia and Africa (Including the Caucasus), Winona Lake, Indiana: Eisenbrauns, Vol. 1, 64-83.

Vol. XXXVIII 2003 103 Rendsburg, G. A. 2003: "Hurvitz Redux: On the Continued Scholarly Inattention to a Simple Principle of Hebrew Philology," in I. Young (ed), Biblical Hebrew: Chronology and Typology, Sheffield: Sheffield Academic Press. (In press.) Sass, B. 1988: The Genesis of the Alphabet and Its Development in the Second Millennium B.C., Agypten und Altes Testament, 13, Wiesbadsen: Otto Harrassowitz. Schroeder, O. 1915: Die Thontafeln von El-Amarna, Vorderasiatische Schriftdenkmaler der Koniglichen Museen zu Berlin, XI-XII, Berlin. (Reprinted: Osnabruck: Zeller, 1973.) Segert, S. 1997: "Phoenician and Punic Phonology," in A. S. Kaye (ed.), Phonologies of Asia and Africa (Including the Caucasus), Winona Lake, Indiana: Eisenbrauns, Vol. 1, 55-64. Sivan, D. 1984: Grammatical Analysis and Glossary of the Northwest Semitic Vocables in Akkadian Texts of the 15th-13th C.B.C. from Canaan and Syria, Alter Orient und Altes Testament, 214, Kevelaer: Butzon und Bercker. Sivan, D., and Z. Cochavi-Rainey. 1992: West Semitic Vocabulary in Egyptian Script of the 14th to the 10th Centuries BCE, Beer-Sheva: Studies by the Department of Bible and Ancient Near East, VI, Beer-Sheva: Ben-Gurion University of the Negev Press. Sokoloff, M. 1990: A Dictionary of Jewish Palestinian Aramaic of the Byzantine Period, Dictionaries of Talmud, Midrash and Targum, 2, Ramat Gan: Bar Ilan University Press. Van Soldt, W. H. 1989: Review of Sivan 1984, Bibliotheca Orientalis 46, 645-651. Van Soldt, W. H. 1997: "Amarna upsu - Ugaritic ps, 'Boundary (Stone)," NABU 1997/3, #90. Spitaler, A. 1952-54: "Zur Frage der Geminatendissimilation im Semitischen," Zeitschrift fur indogermanische Forschungen 61, 257-266. Steiner, R. C. 1980: "Yuqattil, Yaqattil, or Yiqattil: D-Stem Prefix-Vowels and a Constraint on Reduction in Hebrew and Aramaic," Journal of the American Oriental Society 100, 513-518. Tomback, R. S. 1978: A Comparative Semitic Lexicon of the Phoenician and Punic Languages, SBL Dissertation Series, 32, Missoula: Scholars Press. Tropper, J. 2000: Ugaritische Grammatik, Alter Orient und Altes Testament, 273, Munster: Ugarit Verlag. UT: C. H. Gordon, Ugaritic Textbook, Analecta Orientalia, 38, Roma: Pontifical Biblical Institute. Van der Toorn, K. 2000: "Cueniform Documents from Syria-Palestine: Texts, Scribes, and Schools," Zeitschrif des Deutschen Palastina-Vereins 116, 97-113. Waltisberg, M. 1999: "Zum Alter der Sprache des Deboraliedes Ri 5," Zeitschrift fur Althebraistik 12, 218-232. Waltke, B. K., and M. O'Connor. 1990: An Introduction to Biblical Hebrew Syntax, Winona Lake: Eisenbrauns. Wright, W. 1896; 1898: A Grammar of the Arabic Language, third edition revised by W. Robertson Smith and M. J. de Goeje, 2 volumes, Cambridge: University Press.

104 ORIENT