Northwest Semitic in the Second Millennium BCE (MA thesis)

Master’s thesisResMA Linguistics

Leiden University

Date:July 13, 2015

Supervisor:

Prof. Dr. H. Gzella

Second Reader:

Dr. A. Kloekhorst

Northwest Semitic

in the

Second Millennium bce

Jorik Groen

Contents

1 Introduction 11.1 Purpose of the study . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Problems of linguistic classification . . . . . . . . . . . . . . . . . 1

2 Northwest Semitic in the Second Millennium BCE 42.1 The Semitic family . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1.1 From ‘Proto-Semitic’ to ‘Proto-Northwest Semitic’ . . . . 62.1.2 The Northwest Semitic languages . . . . . . . . . . . . . . 8

2.2 Historical and geographical setting . . . . . . . . . . . . . . . . . 122.2.1 Syria-Palestine . . . . . . . . . . . . . . . . . . . . . . . . 122.2.2 A short history . . . . . . . . . . . . . . . . . . . . . . . . 122.2.3 Language in the second millennium BCE . . . . . . . . . 14

2.3 Methodological framework . . . . . . . . . . . . . . . . . . . . . . 162.3.1 Corpora . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.3.2 Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3.3 Research questions . . . . . . . . . . . . . . . . . . . . . . 18

3 Phonology 203.1 Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.1.1 Sibilants, interdentals, and laterals . . . . . . . . . . . . . 203.1.2 Velar fricatives . . . . . . . . . . . . . . . . . . . . . . . . 28

3.2 Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.2.1 Canaanite Shift . . . . . . . . . . . . . . . . . . . . . . . . 303.2.2 Diphthongs . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4 Morphology 364.1 Nouns and pronouns . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.1.1 Masculine plural and dual endings . . . . . . . . . . . . . 364.1.2 First person singular . . . . . . . . . . . . . . . . . . . . . 394.1.3 First person plural . . . . . . . . . . . . . . . . . . . . . . 404.1.4 Personal interrogative pronoun . . . . . . . . . . . . . . . 42

4.2 Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434.2.1 Suffix-conjugation endings . . . . . . . . . . . . . . . . . . 434.2.2 Derived stems . . . . . . . . . . . . . . . . . . . . . . . . . 45

i

5 Conclusions 485.1 Summary of attested developments . . . . . . . . . . . . . . . . . 485.2 Consequences for classification . . . . . . . . . . . . . . . . . . . 49

5.2.1 Aramaic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495.2.2 Amorite . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505.2.3 Ugaritic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505.2.4 Canaanite . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.3 Wave theory vs. family tree . . . . . . . . . . . . . . . . . . . . . 52

ii

iii

Abbreviations

Text resourcesI List of Thutmose III

from KarnakIX List of Amenhotep III

from SolebXa List of Amenhotep III

from Kom el-Het.anXXII List of Ramesses II

from LuxorXXIII List of Ramesses II

from KarnakXXVI List of Ramesses

II from ‘Amara-West/‘Aksha

XXVII List of Ramesses IIIfrom Medinet Habu

XXXIV List of Shoshenq I fromKarnak

Aph AphekARM Archives royales de

MariBSm Bet-ShemeshEA El-Amarna lettersE “Brussels” Execration

Texts, (Posener, 1940)f “Berlin” Execration

Texts, peoples (Sethe,1926)

e “Berlin” ExecrationTexts, rulers (ibid.)

Haz HazorJdg JudgesJos JoshuaKRI Kitchen (1969)Neh NehemiaNum NumberspAn I Papyrus Anastasi IpAn III Papyrus Anastasi IIITaa Taanakh

Tab TaborUrk.IV Sethe and Helck (1927-

1958)

Languages

Akk AkkadianAmr AmoriteANA Ancient North ArabianAr ArabicBA Biblical AramaicBH Biblical HebrewByb ByblianDA Deir ‘AllaEAC El-Amarna CanaaniteEg EgyptianEH Epigraphic HebrewGz G@‘@zIA Imperial AramaicMb MoabiteMSA Modern South ArabianNPn Neo-PunicNWS Northwest SemiticOA Old AramaicOSA Old South ArabianPh PhoenicianPn PunicPS Proto-SemiticSm Sam’alianSyr SyriacUg Ugaritic

iv

Transcription

Throughout this thesis, the material is presented in Latin transcription. Thiseases the comparison between the various languages and additionally makes thework more accessible to those that are less familiar with Semitic languages.However, transcribing different scripts from various fields of study, one needs toclarify on the conventions in the particular fields.

Common Semitic The transcription of the various Semitic languages hasbeen regularised and largely follows the common conventions, such as the useof a dot below a consonant to indicate its ‘emphatic’ nature (whatever its real-isation), e.g. t., s. , s. . This does not apply to h. , which is the pharyngeal fricative(IPA [è]); the other ‘gutturals’ are transcribed as usual, i.e. h

˘for [x], g for [G],

and – and » for [Q] and [P] respectively. Furthermore, the two lateral fricativesappear as s and s. (the latter was previously known as d. or d. ). Most notably isthat I will follow a modern convention of transcribing the interdental fricativesT, D, and T. with their IPA-symbols instead of the original t

¯, d¯

, and t. (cf. Gzella,2012a).

Biblical Hebrew and Aramaic Although it is clear that the authentic pro-nunciation of the vocalisation traditions of the Hebrew Bible did not exhibitcontrasting vowel quantity (i.e. long vs. short vowels), I will transcribe some oftheir vowels with a macron. In this way, the transcription is more representativeof the script as well as the original reflexes, rather than its Tiberian pronunci-ation (cf. Gzella, 2012d). The full system is found in table 1; note that matreslectiones are neglected. Schwa’s are written in the cases were they are needed tobreak up impossible consonant clusters. As is common use, spiranticised stopsare transcribed with a macron below, e.g. b

¯, t¯

, or, in two cases, above the letter,e.g. g and p.

Table 1: The Tiberian vowel sigs and their transliteration as used in this thesis.A qames. in an open syllable is transcribed with a, one in a closed syllable with o.

!Ë !Ë !Ë !Ë !Ë !Ë !Ë !Ë, Ëו! !Ë !Ë Ëי! !uË Ëו!a a a e e e o o o i ı u u

Egyptian The regular system of transliteration is used for the Egyptian conso-nants. That includes those that are written d

¯and t

¯for [c’] and [ch] respectively,

v

which have been very confusing in previous works on Semitic and Egyptian lan-guage contact (e.g. Sivan and Cochavi-Rainey, 1992). This confusion is solvedby the replacement of these characters in their use for the Semitic interdentals.

In this work, the syllabic or ‘group writings’ have been proved to be veryuseful for the reconstruction of vowels. This system uses ‘weak phonemes’ toindicate (long) vowels, viz. w for /˘u/ (and Semitic /o/), y for /ı/,

»ı for /ı/

(sometimes also for /a/), and »» for any vowel (Edel, 1966; Hoch, 1994). Again,I have chosen for a transliteration closer to the native writing system, so withthe original Egyptian values; in that way, infiltration of personal assumptionsare restricted to a minimum. Thus, the name of Hazor appears in ThutmoseIII’s list as H. -d

¯»»-r (I :32); Rainey (1982) renders this as H. a-d

¯u-ra, while none of

the vowels is obviously indicated in writing.

Additionally, cursive text represent actual attestations, text between slasheseither phonemes or phonemic realisation of a form; phonetic transcriptions aregiven in between square brackets. All reconstructed forms are indicated by anasterisk.

vi

Chapter 1

Introduction

1.1 Purpose of the study

The aim of this thesis is to map the linguistic variation of the Northwest Semiticdata of the second millennium BCE. Apart from Ugaritic, these stages of theNorthwest Semitic branch are almost exclusively attested as elements in Akka-dian or Egyptian texts. Moreover, the vast majority consists of personal andtopographical names, which are by far not as useful as lexical and grammaticalitems. The data used is collected in various publications, already being analysedby some of the best experts in the field of (Northwest) Semitic linguistics. Lateron (section 2.3.1), the corpora are introduced and their usefulness for this thesiswill be pointed out.

Although the relevance of a study on the history of the Northwest Semiticlanguage varieties has been pointed out before, it remains a desideratum un-til this day (Huehnergard, 1987a,b; Pardee, 1999). The first-millennium di-alects have been treated and analysed thoroughly by e.g. Harris (1939) andGarr (1985), and the comparative method has provided many insights on thehistory of this branch. Second-millennium evidence has been included occasion-ally to support one’s point, but a profound study and analysis has been confinedto describing analyses of restricted corpora, such as the one by Sivan (1984).

Thus, the very broad main question is: which developments can we perceive inthe Northwest Semitic data from the second millennium BCE and what do theytell us about the classification of this branch? In order to investigate this, anumber of features is selected (2.3.2) and will be searched for in the Middle andLate Bronze Age corpora (2.3.1). Apart from the independent developments ofthese features, it is interesting to see what consequences they have for the currentclassification of Northwest Semitic (cf. section 2.1). Additional sub-questionson this subject will be posed at the end of that section.

1.2 Problems of linguistic classification

The ideal outcome of this study would enable us to draw isoglosses and linkthese to the ones of first-millennium Northwest Semitic. Any clue found can

1

help to gain an insight into the classification of these languages, and this canbe regarded as a secondary goal of the current thesis.

Subsequently, it is very useful to elucidate which characteristics can be usedfor the genetic classification of related languages, particularly in the case of theSemitic languages. First and most important, one has to pay attention to so-called ‘shared innovations’. In any case shared innovations are of much morevalue than shared retentions (Trask and McColl Millar, 2007); after all, sharedretentions can only provide negative statements. Secondly, it is important toconsider the distinctiveness of shared features and the likeliness for a devel-opment to have occurred independently in different languages. For instance,the assimilation of the /n/ to a following consonant is a frequently attestedphenomenon in Semitic languages, occurring in almost all branches of the lan-guage family. The chance that it happens separately in unrelated languagesis thus not negligible, and it should not be used for purposes of classification(cf. Gzella, 2011a). Another example is the shift of original *p to /f/ in thepresumed ‘South Semitic’ (see below), which is “so natural as to be useless forsubgrouping” (Faber, 1997:4). In general, grammatical features are often morereliable than phonological developments, which in turn say more about geneticrelation than lexical arguments (ibid.; cf. Trask and McColl Millar, 2007).

There are, however, several problems that complicate the matter, and someof them are likely to be the cause of the lesser status that is reserved for com-parative and historical linguistics within the various Semitic fields, especiallywhen compared to Indo-European linguistics (Huehnergard, 2002). First of all,the Semitic languages are attested in different eras. This applies to Semitic ingeneral, but also to the Northwest Semitic branch (cf. section 2.1.2): Ugariticwas written mainly in the thirteenth and twelfth centuries BCE, most Canaan-ite dialects are dated in the first half of the first millennium BCE, while theAramaic varieties, attested as early as the tenth century, spread out throughthe entire Near East from around 700 BCE. The lack of contemporaneous data,in particular for Ugaritic and Aramaic, complicates the comparative research(Sivan, 2000).

The lack of consistent indications of vowels in the native scripts of the North-west Semitic languages is a second issue. All varieties are almost exclusively at-tested in variants of the consonantal alphabet that derived from the Canaanitelinear alphabet (Millard, 2012). The use of matres lectiones, i.e. certain con-sonant signs indicating (mostly final) long vowels, in some languages1 helps usonly to a very limited extend. One may consider this a minor problem, but forthe Semitic languages, with their relatively stable consonantal grid, vowels arecrucial in comparative linguistics (cf. Weninger, 2011b).

A third problem is caused by close contact between the languages. Becauseof the dialect continuum that surfaces in the first millennium BCE, it has beenthought that Aramaic and Canaanite did not separate before the end of thesecond millennium (Garr, 1985; Saenz-Badillos, 1993). This view is challenged

1These ‘vowel letters’ are found in the oldest sources of Aramaic and Hebrew, as well asin the writing systems based on these prestigious languages, viz. Moabite, Sam»alian, Deir–Alla, but not regularly in Ammonite and Edomite (?, cf. Vanderhooft, 1995). They arecompletely absent in Phoenician, only becoming widespread in the much younger Punic textsfrom Northern Africa. Originally, -w indicated final /u/, -y stood for word-final /ı/, and -hwas used for the other vowels, viz. /a, e, o/. Later on, phonological developments, such asthe loss of word-final /»/ and the contraction of diphthongs, facilitated the system of matreslectiones to become more extensive.

2

by the current thesis, but it is an indication of the likeness of the NorthwestSemitic varieties. Especially cases like the idioms used at Deir –Alla and Sam»aldemonstrate many influences from other Northwest Semitic languages and theirclassification has been a moot subject for quite some time (cf. Beyer, 2012;Huehnergard, 1991; Noorlander, 2012; Saenz-Badillos, 1993).

The linguistic diversity of Northwest Semitic may therefore be one of thebest examples of numerous ‘waves’ that spread across the area occupied by theselanguages (Blau, 1978; Huehnergard and Rubin, 2011). The traditional tree di-agrams might not be the best way to illustrate the internal relations of languagefamilies (Huehnergard, 2005), especially for the Semitic family, where continuouscontact has facilitated the spread of features across genetic boundaries (Blau,1978; Huehnergard and Rubin, 2011). Many phonological, morphological andlexical features are shared by neighbouring languages, sometimes crossing ge-netic boundaries that have been set up. Thus, language contact could very wellbe the main reason of the close relations in the Northwest Semitic dialect con-tinuum of the early first millennium BCE (Garr, 1985). For the same reason,however, many conclusions are highly ambiguous; for instance, debated varietieslike the ones mentioned above exhibit both Canaanite and Aramaic features,while the characteristics of these two branches depend on the classification ofsuch debated dialects (cf. Huehnergard, 1995). As a result, it is very importantto try to ignore one’s prejudice and approach the data in a open-minded way(Huehnergard and Rubin, 2011).

3

Chapter 2

Northwest Semitic in theSecond Millennium BCE

2.1 The Semitic family

The languages of the Semitic family are among the best-studied languages ofthe world. Like the other great language family of the Ancient World, Indo-European, the relatedness of the Semitic languages had been noticed alreadyin the Middle Ages (Gzella, 2013b). In contrast with Indo-European, the in-ternal classification of the Semitic family has been debated to this day. Eventhough the common origin of the Semitic languages is dated around the sametime as that of the Indo-European languages,1 the contrast between the lan-guages appears to be much smaller and it is much harder to establish differentbranches (Moscati, 1964). This may be caused by the confined geographicalspace in which these languages appear (Lipinski, 1997), as well as the chrono-logical spread of the different languages. Whereas Indo-European has diffusedfrom India to Iceland with several languages that have been attested for overthousands of years, the ‘heartlands’ of all Semitic languages can be found inthe Middle East and the Horn of Africa,2 while only a few languages have beenattested from antiquity into modern times. For these and other reasons it hasbeen proposed that the Semitic language family should be regarded as equal tothe subbranches of Indo-European, e.g. Romance, Germanic, Slavic, with Proto-Indo-European being compared to Proto-Afroasiatic (Lipinski, 1997; Moscati,1964).

The principles of classification, e.g. those introduced in section 1.2, have beenused in Semitic linguistics to draw the family tree for the first time by Hetzron

1The era reconstructed for a Proto-Indo-European cultural and linguistic unity lies between4000 and 2500 BCE (Beekes, 1995; Mallory, 1989), while ‘Common’ or ‘Proto-Semitic’ hasbeen dated in the fourth millennium BCE (cf. Huehnergard and Rubin, 2011; Lipinski, 1997;Saenz-Badillos, 1993). However, the earliest traces of presumed Old-Akkadian loanwords inSumerian are found in the period between 3200 and 2700 BCE already (Streck, 2011a).

2One might refer to Punic, Maltese and various vernacular Arabic varieties in North Africa,but these clearly developed from Phoenician cq. Arabic. These cases can be set against theenormous variety of offshoots of Indo-European languages on every continent of the world,including Brazilian Portuguese, Afrikaans, American English, numerous pidgins and creolelanguages throughout the Caribbean and Oceania, etc. etc.

4

(1972, 1974, 1976). His framework, which is the most accepted nowadays, chal-lenged the traditional idea that Arabic was to be considered a near equivalentof Proto-Semitic (Huehnergard, 2002). Even though a number of modificationshave been incorporated, the basics of his model still stand (Huehnergard, 2002,2005; Huehnergard and Rubin, 2011). The tree in figure 2.1, adopted fromHuehnergard (2005:168), presents the framework in the current state of the art.The details and reasoning behind the most important splits are presented insection 2.1.1.

Figure 2.1: The modern tree of the Semitic language family

Common Semitic

West Semitic

Central Semitic

NWS Arabic OSA

MSA Ethiopian

East Semitic

Eblaite Akkadian

But even though this is the most accepted subdivision of the Semitic languages,there are still many problems. One of the main points is the classification ofArabic and Old South Arabian (OSA, also called ‘Epigraphic South Arabian’ or‘Sayh. adic’); some features seem to link these varieties to Modern South Arabian(MSA) and Ethiopic languages, viz. (1) the shift of PS *p > /f/; (2) extensiveuse of broken plurals (see below); and (3) existence of an L-stem (Huehnergardand Rubin, 2011).3

A totally different alternative has been presented by Lipinski (1997). Be-cause of the unclear distinctions between West and East Semitic in the before2000 BCE, as well as those between Northwest and Southwest Semitic in thefirst millennium, he abandons this framework. He groups together the lan-guages from early northern Syria and central Mesopotamia, including Eblaite, as‘Palaeosyrian’, which is combined with Ugaritic and Amorite as ‘North Semitic’.It provides an interesting view on shared features that have been neglected byothers. But despite the many references to Afro-Asiatic cognates, Lipinski re-frains from using the comparative method, and linguistic reasons for his classi-fication are far from abundant. Within the literature of the current thesis, noneof the scholars has followed Lipinski’s framework.

3A comparable problem in Indo-European linguistics is the so-called centum/satem-opposition (after the Latin and Avestan words for ‘hundred’): a division of the Indo-Europeanlanguages according to the reflexes of the velar series (Beekes, 1995:109-113). It was thoughtthat it split the family into a western and an eastern group, but that view is now abandoned(1995:129). Since the option of an independent development is excluded as well (1995:129),it must have been an areal feature that spreading at a time the affected language groups(Indo-Iranian, Balto-Slavic and Armenian) bordered each other.

5

2.1.1 From ‘Proto-Semitic’ to ‘Proto-Northwest Semitic’

2.1.1.1 Proto-West Semitic

The most important division in the Semitic family is characterised by the de-velopment of the suffix-conjugation in West Semitic. In the original situation,found in Akkadian and Eblaite, there were just verbal adjectives with pronomi-nal suffixes, e.g. Akk nas.r-ata ‘you are/were guarded’ (cf. English ‘your guard-ing’). Such constructions express a perfective aspect or relative past in all WestSemitic languages, becoming a verbal conjugation called ‘perfect’ or ‘suffix-conjugation’ (contrasting with the ‘prefix-conjugation’), e.g. BH nas.arta, [email protected](a), Ar naz.arta, Gz nas. arka ‘you (have) guarded/watched’.

2.1.1.2 Proto-Central Semitic

The subsequent partition of West Semitic has seen many different proposals(Huehnergard and Rubin, 2011), but we will be following Huehnergard (2002,2005) and Faber (1997) in discerning the Central Semitic, Modern South Ara-bian (MSA), and Ethiopian groups as direct descendants of Common WestSemitic. The existence of a Central Semitic branch is warranted by the sharedinnovation of a new prefix-conjugation paradigm based on a pattern yaqtulu inthe imperfective or prefix-conjugation, against Common Semitic yVqattVl: BHyiqtol,4 BA yiqtul, Ar yaqtulu, OSA yaqtulu (Huehnergard, 2005) vs. Akk iqattal,Gz y@qatt@l, Mehri (MSA) y@q´ut@l (Weninger, 2011b).5 This new pattern did notonly change the form of one conjugation, but it also triggered the emergence ofa whole new tense-mood-aspect system, expressed by an alternating final vowel(Huehnergard and Rubin, 2011).

Other features that are shared by all Central Semitic varieties include: theso-called Barth-Ginsberg Law,6 a new formation of the ‘tens’,7 the impersonalinterrogative mah-,8 and two fossilised constructions9 (Huehnergard, 2005).

4Various sound laws have affected the form, such as the spread of the vowel /i/ in the prefix(also in BA), but it is generally accepted that this form must go back to original *yaqtulu(Gzella, 2012d; Jouon, 2006; Moscati, 1964).

5Although Blau (1978) indicated that the original yVqattVl probably goes back to Proto-Afroasiatic, he rejects the idea of Central Semitic, assuming that Proto-Semitic had bothforms to mark the present and/or future. That would mean that yaqtulu in Arabic andNorthwest Semitic is a shared archaism, and of little value for establishing a separate branch.Moreover, his reasoning includes the presumption that OSA is included in a ‘South Arabic’group together with Ethiopic and MSA. There is, however, no reason to postulate the yaqtulupattern for Proto-Semitic, while OSA can be proven to be Central Semitic because of theattestation of it (Huehnergard, 2005).

6I.e. dissimilation of the prefix vowel /a/ to /i/ in G-stem verbs with varying theme vowel,viz. *yaqtul, *yaqtil, but *yiqtal. This whole system is a regularisation from an originalheterogeneous paradigm, cf. Akk taqtul, niqtul. Note, however, that it is not attested inAmorite (see below).

7They are formed as regular (external) masculine plural forms of the corresponding nu-meral, e.g. BH s@losım, OA slsyn /TalaTın/, Ar TalaTuna, OSA TlTy (unclear). CommonSemitic added a suffix *-a, cf. Akk salasa, Gz salasa.

8PS *min(t)-, cf. Akk mınu(m), Gz [email protected] first is BH hallaz(e), Ar allaDı, which consists of a definite article ha-/»al- (not

yet in ‘Proto-Central Semitic’), an asserative particle *la-, and the demonstrative/relativepronoun *Du; “The combination is unusual enough that we may consider these forms to haveoriginated in a common ancestor.” (Huehnergard, 2005:186). The second is constructionof an interrogative particle with a demonstrative directly following, while Semitic languagesoutside the Central Semitic group use the relative pronoun instead of the latter. Thus, BH

6

Some other (presumed) Central Semitic characteristics, e.g. the (unproven) pha-ryngealisation of emphatic consonants (Faber, 1997), are shared retentions orthe result of areal diffusion (Huehnergard, 2005).

Because the OSA evidence regarding the formation of the ‘tens’ and theuse of a demonstrative pronoun after the interrogative particle is not solid, onecould argue that this group branched of from ‘Proto-Central Semitic’ beforeArabic and Northwest Semitic. The shared history of the latter two could thenbe called ‘North Central Semitic’ (ibid.).

2.1.1.3 Proto-Northwest Semitic

The linguistic unity of the Northwest Semitic languages has been accepted for avery long time already, being referred to under the current label for the first timeby Brockelmann (1908). After the discovery of Ugaritic in 1929, this languagewas soon added to this group (Bordreuil and Pardee, 2009; Saenz-Badillos,1993). The most obvious characteristic of these languages is the shift of initial*w- to y-, e.g. BH yeled

¯, Syr yalda, Ug yld ‘child’, vs. Ar walad-, Gz wald ‘id.’,

Akk waldu(m) ‘born’.10

However, such a sound change is not entirely convincing as a genetic dis-tinguishing feature, as we have seen with the shift *p > *f in ‘South Semitic’(footnote 2.1).11 A more significant12 shared innovation is the insertion of avowel /a/ in the plural of monosyllabic qVtl- nouns, resulting in a ‘double plu-ral marking’ (Faber, 1997:9) of the form qVtal- plus external plural morpheme(already noticed by Harris, 1939). Examples include Ug r

»ıs /ra»s-/, pl r

»asm

/ra»asuma/ ‘head’, BH geb¯

er ‘man’, pl g@b¯

arım, BA melek¯

‘king’, pl malk¯

ın.13

It is the only form of the original ‘broken plural’ system, which is very produc-tive in the other West Semitic languages, that survived in Northwest Semitic.Although similar patterns are attested in all other (West) Semitic languages,14

the fact that it is “both restricted and obligatory in qutl nouns” and accom-panied by a external plural marker is a development only found in NorthwestSemitic (Huehnergard, 1991).

Three other features probably can be assigned to ‘Proto-Northwest Semitic’,viz. the assimilation of the first radical of the verb lqh. ‘to take’ to a followingconsonant,15 the metathesis of /t/ in tD- and tG-stems with sibilant-initial

mı ze, ma-zze, etc., corresponds to Ar man Da, ma Da, etc., and Syr man (< *ma den), mana(< *ma dna); these contrast with Akk mannum sa and Gz mannu za-, etc.

10This sound law does not affect the conjunction *wa- ‘and’, as well as the name of theletter *waw ‘hook’.

11In fact, it appears to be an areal feature for Lipinski (1997:115), who classifies Ugariticand Amorite as ‘North Semitic’. Note that an additional attestation of the shift of initial *w-to y- in Safaitic (ANA) as in yrh

˘∼ wrh

˘‘month’ does not occur invariably and in various

positions and is not related to this NWS characteristic (Huehnergard and Hasselbach, 2007).12Thus Huehnergard (1991, 1992b).13Although the inserted /a/ is not visible anymore, the spirantisation of the root-final

consonant indicates that it must have followed a vowel, i.e. *malVk-ın (Rosenthal, 1983:27).For OA and IA, cf. Folmer (2012) and Gzella (2011b).

14For Akkadian, cf. Huehnergard (1987c:183-188).15Ug yqh. /yiqqah. -/, BH yiqqah. , OA yqh. /yeqqah. /. The unusual assimilation of /l/ is

thought to have happened analogous to the /n/ from the antonym ntn ‘to give’ (Folmer,2012; Jouon, 2006), but it also occurs in languages that make use of a root ytn for ‘to give’,e.g. Ugaritic. In OA, the /l/ is more often written than not, but so is the /n/ from I-n verbs.The argument that it is an secondary spelling is supported by the imperative qh. (Beyer, 1984;Gzella, 2011a), cf. Ug qh. /qah. /, BH qah. (as reanalysis of the imperfective yiqqah. : *qqah. >

7

verbs,16 and possibly the prefixing of »a- to the first person plural independentpronoun (Rubin, 2008).17

Note that the ‘Arabo-Canaanite’ of Hetzron (1974), which was characterisedby the feminine plural verb ending *-na18 has been disproved by the attestationof a reflex of this ending in early Aramaic (Huehnergard, 1991, 2005).

2.1.2 The Northwest Semitic languages

Having established the reconstructed history of the Northwest Semitic languagegroup, this section will focus on the diversity within the group. Until the discov-ery of Ugaritic, the internal sub-classification of Northwest Semitic was quitestraightforward: any dialect is either Canaanite or Aramaic (Moscati, 1964;Saenz-Badillos, 1993). This view has been retained by several scholars (e.g.Ginsberg, 1970; Tropper, 1994), even though there are at least three ambigu-ous cases: Ugaritic, Sam»alian (Sm, sometimes called ‘Ya’udic’) and Deir –Alla(DA). The classification of another assumed Northwest Semitic language, theso-called ‘Amorite’, is even more problematic.

2.1.2.1 Amorite

The oldest West Semitic language attested is Amorite. Its corpus does not con-tain a single text, but a very large collection of names and loanwords in mainlyAkkadian texts provide enough grammatical information to conclude that theseelements must come from a West Semitic language. Its name, ‘Amorite’, isderived from the Amorites (Akk Amurru), a people that inhabited NorthernMesopotamia and the Syrian inland from at least 2500 until 1200 BCE (Streck,2011b). One of their main cities was Mari, which had a considerable archiveof texts written in Akkadian cuneiform with a large amount of non-Akkadianelements.

Although the definition of Amorite is often a negative one,19 its linguistic

qah. , cf. yaqtul- : qtul-). NPn nlqh.» and similar spellings are probably caused by reanalysis as

well, cf. Pn mtnt : Ph mtt.16Ug yst

»al /yista»»al-/ (regular yt–dd /yit–addad-/), BH histammer (regular hit

¯qaddes),

IA/OA »zdhrw /»ezdaharu/ (< */-zt-/; Folmer, 2012, regular /»etkattab/), vs. Gz y@sse»alu (<*y@tse»alu; Tropper, 2002). This metathesis only affects original /tS/-clusters and thereforedoes not occur in Arabic (Dt/V.: yataqattalu, Gt/VIII.: yaqtatilu), nor in the Gt-stem ofUgaritic (/yiktatib/, Gianto, 2012), nor in any Ct-stem.

17BH »anah. nu, Ph »nh. n /»anah. nu/, OA/IA »nh. n /»anah. na/, vs. Ar nah. nu, Gz n@h. na, Akknınu. The pronoun is unattested in Ugaritic. Epigraphic Hebrew has only nh. nw /nah. nu/,which occurs very rarely (only five times) in BH. The development is well-explicable as ananalogy to all other first- and second-person pronouns, but a comparable form is only foundin Chadian Arabic: anına (Luffin, 2011). (The loss of initial /n/ in many Arabic and Ethiopicvarieties might be due to the same analogy, cf. Egyptian Arabic ih. na (Dickins, 2011), Amharic@nna (Meyer, 2011).) This feature cannot be established with certainty for ‘Proto-NorthwestSemitic’ without attestation in Ugaritic and suitable explanation for Hebrew nah. nu.

18On the length of the vowel, cf. Hasselbach (2004) and al-Jallad (2014). According to them,the final vowels of the various pronominal suffixes, including the suffix-conjugation endings,were originally short, but by various analogies and complex sound shifts, many appeared asreflexes of long vowels in the attested languages. The only exceptions are the 1csg suffixes *-ı(genitive) and *-nı (accusative), for which a long /ı/ can be regularly reconstructed throughoutSemitic. Henceforth, these vowels will be transcribed as short.

19“In the modern definition adopted here, the term ‘Amorite’ designates the language ofall names and loan words in Akkadian and Sumerian cuneiform texts from the mid-3rd mil-

8

unity is likely for several reasons (Streck, 2000, 2011b). These include theassumption that Amorite is a Northwest Semitic language, a view justified bythe attestation of at least two Proto-Northwest Semitic innovations.20

However, the attestation of the West Semitic characteristic suffix-conjugationis highly debated; Streck (2000, 2011b), e.g., argues that Amorite only had astative inflection like Akkadian. Forms that have been proposed as perfective byothers (e.g. Huffmon, 1965; Mugnaioni, 2000) have to be interpreted otherwise(Streck, 1998b, 2011b).21 A complete absence of the suffix-conjugation wouldof course conflict with the current classification of Semitic (cf. section 2.1.1.1).And as far as we can see, Amorite did participate in all innovations that lead tothe unicity of Northwest Semitic, e.g. prefix-conjugation yaqtul in /Yas.duq-»Il/‘The god/El has proved to be just’ (Gelb, 1980; Huffmon, 1965). Therefore, wemust assume on historical-comparative grounds that Amorite did have a per-fective conjugation, whether it penetrated the native onomasticon or not (cf.Streck, 2000:143-144).

Within the Northwest Semitic language group Amorite is distinguished byits date rather by linguistic characteristics.22 Being one of the most prominentNorthwest Semitic sources in the second millennium BCE, its features will beinvestigated on significant forms in this thesis.

2.1.2.2 Ugaritic

Ugaritic is the first Northwest Semitic language to be directly attested. Since itsdiscovery in 1929 the language has been studied extensively, leading to a fairlygood understanding of the language. A far more problematic question is theclassification of Ugaritic. The main camps consider the language respectively as(‘archaic’ or ‘Northern’) Canaanite (e.g. Ginsberg, 1970; Harris, 1939; Tropper,1994) or as a separate branch of Northwest Semitic (e.g. Blau, 1978; Huehner-gard, 1991, 1992b; Rainey, 1998; Sivan, 2000). The latter opinion is the mostwidely accepted (Gzella, 2011a:427), albeit often added that Ugaritic is “moreclosely related to the Canaanite languages than to Aramaic” (Pardee, 2012:25).The problem is that on either side of the discussion distinctive features are be-ing ignored or regarded irrelevant for genetic relation, while others, supportingone’s point of view, are emphasised as being most significant (cf. Sivan, 2000;Tropper, 1994).

Nevertheless, we will agree with the view that Ugaritic is an independentlanguage for the time being. The reason is that contemporary Canaanite, as at-tested in the El-Amarna letters,23 diverges in a number of features from Ugaritic.

lennium BCE until about 1200 BCE that are Semitic but not Akkadian.” (Streck, 2011b:452)20Viz. initial /y-/ for etymological *w-, e.g. Ya-du/Ya-an-du > *wdd, (Gordon, 1997), and

qVtal-patterned plurals for singular qVtl, e.g. in loanwords as s.amaratu ‘a kind of sheep’ (cf.BH s. emer ‘wool’, s.amrı ‘my wool’) and rababatu ‘ten thousand’ (cf. Ug rbbt, BH r@b

¯ab¯

a ‘id.’;Knudsen, 2004; Streck, 2000, 2011b).

21He states that the suffix-conjugation becomes widespread in West Semitic personal namesonly in the first millennium BCE (Streck, 1998b), even though such forms are clearly attestedin textual sources of the Late Bronze Age already.

22Although it can be noted that the so-called Barth-Ginsberg law, concerning the pre-fix vowel of the various prefix-conjugation patterns (yaqtul, yaqtil vs. yiqtal), is widespreadthroughout Northwest Semitic, but absent in Amorite Huffmon; Jouon’s (1965; 2006).

23In fact, the El-Amarna corpus (14th century) pre-dates the majority of the texts inUgaritic by more than a century (Bordreuil and Pardee, 2009), but in the light of the scarcityof second-millennium Northwest Semitic material, this can be regarded as more or less ‘con-

9

These are, according to Huehnergard (1991):

1. /i/ instead of /a/ in the first syllable of the suffix-conjugation of the D-and C-stems, i.e. *qittil/*hiqtil instead of *qattil/*saqtil (see section 4.2.2);

2. the Canaanite Shift *a > /o/ (see section 3.2.1);24 examples are numer-ous, e.g. h

˘u-mi-tu /h. omit-/ (EA 141:44) vs. Ug h

˘a-mi-ti /h. amıt-/ ‘wall’

(Huehnergard, 1987d);

3. the 1csg pronoun, viz. a-nu-ki /»anokı/ (EA 287:66,69) vs. Ug a-na-ku/»anak˘u/ (see section 4.1.2).

Some other presumed Canaanite characteristics (see section 2.1.2.3) have noattested counterparts in Ugaritic.

2.1.2.3 Canaanite

All Canaanite languages are attested in the first millennium BCE, while onlyHebrew survived until later times. Their contemporaneity facilitates the re-search of the linguistic variety of this group, which has been done by Harris(1939) and Garr (1985). However, most of the features that are said to beCanaanite characteristics cannot be postulated for all varieties.

1. /i/ instead of /a/ in the first syllable of the suffix-conjugation of D- andC-stems, viz. *qittil/*hiqtil, attested in El-Amarna Canaanite, Phoenicianand Biblical Hebrew, no evidence from elsewhere (see section 4.2.2);

2. the Canaanite shift *a > /o/, attested in El-Amarna Canaanite, Phoeni-cian and Biblical Hebrew, but in Ammonite and Moabite not before theseventh century BCE, according to Beyer (2012) (see section 3.2.1);

3. the 1csg ending of the suffix-conjugation *-tı,25 attested in El-AmarnaCanaanite, Phoenician (Punic), Hebrew and Moabite (see section 4.2.1);

4. the generalisation of *-nu in all 1cpl pronominal forms and suffix conju-gation, only confirmed for Hebrew and El-Amarna Canaanite (see section4.1.3 and 4.2.1);

Only the third feature can be assigned with moderate certainty to all Canaan-ite languages, including Transjordanian. It is thus questionable whether thevarieties normally grouped together as ‘Canaanite’ really stem from a singleproto-language.

temporary’.24“Not in itself a significant feature” (Huehnergard, 1991:285), but it is thought to have

triggered the change of the final vowel of the 1c.sg pronoun, see the next point. Anotherargument against the value of this sound change is that Ammonite and Moabite, classified asCanaanite languages, were not affected by the Canaanite Shift in the first place either (Beyer,2012; see below). So if Ugaritic is said to be not a Canaanite language just because of thepreservation of original *a, one has to conclude that Transjordanian is not either.

25It is quite certain that this is a long vowel. First of all, while short vowels are lost inCanaanite, this ending is preserved, and even written in Hebrew and Moabite. Secondly, thereis a clear connection with the pronominal suffixes of the same person *-ı and *-nı, which wereunquestionably long in Proto-Semitic (Hasselbach, 2004; cf. footnote 18 above).

10

2.1.2.4 Aramaic

The attestation of ‘Aramaic’ starts with the so-called Old Aramaic inscriptionsfrom the ninth to seventh century BCE (Fales, 2011; Gzella, 2015). But ratherthan a linguistic unity, the label ‘Old Aramaic’ refers to the earlier stages of thisbranch before the relatively standardised variety of Official Aramaic (Huehner-gard, 1995). It is distinguishable from contemporary Canaanite (Gzella, 2011a,2015), but the innovations shared by all early Aramaic dialects are more in therange of lexicon and phonology than structural or morphological changes, andtherefore not quite convincing as genetic features (cf. section 1.2). As a result ithas been suggested that Aramaic represents Common-Northwest Semitic afterthe split-of of Canaanite and Ugaritic (Beyer, 2012; Gzella, 2014; Huehnergard,1995). These are:

1. preservation of PS *s. , graphically represented first by <q>, later mergingwith /–/;26

2. the change of *n"> *r in the words for ‘son’ and ‘two’;27

3. reduction of *»ah. ad- > h. d /h. ad/ ‘one’;

4. 3m.sg suffix on plural nouns -awh(˘u).28

These developments are the only that set apart Aramaic, Sam»alian and Deir–Alla as one group against the other Northwest Semitic languages. Other char-acteristics can be found in Aramaic, such as the loss of the N-stem, the devel-opment of an article -a», and a feminine plural ending -an, but not in Sam»alianand Deir –Alla (Beyer, 2012; Huehnergard, 1995). The most rational solution tothis problem is to regard these dialects as earlier derivatives from the Aramaic(or ‘Syrian’ or ‘Aramoid’) branch (Huehnergard, 1991, 1995), but the questionremains whether this is the most factual account of the situation and origin ofSam»alian and Deir –Alla.29

26E.g. OA –rq, BA »ara– vs. Ar »ard. -, BH »eres. ‘earth’. The Old Aramaic outcome isrepresented in cuneiform by either <q> or <h

˘>; the realisation /g/ is, according to Beyer

(2012) and Folmer (2012), something similar to [G’] ∼ [K’].27E.g. *bn

"‘son’ (cf. Ar »ibn, BH b@nı ‘my son’) > OA br /bar/, and *Tn

"- ‘two’, often with

dual ending (cf. Ar »it¯

n-ani, BH s@n-ayim) > BA tr-en. A similar though independent shiftis attested in MSA, cf. Mehri b@r ‘son’ and troh/trıt ‘two’.

28OA -wh, but IA -why /-aw-hı/ (Folmer, 2012). The Canaanite languages probably basedtheir suffixes on *-ay-hu (contra Garr, 1985): BH -aw, EH -(y)h(w) /-e-hu/ and -(y)w /-ew/,Byb -w /-ew/, Ph -y /-e-yu/, Mb -h /-e-hu

¯/ (Beyer, 2012; Gzella, 2012c,d).

29Some features, e.g. as the attestation of »nk(y) in Sam»alian and the attestation of aconsecutive imperfect in Deir –Alla, have been assumed to be borrowed from Canaanite (cf.Gzella, 2011a, 2014; Noorlander, 2012). Others, such as the preservation of diphthongs, cannotbe regarded as genetic characteristics, as contraction and the lack of it occur side by side inother closely related dialects, e.g. southern (Judean) and northern (Israelite) Hebrew (Garr,1985; cf. Sivan, 2000).

11

2.2 Historical and geographical setting

2.2.1 Syria-Palestine

In the world of the Semitic languages, with Mesopotamia in the (north-)east,and the Arabic peninsula and Ethiopia in the south(-west) (Moscati, 1964),‘Northwest Semitic’ is a proper label for the languages attested in an area alsoknown as ‘Syria-Palestine’.30 In the broad sense of the word, “‘Syria-Palestine’designates all areas west and north of the Syrian desert in which a NorthwestSemitic dialect was spoken” (Garr, 1985:6-7). Historically, the cultural centre ofthe area is the Mediterranean coast, represented in Northwest Semitic by theearly Ugaritic and prestigious Phoenician. Little is known from the origins ofthe presumably Aramaic-speaking tribes in the Syrian desert in the east, whilethe southern boundary of the ‘geographical unit’ is constituted by Egypt (theSinai desert) and the Arabian desert. In the north, there are the mountainsof Anatolia and North Mesopotamia inhabited by speakers of non-Semitic lan-guages like Hittite, Hurrian, and later Urartian.

2.2.2 A short history

2.2.2.1 Middle Bronze Age (2100-1550 BCE)

The earliest references to the Amorites come from Mesopotamia around the sec-ond half of the third millennium BCE. Their heartland appears to be at the JabalBishri mountain range in central Syria, from where they will invade Babyloniaat the turn of the millennium. The following Old Babylonian Period is char-acterised by Amorite rule of Babylonia. Initially, thousands of Amorite namesappear throughout Mesopotamia (Gelb, 1961), though the amount decreases astime goes on; Streck (2000) ascribes this to a slow process of assimilation to thelanguage and culture of the Babylonians.

In the west, the Levant exhibited an urban system dominated by a number ofcity-states. The Execration Texts show that Egypt had connections with thosein Canaan, but they presumably did not fell into the range of influence of thePharaoh’s of the Middle Kingdom (Rainey, 1972). The Syrian cities, such asYamh

˘ad (later Aleppo), Qat.na, Karkemish, and Alalakh came under Amorite

rule during a second ‘phase of infiltration’ (Streck, 2000:39). The most impor-tant source of Amorite names and lexical borrowings into Akkadian, however, isthe Mari archive, which witnesses a large Amorite population during the eigh-teenth century. The rulers of the cities were confronted with a constant threatfrom neighbouring nomadic tribes, including some with West Semitic/Amoritenames (Gelb, 1961). But at the time a treaty between these two parties wassigned (ca. 1700 BCE), Hammurabi, the well-known Amorite ruler of Babylon,destroys Mari. An Amorite dynasty in Yamh

˘ad/Aleppo survives until the end

of the Middle Bronze Age, but by that time, the Amorites slowly fade into thebackground (Streck, 2000).31

30Exception to this straightforward division of the Semitic languages is Eblaite, spoken inwestern Syria, but probably an East Semitic language (cf. Saenz-Badillos, 1993).

31It is unclear whether the land of Amurru in the El-Amarna texts was really inhabitedby ethnic Amorites. Streck (2000) notes the westward movement of the area with the label‘Amurru’: it starts in Central Syria, but ends up as far west as the borders of the Lebanese

12

Not only the once so powerful Amorite tribes are in decline; Na’aman (1994)describes how the Middle Bronze Age ended violently, with many destructionsthroughout Palestine. Two of the main causes of this downfall include the fall ofthe foreign Hyksos Dynasty, an ally to Canaanite city-states, in Egypt, and theinfiltration and growing domination of a new people in the north: the Hurrians(cf. Gelb, 1961).

2.2.2.2 Late Bronze Age (1550-1200 BCE)

Thus, Syria and Palestine entered the Late Bronze Age as a buffer zone betweentwo superpowers. The Hurrian Mittani-kingdom rapidly grew to its maximum,taking control over most of Syria (including Ugarit), while kings with Hur-rian names are ruling various Canaanite city-states, including Jerusalem (Hesse,2008; Na’aman, 1994). The fifteenth-century campaign by Thutmose III, well-described at his topographical list, is aimed in particular to fight a Hurrian-supported coalition in Northern Canaan (Aharoni, 1976; Na’aman, 1994). Thisrevived interest of the New Kingdom of Egypt towards Asia reaches its zenith inthe El-Amarna letters, a correspondence between various Canaanite and Syrianvassals and Amenhotep IV, alias Akhnaton. At that time (i.e. mid-14th c.),Egypt controlled all of Canaan, parts of Syria, and the Mediterranean coast asfar north as Ugarit (van Soldt, 2010; Wilhelm, 1995).

With the Mittani and Egypt on terms after their struggles, the ‘Age of In-ternationalism’ (Killebrew, 2005) headed for an era of relative peace, had itnot been for the Hittite king Suppiluliuma I, who blasted the Mittani Empirefrom the scene after defeating its Syrian vassals. Some kingdoms in the north-ern Levant that were loyal to the Egyptian Pharaoh defected to the Hittites(Moran, 1992; van Soldt, 2010; Wilhelm, 1995). Although both the Hittitesand Egyptians tried to extend their territories at the cost of the other (cf. theBattle of Qades around 1275 BCE), the situation did not alter much in the lastone and a half century of the Late Bronze Age; Canaan (including Phoenicia)remained under Egyptian control, while western Syria (including Ugarit) wasruled by Hittite vassals (Killebrew, 2005; van Soldt, 2010).

2.2.2.3 Bronze Age collapse and beyond (from 1200 BCE)

The end of the Late Bronze Age witnesses dramatic happenings causing a com-plete change of the scene. The cause of these events, referred to as ‘the BronzeAge collapse’ is probably to be established as a combination of several factorsthat led to massive migrations and chaos (Cline, 2014; Killebrew, 2005). TheHittite Empire was completely destroyed, while Egypt had to give up theirprominent position in the Levant. Many cities in Syria-Palestine, includingUgarit, Emar, Megiddo, and Lakis, were destroyed. Those former city-statesthat were left unaffected or reoccupied grew into independent ‘countries’. Theywould form the basis of the Iron Age kingdoms, although the details of this evo-lution are not yet fully understood (Gzella, 2012b). Another important changeis the arrival of the Philistines (Eg Plst) from the Aegean region with a totallydifferent culture and language,32 establishing themselves in southwest Canaan

Beqaa Valley during the Late Bronze Age.32They are one of the so-called ‘Sea Peoples’, i.e. groups of warriors or complete tribes on

the move in the entire eastern Mediterranean in the beginning of the twelfth century. They

13

(Killebrew, 2005).Due to the crises, the Iron Age started with a period of few textual sources,

which in Greek history became known as the ‘Dark Ages’. The decline of theEgyptian and Hittite powers resulted in new possibilities for the NorthwestSemitic populations of Syria-Palestine (Cline, 2014; Killebrew, 2005). Ulti-mately, it facilitated Phoenicia to become a prominent player on the worldstage, the literary culture of Hebrew to form the Old Testament, while Aramaicwould soon replace Akkadian as lingua franca of the Near East. Although thereis no direct evidence for these developments in the second millennium BCE, thesocio-political, cultural and linguistic landscape that emerged from mid-eleventhcentury onward must have been shaped in these years (Gzella, 2012b).

2.2.3 Language in the second millennium BCE

2.2.3.1 Prestige languages and writing systems

Northwest Semitic languages, appearing to be so prestigious in the first mil-lennium, play a minor role in the Middle and Late Bronze Age. Whereas theSumerian cuneiform system was rapidly picked up by Akkadian and adaptedto that language, the Amorite scribes chose to write their archives in Akka-dian rather than their own language. Similarly, the scribes at Middle and LateBronze Age Palestine prefer to write in Akkadian. The occasional infiltration ofnative West Semitic elements is considered amateurish (Horowitz, Oshima, andSanders, 2006).

Writing was restricted to certain administrative centres with archives in citiesthat had a regional function, e.g. Mari, Alalakh, Hazor, Ugarit. Textual gen-res written at these centres were always official of nature, whether it concernsadministrative and legal texts or diplomatic letters (van Soldt, 2013). Subse-quently, literacy was only granted for trained scribes and not for the commonman. Thus there was ample need to write in the native dialects. Moreover,writing in a common language like Akkadian had obvious advantages in inter-national communication, as is shown by the El-Amarna correspondence.33

Such being the case, it is very exceptional that Ugaritic did not only expressits own language in the syllabic cuneiform, but even more that it started to usethe cuneiform alphabet (cf. Gzella, 2012b; van Soldt, 2013). The emergenceof a new writing system is possibly connected to the appearance of Ugariticepic writing. Being originally composed in the native language, this particular

are sometimes held responsible for all destructions throughout the area, but that might notbe true for all of these peoples, who appear to be of varying European origins (Cline, 2014).

33Although it should be noted that the Akkadian (or more precisely: Babylonian) wasincreasingly influenced by native Hurrian and West Semitic elements, some of which ultimatelybecome the rule in the scribal culture of Syria and Palestine. Because the Mesopotamian EastSemitic dialects developed as well, the language employed outside Babylonia and Assyria iscalled ‘Peripheral Akkadian’ (van Soldt, 2013). The language of the El-Amarna letters isone of the most altered peripheral Akkadian varieties, affecting almost the entire morphology,while most of the Babylonian lexicon is preserved (Horowitz, Oshima, and Sanders, 2006;Moran, 1992). Furthermore, the Hurrites and Hittites already adapted the syllabic cuneiformto their own languages at an early point in time, although the former appear to continue towrite in Akkadian as well. The most obvious reason for the development of a scribal culture ofthese cultures is the fact that their languages deviate in many aspects from Akkadian, whilea West Semitic scribe could pick up the grammar and lexicon of the East Semitic languagefar more easily.

14

genre probably facilitated the development of alphabetic cuneiform writing (cf.Bordreuil and Pardee, 2009).

The cuneiform alphabet is not only attested in Ugarit; within a century, thereare three more ‘texts’ from different locations in the southern Levant written inthis script. Meanwhile, the Mesopotamian syllabic cuneiform totally ceases tobe used by Canaanite scribes at the end of the Late Bronze Age (Gzella, 2015;Horowitz, Oshima, and Sanders, 2006). And as many of the peripheral Akkadianwritings went out from official scribal centres, the loss of such bureaucracies ledto a downfall of written texts in general. When writing reappears in Iron AgePhoenicia, Judea, and other newly formed kingdoms, it does so in a new scribalculture, expressing the well-known Northwest Semitic languages in their own(version of the) Phoenician linear alphabet (Gzella, 2012b, 2015).

2.2.3.2 Language contact

2.2.3.2.1 Hurrian In second-millennium Syria-Palestine there were a num-ber of languages spoken next to Northwest Semitic. The most prominent of theseis Hurrian, a near-language isolate with only Urartian of the first millenniumBCE as attested relative. Originally coming from the north, they establishedthemselves in northern Syria, from where they spread throughout the regionduring the late Middle Bronze Age. Although Hurrian names appear as southas Jerusalem at a certain point in time (Na’aman, 1994, see above), Hurrianlinguistic sources are very scarce and only appear in Syria (Gzella, 2012b). No-table is the attestation of Hurrian names and texts in Ugarit, some of which arewritten in the city’s native alphabetic cuneiform. Except for a handful mostlyhistorical references in the Hebrew Bible, there is no evidence for Hurrian inIron Age Canaan.

2.2.3.2.2 Hittite Second, there is the Indo-European Hittite. The Hittitesalready visited the Fertile Crescent in the sixteenth century, during the cam-paigns of H

˘attusili I and Mursili I, but only managed to extend their Anatolian

kingdom to Syria under Suppiluliuma I (see above). Earlier, Hittite people mayhave migrated into Canaan during the chaotic ending of the Middle Bronze Age(Na’aman, 1994), but there is not much evidence for Hittite-speaking commu-nities outside Syria. Despite the considerable influence the Hittites exposedin north-western Syria and the establishment of Anatolian people in that area,there is only one Hittite text found at Ugarit (van Soldt, 2010). Communicationbetween the satellite-states and the Hittite king was in Akkadian.

2.2.3.2.3 Egyptian During the entire second millennium, Canaan was moreorientated at Egypt rather than the Hittite and Mittani empires. The tale ofSinuhe, set in the mid-twentieth century, tells of an Egyptian official fleeing toByblos and making career in northern Canaan. What is clear from the use ofwords in this text, is that the Egyptian culture was considered superior over theCanaanite (Rainey, 1972). The contempt of the Egyptians towards Canaanitesociety may be the reason why there is so little information on Egyptians livingin Canaan. In comparison with the Hurrians and Hittites, the Egyptians seem toregard their Levantine vassals more as an occupied region and valuable economicsource rather than an expansion of their own state (cf. Na’aman, 1981), i.e. withan ‘imperialistic’ attitude instead of ‘colonisation’ (Killebrew, 2005). Because

15

it is likely that only some officials and military units were stationed in Canaan,the effect of the Egyptian language on Canaanite was probably confined to aminimum.

2.3 Methodological framework

2.3.1 Corpora

Second-millennium Northwest Semitic comes down to us in a number of sources.The majority of these are onomastic corpora, some contain lexical and morpho-logical elements, but only one, the Ugaritic corpus, contains texts with completesentences. All data has been analysed before. In

Execration Texts (e, f /E) These ‘texts’ from the Egyptian Middle King-dom contain names of several localities and their rulers in the southern Levant(i.e. Canaan). They are published in two series, subsequently known as the‘Berlin’ (e: list of rulers, f : list of peoples, Sethe, 1926) and ‘Brussels’ (E ,Posener, 1940) groups. The first is dated around the start of the twentieth

century (i.e. the end of the eleventh dynasty), the second probably originatesin the mid-nineteenth century BCE. Although toponyms and personal namesare not the most useful data, their antiquity provides us a unique insight in theearliest phase of Semitic data in Phoenicia and Palestine.

Amorite Although Amorite names appear as early as the mid-third millen-nium BCE, the great majority of this West Semitic language dates to the OldBabylonian period, i.e. nineteenth to sixteenth century BCE. The onomasticonis collected by Gelb (1980) and analysed by Streck (2000); the few lexical itemsthat were borrowed in Akkadian are found in Knudsen (2004) and Streck (2000)as well. Additional grammatical overviews consulted are those by Huffmon(1965), Mugnaioni (2000), and Streck (2011b).

Egyptian topographical lists (div.) A number of other Egyptian lists pro-vide the names of Syro-Palestinian cities and villages conquered by Pharaohsof the New Kingdom and the early Third Intermediate Period (15th-10th c.BCE). The most important ones date to the reigns of Thutmose III (I ; mid-15th c.), Amenhotep III (Xa; first half of the 14th c.), Ramesses II (XXII ,XXIII , XXVI ; 13th c.), and Shishak/Shoshenq I (XXXIV ; second half 10th c.).They are referred to by their arrangement by Ah. ituv (1984), who in turn basedhis numbering on Simons (1937). Additional commentaries that have been usedare Rainey (1982:list I ) and Edel (1966:list Xa), as well as a number of articles.

El-Amarna letters (EA) The fourteenth-century correspondences of Egyp-tian Pharaohs and officials with their vassals and allies consist for a major partof letters from Canaanite city-state rulers. It is very clear that the languageof these letters is heavily influenced by the native Northwest Semitic idiom ofthe authors, although the exact nature remains a matter of debate (cf. Izre’el,2007:, with bibliography). Nevertheless, numerous glosses and obvious North-west Semitic words serve as the most important testimony of ‘Proto-Canaanite’.The texts are most recently published by Rainey (2015), who also compiled a

16

major work on the Canaanite (or non-Akkadian, cf. Pardee, 1999) elements inthe corpus (Rainey, 1996). Additionally, there is a substantial amount of otherpublications which will be used when necessary.

Ugaritic The only Northwest Semitic language directly attested in the secondmillennium BCE is found in the ancient city of Ugarit. The language is writtenin its own cuneiform alphabetic script, and the city’s ruins provide a relativelylarge corpus of various texts. Although Ugaritic shows unique linguistic featuresin several ways, some interesting details about the language remain obscure (cf.section 2.1.2.2).

Cuneiform in Canaanite (div.) A collection of cuneiform texts from thesouthern Levant have been collected by Horowitz, Oshima, and Sanders (2006).These texts are found throughout the area and contain a few Northwest Semiticglosses and names. For the second-millennium texts, dating as early as thelate Middle Bronze Age, the most important provenances are Hazor (Haz ) andTaanakh (Taa).34

Occasionally, items from other sources are included. This applies to one namefrom Ebla (second half third millennium), two or three toponyms mentionedin the Mari-archives (18th c., also the source of most Amorite data), and a to-ponym attested at Alalakh (level IV, Late Bronze Age). The corpora of theeleventh- and tenth-century arrowheads from Phoenicia (Hess, 2007) are notincluded in the current thesis. First of all, they can be considered as part ofIron Age Northwest Semitic, different from the majority of the second millen-nium. Secondly, the linguistic value of these names is not of additional value incomparison with earlier and later evidence.

2.3.2 Features

The features that will be investigated are selected on the basis of a number ofarguments. The most important of these is that it must be contrastive for atleast one Northwest Semitic branch. Secondly, there must be a reason to assumea development in the second millennium BCE.

Sibilants, interdentals, and laterals The treatment of the interdentalscharacterised the first-millennium Northwest Semitic branches, while the preser-vation of the laterals */s/ and */s./ constituted an isogloss within the Canaanitegroup. The outcome of sibilants is interesting in the light of the Proto-Semiticreconstruction.

Velar fricatives There is some clue that the phonemes */h˘/ and */g/ were

preserved in the languages of the Iron Age, but the Phoenician alphabet wasnot able to represent them. Were they lost during the second millennium?

34A few texts used in this thesis are from other localities: Aphek (Aph), Bet-Shemesh(BSm), and Tabor (Tab). All numbers accompanying the abbreviations correspond to thenumbers of the text in Horowitz, Oshima, and Sanders (2006).

17

Canaanite Shift Although its use for the establishment of genetic relationsbetween Northwest Semitic languages has been debated, the Canaanite Shift oforiginal */a/ to */o/ is an innovation that diversifies the various dialects of thisbranch. From when and where did it spread?

Diphthongs All first-millennium languages are characterised by the contrac-tion or the retention of the original diphthongs */aw/ and */ay/. It is interestingto see how these differences emerged in the earlier stages.

Masculine plural and dual endings There are various plural and dualendings of masculine nouns attested in Iron Age Northwest Semitic. The varietyhas triggered discussions on the classification of some of them, e.g. Sam»alianand Moabite. Are these different endings truly genetic features?

First person singular pronoun Multiple developments have resulted in theBH form »anok

¯ı, but it is still a matter of debate in which order they took place.

Can we trace its origin and its spread in the second millennium already?

First person plural The heterogeneous endings of the 1cpl pronominalforms have been generalised differently in Hebrew and Aramaic. This has beenregarded a significant feature (see above), despite the fact that it cannot beestablished with certainty for other Northwest Semitic languages. At what timedid either of these developments occur?

Personal interrogative pronoun Proto-Semitic *man(n)- ‘who’ was notpreserved in the Canaanite languages Hebrew and Phoenician, which developeda new form. Again, it is interesting to see how early this form appeared andinto which languages it spread.

Suffix-conjugation endings Like the pronominal forms of the first personsingular and plural, the accompanying verbal suffixes are of equal interest. Be-cause in West Semitic, these suffixes belong to verbal morphology, they havebeen awarded a separate section, but the innovations of these endings must beapproached in connection with their pronominal counterparts.

Derived stems According to Huehnergard (1991, 1992a), the vocalisationpattern of the C- and D-stem is a significant feature in the classification ofNorthwest Semitic. At the same time, the prefix of the causative has altered inall varieties of the first millennium BCE, but not in Ugaritic.

2.3.3 Research questions

In order the main question posed in section 1.1, a number of sub-questions havebeen formulated to investigate the relation between second- and first-millenniumNorthwest Semitic. It may be clear that the classification of these languages isthe main theme of the current thesis, and although the following sections willfocus on linguistic developments only, the ultimate goal is to elucidate the evo-lution of the genetic relations of the Northwest Semitic languages. Furthermore,

18

the questions are mostly from a first-millennium perspective, as the Iron Ageis our frame of reference concerning the linguistic variety in this set of Semiticlanguages. They are:

• Which features should be regarded significant for genetic sub-grouping?

• At what time did Aramaic and Canaanite separate? Or did they hardlydiverge before the end of the second millennium and is it impossible tospeak of different branches before that time?

• Should we re-analyse the classification of Ugaritic? Is there any clue thatit was either an Amorite or a Canaanite dialect?

Of course, it applies to all questions that the limited data from the secondmillennium may not provide well-defined answers. One should not expect thatthis thesis takes away all doubt on the classification. After all, the data underconsideration are accessible to everyone that has shed his light on this matter.The most important distinction between this research and earlier publications isthe collection of sources, being as complete as possible with the data currentlyknown. This way, we can pursue the best available image of second-millenniumNorthwest Semitic, its developments, and its linguistic diversity.

19

Chapter 3

Phonology

3.1 Consonants

3.1.1 Sibilants, interdentals, and laterals

3.1.1.1 Proto-Semitic

Traditionally, the Proto-Semitic sibilants and interdentals were reconstructed ina system like the one in table 3.1a. Especially on the basis of Modern SouthArabian evidence, it turned out that *s and *d. were not ‘lateralised’ (which was

even doubted by Moscati, 1964), but real laterals, viz. [ì] and [ì’].1 The affricatepronunciation of the original ‘plain’ sibilant series s—z—s. is based on variouspieces of evidence (Kogan, 2011b), some of which we will encounter furtheron. Subsequently, *s is the ‘normal’ sibilant, and it is questioned whether itsrealisation was a palato-alveolar sound; it is typologically most improbable tohave such a asymmetrical system. Reconstructing it as [s], it possibly shiftedto [S] when *s [

>ts] de-affricated to [s], as in most languages, or merged with

it, like in Arabic. The second-millennium evidence may elucidate the nature ofthis phoneme and its development in Northwest Semitic.

Table 3.1: The system of sibilants in the traditional (a; Moscati, 1964) and themodern (b; Kogan, 2011b) framework.

Fricativet¯

s sd¯

zt. s.

Lateralized? d. s

(a)

Fricative T D T.Affricate s z s.Lateral s s.Sibilant s

(b)

3.1.1.2 First millennium BCE

The short variety of the alphabet, used by all Northwest Semitic languages ofthe Iron Age, has three signs for sibilants: s/ ס! s, x/!|צ s. and S/ ש! s. In con-trast with the longer alphabet, used by Ugaritic and Old South Arabian, there

1In order to create another triad, one could propose that *s and *s. were the unvoiced andglottalised counterpart of *l.

20

Table 3.2: The reflexes of the Proto-Semitic sibilants, interdentals, and lateralsin the documented Northwest Semitic languages.

*s *s *z *s. *s *s. *T *D *T.Ug s s z s. s s. T D T./g4

Ph/EH s s z s. s s. s z s.BH s s z s. s s. s z s.OA s s z s. s q s z s.IA s s z s. s/s – t d t.

are no separate signs for interdentals. However, it is not unlikely that the lim-ited number of signs prompted the scribes of these languages to write differentsounds with the same letter. The lateral sibilant *s appears to be preserved atleast in Aramaic, Hebrew and Moabite,2 although it was originally spelled likes. Table 3.2 presents the spelling of all sibilants and interdentals in the variousNorthwest Semitic varieties.3 As can be seen from this table, the orthographydoes only vary for the laterals and interdentals. However, as has been explainedin the previous section, the phonetic realisation of other phonemes may differper language and period.

3.1.1.3 Second millennium BCE

3.1.1.3.1 */s/ Although the system as proposed in section 3.1.1.1 points toan original realisation as [s], this is not what we find in the earliest West Semiticsources. Ancient Egyptian had both [s] and [S] (Loprieno and Muller, 2012),but they chose the signs for the latter for Semitic /s/, e.g. B-w-t S-m-s-w (E 60)for */Bayt-Samsu/, Ma-s-»a-»»a (E 13) for */Mas»ala/. Hence, we must concludethat the pronunciation was closer to [S]. The use of s-signs for this phoneme ismaintained in later Egyptian spellings of West Semitic toponyms, as well as mostlexical borrowings (Hoch, 1994; Sivan and Cochavi-Rainey, 1992). However, anumber of cases where */s/ is found with Egyptian <s> may represent caseswhere it merged with /s/ (Hoch, 1994).

In Amorite orthography, the outcome of PS *s is spelled with the S-series,confusingly transcribed as /s/. Although the underlying phonetic relation ofthe S-series cannot be established with entire certainty, it is most unlikely thatit was [S], which would have been written with the available S-series. Hence,we may assume that PS *s was still pronounced [s] in Amorite.

Regularly, /s/ was represented by the S-series in the El-Amarna letters. Anexception to this is found in toponyms in the letters from Jerusalem (EA 285-291), where it is written with the S-series.5 This can only mean that /s/ was

2Already in Imperial Aramaic *s is spelled like *s, while *s does not change. BiblicalHebrew shows a difference between /s/ and /s/ indicated by dots: ש! for /s/ and ש! for /s/.A cuneiform transcription of the name /Kamas-–Asa/ as Ka-ma-as-H

˘a-al-ta-a establishes a

lateral pronunciation for Moabite (Knauf and Ma’ani, 1987).3The spellings in Ammonite, Moabite, Edomite and Philistine are similar to those of

Phoenician and Epigraphic Hebrew; Deir –Alla and Sam»alian spellings correspond to OldAramaic.

4The shift from *T. to /g/, affecting only five lexical roots, is most remarkable. No satisfyingexplanation has been found for it yet.

5The three examples are: uruU-ru-sa-lim (multiple occurrences), uruSı-lu-uki (EA 288:42,

46), uruLa-ki-siki, and E-sa-a-an-i (EA 289:20), corresponding to BH Y@rusalayim, Silo, Lak¯

ıs,

21

still pronounced [s] at this time in Jerusalem.6 Unfortunately, other letters orspellings do not invite for conclusions or isoglosses.

Sivan (1984) mentions that there is an interchange of s and s in Ugariticproper nouns, but he suggests that the names with [s] are Amorite. Althoughthis turned out to be the original value rather than a shift, it might prove that*/s/ was pronounced [S] in Ugaritic. Additional evidence comes from foreignwords in Ugaritic: the ‘plain’ sibilant of both Hittite and Hurrian was spelledwith the S-series, but they both appeared in the Ugaritic syllabic script as <T>.7

However, the phonetic value of Hittite and Hurrian /s/ was [s], the S-series beingvery rarely used in both languages. In the alphabetic writing, <T> was proba-bly closest to [s], which means that PS *s must have shifted to [S] at this stage.8

The [s]-value of PS *s surviving into the Jerusalem-letters from the El-Amarnacorpus conflicts with the Egyptian s-spellings in the Execration Texts: uruU-ru-sa-lim vs. Ru-sa-»»-m-m. If we ideally may assume that this place-name wasstable, this contrast can be only explained by assuming that *s in twentieth-century Northwest Semitic was an apico-alveolar, i.e. [s„].9 Furthermore, onlyUgaritic shows a clear shift to [S], while it is probable, though not certain, in‘Common El-Amarna Canaanite’. The letters from Jerusalem, on the otherhand, show that /s/ survived as plain sibilant until at least the fourteenthcentury.

and Bet¯

-S@»an. That only these words show this peculiar phenomenon is not that strange, asthese are probably the only native elements in the Jerusalem-letters; even the glosses, or atleast those with /s/, are Akkadian words, cf. \sa-ak-na-ni (EA 286:10) vs. NWS skn.

6Sivan (1984) proposes to read these names differently, viz. with sa10 instead of sa, and sıinstead of si, but I cannot agree with him on this. The values /sa/ and /si/ are consistentlyspelled with the signs SA and SI elsewhere in these letters, and there are no examples of sa10

and sı (i.e. SA and SI as /sa/ and /si/). Why would the scribe use these different readingsonly in these toponyms?

The case of sa10 is in any case confusing; it appears almost exclusively in the word for‘heaven’ in the Pharaoh’s epithet dUTU is-tu AN sa10-mı ‘the sun god from heaven’. Moreover,this word is spelled 94 times with SA against not more than ten times with SA, while bothspellings occur side by side in EA 232 and 301. For the other seven attestations of SA for/sa/ all occur in regular Akkadian words, while Rainey (1996) has pointed out a number ofcases where S-signs are used for /s/. The interchange of sibilants is very likely to be causedby the confusion of languages underlying the corpus.

7E.g. Hittite /Sut(t)arna/ is Ttrn; the Hurrian sun god /Simigi/ is known as Tmg in Ugarit(DUL). On the other hand, Hurrian and Hittite names with /s/ appear in Egyptian textswith <s>, e.g. Tı-su-pi for Hurrian /Tessub/; H

˘a-tu-si-ra for Hittite /H

˘attusilis/, not <s>

(Cross, 1962).8Cf. also the spelling of /Simigi/ in a lexical list from Ugarit as Si-mi-gi (Del Olmo Lete

and Sanmartın, 2003): syllabic Ugaritic <s> stood for [s] (or [>ts]) and was never used for /T/

or /s/.9Personal experience learns that this sound, occurring in Basque as /s/ and conflicting with

a laminal alveolar [s”] /z/, is perceived differently according to the native phoneme inventoryof the listener. So for speakers of Castilian Spanish, where the ‘plain’ /s/ is pronounced [s„],the Basque /z/ is the marked sibilant, while /s/ [s„] and /x/ [S] are obviously contrastive tothem. In contrast, for speakers of languages like English, Dutch and French (/s/ = [s”]) itwas difficult to hear the difference between /s/ [s„] and /x/ [S], and the sound of /z/ [s”] wasfamiliar to them. The case of Basque, where these two sounds are contrastive, is exceptional,and the most common inventory of two (voiceless) sibilants is [s] and [S], as presumably inEgyptian. Proto-Semitic had only one sibilant [s„], which sounded like Egyptian /s/, but wasjust a ‘plain’ sibilant for most Semitic languages. For this reason and for the sake of clarity,I still will use the transcription [s] for PS *s.

22

3.1.1.3.2 */s/ Consistent spelling in toponyms and lexical items is foundin the Egyptian rendering of West Semitic /s/ as /t

¯/, i.e. a voiceless palatal af-

fricate (IPA [c]). This is one of the main arguments for the affricate-hypothesis.10

The single exception is Su-su (E 57), –-y-n Sa-su (Xa:Cn l.11), if this resem-bles BH sosim ‘plunderers’ (Eg /s/ is normally used for NWS */s/ or */T/, seebelow).

Additional evidence supporting the affricate-theory is provided by Amorite,where */s/ is commonly written with the Z-signs. Since this set most certainlyreflected affricates, Streck (2000) is right to assume a value [

>ts] for this sound.

This phoneme poses no problems in the El-Amarna letters as well as in theUgaritic corpus. In both cases the native /s/ is always reflected by cuneiformS-signs, but this does not necessarily mean that they were de-affricated. ForUgaritic, the foreign evidence again comes to help: Hurrian and Hittite [s] isrepresented in the alphabetic writings as <T>, not as <s>. This cannot beexplained unless Ugaritic /s/ was still an affricate.

The alphabetic cuneiform inscription from Tabor (Tab 1) contains a namethat appears as Pls.-B–l but is interpreted as /Palsı-Ba–l/. It has been suggestedthat this reflects a merger of /s./ and /s/ in the variety expressed at this knifeblade (Horowitz, Oshima, and Sanders, 2006). This is problematic anyhow be-cause loss of glottalisation is quite rare in Semitic languages, and not paralleledin any Northwest Semitic variety of the second or first millennium BCE. Andalthough the interpretation seems justified (the name Pls-B–l is also found inUgaritic), it seems more likely to me that it is a scribal error; note that thedifference between alphabetic cuneiform <s> and <s.> is not much more thanan extra wedge.

As an independent phoneme, PS *s was stable in all Northwest Semitic lan-guages. Its realisation, however, shifted from [

>ts] (or the like) to [s], but this

did not happen in the second millennium. The correspondence of Phoenician s∼ Greek Ξ may indicate it was still a fricative in the ninth century (cf. Millard,2012), even though Egyptian spellings render */s/ with their s-signs from thestart of the first millennium onward (Kogan, 2011b).

3.1.1.3.3 */z/ Just like /s/, the outcome of PS *z was quite fixed and theonly question is to what extend we can reconstruct its phonetic realisation. Itappears in Egyptian transcriptions always as /d

¯/, i.e. [c’],11 as late as the tenth

century (Ah. ituv, 1984). In Amorite, syllabic Ugaritic and the El-Amarna corpusthe Z-series reflects /z/ and shows us that it was affricated.

Thus /z/ did not take a course very different from its voiceless counterpart.Affrication seems to have been present throughout the second millennium, whilethe Greek letter Ζ probably stands for /zd/. Likewise, Phoenician z /z/ maystill have been an affricate.

10In my opinion, this Egyptian spelling also speaks in favour of an apico-alveolar sound.The reconstructed sound [

>ts] resembles [ch], but they are not identical; a more similar sound

would have been the combination of /t/ and /s/, but such writings have not been found.However, if the plain sibilant of Proto-Semitic was a [s„], its affricate counterpart would havebeen ts„] (i.e. Basque /ts/). Just like Semitic /s/, it sounded to the Egyptians more like theirpalatal phoneme /t

¯/ than like an alveolar affricate [

>ts]. On the other hand, while */T/ appears

in Egyptian as /s/, */T./ is written as /d¯/ as well.

11Either the glottalisation is a way to cope with the lack of voicing, or the Egyptian conso-nant was really voiced (Hoch, 1994).

23

3.1.1.3.4 */s./ There is not much to say about PS *s. , as it survived in allNorthwest Semitic varieties. Regardless of the affrication theory, this soundhad most likely an occlusive onset to facilitate the glottalised realisation of theconsonant. Hence, it corresponds to Egyptian /d

¯/ [c’], while it is written with

S. -signs in Amorite, syllabic Ugaritic and the El-Amarna letters.

3.1.1.3.5 */s/ The voiceless lateral is the first sound that establishes iso-glosses in the Northwest Semitic speech area. However, useful evidence fromthe time of the Egyptian Middle Kingdom lacks.12

Amorite /s/, i.e. the consonant spelled with the S-set, can only be verified asa graphemic merger of *s and *s. It cannot be told whether the sign representedtwo independent phonemes (Streck, 2000).

Returning to Egyptian evidence, Late Bronze Age lexical and onomastic datashow that the Northwest Semitic /s/ resembled Egyptian /s/ the most. Hence,it contrasts with /s/, which is written with s-signs. A nice shift can be witnessedin the name of Soko, which is written with s-signs until the tenth-century listof Shoshenq I: Sa-w-ko (XXXIV :38). The city was probably situated on theSharon Plain of western Israel (Ah. ituv, 1984).

Similar to */s/, */s/ shows peculiarities in the ‘Jerusalem-corpus’ of theEl-Amarna letters: while /s/ is spelled with S-signs, /s/ is represented by theS-signs.13 This is the series commonly used for */s/, but it does not contrastwith */s/ (regularly <s>)outside Jerusalem.

The alphabets found in the Arabian peninsula show that there was a let-ter for the lateral sibilant in the original long variety, but it was not used inUgaritic. Both in alphabetic and syllabic writing, */s/ is rendered as <s>, andit is generally assumed that the two sounds merged. The same goes for thecuneiform alphabet from thirteenth-century Beth-Shemesh, where it is one ofthe main linguistic arguments to reject a connection with Ancient South Ara-bian (Horowitz, Oshima, and Sanders, 2006).

The only evidence for a merger of */s/ with */s/ is found in Ugaritic, Amor-ite, and the Beth-Shemesh abecedary. The El-Amarna data provide very littleattestations of this consonant, but those that are found deviate from /s/. Itremains a question whether and when the merger occurred in Phoenician, butit was contrastive in at least Hebrew, Aramaic and Moabite. Possibly, the va-riety used in the fragment from Beth-Shemesh, in the later Israelite Kingdom,joins Ugaritic, but it does not provide much information on a broader isoglossin second-millennium Northwest Semitic.

3.1.1.3.6 */s./ The emphatic lateral [ì’] is perhaps the most unstable soundof the Semitic phoneme inventory. It has no separate sign in neither of thewriting systems used for Northwest Semitic. The only evidence for its existenceas independent phoneme comes from Aramaic (including Sam»alian and Deir–Alla), where it appeared first as <q>, but later as <–>. In all other languagesit merged, at least graphemically, with /s./.

12The connection that has been made between Sa-»»-»ı-nu (E30) and BH Siryon (Ah. ituv,1984; Rainey, 1972) is not certain and problematic.

13The examples are: sa-de4-e ‘field’ (EA 287:56; BH sad¯

e), kur.mesSe-e-riki TN (EA 287:56;

BH Se –ır). Again, Sivan (1984) suggests different readings, viz. SEx and SAx, but this cannotbe justified.

24

One toponym that has a certain etymology with *s. is Hazor, probably fromH. as. ar- (cf. Ar h. ad. ara ‘to be present; to settle, dwell’), which appears in theExecration Texts as H. -d

¯w-

»ı-»»a. The consonant under discussion is written as

Egyptian /d¯/, just like original *s. . The same spelling occurs in later Egyptian

sources, and its phonemic independence from /s./ therefore cannot be deter-mined (Hoch, 1994).

The same problem occurs with the Amorite data, where the Z-series is usedfor *s. . Neither does the use of S. -signs in ‘El-Amarna Canaanite’ provide anyinformation on its phonetic realisation.

Thus, Ugaritic is the first and only language variety of the second millen-nium that exhibits the loss of *s. at the cost of *s. , a development parallel to themerger of *s and *s.14

Graphemically, there is no indication that *s. survived in any second-millenniumNorthwest Semitic variety, except for the Jerusalem sub-corpus of the El-Amarnaletters.

3.1.1.3.7 */T/ The voiceless dental fricative did not have its own sign inthe short Phoenician alphabet. However, Aramaic varieties show that S s wasused for the /T/, which only later appeared as <t>. To what extend otherfirst-millennium languages still preserved this sound is not known.15

Toponyms with /T/ are relatively frequent, and thus the spelling with Egyp-tian s-signs can be considered regular and consequent. Considering that */s/corresponded to Egyptian /s/, /T/ must have been a separate phoneme duringthe entire second millennium BCE. The idea that /T/ merged with */s/, whichwas spelled as Egyptian /s/ as well, must be rejected (Sivan, 1984).

Amorite has also a division of S- and S-signs that points to the preserva-tion of /T/: /s/ is written with S-signs, /T/ with S-signs (Streck, 2000, contraSivan, 1984). All El-Amarna letters display the same orthography with /T/being spelled with S-signs, but because /s/ is only spelled with the S-seriesat Jerusalem the voiceless interdental fricative can only be established as aphoneme in those six letters.

There is no doubt about the existence of /T/ in Ugaritic, which had its ownsign in the cuneiform alphabet. Some scholars think it merged with /s/ at somepoint in time (one of the first being Cross, 1962), based on its correspondenceto Hurrian and Hittite /s/ as well as Egyptian /s/ (all probably [s], see above).In my opinion, this agreement tells more about the nature of Ugaritic /s/ and/s/, which were apparently not as close to [s] as /T/.16 The fact that the‘smaller’ alphabets are attested in some texts from outside the territory of Ugaritmay indicate that /T/ merged with /s/ (< *s and *s) in this presumed earlyPhoenician language (Bordreuil and Pardee, 2009; Millard, 2012).

14In the light of the merger of *s. and *s. , it is remarkable that the non-glottalised *s mergedwith *s, not with *s. This is additional evidence for the fact that *s was not a plain sibilant,and this parallel may indicate that /s/ was still affricated in ‘pre-Ugaritic’.

15Cross (1962:245,n.96) remarks: “At all events it must be insisted that the failure of thePhoenician alphabet to distinguish t

¯[=T], s, and s, has no bearing on the shift of the sibilants

in other dialects.”16Note that Egyptian /s/ corresponded to early Canaanite /T/ and /s/, but not to /s/.

Moreover, in contrast with Hurrian/Hittite /s/ [s], Ugaritic /T/ is never spelled with S-signsin syllabic cuneiform (Huehnergard, 1987d).

25

Of the three other cuneiform alphabetic texts, viz. those treated by Horowitz,Oshima, and Sanders (2006), the one from Beth-Shemesh has signs for both /T/and /s/, while the one from Taanakh (Taa 15) has a <T> that must be read/s/; Tab 1 has no original *T.

First traces of a changing /T/ do not appear earlier than the emergence of theshort alphabet in the twelfth century BCE (cf. Cross, 1962). This same alphabetis probably also used in Taanakh, which might indicate that /T/ shifted with/s/ in the areas later belonging to Phoenicia and Israel. Elsewhere the phonemeis preserved. Fortition to and merger with /t/ as in later Aramaic varieties isexpectedly not found in the second millennium.

3.1.1.3.8 */D/ In most cases, /D/ behaves like its voiceless counterpart.Thus, it appears as <z> in first-millennium Northwest Semitic, but as <d>in Aramaic from the seventh century onward (though historical spellings with<z> still occur).

The phoneme /D/ is particularly rare, and there is no attestation of it inthe Execration Texts. Of all other toponyms that are presented in Egyptiantopographical lists, only */Ma»h

˘aDi/ contains an original voiced interdental.

It is initially written with Egyptian s-signs, viz. M-»ı-h˘

a-si (I :61), but in thethirteenth-century list of Ramesses II it occurs as Mu-h

˘a-d

¯i (XXVI :69). Re-

garding that /z/ was always transcribed with Egyptian /d¯/, this may indicate

that /D/ had merged with /z/ at that time and place (close to Yavne-Yam,Ah. ituv, 1984).

In contrast with this toponym, /D/ appears in lexical items almost as oftenwith /t

¯/ as with /d

¯/ (/z/ only as /d

¯/). Hoch (1994) notes that transcriptions

with /t¯/, which suggests a retention of Northwest Semitic /D/, appear still in

eleventh-century Byblos. He also admits, however, that the ambiguity of both*/D/ and */z/ being rendered as Egyptian /d

¯/ cannot provide more indications

on a merge of these phonemes.Similar ambiguities are found in Amorite, where the independence of /D/

is only established by alternative spellings with D-signs in 16% of the cases ofetymological *D, although the Z-series is far more common (79%, Streck, 2000).It is only by the fact that both spellings occur that we can assume /D/ waspreserved in Amorite, but we cannot be absolutely sure about any shifts inpronunciation.

The toponym treated above appears in EA 298:25 as uruMu-u[»]-h˘

a-zi, andall other instances of etymological *D are likewise graphemically merged withoriginal *z. Thus we cannot decide whether these two also merged phonemically,but at least *D did not become /d/ as in Ugaritic: e.g. Ma-a(h

˘)-h˘

a-di, m»a/

»ıh˘

d‘harbour’. In the earliest stages of the language there is still a grapheme for/D/, but it is replaced by <d> later on. It is widely accepted that *D did indeedmerge with *d, as in later Aramaic.

In the Beth-Shemesh abecedary, the sign for /D/ is absent from the positionwhere it occurs in the Ugaritic abecedaries. However, Horowitz, Oshima, andSanders (2006:159, n.10) note that its position was not entirely set, occurringnear the end in South Arabian alphabets. Because the order does show moresimilarities with the South Arabian version, the authors seem to assume thephoneme was preserved.

26

The variety of voiceless sibilants and interdentals facilitated the distinction be-tween e.g. /s/ and /T/ in the foreign transcriptions of second-millennium North-west Semitic. Unfortunately, there are only two voiced variants and they areoften written in the same way. Ambiguous spellings sometimes point to two sep-arate phonemes, but they complicate the detection of linguistic variety. Ugariticshows that we cannot take over our findings from */T/, so we have to work withwhat we have. That leads us to the following conclusions: (1) in Ugaritic*/D/ merged with */d/; (2) */D/ was preserved at least in proper names inthirteenth-century Phoenicia, but (3) it possibly merged with */z/ in coastalsouthern Canaan (based on Mu-h

˘a-d

¯i and Egyptian lexical evidence).

3.1.1.3.9 */T./ The last interdental is the glottalised variant. Problems intranscription are possibly worse than with */D/ and similar to */s./. As theinterdentals merged with the sibilants in Canaanite (writing) after 1100 BCE,original *T. appears as <s.>; after 700 BCE, however, as <t.> in Aramaic. Similarto the other glottalised continuants (i.e. ‘non-stops’), the phonological realisa-tion of this sound may have contained a small occlusive onset, e.g. [tT’], tofacilitate the glottalisation.

As the other interdentals appeared as sibilants in Egyptian renderings, theclosest glottalised variant is Eg /d

¯/ [c’] (with affrication caused by the glottali-

sation), e.g. D¯

w-»»w-y (E35) for */Θ. uruya/ ‘the Tyrian’. This use is continuedin later times, but one of the two certain lexical items has <t

¯> (= [ch]): r-w»»-t

¯»»

from *rwΘ. ‘to run’. According to Hoch (1994), this implies that it was stillpronounced as a fricative, as */s./ is always represented by <d

¯>.

Streck (2000) argues that it cannot be excluded that PS *T. was preservedin Amorite: while /s./ (< *s. , *s.) is always spelled with Z-signs,17 */T./ is rep-resented by signs from the S. -series in 14% of the cases. Parallel to /D/, wherea small group of D-signs establishes its interdental nature for Amorite, we mayconclude /T./ was present in the language.

But as S. -signs are used for /s./ to an increasing extent, the syllabic cuneiformof the El-Amarna corpus does not provide any information on whether */T./was preserved or not. Luckily, Ugaritic was also written in an alphabetic script,which shows us a special character for what is often transcribed as /z./, butwhich was pronounced [tT’] (Dietrich, Loretz, and Sanmartın, 1975). A veryconfusing and not yet explained development is the shift of *T. to /g/ in no morethan five lexical roots.

In the Beth-Shemesh alphabet, the sign for */T./ is absent, but it appears inthe South Arabian abecedaries as the very last sign.

The evidence on */T./ is rather scarce and highly ambiguous. It is visibly re-tained in Ugaritic and Amorite, but that is as far as it goes for the attestationof this phoneme.

3.1.1.4 Conclusions

The summary of this section is presented in table 3.3. Except for occasionalshifts, e.g. the merger of */D/ with */d/ in Ugaritic, the majority of the dental-

17There is only one out of 28 occurrences of this phoneme spelled with a S. -sign.

27

Table 3.3: The reflexes of the Proto-Semitic sibilants, interdentals, and laterals inthe various second-millennium corpora. Wherever possible, a difference betweenspelling and probable realisation has been made.

MK Eg Amr NK Eg EA EA Jer Ug BSm1 Taa15

*/s/s S s/s S S s s T[s„] [s] [s„]? [?] [s] [S]? [?] [S]

*/s/t¯

Z t¯

S—

s— —

[>ts] [>ts] [>ts] [?] [>ts]

*/z/d¯

Z d¯

Z—

z z—

[>dz] [

>dz] [

>dz] [?] [

>dz]? [?]

*/s./d¯

Z d¯

S. S. s. —s.

[>ts’] [>ts’] [>ts’] [>ts’]? [>ts’] [>ts’] [>ts’]

*/s/ —S s S S s s

—[s/ì] [ì] [?] [ì]? [S]? [?]

*/s./d¯

Z d¯

S. —s. — —

[tì’] [?] [tì’] [?] [>ts’]

*/T/s S s S

—T T —

[T] [T] [T] [?] [T] [T] [S]

*/D/s Z/D d

¯/t¯

Z Z D > d— —

[D] [D] [?] [?] [?] [D], [d]

*/T./d¯

S. d¯

/t¯

S. —T. — —

[tT’] [tT’] [tT’] [?] [tT’]

alveolar fricatives have been preserved throughout the second millennium. Theisogloss concerning the loss of */s/, which was still present in Iron Age Hebrew,Moabite and Aramaic, might have been shaped in the thirteenth century asUgaritic and the variety written at Beth-Shemesh use <s>. A similar develop-ment in Amorite seems unrelated to me.

The interdentals (or at least */T/) did not survive into twelfth-centuryTaanakh, while they also show a growing amount of s-spellings in Egyptianorthography. It is impossible to determine whether the latter reflects a phoneticshift in Northwest Semitic or in Egyptian (Hoch, 1994). There is no evidencefor the interdental fricatives in first-millennium Canaanite, but they did survivein the earliest stages of Aramaic, including Sam»alian and Deir –Alla.

Interesting is the preservation of the affricated realisation of the series s—z—s. .However, we must admit that the writing systems do not always allow us to per-ceive all the phonetic details, which is particularly true for the El-Amarna letters(cf. Izre’el, 2003).

3.1.2 Velar fricatives

3.1.2.1 First millennium BCE

Another characteristic of Iron Age Northwest Semitic is the loss of the velarfricatives /h

˘/ [x] and /g/ [G].18 They seem to merge with their pharyngeal

18Some reconstruct a uvular pronunciation (e.g. Kogan, 2011b), but in line with the velarstop series /k, g, q/ [k, g, k’] it seems most likely to me that these fricatives were velar as well.A third velar fricative, the emphatic /x./ [x’], is reconstructed by Huehnergard (2003) in orderto account for the correspondence of Ar /h. / : Akk /h

˘/. Its existence is not widely excepted.

28

counterparts /h. / [è] and /–/ [Q]. Some transcriptions from the Greek Septuagint,though, may indicate that the velar fricatives were preserved in the earlieststages of Hebrew, if they are not historic spellings (Gzella, 2012d). Similar tothe sibilants and interdentals, the Phoenician alphabet does not allow us toestablish the outcome of PS *h

˘and *g for every Northwest Semitic language of

the first millennium, as it had only signs for /h. / (H/ (ח! and /–/ (o/!ע). Becausethese two phonemes were preserved, only the velar fricatives will be examinedin the following sections.

3.1.2.2 Second millennium BCE

3.1.2.2.1 */h˘/ A fortunate thing is that the Egyptian language had almost

the same series of post-palatal fricatives as Proto-Semitic. Subsequently, Egyp-tian transcriptions can be used pre-eminently for the reconstruction of the velarfricatives of Northwest Semitic. Thus, /h

˘/ is found multiple times in Canaan-

ite toponyms in Egyptian sources from the nineteenth until the twelfth centuryBCE, always corresponding to Egyptian /h

˘/.19

The Mesopotamian cuneiform does not provide such a wide range of gut-turals; there are only H

˘- and Ø-signs. Nevertheless, Streck (2000) reconstructs a

full system of velar and pharyngeal fricatives. Amorite /h˘/ is exclusively spelled

as <H˘>, whereas the outcome of PS *h. (with which /h

˘/ merged in Iron Age

Northwest Semitic) is spelled with both H˘

- and Ø-signs.The same problem occurs in the El-Amarna letters, but there are no clues

in the orthography, as both */h. / and */h˘/ are always spelled with H

˘-signs in

West Semitic glosses en toponyms. Hence, we cannot say whether the scribes ofthese letters still knew the distinction between velar and pharyngeal fricatives.

It is possible to confirm that for the speakers of Ugaritic. In alphabeticcuneiform, there are separate signs for /h. / and /h

˘/, which makes the language

unique among the Northwest Semitic languages that are directly attested.Furthermore, there are some extra cuneiform alphabetic texts from Canaan

that provide information on the retention of the velar fricatives. The thirteenth-century abecedary from Beth-Shemesh has, just like Ugaritic, signs for both /h. /and /h

˘/. On the other hand, the Taanakh text from a century later (Taa 15)

contains the word yh˘

tk. This word is interpreted as a form of the root *h.tk ‘torule, fix, decide’, which implies a merger of */h. / and */h

˘/ (Horowitz, Oshima,

and Sanders, 2006).20

3.1.2.2.2 */g/ The Egyptians did not know the sound [G], but they stilldistinguished it from the voiced pharyngeal fricative [Q]. The latter correspondsalways to Egyptian /–/, while the velar variant is transcribed with either g- orq-signs. Thus we find both Ga-d

¯-tu (Urk.IV 648:10-11) and Qa-d

¯a-tu (pAn I

Furthermore, it is irrelevant for the current research and it will therefore be neglected.19Examples include –-h

˘u-Mwt (E26) for */»Ah

˘u-Mawt/ ‘My brother is Mawt/Mot’ (cf. Ug

»ah˘

, BH »ah. ‘brother’), M-»ı-h˘

i-sa (I :61)/Mu-h˘

i-d¯

a (XXVI a:69) for */Ma»h˘iDa/ (cf. Ug m

»ı/

»ah˘

d‘harbour’, BH »ah. az ‘to grasp, seize’); T»»/»

ı-h˘

i-si (var. from 15th until mid-12th c.) */Tah˘Ti/

(= EAC Tah˘

si, BH Tah. as).20It does not mean that the outcome of this merger was pronounced [x]; we saw in the

previous section that */s/ and */T/ merged in the same text from Taanakh, but it is writtenas <T>. This is caused by the fact that the author just had no idea which of the two signsrepresents which sound; both <s> and <T> sound as /s/, just like both <h

˘> and <h.> sound

as /h. /.

29

27:8) for */Gazzatu/. The few attestations of */g/ in Egyptian topographicallists all use either of these spellings, even as late as the reign of pharaoh ShosenqI (10th century).21

Were the single guttural fricative /h˘/ had its own sign in the Mesopotamian

cuneiform, matters for the distinction between */g/ and */–/ are a bit morecomplicated. Both are represented by either H

˘- or Ø-series, and the existence

of /g/ as a separate phoneme cannot be anything but a reconstruction. Streck(2000) provides two arguments, the second particularly stronger than the first:(1) /g/ is preserved in Ugaritic, which is perhaps the closest relative of Amorite;(2) /h

˘/ is obviously preserved and therefore it is very likely that its voiced

counterpart did as well.In the El-Amarna corpus, we see both H

˘- and Ø-spellings for both /g/ and

/–/ as well, although the Ø-writing for /g/ only occurs with the toponym ofGaza: Az-za-ti (EA 129:84; 296:32) vs. H

˘a-za-ti (EA 289:17, 33, 40; Taa 6:12).

The cuneiform alphabets of Ugarit and Beth-Shemesh contain graphemesfor both /g/ and /–/, so we can safely assume that both sounds co-existed ineither variety.

3.1.2.3 Conclusions

Given that Egyptian sources, being the most capable of recording the post-alveolar fricatives, still show a complete system far into the first millennium,the velar fricatives were probably preserved in second-millennium NorthwestSemitic. The only possible loss of these phonemes, */h

˘/ in particular, is at-

tested in the alphabetic cuneiform of Taanakh. This may point to an isoglossbetween Northern and Southern Hebrew, the latter of which has been provedto preserve the velar fricatives in the Septuagint.

3.2 Vowels

3.2.1 Canaanite Shift

3.2.1.1 First millennium BCE

Thought to have been the main characteristic of the Canaanite language group,the Canaanite Shift of *a > /o/ is attested in both Phoenician and Hebrew.It distinguishes the Canaanite languages from Aramaic varieties, where original*a is preserved (Folmer, 2012). However, Beyer (2012) stated that Moabite andAmmonite, both classified as Canaanite, did not undergo this shift until theseventh century BCE. Any second-millennium evidence on this shift might beeasily linked with these isoglosses.

3.2.1.2 Second millennium BCE

In contrast with Harris (1939) and Sivan (1984), who attribute the emergence ofthe Canaanite Shift to the 16th and 15th-14th centuries respectively, the earliest

21This late date can only be verified for the name of the city of Gaza. The two otherexamples are: M-q-r-tu (I :106) for */Magaratu/ (cf. BH m@–ara ‘cave’, Ug mgrt ‘TN’), whichmust be situated not too far from Gezer, and R-gi-d

¯u/i (I :79, and on copies of this list, Xa:Bn

l.2) for */Rogizi/ (EA 53:36, 56; 191:2), north of Damascus.

30

attestation of the Canaanite Shift can be found as early as the nineteenth-century Execration Texts: S?w-sw */Sosu/ (?) (E57, cf. BH sosım ‘plunderers’or Eg s»»sw ‘bedouins’?; location unknown), and H. -d

¯-w-

»ı-»»-»ı */H. as.or-/ (E15, BH

H. as. or).22 The spread of the shift was very limited, as cities situated relativelyclose to Hazor point to a preservation of /a/.23 Similarly, we find the firsttoponym in the Mari-archives as H

˘a-s. u-ra-a (ARM VI, 78:15), but also Ra-h

˘i-

s. i-im for */Ragis.i-/ (see below).There is no reason to assume that the Canaanite Shift occurred in Amorite,

neither in personal names nor in lexical borrowings into Akkadian (Streck, 2000,2011b).24

In a sixteenth-century text from Hazor itself both H˘

a-s. u-ra (Haz 5:5) andH˘

a-nu-ta ‘Anat’ (Haz 5:1,2)25 appear to show reflexes of the Canaanite Shift.Most proper nouns from Taanakh (15th c.) seem to retain *a (e.g. Ra-h

˘a-bi

/Rah. abi/, Taa 2:22), but a few others do have spellings with <u> (= /o/, e.g.Ru-bu-ti /Rub(b)oti/, Taa 1:26).

From the same fifteenth century onward, Egyptian scribes regularly translit-erate original *a with u-writings. Most of them are located in the north, e.g. inth Beqaa Valley, the Bashan Heights, the Plains of Jezreel and Sharon, and thecoastal areas. Only a few toponyms seem te preserve an /a/: R-b»»-tw (!; I :105,BH ha-Rabba),26 [M]-k-t

»ı (I :2b, BH M@giddo),27 and –»»-k»» (I :47, BH –Akko).28

22Etymological H. as. ar- (Gelb, 1961; Rainey, 1982), cf. Ar h. ad. ara ‘to be (present), attend,dwell’ (Lane 588). Other presumed occurrences of /o/ include –-y-nu (E18, BH –Iyyon), andS-mu-–-nu (E55, BH Sim –on/Simron; Rainey, 1972). In both cases it concerns the suffix*-anu, which is found very frequently in toponyms. However, when the long vowel of thissuffix is backed to /o/, the final vowel regularly dissimilates: *-anu > -onu (Sivan, 1984).This phenomenon is well-attested in the Egyptian transcriptions, e.g.

»ı-s-q-»»-nu (e23-(25),

f 14) */»ATqalanu/ >»ı-s-q»»-ru-n»» (IX :7α5) */»ATqalona/. In case of the two suggested place-

names, the written -nu shows that the suffix was still -anu at this stage. The mu-spellingof */Sam–anu/ reflects an epenthetic vowel which resounds /u/ because of its vicinity to thelabial /m/.

23These include the earlier mentioned */–Iyyanu/ and */Sam–anu/, as well as»ı-bu-»»-m

*/»Abulim/ (E47; later occurrences of this name show dissimilation, e.g.»ıu-bi-r */»Obılu/,

I :99), and somewhat further away»ı-»»-h. -bu */»Arh. abu/ (E14, BH R@h. ob

¯).

24The form rusum has been explained as *ros- < *ra »s- ‘head’, but it stands alone asexample. Moreover, the spelling with an S-sign instead of S (cf. section 3.1.1.3.1) points to anative Akkadian word, and it probably reflects a qutl-pattern (Streck, 2000).

25A problem with the connection with the goddess Anat, proposed by Horowitz, Oshima,and Sanders (2006), is that there is not many additional evidence in favour of a long vowel:

Ug da-na-tu4 and even da-an-t[u4] (DUL 171), EAC A-na-ti (EA 170:43), BH –Anat¯

. On the

other hand, there is Bet¯

-–Anot¯

(Jos 15:59, next to Bet¯

-–-Anat¯

in Jos 19:38), indicating thatforms with /o/ might have existed. Na’aman (1994) takes H

˘anuta as a Hurrian name, but he

does not explain his assumption.26Rainey (1982) reads R(a)-bu-tu, but the script does indicate any vowel in particular; b»»

could be read either /ba/, /bi/ or /bu/ (Edel, 1966). I suggest that it should be read /ba/ here,because of the final vowel of the toponym. As we have seen with the ending *-anu, the casevowel often dissimilates when /a/ becomes /o/ (see footnote 22). A similar development canbe witnessed in the name under consideration: Ru-bu-ti (Taa 1:26), Ru-bu-te (EA 290:11), andRw-b»»-t»ı (XXXIV :13). Since the form attested in Thutmose III’s list exhibits a final vowel -u,the penultimate vowel was probably not /o/, but still /a/. On the location and identificationof this toponym with other occurrences of a similar name, as well as the connection with BH»Arubot

¯, see Na’aman (2000).

27The spelling in list I a may provide a clue on the origin of this name: M-k-t-y. Possiblythe ending -a is the outcome of a contracted triphthong *-aya, while this secondarily formed/a/ was affected not before the tenth century (XXXIV :27: M-k-d-

»ı-w).

28The Execration Texts contain a spelling –-k-y (E49), which can be either the original formof the name, or a form with a nisbeh suffix, i.e. ‘the –Akkayan’, referring to the ruler cursed

31

Regarding the geographic proximity of the latter two, one could opt for a dialec-tal preservation of */a/. That isogloss would encompass the Plain of Acco andthe northern cities of the Jezreel Plain.29 Additionally, there is –-n S»»-w(

»ı)30

(I :5, cf. BH sosım ‘plunderers’), but this is likely to be a form influenced by theEgyptian word s»»sw ‘bedouins’ (Ah. ituv, 1984; Rainey, 1982).

Examples of the Canaanite Shift are abundant in the lexicon of the El-Amarna letters. Two possible exceptions can be interpreted as Akkadianisms,31

all others are proper names. Some northern borders of the spread of the Canaan-ite Shift can be visualised by the name of Damascus-region: the original *»Apu-(cf. E33-34

»ı-pw-m) is found in native letters (EA 197: A-pı), while it appears in

letters from Qat.na (EA 53) and Qadesh (EA 189) as U-pe/ı. In my opinion, thiscannot reflect anything else than the Canaanite Shift being active in southwestSyria.

Despite the relative proximity of Qat.na, Ugarit was not reached by theCanaanite Shift, cf. alphabetic

»ar and syllabic a-ri- for /»ar-/ ‘light’ (BH »or).

The Egyptian sources of the thirteenth century and later show additionalinteresting transliterations: the name of *»Apu (see above) is always written

»ıw-

p»», i.e. with /o/, while the Canaanite Shift is also found in Mw-»ı-b */Mo»ab/ (e.g.

XXII :b10, KRI II 180:2) and its city Bw-t»»ı-r-t»»ı */Botirati/ (KRI II 180:2).Given that other localities in Transjordania also show /o/ instead of */a/ inthe fifteenth and fourteenth centuries already,32 Beyer’s (2010; 2012) view ischallenged.

Additionally, Hoch (1994) remarks that /o/ is the standard in Semitic bor-rowings in Egyptian.

3.2.1.3 Conclusions

The Canaanite Shift is a classic example of an areal spread: the first stage(Execration Texts) only records an /o/ in Hazor, but a diffusion of this feature

(cf. –-q-»»-y */–Aqlaya/, M-–-k-»ı-y */Ma–kaya/, etc.).

29Evidence from the fifteenth to thirteenth century includes: R-h. -bu */Rah. abu/ (I :87,XXVI :77,104; vs. Rehob of the Jordan valley: R-h. w-bw), B»»-y-t-»»ı D»»-q»»-n»» */Bayti Dagana/(XXVII :72, BH Bet

¯-Dagon), and possibly R-b»»-n-t

¯*/Libinatu/ (XXVII :71; cf. BH Sih. or-

Lib¯

nat¯

), which is probably a plural form, considering the Egyptian plural sign. Megiddo andTaanakh are situated in the adjacent parts of the Jezreel Valley. The only toponym in thisarea that does reflect the Canaanite shift is S»»-m-–-n»» */Sam–ona/ (I :35). (Many of thesenames show a lack, or, in the case of the latter, an example of the dissimilation as foundabove, see footnote 22. However, counterexamples are presented here as well, viz. R-h. w-bwfor /Rah. obu/ (but note the labial /b/), and possibly D»»-q»»-n»» for /Dagana/.)

30According to Rainey (1982) erroneous for –En-Sa-si, or one could add –-n Sa-<su>-wi(Ah. ituv, 1984).

31This applies to s.a-pa-ni-su ‘to hide him’ (EA 147:10) and h˘

a-an-pa ’villainy’ (EA 288:7).Both words do not exist in Akkadian, but the first clearly is an Akkadian infinitive-form of aNorthwest Semitic root (Sivan, 1984), while the second is probably a hypochoristicon. Notethat both are not accompanied by a Glossenkeil. The form h

˘a-s. i-ri (EA 138:80, 130) is said

to reflect a qatil-pattern instead of the more usual qatil (ibid.).32For */Bus.rona/, the suffix *-ana is found in the Execration Texts (Bu-d

¯-»»-nu, E27), which

shifted to -ona (B»»-d¯

-<r>-n»», I :23; B»»-d¯

-ru-n»», Xa:Bn r.5; cf. uruBu-us.-ru-na, EA 197:13;199:13). It is less certain for –»»-ru-n»»/H

˘a-lu-un-ni /–Alvni/ (ibid.:< *–ly), but cf. the modern

toponym Nahr el-–Allan (Ah. ituv, 1984). South of Damascus, there is M-s»»-h˘

»»-[na]? (I :25)/Mu-si-h

˘u-na (EA 182:2, 183:4, 184:4) /MoTih. vna/. Because this noun-formation appears to be

an active participle, it goes back to *masih. -, in which case *a did shift to /o/ (Sivan, 1984).Furthermore, there is the geographical name Tu-b-y (I :22)/T. u-bu (EA 205:3) /T. obu/ (cf. Ugt.a-bu).

32

is as good as unattested until the fifteenth century. From that time on it hassuddenly spread from north of Philistia to Transjordania, reaching northwardsas far as the southern regions of West-Syria (viz. Qat.na and Qadesh) duringthe Amarna Age. Notable exceptions, where */a/ is preserved, include theDamascus region and possibly the Plain of Acco33 and northern Jezreel Valley.In later documents, the shift is universal in Canaan, which is in our case is equalto ‘everywhere except Ugarit’.

The fact that we can see this development happening and spreading through-out time confirms the believe that the Canaanite Shift was not a Proto-Canaanitesound law. It spread throughout the area, seemingly regardless of linguisticboundaries.34

3.2.2 Diphthongs

3.2.2.1 First millennium BCE

Another important isogloss of first-millennium Northwest Semitic is the outcomeof the Proto-Semitic diphthongs *aw and *ay. In most languages they contractedto /o/ and /e/ respectively, but they were preserved in the earliest texts ofSouthern (Judean) Hebrew, Edomite, Ammonite (possibly only /aw/, Beyer,2012), and in the idioms used at Sam»al and Deir –Alla.

3.2.2.2 Second millennium BCE

Problematic for the quest for diphthongs in Egyptian transliterations is thatEgypt scribes do use w- and y-signs as matres lectiones, as we have seen mul-tiple times in the previous paragraph. Nevertheless, there are some contrastivespellings that may give us a clue on the realisation of the original diphthongs.

The oldest West Semitic of the second millennium, the Execration Texts,contain a diphthong *ay in the place-names B-w-t S-m-su (E60, BH Bet

¯-Semes),

»»w-w-s-y (E59 BH Layis/Lesem),35 –-n-y (E10, cf. –ain- ‘spring’), and possiblyM-k-t-r-y-n (E5, with ending *-aynu). The w-spelling of the first two toponymsshould be interpreted in the light of the Egyptian sound shift */u/ > */e/(Loprieno and Muller, 2012); it reflects probably a contracted *ay > /e/ (Rainey,1982).36 The two others appear to show a contracted and a preserved diphthongrespectively. The personal name –-h

˘w-Mwt ‘(My) brother is Mot’ could be read

either /»Ah.u-Mawta/ or /»Ah.u-Mota/ (Hoch, 1994; Rainey, 1972).According to Streck (2000, 2011b), the diphthongs were usually preserved in

Amorite. Original *aw sometimes contracts in the vicinity of the labial /m/,37

33Interestingly enough very close to the origin of the Canaanite Shift at Hazor, but geo-graphically a clearly distinct area.

34One could even link it to a similar shift of /a/ to /o/ in Late Egyptian (cf. Loprieno andMuller, 2012).

35If this is not a triphthong */LawiTa/, cf. R-wi-Ta in I :31. It is the old name of the cityof Dan, but it cannot be the same locality as the one referred to as as La-wi/yi-si-imki in theMari archives (contra Rainey, 1982), which must be situated on the road between Aleppo andUgarit (Sasson, 1984).

36Hoch (1994) proposes the possibility of an unparalleled vowel metathesis for what he readsas Bw-t

»ı */Buti/ for */Bıtu/.

37E.g. Mu-zi-ia /Mus.ıya/ (< *ws.»). It also occurs several times in the word *yawm- ‘day’,

e.g. Su-mu-la-u-mu /Sumu-la-yumu/, Yu-mi-i-la /Yumi-»Ila/.

33

while the occasions where *ay appears as a monophthong are considered Akka-dianisms (Streck, 2000).

Moving on to the fifteenth-century Egyptian topographical lists, Edel (1966)

has pointed out that a spelling contrast between y <y> and ¸ or || <i> reflectsdifferent outcomes of original diphthongs.38 Nevertheless, the early data onlyallow for some tendencies: (1) practically all place-names with historical diph-thongs, whether contracted or not, can be located in northern Canaan, i.e. IronAge Phoenicia and Israel; (2) every instance of *bayt- ‘house’ and *–ayn- ‘spring’in a compound name shows contraction;39 and (3) the suffix *-ayn and *-aymare always preserved in the early list of Thutmose III, while two out of three inAmenhotep III’s list have a contracted diphthong.

Both Ugaritic and the Canaanite of the El-Amarna letters show the contrac-tion of *aw and *ay, e.g. *qayT. - ‘summer’ > Ug qT. /qeT.-/, EAC \qe-e-s. i (EA131:15); and *mawt- > Ug mt /mot-/ ‘death’. For EAC there are only exam-ples of unaccented aw, which also contracted in Biblical Hebrew; e.g. *maws. a»

> EAC mu-u-s.a (EA 85:35), BH mos. a (Sivan, 1984). However, it should benoted that all occurrences of diphthongs in the West Semitic of the El-Amarnacorpus originate in northern Canaan, the southern-most provenance being thecity of Megiddo.40

Additionally, the trilingual lexical list from Aphek (Aph 3, ca. 1230 BCE)includes the word \ye-nu ‘wine’ (> *yaynu). The case of mu-mi (Aph 3:1)is explained by Sivan (ibid.) as being an original *mayu- with enclitic -mi.Alternatively, one could see it as *may- with a plural suffix -umi (cf. Ug -uma).Either way, the origin of the contracted vowel was probably a triphthong.

In the increasing amount of Egyptian lists and texts from the thirteenth untilthe tenth century BCE we find a handful of uncontracted diphthongs.41 Again,they do not allow drastic conclusions, but it can be perceived that even mostdiphthongs in the south (Iron Age Judea, Ammon and Edom) are contracted.

38E.g. Tu-t-y-ni (I :9) vs. Du-ti-ni (Xa:Cn l.13) for /Dotayn/, and Ta-y-»ı (I :74) vs. Di-

»ı

(Xa:Bn l.8) for /Tay»i/ ∼ /Te»i/ (cf. uruTa-ia-e(ki) in Alalakh, Rainey, 1982).39The only exception is –-y-n Sa-sw (Xa:Cn l.11) for /–Ayn-Sasu/, but this name is found

with /–En-/ in both Thutmose III’s list and in the El-Amarna letters. Rainey (1975) suggeststhat the Egyptian scribe used a matres lectiones, but that would cast doubt on every instanceof a diphthong. Another option he mentions is that an informant has been involved, in whoseown dialect the diphthongs were retained.

40The form \ia-s. ı-ni (> *ws.h), attested in EA 282:14 from Gath (Iron Age Philistia),may be an unusual spelling for /yos.ı(

»)-nı/ (Sivan, 1984:12, n.13), cf. BH yos. ı»-enı (C-stem).Optionally, it does reflect /yas.ı

»-nı/ with the first consonant of the root elided, as happens moreoften with I-w/y verbs, e.g. Ug

»ard /»arid-/ (Bordreuil and Pardee, 2009) and EH */yarid/

(Gzella, 2012d) from *wrd (forms like BH yered¯

are secondary innovations, Jouon, 2006).This can, however, not be confirmed for C-stems, but the attestation of yshlk /yusahliku/(hlk behaves like a I-w/y verb) makes it likely that the first radical was preserved in thisstem, in accordance with the reconstructions of both Bordreuil and Pardee (2009) and Gianto(2012): /»asos.i

»u/ < *»asaws.i»u. Thus this form remains problematic.41The only occurrences of original *aw are B»»-t H. -wr-rw-n */Bet-H. awronu/ (XXXIV :24)

and Q»»-w-t¯

»ı-s-rw */Qaws-»ATiru/ (XXIII :8, XXVII :103). Especially the first example does

not unambiguously show preservation of the diphthongs, as <w> could also indicate a longvowel /o/ or /u/. The first locality was situated in the Judean mountains, the first nearQadesh in the north.

34

3.2.2.3 Conclusions

In the first place, the data are scarce on the reflexes of original *aw, and thedata available do not allow us to make any conclusions. As for *ay, its con-traction is active in the fifteenth century BCE, having affected all unaccentedoccurrences in the toponyms of Thutmose III’s list (i.e. the construct state of*bayt- and *–ayn-). In syllables under stress, *ay is usually preserved: Dotayna,Mar˘amayma, etc. This is particularly interesting as Biblical Hebrew shows thesame pattern. Later data show some exceptions, but these can be caused bythe ambiguity of the Egyptian transliteration system at those times.

The lexical evidence, appearing from the fourteenth century, only providescontracted diphthongs, but this is not unexpected because of the geographicalprovenance of the data.

Something to note here is the ambiguity of the data. Much more than conso-nants, vowels are dependent of their relative positions. This can vary phonolog-ically among languages, even though they may have the same phonemic vowelinventory. In a system with three original vowels, as in second-millenniumNorthwest Semitic, the phonological range of all three is relatively large; in lan-guages with more vocalic phonemes, vowels must be pronounced more preciselyin order to avoid confusion with a ‘neighbouring’ vowel. Thus, if NorthwestSemitic /ay/ was pronounced something like [

>æi] (or [

>æj]), still far away from

native /ı/ [i:] and thus contrasting, it may be perceived by Egyptians as their

/e/ [e:] rather than /ay/ [>aj] (the palatal coda can be considered a negligible

secondary articulation for front vowels). These phonological details influencedthe choices of the scribes and subsequently blurs our view into Northwest Semiticphonetics.

35

Chapter 4

Morphology

4.1 Nouns and pronouns

4.1.1 Masculine plural and dual endings

4.1.1.1 Proto-Semitic

Although we cannot be sure whether Proto-Semitic had external plural markingin the form of a nasal attached to the case ending (Huehnergard, 2006), it is clearthat the dual originally ended in *-a-n(a)/**-ay-n(a) (nominative/oblique).Furthermore, it is most probable that, in contrast with this nunation of thedual, endings in -m (i.e. mimation) were confined to the singular, as it is at-tested in (Old) Akkadian and Sabaean (OSA; ibid.). With regard to the plural,Akkadian had a vocalic suffix -u/-ı only, while Sabaean did apply an /n/ tothese endings. Note, however, that nunation/mimation is never applied to anynoun in the construct state.

If we take a look at the other Semitic languages, we see that nasal endingswere lost in ‘Proto-Ethiopic’, leaving ample clear traces in G@–@z (cf. al-Jallad,2014). Arabic uses nunation mainly in two environments: (1) as markers ofthe dual (-ani/-ayni) and external plural (-una/-ına); and (2) as marker ofindefiniteness for the singular, e.g. rajulun ‘a man’ vs. ar-rajulu ‘the man’, rajulual-bayti ‘the man of the house’.

4.1.1.2 First millennium BCE

Different forms of the masculine plural and dual have been thought to be distinc-tive in the division of Northwest Semitic: Aramaic languages form their mas-culine plural endings with -ın, Canaanite languages with -ım (Saenz-Badillos,1993). Ugaritic, with mimation, would then be Canaanite, Moabite has -n, andSam»alian, similar to Akkadian, had the vocalic -w /-u/ and -y /-ı/ (Garr, 1985).

So, where we were confident in reconstructing mimation only for the singu-lar, Canaanite and Ugaritic seem to have mimation everywhere except in thesingular. Did Sam»alian lose the mimation/nunation in the plural, or shouldits vocalic endings, in line with ‘Proto-East Semitic’, be considered original? Isthe nunation of Moabite influence from Aramaic or a retention of the commonWest Semitic situation? There are many questions for which the developmentsin second-millennium Northwest Semitic are interesting.

36

4.1.1.3 Second millennium BCE

Mimation is widely attested already in the nineteenth-century Execration Texts,e.g. P

»ı-h. -»»u-m */Pih. ilum/ (E8, cf. later P»»-h. i-ra, I :33) and

»ı-p-qu-m */»Ap(i)qum/

(E9, cf. later»ı-p»»-q»», XXIII :30). Since these toponyms cannot be considered

dual or plural, it is probable that this supports the theory that mimation wasused for singular nouns in early Northwest Semitic, just like in East Semitic (seeabove). Nunation, however, does not occur at this stage yet,1 but there are fewtoponyms that could be interpreted as duals or plurals. Two forms are interest-ing: Sw-sw */Sosu/ (E57) might be a plural form (cf. BH sosım ‘plunderers’),but we cannot exclude influence of Eg s»»sw ‘beduin’ (Rainey, 1972);2 and S-k-m

»ı-m

»ı */Θakmaymi/ (E6), which ending is not the dual, but a phonologically

identical locative suffix (Ah. ituv, 1984; Edel, 1966; Rainey, 1972) added to theroot *Takm- ‘shoulder’ (Ug Tkm, BH sek

¯em).

There are no certain examples of Amorite plural forms, so we cannot de-cide whether it was formed with nunation, mimation, or neither. Strikingly, adual form is found in the toponym /Dimtan/, genitive /Dimten/. The formsare identical to Akkadian (Streck, 2000), but to the reconstructed ‘Proto-WestSemitic’ as well. Forms with mimation are numerous and may still be a suffixfor the singular only. A some point in time, the ending -um became *-uw > -u,a development that appears to be more common in the West than in Babylonia(Golinets, 2010).

From the second half of the millennium on, mimation has disappeared fromthe singular. In toponyms that seem to contain plural or dual marking, as wellas the toponomic suffix *-aym/n, mimation and nunation occur side by side,e.g. H-y-k-ra-y-m (I :89) vs. Hw-y-ka-ra-y-na (Xa:Bn l.10). In the topographiclists of the Eighteenth Dynasty (15th-14th c. BCE), nunation appears fromSouthern Palestine to Phoenicia, while mimation is interestingly restricted toGalilee.3 Sivan (1984) finds a mimated dual-ending attached to an Akkadianword in 2 gisma-ga-ri/e-ma ‘two wheels’ (Taa 2:8), but this and the following2 tugup-pa-as-ia-ni-ma could be Akkadian formations with the enclitic -ma aswell.

In the scribal languages of the Late Bronze Age, however, we see a dif-ferent picture: ‘El-Amarna Canaanite’, the Tanaakh-texts, and Ugaritic useexclusively mimation. The plural endings are the well-attested -uma4/-ıma,the dual -ami/-emi (the first only in Ugaritic). There are two attestations ofnunated plural endings in Ugaritic, viz. bi-da-lu-na /bidaluna/ ‘merchants’ andksmn /kus(s)umu/ına/ ‘spelt’ (Sivan, 2000; Tropper, 2000). Both are also at-tested with mimated endings. Their origin is disputed: Huehnergard (1987d)

1Noun suffixes like -anu are of a different kind; they are not dropped in later stages and,more importantly, they precede the case vowel. It is interesting to note that mimation doesnot co-occur with these suffixes.

2Note that NWS /s/ regularly corresponds to Eg /t¯/ (see section 3.1.1.3.2).

3Of course, this could be just coincidence; only the location of»ı-t-m-m */»Adomım-/ (I :36)

and Ma-r-ma-»ı»»m */Maromaym-/ (I :85) can be established on the basis of their biblical

attestation. The situation of H-y-k-r-y-m */Haykalaym-/ (I :89; with nunation in Xa:Bn l.10)cannot be further specified then ‘northern’ (Ah. ituv, 1984), although the surrounding placesare all Galilean (cf. Aharoni, 1976). The fourth place-name, K»»-mw-ri-m (Xa:Bn l.7), isunidentified (Ah. ituv, 1984), but a northern location is again favoured because of its positionin the list.

4Scarce in the El-Amarna letters, but Sivan (1984) has found sa-mu-ma /samuma/ ‘heav-ens’ (EA 211:17).

37

proposes a dialectal feature, Tropper (2000) suggest phonological dissimilation,while Sivan (2000) sees them as retentions of the original West Semitic suffix.

Nunation still occurs in the Egyptian topographical lists of the thirteenthcentury and later, but it is clearly disappearing from Northwest Semitic. Thelatest example of second-millennium nunation from Canaan can be dated tothe early twelfth century, e.g. K»»-r-m-y-na (KRI II 156:16, 182:6), which isK»»-r-m-y-m a few decades later (XXVII :74). Some names outside Canaan stillpreserve it after that time, e.g. N-h-r-(y)-n for */Nah(a)rayna/ (numerous at-testations, cf. BH Naharayim ‘Mesopotamia’, litt. ‘two rivers’).5 Endings in/n/ are presumably also preserved in Semitic lexical borrowings in Egyptian,as Hoch (1994) concludes that nunation is more frequent than mimation (7:2);vocalic -u/-ı, however, is far more common (21 occurrences), but is not clearwhether these are real plural forms of the absolute state.

4.1.1.4 Conclusions

Coastal Northwest Semitic has generalised mimation in the endings of the mas-culine plural at some time between the eighteenth and fifteenth century BCE(Harris, 1939). Strikingly, this appears to have happened simultaneously withthe loss of mimation in the singular, which is not found outside Amorite and theExecration Texts. But where the El-Amarna letters, Ugaritic, and the Egyptianevidence show endings with /m/ in the plural, mimation in the dual can onlybe verified for Ugaritic. The original dual ending *-ayna is preserved in theEgyptian topographical lists (Hoch, 1994), while there is no attestation of anabsolute state dual in the El-Amarna corpus.

Before mimation evidently has driven away nunation in the Canaanite ofthe early Iron Age (Moabite excluded), Egyptian records from the NineteenthDynasty (13th c.) witness a preference for nunated forms in plural and dual,while lexical borrowings exhibit a greater amount of plural forms in -ın as well.Hoch (ibid.) suggests that a peripheral Northwest Semitic language variety wasthe source of these unexpected forms; he refers to Garr (1985), who draws anisogloss for nunation in Philistia, but unfortunately without substantiating hisaction.6

Moabite could have been one of the peripheral varieties that did preservenunation, but there is no second-millennium attestation from that area.7 Nei-ther is there any form from inland Syria to clarify anything on the generalisationof nunation in these areas. A single exception is *Nah(a)rayna mentioned above,but this would be the expected form in many respects.

The probable developments of mimation and nunation can be summarised asfollows:

5Although this case could be the regular Egyptian name of Mesopotamia, rather than anaccurate reflection of the contemporaneous form in Northwest Semitic. The name, the mostfrequent in Egyptian topographical lists, is attested for the first time in the early fourteenthcentury, and its form is still the same in the seventh-century list of Taharqa (Simons, 1937).Mimated forms occur exclusively in the El-Amarna letters.

6Inscriptions from Iron Age Eqron do not contain any masculine plural suffixes (Gitin,Dothan, and Naveh, 1997).

7Regarding the fact that mimation spread from the north, it is likely that Edomite, situatedeven more to the south than Moab, did preserve the original endings as well. There are,however, no abstract state plural forms found in Edomite in order to establish this.

38

Phase I mimation in the singular, nunation in the dual, nunated endings inthe plural (?)8 (Amorite and the Execration Texts);

Phase IIa mimation in the singular is lost, nunation in the dual and plural pre-served (Egyptian evidence from the 19th Dyn., first-millennium Aramaic,Moabite);

Phase IIb mimation in the singular is lost, but it re-appears in the plural (ina related development or not; fifteenth-century Galilee (?), El-AmarnaCanaanite (?), Egyptian evidence from the 18th Dyn.);

Phase III mimation of the plural is generalised in the dual (Ugaritic,9 Egyp-tian evidence from 20th Dyn., first-millennium Phoenician, Hebrew, Am-monite);

Of course, it is difficult to determine under which conditions mimation andnunation developed in the various languages. In this case, Moabite really wouldbe separated from the other Canaanite languages halfway the second millenniumalready, while Ugaritic shared the innovation with Canaanite. Furthermore, thevocalic endings found in Egyptian borrowings as well as in Sam»alian could beexplained easier if one assumes those to be Proto-Semitic.

4.1.2 First person singular

4.1.2.1 First millennium BCE

The independent pronoun of the first person singular can be reconstructed inProto-Semitic as *»anaku, with a secondary form *»ana (Moscati, 1964; Weninger,2011b). The first one is attested only in Akkadian, Ugaritic and Canaanite.Aramaic and Epigraphic Hebrew only have the second, just like the other WestSemitic languages. Nevertheless, the first form is the most relevant for the cur-rent research. The Biblical Hebrew »anok

¯ı shows that various sound changes

have occurred, and it is interesting to see in which order they occurred. Thereare two possible pathways: (1) final /˘u/ became /ı/ by dissimilation after theCanaanite Shift had occurred (*»anak˘u > *»anok˘u > »anokı, Huehnergard, 1991);(2) final /˘u/ became /ı/ in analogy with the pronominal suffixes -ı, -nı, as hap-pened with the perfective ending -tı (< *-tu, see section 4.2.1).10

In sum, there is Ph »nk /»anokı/ (Gzella, 2012c, contra Krahmalkov, 2001),Mb »nk /»anak˘u/ (Beyer, 2012), Sm »nk(y) /»anokı/ (probably borrowed from

8Cf. Huehnergard (2006). This explains the attestation of masculine plural endings with/n/ in Northwest Semitic, Arabic, and Sabaean, i.e. throughout Central Semitic. Still, heconsiders the masculine external plural ending -an of Ethiopian to be “unrelated”.

9Note that in this framework, the nunation in /kus(s)umu/ına/ and /bidaluna/ could easilyreflect dialectal archaisms (cf. Huehnergard, 1987d; Sivan, 2000).

10Regarding the situation in Moabite, with a perfective ending -ty /-tı/ but a pronoun »nk/»anak˘u/ (Beyer, 2012), the first seems most likely. On the other hand, Hasselbach (2004)and al-Jallad (2014) have shown that it is most probable that the word-final vowels of thepronominal forms were originally short, whereas Hebrew and Sam»alian clearly show that /ı/was long in the first millennium. Thus, an origin in the pronominal suffixes -ı and -nı, whichdid have long vowels, would be more likely. In theory, it is even possible that Moabite »nkreflects /»anoki/ and that the lengthening in Hebrew and Sam»alian (and thus probably alsoin Phoenician) is secondary.

39

Phoenician, Noorlander, 2012). Reflexes of the short variant are found in EH»ny /»anı/ and OA/IA »nh /»ana/.

4.1.2.2 Second millennium BCE

Amorite can be included in the small group of languages that have preservedboth the short and long form of the first person singular pronoun: e.g. A-na-ma-DINGIR /»Ana-ma-»Il/ ‘I am truly God’, A-na-ku-DINGIR-lam/A-na-ku-i-la-ma /»Anaku-»Ila-ma/ ‘id.’ (Gelb, 1980; Mugnaioni, 2000).

The first Northwest Semitic textual sources from the Late Bronze Age dogive us some evidence. The form a-na-ku is by far the most common, but thisis probably the Akkadian form. More interesting is the form a-nu-ki, attestedtwice in a letter from Jerusalem (EA 287:66, 69). This must reflect the nativeform /»anokı/, already showing the Canaanite Shift as well as the change of thelast vowel from /-˘u/ to /-ı/. All proposals regarding the existence of a shortform /»an˘u/ have been rejected by Rainey (1996).11

Alphabetic sources of Ugaritic show both the forms»ank and

»an, the first of

which is also found in syllabic transcription: a-na-ku /»anaku/ (Huehnergard,1987d).

4.1.2.3 Conclusions

All in all, the only significant form attested in the second millennium is thea-nu-ki of EA 287. Since contemporary perfective endings always have -tı (seesection 4.2.1), the order of the sound shifts cannot be established. Moreover, itis impossible to postulate the final vowel of the pronoun outside Jerusalem; intheory, forms like *»anoku and *»anakı could have existed, while the El-AmarnaCanaanite form could reflect */»anoki/; all of these would be explainable in thelight of the developments described above.

4.1.3 First person plural

4.1.3.1 Proto-Semitic

Another division in Iron Age Northwest Semitic concerns the endings of thefirst person plural pronominal suffix and the last syllable of the independentpronoun (see section 4.1.3). Proto-Semitic had *nih. nu as independent pronoun,while the suffix of the same person was originally *-na. In many languages, butnot in Arabic, analogy caused the system to alter, generalising either -a or -u inall pronominal forms of the first person. Akkadian and Eblaite even developeda new genitive suffix -ni (Lipinski, 1997; Moscati, 1964).

11The most convincing example is the parallel of a-nu i-de-su (EA 89:38-39) and a-na-kui-de-su (EA 89:47); Rainey suggests this has to be either a-nu-ki (1996) or a-nu-u (2015).Furthermore, the form *»an˘u is nowhere else attested as first person pronoun. In theory, itcould be the regular Canaanite reflex of PS *»ana, viz. /»ano/, but this is rather improbable.We have to agree with Rainey (1996:I.48) that “more convincing evidence [...] is necessaryto establish the short form in Amarna Canaanite.”

40

4.1.3.2 First millennium BCE

As we have seen, the pronominal forms of the first person plural used two differ-ent final vowels: /u/ in the independent pronoun, /a/ in the various suffixes. Itis obvious that Aramaic, like G@–@z, generalised the latter: BA -na and »anah. na.This contrasts with BH -nu and ( »a)nah. nu/ »anu, considered to be representa-tive for all Canaanite languages (Hackett and Pat-El, 2010). Are there anysecond-millennium forms that allow us to date these generalisations?

4.1.3.3 Second millennium BCE

Amorite has -na as pronominal suffix of the first person plural, e.g. La-na-dIM/Lana-Hadda/ ‘Hadda is for us’ (Gelb, 1980; Streck, 2011b), Is-h

˘i-lu-na */YiT–ı-

»Iluna/ ‘My help is our god’ (Lipinski, 1997; Streck, 2000). The form thuspreserves the original ending. There is no parallel of the independent pronounto prove whether Amorite has generalised this suffix.

There are various spellings of the independent pronoun of the first personin the El-Amarna letters, all indicating the Akkadian nınu.12 However, thehybrid forms of the dative and accusative cases, ia-si-nu (dative) and ia-ti-nu(accusative), add a suffix -nu to the Akkadian forms ia-si and ia-ti (von Soden,1995). In my opinion, this must be influence of the native languages of thescribes, but whether the ending originates in the independent pronoun or thepronominal suffixes is hard to tell. An ending -na can be ruled out anyway.

Far more valuable is the attestation of pronominal suffixes themselves. Beingreconstructed as *-na in Proto-Semitic, the glosses \ru-su-nu /ro(»)su=nu/ ‘ourhead’ (EA 264:18) and \ti-mi-tu-na-nu /timıtuna=nu/ ‘you (pl.) will kill us’(EA 238:33) clearly show that both the genitive and the accusative suffixesshifted to /-nu/ in Late Bronze Age Canaanite. The genitive suffix occurs onAkkadian words, having replaced the original -ni of that language. Notable andsystematic exceptions, where the Akkadian suffix was used, appear in lettersfrom S. iri-Bashani (EA 200), Amurru (EA 170), and Tunip (EA 59). It seemssuspicious that these are all located at the fringes of Canaan, but it is unlikelythat the dialects from these places had -ni instead of -nu or -na. The variantwith /i/ is not attested outside East Semitic, and the only conclusion that can bedrawn is that the authors of these letters stood true to Akkadian orthography.

In Ugaritic, there is unfortunately no syllabic occurrence of either the inde-pendent or the dependent pronoun of the first person plural. The independentpronoun is not even attested in the alphabetic cuneiform, while the vowel of thepossessive -n cannot be determined, but Tropper (2000) states that, because ofa rare form -ny, a vocalisation /-n˘u/ is the most unlikely of all.13

12Interpreting these forms as West Semitic nih. nu is not impossible, cf. ba-t.ı-i-ti for /bat.ih. ti/‘I am at rest’ (EA 147:56) and la-qı-ti (EA 109:30)/la-qı-ti7 (EA 251:1) for /laqih. ti/ ‘I havetaken’. The vowel /i/ in the first syllable of the pronoun is also found in Gz n@h. na (/@/ goesback to *i or *u) and must be original (Lipinski, 1997; Moscati, 1964). The backing of /i/ to/a/ in the neighbourhood of a guttural is such a common sound change Lipinski, 1997 that iteasily could have happened independently in the various Northwest Semitic languages.

13He proposes, on the basis of that same form, a contracted diphthong: /-ne/. Despitehis link with Akk -niasi(m)/-niati(m), this is very improbable in the light of comparativeevidence.

41

4.1.3.4 Conclusions

The data on the pronominal forms of the first person plural are quite meager.Amorite had -na as possessive pronoun, corresponding to Aramaic, but in con-trast with Akkadian (-ni), Hebrew and El Amarna Canaanite (both -n˘u). Thegeneralisation of the latter seems to have been active throughout Canaanite inthe fourteenth century, which supports the idea that the languages of Canaanwere already contrasting with the Aramaic branch at that time. Furthermore,because the majority of the examples of the suffix -nu comes from Phoenicia,we can be sure that the later Phoenician exhibited the same generalisation asHebrew.

4.1.4 Personal interrogative pronoun

4.1.4.1 Proto-Semitic

Throughout the Semitic languages the element *man- is used for the interrog-ative ‘who’, e.g. Akk mannum, Gz mannu, and Ar man. It originally had animpersonal parallel in *min(t)- (Akk mınum, Gz m@nt), but that was replacedin Central Semitic by *mah- (Huehnergard, 2005), e.g. Ar ma-, BH ma-, Ug mh/mahu/, BA ma.

4.1.4.2 First millennium BCE

Of interest is the personal interrogative, as it differs between Aramaic andCanaanite in the first millennium. Aramaic varieties, including Sam»alian, usethe common man- ‘who’. In Ugaritic and Canaanite, however, reflexes of theelement *miya- are found, e.g. BH mı, Ph my /mı/ or /mıya/ (Garr, 1985).14

4.1.4.3 Second millennium BCE

A very interesting personal name found in the western Syrian city of Ebla fromthe third millennium BCE is Mi-ka-il /Mı-ka-»Il/ ‘Who is like god/El?’ (Streck,2011a; cf. Lipinski, 1997). Because of the elements mı ‘who’ and ka like, thismust be a Northwest Semitic name (cf. Akk man(num) ‘who’ and kı(ma) ‘how’,Krebernik, n.d.). Being one of the earliest attestations of Northwest Semitic ingeneral, the antiquity of the interrogative pronoun is astonishing.

The name Ma-an-na-ba-al-ti-DINGIR /Manna-balti-»El/ ‘Who is withoutthe God?’ and its alternative appearances show that the Amorite preservedthe common Semitic personal interrogative *man(n)- (Gelb, 1980). There is notrace of the element *miya-.

Rainey (1996) provides a good overview of the various interrogative pronounsattested in the El-Amarna letters. The most frequent personal interrogative ismıya, which must be the Canaanite element that appears in later Phoenicianand Hebrew.15 Occurrences are found in letters from various provenances, from

14It is unknown how the Canaanite forms relate with Gz mi ‘what’, mostly in connectionwith adverbs, e.g. mi-mat.ana ‘how much?’, mi-ke ‘how?’ (Tropper, 2002, CDG).

15In contrast with the personal pronoun, the impersonal mah-, typical for Central Semitic,is not attested in ‘El-Amarna Canaanite’. The use of mannu for ‘what’ could be influenced bythis element, but according to Rainey (1996) this is mere speculation. Likewise, one could seethe use of mınum, impersonal in Akkadian, for ‘who’ as contamination by Canaanite mıya.

42

Damascus to Ashqelon. Additionally, there are about ten cases of mannu/mannifor ‘who’. Because the majority can be paralleled with an attestation of mıyain a letter from the same locality, these are most probably just Akkadian.

Ugaritic has both mn /mannu/ and my /mıya/, but the second of these isobviously used as personal interrogative pronoun. The first is commonly trans-lated as ‘which’ (Gianto, 2012), but possibly also in the context of ‘whatever,whoever’ (Bordreuil and Pardee, 2009).

4.1.4.4 Conclusion

Occurring already in the third millennium BCE, the interrogative pronoun*mıya is a typical Northwest Semitic innovation. Although we do not knowthe range of its use at that time, it is obvious that it ultimately spread intoUgaritic and Canaanite. The attestation of *man- in Amorite and Iron AgeAramaic show, however, that it is not a common Northwest Semitic form. Sub-sequently, the very early date of the innovative form tempts to assign a veryearly date for a split in Northwest Semitic, but the lack of additional attesta-tions from before the Late Bronze Age complicates the matter.

4.2 Verbs

4.2.1 Suffix-conjugation endings

4.2.1.1 Proto-Semitic

The endings of the perfective inflection are undeniably connected to the pronom-inal suffixes added to nouns. For the current research, we will focus on the firstperson singular and plural.

On the basis of Akk -aku and Gz -ku the ending *-ku can be reconstructedfor the first person singular in Proto-Semitic (Lipinski, 1997). However, Arabicand Northwest Semitic extended the endings with /t/ from the second person,e.g. Ar -tu (cf. Huehnergard, 2005).

The first person plural ending is -nu in Akkadian and Hebrew, but -na inEthiopic and Arabic. The suffix-conjugation probably had the ending -nu, whilethe suffixes indicating the direct object of a verb and the possessor of a nounwere expressed by *-na. This situation was preserved into ‘Proto-NorthwestSemitic’ (Huehnergard, 1991).

4.2.1.2 First millennium BCE

Of these two endings, only the first person plural is visibly contrasting in first-millennium Northwest Semitic: BH -nu vs. BA -na. Phoenician and Ugaritic -nprovide no information on the vowel of the suffix. For the first person singular,the ending -tı is certain for (Biblical) Hebrew, Moabite and Phoenician, but avowel (probably /u/) was lost in Biblical Aramaic16 and unknown in Ugaritic.

16Because of the loss, it can be argued that the vowel remained unchanged, i.e. *-tu: sincethe new ending -tı was based on the pronominal genitive and accusative suffixes of the firstperson singular -ı and -nı, it was most certainly long (cf. Hasselbach, 2004). Only final shortvowels are commonly lost throughout first-millennium Northwest Semitic, while word-final /ı/was preserved and written in Aramaic, cf. OA/IA -y and -ny for the 1csg pronominal suffixes.

43

4.2.1.3 Second millennium BCE

4.2.1.3.1 First person singular The discussion on the existence of a per-fective conjugation in Amorite has been dealt with above (section 2.1.2.1). Re-garding Streck’s (2000) point of view, his rejection of the first person perfectiveending -tı found by Huffmon (1965) in Ya-sa-ar-ti-dingir /Yasartı-»Il(u)/ isunsurprising. In his view, this form is a female abstract noun ‘(my) Justice’. Itis indeed problematic to regard this as a finite verb form for semantic reasons,let alone as a first singular.

On the other hand, the ending -ti is extremely common in the El-Amarnaletters, providing nearly one hundred instances of this Northwest Semitic suffix.This cannot be an Akkadian form, thus it must reflect the actual Canaanitemorpheme. Three possible instances of a suffix *-tu have been disproved assuch by Moran (1992) and Rainey (1996).

Alphabetic Ugaritic has -t, but its vowel is unattested. Nevertheless, mostscholars (Bordreuil and Pardee, 2009; Gianto, 2012; Tropper, 2000) recon-structed */-tu/.17

The only certain perfective form in Egyptian sources contains a first personsingular suffix, viz.

»ı-b»»-t»ı /»ab(b)adtı/ (Hoch, 1994).18 The exact linguistic

origin of this thirteenth-century phrase is unknown.

4.2.1.3.2 First person plural The lack of a first person plural suffix pre-cludes any need for remarks on the moot subject of the perfect in Amorite atthis punt.

Only one example of a first person plural perfective form can be found in theEl-Amarna corpus, viz. da-na-nu-u16 /dannanu/ ‘we are strong’ (EAC 362:27).However, the root dnn is not attested in any Northwest Semitic variety, andthe ending could be Akkadian as well.19

There is no attestation of this ending in Ugaritic (Tropper, 2000).

4.2.1.4 Conclusion

The results of the search for perfective suffixes in second-millennium NorthwestSemitic unfortunately do not provide anything more than the attestation of -tı

Moreover, the other endings of the perfective paradigm became lengthened in Aramaic, e.g.2fsg -ty /-tı/. This may appear as a problem for the assumption of an ending -tu in ‘Proto-Aramaic’, but there are various ways to account for a special status of word-final /u/: (1)short /u/ one of the most unstable vowels of the Semitic language, especially when it is thefinal sound of an utterance (e.g. Gz *u > /@/, lost in word-final position); (2) long /u/ is verymuch associated with the plural throughout (Northwest) Semitic; (3) the vowel of the ending-tu has no parallel in the prefix-conjugation, while there is, e.g., */tiktubı/ for 2fsg;.

17Tropper (2000:465) gives three arguments: (1) the independent pronoun ends also in/u/ (see above); (2) Canaanite /i/ is a secondary development; and (3) the 2fsg perfectiveending is, on comparative basis, reconstructed as */-ti/, and it is improbable that the suffixeswere identical. The last is interesting, but equal endings for these two persons are found inCanaanite, e.g. BH w-yarad

¯tı ‘and you will descend’ (Ruth 3:3, qre: w-yarad

¯t) vs. w-yarad

¯tı

‘and I will descend’ (Num 11:17, Jdg 11:37, Neh 6:3); for Phoenician/Punic, see Gzella (2012c)and Krahmalkov (2001) (both reconstruct -tı for both endings.)

18Semantically, a D-stem is favourable, but we would expect at least an /i/ in the secondsyllable. The form remains disputable.

19Moreover, the perfective form of II-III verbs in Biblical Hebrew shows a different vo-calisation: sabbonu, which probably goes back as far as a pattern similar to Akkadiandannanu (Jouon, 2006). Ugaritic has both consonants when the suffix starts with a con-sonant: *sababna/u (Gianto, 2012), similar to BA –alalna (Rosenthal, 1983).

44

in Late Bronze Age Canaanite. Therefore, no dates can be postulated for thedevelopments that diverged Northwest Semitic in the first millennium BCE.

4.2.2 Derived stems

4.2.2.1 Proto-Semitic

Two features of the morphology of the derived stems are relevant for the currentresearch, viz. the consonant of the C-stem prefix, and the vocalisation patternof the perfective forms of the C- and D-stem. It is generally assumed thatthe first was originally *s-, attested in Akkadian, Ugaritic and Ancient SouthArabian (Moscati, 1964; Weninger, 2011b). The sibilant shifted to /h/ in mostWest-Semitic languages, including most Northwest Semitic varieties, lenitingeven further to /»/ in later Aramaic, Arabic and Ethiopic.

The vocalisation pattern of the suffix-conjugation20 is definitely less clear.Akkadian has u-u in D-stem quttul- and C-stem suqtul- stative inflections, thelatter of which is the basis of the El Amarna hybrid form suqtil- (Huehnergard,1998, contra Rainey, 1996 and Tropper, 1994). These vowels are not found inWest Semitic, where both G@–@z and Arabic have qattala/»aqtala. However, thisis not the ‘Proto-West Semitic’ form, as is shown by Northwest Semitic evidence.

4.2.2.2 First millennium BCE

All Northwest Semitic languages favour for an /i/ in the second syllable inboth stems: Ug */qattila/, */saqtila/ (Tropper, 2000), BH qittel/hiqtıl, BAkatteb

¯/haqtel. Therefore, it is most likely that the original West Semitic forms

were identical to the ones reconstructed for Ugaritic.21 Hebrew and Phoenicianassimilated the first vowel to the second: *qittil/*hiqtil.22 For want of anythingbetter, reconstructions of the vocalisation of this form in the Transjordanianvarieties are based on comparative evidence (cf. Beyer, 2012 vs. Vanderhooft,1995).

20In the prefix-conjugation there are no important differences among the Northwest Semiticlanguages. Most interesting is the prefix vowel, which is reconstructed for Ugaritic as both*/yu-/ (Gianto, 2012) and */ya-/ (Bordreuil and Pardee, 2009). The reconstruction of /u/ inthe first syllable is supported by Arabic yuktabu and Akkadian usaqba/ir, but the only solidevidence, for the first person singular, shows an /a/: e.g.

»ansq /»anassiq/ ‘I will assault’, and

»asspr /»asaspiru/ ‘I will make [you] count’ (ibid.). Tropper (2000) suggests that the original/u/ was reduced to [@], which was coloured by the glottal stop of the 1csg prefix to /a/.

As for the other Northwest Semitic forms, they all correspond to each other: BH yaqtıl,Ph */yaqtil/ (Gzella, 2012c), BA yaqtel. The /a/ probably comes from the causative prefixha-, as is shown Aramaic forms like BA y@haqtel (Folmer, 2012). There are some imperfectiveforms of the C-stem found in the El-Amarna letters, but most of them present difficulties withthe prefix vowel (Sivan, 1984). As this is not very significant for Iron Age Northwest Semitic,we will ignore them in this thesis.

21Vowel assimilation from a-i to Gz, Ar a-a is more probable than the vowel dissimilationof presumed a-a to a-i (Huehnergard, 1992a, contra Bordreuil and Pardee, 2009). Moreover,this is such a natural development that it could have happened independently in Arabic andEthiopic. Note also that these two languages point to an /i/ in the D-stem prefix-conjugationand participles as well: Ar yukattibu and mukattibu, Gz y@kett@b and makatt@b.

22Evidence for this shift in Phoenician is provided by (1) the palatalisation of the causativesuffix (see below), and (2) (Neo-)Punic spellings and transcriptions in the Latin alphabet,e.g. Pn h. ds for /h. iddis/, NPn bycys for /biqqis/ ‘he requested’ (both D-stem); Pn ypyq» for/yipıqo/ ‘he found it’. In later times, /yi-/ developed further into /»i-/: Pn »yb» for /»ibı(»)/‘he braught’, NPn intseb for /»ins.ib/ ‘he erected’ (Krahmalkov, 2001).

45

We find three different causative morphemes in Iron Age Northwest Semitic:y- in Phoenician, »- in Imperial and later Aramaic, and h- in Hebrew, all Tran-sjordanian languages, and early Aramaic varieties (OA, BA, Sm, DA). The firsttwo obviously derive from the latter; Phoenician /y-/ is caused by palatalisa-tion (*/hi-/ > /yi-/, Garr, 1985; Gzella, 2012c), while the loss of /h-/ in laterAramaic is paralleled in Arabic and Ethiopic.

4.2.2.3 Second millennium BCE

4.2.2.3.1 C-stem morpheme Gelb (1980) lists various Amorite nameswith prefixes s- and st- of the C- and Ct-stems. The problem with these is thatAmorite /s/ is the outcome of PS *T, while Ugaritic shows that the causativeprefix must go back to PS s (Amorite /s/). Some time before Gelb, Huffmon(1965) promoted a number of prefix-conjugation verb forms as C-stems on thebasis of their vowels and the preservation of the first radical of I-w/y.23 Thelack of a C-stem morpheme implies it was a ‘weak’ consonant, like /h/ or /»/ (cf.ibid.:71, fn.47). The causative morpheme *s is preserved in all environments inthe languages that use it, e.g. Ug

»ass.

»u /»asos.i

»u/ (< *ws.», Gianto, 2012).

However, Streck (2011b) stresses that a ‘H-causative’ is highly improbable.There is indeed a clear parallel in the development of the phoneme *s in theC-stem morpheme and in the pronominal forms of the third person singular.In Akkadian, both appear as /s/. Outside lexical roots *s lenited to /h/ (andfurther) throughout West Semitic (except some OSA languages). Ugaritic, how-ever, has causatives with /s/, but pronominal forms with /h/. In that light it isunlikely that Amorite, preserving the sibilant in the independent pronouns /su/‘him’ and /sı/ ‘her’ (Mugnaioni, 2000; Streck, 2011b; Golinets, 2010 recentlydisproved the use of U for a possessive suffix /-hu/), would have lost it in thecausative. Streck (2011b) cites two possible forms of C-stem derivatives with/s/ ([s] < *s), viz. Sah. baru, Sah. baratum ‘Ally’ (< *h.br, cf. Ug h. br ‘compan-ion’), and Saklalu ‘Perfect(ly made)(?)’ (< *kll, cf. Ug skllt ‘emaciated’, litt.‘perfect’).

The gloss in EA 256:7, sent from Transjordanian Pella (Pih˘ilu), is the most

famous example of a West Semitic C-stem in the Late Bronze Age. It reads h˘

i-ih˘

-bi-e, to be interpreted as /hih.bi»a/, a parallel to BH heh. bı(») (Huehnergard,1992a:219, n.43). A second one has been found by Greenstein (2004) in a letterfrom Hazor (Haz 10:19): h

˘e-te-qu for /hi–tiqu/ (cf. BH he–tıqu). In both cases,

the causative morpheme is h-. Furthermore, it can be noted that the /h-/is syncopated in imperfective forms, e.g. ti-mi-tu-na-nu /timıtuna=nu/ < *t-hamıtuna ‘you (pl.) will kill us’ (EA 238:33).24

Even though the sound shift */s/ > /h/ has affected its pronouns of thethird person, Ugaritic retains the archaic s-prefix. Suspicious forms that gaverise to the early assumption that the language exhibited both a saf–el and ahaf–el/»af–el are all rejected (Tropper, 1990).

Interesting is that Semitic loans in Late Egyptian contain more instances ofC-stems with a sibilant than with a /h-/ in the prefix. All attestations postdatethe thirteenth century, but it should be noted that two out of three occurrences

23E.g. ya-aw-si-bu /yawTibu/ corresponds to BH yoseb¯

(< *yawTib-), contrasting to theG-stem forms Ug

»ard /»arid-/ (1csg); Biblical Hebrew forms like yeseb

¯have developed from

a secondarily constructed *yaysib Jouon’s (2006). Cf. chapter 2, footnote 40.24The /i/ in the prefix is caused by vowel assimilation, according to Sivan, 1984.

46

of the prefix /sa-/ are written with Egyptian s-signs instead of the expecteds-signs. Considering that the origin of the loans cannot be retrieved, we cannotjump to too many conclusions. Furthermore, Hoch (1994) also finds some formsof a CtD-stem, which do have a h-prefix instead of their causative-only parallels;they are, however, not unproblematic.25

4.2.2.3.2 Vocalisation pattern of the D- and C-stem The nominal C-stem forms provided by Streck (2011b) cited above do not provide informationon the vocalisation of the suffix-conjugation (if it exists in Amorite), while thereare no potential perfective forms of the D-stem either.

The Late Bronze Age examples from Pella and Hazor, on the other hand,show the Canaanite pattern i-i. The spelling of a vowel /e/ in h

˘e-te-qu proba-

bly reflects allophonic lowering of /i/ in the neighbourhood of the pharyngeal/–/ (Greenstein, 2004). No suffix-conjugation forms of the D-stem have beenidentified (yet).

For Ugaritic, we only have a D-stem in sa/sal-li-ma (Huehnergard, 1987d),showing an /a/ in the first syllable.26 Regarding that the vowel patterns of theD- and C-stem agree with each other throughout the Semitic language family, wecan be confident in reconstructing a pattern a-i for the Ugaritic causative as well.Furthermore, III-y verbs preserve their final radical in the prefix-conjugation ofboth stem formations, which happens only after a vowel /i/ (Tropper, 2000).

The data from Egyptian loanwords, like the Amorite words, only containnon-finite forms. The transcribed

»ı-b»»-t»ı (see above), if reflecting a D-stem

*/»abbadtı/, would be unique in Northwest Semitic with a pattern a-a.27 How-ever, if Arabic and G@–@z applied vowel harmony to these forms independentlyfrom each other, so could a Northwest Semitic variety. In that case, a Canaaniteorigin can be considered impossible for this phrase.

4.2.2.4 Conclusions

The sibilant *s of the prefix of the C-stem is still found in Amorite and Ugaritic,and there are no traces of a lenition to /h/ before the two Late Bronze Age ex-amples from Israel. In the same two forms, we witness the Canaanite innovativepattern i-i, which is unattested elsewhere. Ugaritic has preserved the originala-i. No areal spreads or moving isoglosses can be postulated for the two featuresof the derived stems.

25The attestation of a CtD-stem is highly remarkable. Hoch (1994) does not hesitate tolink it to the Hebrew hitpa–el, but this stem-formation probably corresponds to the commonSemitic Dt-stem, while the initial /h/ is only added secondarily (Gzella, 2012d; Jouon, 2006;Moscati, 1964). And where Hoch states that all these forms exhibit a metathesis of theprefixed /t/, infixation of this morpheme is more frequently attested throughout Semitic. Iwould think that there must be a different explanation of this forms, but the treatment ofsuch difficult transcriptions would go beyond the scope of this thesis.

26Bordreuil and Pardee (2009) proposes qittala (but C-stem saqtila), because of»ıhb

/»ihhaba/ ‘he loved’ (but this form is regarded a G-stem by DUL). According to them,/sallima/ is an Akkadianism, while Ugaritic shares with Canaanite a dissimilation i-a from‘Proto-West Semitic’ a-a. Aramaic then contrasts with a dissmilated a-i. However, the ev-idence of the reconstructed causative form and the theory of the more probable vowel as-similation instead of dissimilation are in favour for Huehnergard’s (1992) reconstructions (cf.Tropper, 2000).

27Possibly, such a pattern existed as allomorph in Ugaritic as well (ibid.).

47

Chapter 5

Conclusions

At the very beginning of the current research we have posed the following ques-tion: which developments can we perceive in the Northwest Semitic data fromthe second millennium BCE and what do they tell us about the classificationof this branch? Being a twofold question, it will be answered in two differentsections. The first presents the most interesting developments we have foundin the previous chapters, the second puts them in the perspective of the clas-sification of Northwest Semitic in general. The sub-questions of section 2.3.3,although not being particularly addressed, are answered in that section.

5.1 Summary of attested developments

In accordance with the expectations, some features cannot be checked for everystage of Northwest Semitic, while the foreign nature of the corpora does notcontribute to the value of the forms. Nevertheless, there are some interestingresults which can be of use in the classification of Northwest Semitic.

We have seen that the Proto-Semitic dental-alveolar fricatives in general havebeen preserved throughout the second millennium BCE. Exceptions include theloss of */s/ in Ugaritic, Beth-Shemesh, and possibly also in Amorite, as well asthe merger of */T/ and */s/ in twelfth-century Taanakh, and of */D/ and */d/in Ugaritic. However, many details remain obscure, in particular with regard tothe emphatic series. Neither of the attested developments seems significant forthe classification of the Northwest Semitic languages.

Similarly, the survival of the velar fricatives in all second-millennium sourcesdoes not help us very much further. A merger of /h

˘/ can only be established

for the alphabet used in Taanakh in the twelfth century.The Canaanite Shift, presented as a typical example of a spreading inno-

vation, can be traced back to nineteenth century Galilee (Hazor). It diffusedfrom there, ultimately affecting all idioms between Moab and Qat.na. Possibly,it took some time to spread to northwestern Israel, but only Ugaritic preserves*/a/ at the end of the Late Bronze Age.

The data on the diphthongs has turned out to be too complex and ambiguousto draw any conclusions.

Another very interesting development is the spread of mimation. First at-

48

tested in northern Israel, it quickly extends to most of Canaan as well as Ugarit.There is no evidence on the situation in Moab.

Pronominal forms lack in widespread attestation, and there are hardly anyconsequences we can connect to the few forms attested. The altered form»an=økı, the generalisation of -nu for the 1cpl forms, and the suffix -tı areall attested in the El-Amarna letters, but they cannot be put in a broaderpicture.

Interesting is the third-millennium occurrence of the interrogative pronoun*mı(ya) ‘who’, which is also found in El-Amarna Canaanite and Ugaritic.

The original S-causatives are still attested in Amorite, Ugaritic, and someunknown sources of Egyptian lexical borrowings. Only two forms attest a prefixhi- and a pattern i-i in the derived stems, the earliest dating to the sixteenthcentury.

5.2 Consequences for classification

Besides linguistic features, the only link we can establish between the variousfirst-millennium Northwest Semitic languages and their predecessors from thesecond millennium is their geographical location. Even though we know thatvarious migrations took place during the Bronze Age collapse, we have to trustthat the speech variety spoken in e.g. Jerusalem was the same before and afterthat time.1 Thus, we have searched for Moabite features in second-millenniumMoab, Southern Hebrew in the area around Jerusalem, Phoenician in the areaof modern-day Lebanon, etc. All of these are considered ‘Canaanite’ and willbe treated further on. On the contrary, no Aramaic characteristics were foundto the east.

5.2.1 Aramaic

It is striking that regions that are Aramaic-speaking areas in the Iron Ageappear to participate in linguistic changes that are thought to be Canaanite,e.g. the Canaanite Shift, the use of mıya, plural suffix -ıma, 1csg ending -tı,etc. Apart from an occasional attestation of their name,2 the Arameans seemto be completely absent before the start of the Iron Age.

It has been proposed that, similar to the Amorites, the Arameans consti-tuted (semi-)nomadic tribes that invaded the Levant from their homeland inCentral-Syria, near the Jabal Bishri mountain range (cf. section 2.2.2; LawsonYounger Jr., 2007). They are thought to have taken advantage of the powervacuum that arose from the Bronze Age collapse. Within a few centuries, theyestablished or took over a number of kingdoms throughout Syria and northernPalestine, one of the most important being Damascus (Gzella, 2015; Pitard,

1In some cases, we do know that a different people moved in, e.g. the Philistines in south-west Canaan (see section 2.2.2.3).

2E.g. Eg p»»-»ı-r-mu ‘land of the Arameans’ (Xa:Dn l.7, pAn III 5:5) and Akk eqletmes

a-ra-mi-ma ‘fields of the Arameans’ (from Ugarit; Edel, 1966; Lawson Younger Jr., 2007;Pitard, 1987). Whether these are really connected to the Arameans of the first millennium isan ongoing debate; the first indisputable occurrence is in Tiglath-Pileser I’s annals (between1116 and 1076 BCE; Lawson Younger Jr., 2007; Pitard, 1987).

49

1987). To what extent the Arameans form a ethnic unity with the groups men-tioned in the late second millennium cannot be established with certainty, buttheir language is clearly distinct from other (contemporary) Northwest Semiticlanguages (Gzella, 2015).

Now that we know that the Arameans were absent from the (linguistic) North-west Semitic scene throughout the second millennium BCE, the question is whatimpact it has on the classification of the Northwest Semitic languages. Interest-ing in this matter is to see whether any of the second-millennium developmentsdid reach the Aramaic language; is there any reason to assume (intensive) con-tact between Aramaic and other Northwest Semitic varieties?

Most of the observed changes, however, lack in Aramaic during the firstmillennium. Only two shared innovations are found:

• loss of mimation in singular nouns;

• lenition of */s/ to /h/ in the prefix of the C-stem.

Both of these are obvious areal features and common enough to be found inother West Semitic languages as well. Thus, we can conclude that the ‘Proto-Arameans’ were distant enough from the Levantine linguistic centres to be un-affected by most of the changes. The claim that Aramaic and Canaanite formeda linguistic unity until the end of the second century (see above; Saenz-Badillos,1993) is hereby untenable.

It is nevertheless hard to establish a date for the linguistic separation ofAramaic; various arguments can lead to different points in time. The sixteenthto fifteenth centuries, with mimation in the plural and dual, an i-i vowel patternfor derived stems, and a rapidly spreading Canaanite Shift, can be set as aterminus ante quem. A much earlier date could be feasible in the light of third-millennium attestation of *mı(ya) ‘who’, but we should not draw too drasticconclusions on the base of such an isolated finding alone.

5.2.2 Amorite

Among the Northwest Semitic branches, Amorite is clearly different from theother languages in its many archaisms, e.g. the preservation of mimation onsingular nouns. There is no significant innovation that would combine Amoritewith any other Northwest Semitic language, as a development like the presumedmerger of */s/ and */s/ is too common to be meaningful for purposes of clas-sification.

5.2.3 Ugaritic

The most difficult case in terms of the classification of Northwest Semitic isUgaritic. The most significant innovation we encounter in Ugaritic are thefollowing:

• loss of */D/, fusing with /d/ (observable within the corpus);

• application of mimation in dual and plural (only two examples of nuna-tion);

50

• use of my /mıya/ for ‘who’ (vs. mn /mannu/ ‘whoever, whatever’).

The first cannot be a ‘Proto-Ugaritic’ phenomenon, as it happens before oureyes. The other two are attacked as such by Sivan (2000), but neither veryconvincingly. The attestation of mn does not necessarily devalue the signifi-cance of this innovation. Moreover, /mannu/ is only occasionally attested, andits meaning seems to be very specific (DUL; Tropper, 2000). Considering theEblaite attestation of *mı(ya), the date of this innovation might be searched forin the third millennium, and that would point to a shared history of Ugariticand Canaanite. Moreover, it reduces the possibility that language of the city ofUgarit is closely related to or even a dialect of Amorite, as the latter does notuse *mıya as personal interrogative.3

The second and last feature that Sivan (2000) treats concerns the mimatedendings. He refutes the argument by presenting the two nunated forms pre-sented above. But whatever their origin, they are not at all representative forUgaritic. As I have proposed, mimation spread around the sixteenth centuryfrom northern Israel, and it is in theory possible that there was a dialectal vari-ety of Ugarit that was not affected by it. However, if mimation is really such ayoung innovation, it cannot be considered significant for linguistic classification.We have seen in section 2.1.2.2 that fourteenth-century Canaanite differed inmany aspects from Ugaritic, and the split of these branches must have happenedsome time before that. The sixteenth-century attestation of a hiqtil-form at Ha-zor vs. Ug saqtil may be one of the earlier isoglosses that separated Canaanitefrom Ugaritic (cf. Huehnergard, 1992a).

5.2.4 Canaanite

As we have set apart the other Northwest Semitic languages, Canaanite remainsas an undisputed, independent branch. There are quite some developmentsvisible in the second millennium that set it apart from the rest:

• as good as universal adoption of the Canaanite Shift;

• application of mimation in dual and plural;

• the altered form »anokı of the 1csg pronoun;

• 1cpl suffix -nu in analogy to the independent pronoun;

• replacement of original *man(n)- with mıya ‘who’;

• perfective ending 1csg changed into -tı in analogy with pronominal forms;

• C-stem morpheme h-;

• a vocalisation pattern i-i in the D- and C-stem.

Therefore, Canaanite can be considered the most innovative of the NorthwestSemitic branches and it has obviously set itself apart from the others in thesecond millennium already. Some of its innovations are shared by Ugaritic, butthe developments that are unique in Canaanite are the most interesting for now;

3But this is very hard to substantiate any further, due to the considerable distance in timebetween the two languages.

51

these can tell us how the Ugaritic and Canaanite branches relate to each other.However, three of them, viz. the 1csg pronoun »anokı, the C-stem prefix h andthe derived-stem pattern qittil/hiqtil, are only attested in one or two examples.It is not possible to establish their spread in the second millennium, and thuswe cannot state whether they are representative for Canaanite. These develop-ments, however, do allow us to state that Canaanite was at least beginning todifferentiate from Ugaritic from the sixteenth century on. Note that mimationpresumably spread from around the same time as well, and did reach the cityof Ugarit.

Concerning the fourth, the famous Canaanite Shift, we can observe its spreadthroughout the southern Levant (i.e. Canaan) already in the fourteenth century,reaching as far as Qat.na in the north. It is likely that Ugaritic was at the timealready distinct enough to remain unaffected by the shift, creating an obviousisogloss that can be verified for all of Canaan, including Transjordania.

5.3 Wave theory vs. family tree

The question that remains unanswered is: which features should be regardedsignificant for genetic sub-grouping? In the past, various attempts have beenmade to bring the Canaanite languages together as descendants of a single‘Proto-Canaanite’. The current thesis has shown once more that this is not aneasy task. Huehnergard (1992a) makes a great deal of his ‘Proto-Canaanite’sound-law that triggered the i-i vocalisation pattern of the D- and C-stem, butthis can only be confirmed for Phoenician and Hebrew. Its postulation in Trans-jordanian languages is only based on their classification as Canaanite, which is acircular argument. The mimation of the masculine plural endings, proposed byTropper (1994), amongst others, did not spread to Moabite (and Edomite?), butdid affect Ugaritic, while the Canaanite Shift, lacking in Ugaritic, is attested insecond-millennium ‘Moabite’ (contra Beyer, 2012). However, both innovationsvisibly diffuse during the early Late Bronze Age BCE. Huehnergard’s sound-lawis first attested around the same time, but its ‘pan-Canaaniteness’ cannot beestablished. Then what makes Moabite a Canaanite language and Ugaritic not?

This is exactly the grey area between the family tree and wave theory. Tak-ing snapshots at certain points in time, as with comparing e.g. El-AmarnaCanaanite, Ugaritic, and Moabite, results in obvious differences and facilitatesthe drawing of a family tree. This has been common practice in many cases,such as the greater Semitic and Indo-European families, while we are very oftenguided by historical and geographical factors as well. In the current case, how-ever, there is no evidence for sudden changes that trigger the separation of alanguage group. On the contrary, the best attested innovations diffuse through-out the area, disregarding presumed genetic boundaries. But since we simplycannot know whether second-millennium Moabite would have had a hiqtil orsaqtil causative, or whether Iron Age Ugaritic would have been affected by theCanaanite Shift, it is impossible to be sure of the genetic classification of theselanguages.

As an outcome of these considerations, I would propose that Canaanite’sbranching off from its predecessor, whether Ugarito-Canaanite (Tropper, 1994)or common Northwest Semitic, was not a sudden event. Rather than that, the

52

innovations gradually diverged the speech varieties of Canaan from contempo-rary Northwest Semitic languages, such as Ugaritic. In this growth, there wasan obvious core area in later Israel and Phoenicia, where the languages wereaffected by most changes. But not every innovation reached to peripheral areas,such as Moab. It is true that Late Bronze Age Ugaritic differs from contem-porary ‘core-Canaanite’, but so does Moabite in its own era. In this view, theclassification of both languages is a highly arbitrary decision. And it has turnedout to be so, as scholars seem to chose random features to confirm their conclu-sions.

The thing we should learn from this discussion, in my opinion, is that the clas-sification of languages cannot depend on one or a few linguistic characteristicsalone. Geographic, ethnic, and historical arguments can be of great help, butin absence of these, one must not hasten to draw any conclusions. As for theconnection between Canaanite and Ugaritic, it is obvious that ‘core-Canaanite’underwent a set of changes from the late Middle Bronze Age on, but whetherthis is representative for Canaan in its entirety is unknown. Perhaps it is moreaccurate to see the Late Bronze Age Levant as a dialect continuum, similar tothe Canaanite-Aramaic diversity of the first millennium BCE. Garr (1985) hasshown that it is difficult, if not impossible, to draw firm lines, and his analysisof linguistic centres and peripheries may be a scheme we can postulate for thesecond millennium as well.

53

Epilogue

Further research and desiderata

The intriguing second millennium BCE provides interesting clues on the devel-opment of the Northwest Semitic languages. In the previous section, we haveseen that the answers also raise new questions. As we were able to date some ofthe innovations that were believed to be of significant value for the classificationof the four branches, it turned out that the drawing of an actual family tree hasbecome hardly any easier.

For instance, there is no obvious evidence that precludes a relation betweenAramaic and Amorite. Both underwent almost no drastic changes and theircommon origin in the Syrian desert, about which very little is known frombefore the Iron Age, is tempting. Previous suggestions on a common originhave been rejected mostly for the sake of their name (cf. Lawson Younger Jr.,2007). I do not say the Arameans were Amorites or the other way around, buta shared cultural and linguistic history is plausible for as far as we have seen inthis paper.

The most welcome addition to this topic is, of course, new texts and at-testations of Northwest Semitic elements, in particular pronominal and verbalforms. Other fields of interest could be involved more than has been done inthis study, e.g. the historical and cultural background and the origin of theNorthwest Semitic people in the Levant.

The subject of this thesis will probably never cease to be am interesting andhighly debated topic in Semitic linguistics. My analysis hopefully has shed somelight on it, although it does not provide very many new answers. Nevertheless,it is always a better thing to add useful observations to a discussion, rather thanto draw non-substantiated conclusions.

Acknowledgements

First of all, I would like express my sincere gratitude to my supervisor, prof.dr. Holger Gzella, for his contagious enthusiasm during the last years, as wellas for his helpful guidance during the process of this thesis; also to the secondreader, dr. Alwin Kloekhorst, who already inspired me during the first years ofmy bachelor; to my mother, who profoundly went through my work; and, lastbut not least, to my fiancee, family, close friends, and my housemates for theirmoral support, interest, patience, and coffee. Thank you.

54

Bibliography

Abbreviations

AOAT Alter Orient und Altes TestamentAO Analecta OrientaliaBASOR Bulletin of the American Schools of Oriental ResearchBZAW Beihefte zur Zeitschrift fr die alttestamentliche Wis-

senschaftHSS Harvard Semitic StudiesSH Scripta HierosolymitanaUBL Ugarit-Biblische LiteraturUF Ugarit-ForschungenAF Altorientalische ForschungenBA the Biblical ArchaeologistBiOr Bibliotheca OrientalisDS Dutch StudiesEALL Encyclopedia of Arabic Language and LinguisticsEHLL Encyclopedia of Hebrew Language and LinguisticsHAR Hebrew Annual ReviewHTR Harvard Theological ReviewIEJ Israel Exploration JournalIOS Israel Oriental StudiesJAOS Journal of the American SocietyJBL Journal of Biblical LiteratureJCS Journal of Cuneiform StudiesJNES Journal of Near Eastern StudiesJSS Journal of Semitic StudiesKUSATU Kleine Untersuchungen zur Sprache des Alten Testa-

ments und seiner UmweltLLC Language and Linguistics CompassRlA Reallexikon der Assyriologie und Vorderasiatischen

ArchaologieTA Tel Aviv

Aharoni, Yohanan1976 – The Land of the Bible: A Historical Geography. Trans. by

Anson F. Rainey. London: Burns & OatesAh. ituv, Shmuel

1984 – Canaanite Toponyms in Ancient Egyptian Documents. Jerusalem:The Magnes Press

55

Albright, William Fox1940 – “Review of (Harris, 1939)”. In: JAOS 60.3, pp. 414–4221943 – “Two Little Understood Amarna Letters from the Middle Jordan

Valley”. In: BASOR 89, pp. 7–17Beekes, Robert S. P.

1995 – Comparative Indo-European Linguistics: An Introduction.Amsterdam: John Benjamins

Beyer, Klaus1984 – Die Aramaischen Texte vom Toten Meer. Gottingen: Vandenhoeck &

Ruprecht2010 – “Die Sprache der Moabitischen Inschriften”. In: KUSATU 11,

pp. 5–412012 – “The Languages of Transjordan”. In: Languages from the World of

the Bible. Ed. by Holger Gzella. Berlin: De Gruyter, pp. 111–127Blau, Joshua

1978 – “Hebrew and North West Semitic: Reflections on the Classification ofthe Semitic Languages”. In: HAR 1, pp. 21–44

Bordreuil, Pierre and Pardee, Dennis2009 – A Manual of Ugaritic. Ed. by M. O’Connor and Cynthia L. Miller.

Linguistic Studies in Ancient West Semitic. Winona Lake:Eisenbrauns

Brockelmann, Carl1908 – Grundriss der Vergleichenden Grammatik der Semitischen Sprachen.

Berlin: Reuther & ReichardBrown, Francis, Driver, Samuel R., and Briggs, Charles A.

1907 – A Hebrew and English Lexicon of the Old Testament. Oxford:Clarendon Press. (= BDB)

Cline, Eric H.2014 – 1177 B.C.: The Year Civilization Collapsed. Princeton/Oxford:

Princeton University PressCross, Frank Moore

1962 – “Yahweh and the God of the Patriarchs”. In: HTR 55, pp. 225–259Del Olmo Lete, Gregorio and Sanmartın, Joaquın

2003 – A Dictionary of the Ugaritic Language in the Alphabetic Tradition (2vols.) Leiden: Brill. (= DUL)

Dickins, James2011 – “Dialects of Egypt and Sudan”. In: The Semitic Languages: An

International Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 54, pp. 935–953

Dietrich, Manfried, Loretz, Oswald, and Sanmartın, Joaquın1975 – “Untersuchungen zur Schrift- und Lautlehre des Ugaritischen (III)”.

In: UF 7, pp. 103–108Edel, Elmar

1966 – Die Ortsnamenlisten aus dem Totentempel Amenophis III. Vol. 25.Bonner Biblische Beitrage. Bonn: Peter Hanstein

Faber, Alice1997 – “Genetic Subgrouping of the Semitic Languages”. In: The Semitic

Languages. Ed. by Robert Hetzron. New York: Routledge. Chap. 1,pp. 3–15

56

Fales, Frederick M.2011 – “Old Aramaic”. In: The Semitic Languages: An International

Handbook. Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach-und Kommunikationswissenschaft. Berlin: De Gruyter-Mouton.Chap. 27, pp. 555–573

Folmer, Margaretha2012 – “Old and Imperial Aramaic”. In: Languages from the World of the

Bible. Ed. by Holger Gzella. Berlin: De Gruyter, pp. 128–159Gardiner, Alan

1957 – Egyptian Grammar. 3rd edn. Cambridge: University PressGarr, W. Randall

1985 – Dialect Geography of Syria-Palestine, 1000-586 b.c.e. Philadelphia:University of Pennsylvania Press

Gelb, I. J.1961 – “The Early History of the West Semitic Peoples”. In: JCS 15.1,

pp. 27–47Gelb, Ignace J.

1980 – Computer-Aided Analysis of Amorite. Vol. 21. Assyriological Studies.Chicago: The Oriental Institute of the University of Chicago

Gianto, Agustinus2012 – “Ugaritic”. In: Languages from the World of the Bible. Ed. by

Holger Gzella. Berlin: De Gruyter, pp. 28–54Ginsberg, Harold L.

1970 – “The Northwest Semitic Languages”. In: The World History of theJewish People: II. The Patriarchs. Ed. by B. Mazar. London: Allen,pp. 102–124

Gitin, Seymour, Dothan, Trude, and Naveh, Joseph1997 – “A Royal Dedicatory Inscription from Ekron”. In: IEJ 47.1-2,

pp. 1–16Golinets, Viktor

2010 – “Amorite Names written with the Sign U and the Issue of SuffixedThird Person Masculine Singular Pronoun in Amorite”. In:Proceedings of the 53e Rencontre Assyriologique Internationale, Vol.1, Part 2. Ed. by Leonid Kogan et al. Vol. 4/5. Babel und Bibel.Winonan Lake: Eisenbrauns

Gordon, Cyrus H.1997 – “Amorite and Eblaite”. In: The Semitic Languages. Ed. by

Robert Hetzron. New York: Routledge. Chap. 6, pp. 100–113Goren, Yuval, Finkelstein, Israel, and Na’aman, Nadav

2004 – Inscribed in Clay: Provenance Study of the Amarna Tablets andOther Ancient Near Eastern Texts. Vol. 23. Monograph Series. TelAviv: Tel Aviv University/Institute of Archaeology

Greenstein, Edward L.2004 – “Another Case of Hiphil in Amarna Age Canaanite”. In: Sefer

Moshe: The Moshe Weinfeld Jubilee Volume. Ed. by Chaim Cohen,Avi Hurvitz, and Shalom M. Paul. Winona Lake: Eisenbrauns,pp. 351–360

Gzella, Holger2011a – “Northwest Semitic in General”. In: The Semitic Languages: An

International Handbook. Ed. by Stefan Weninger. Vol. 36.

57

Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 18, pp. 425–451

2011b – “Imperial Aramaic”. In: The Semitic Languages: An InternationalHandbook. Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach-und Kommunikationswissenschaft. Berlin: De Gruyter-Mouton.Chap. 28, pp. 574–586

2012a – Languages from the World of the Bible. Ed. by Holger Gzella.Berlin: De Gruyter

2012b – “Introduction”. In: Languages from the World of the Bible. Ed. byHolger Gzella. Berlin: De Gruyter, pp. 1–13

2012c – “Phoenician”. In: Languages from the World of the Bible. Ed. byHolger Gzella. Berlin: De Gruyter, pp. 55–75

2012d – “Ancient Hebrew”. In: Languages from the World of the Bible. Ed. byHolger Gzella. Berlin: De Gruyter, pp. 76–110

2013a – “Northwest Semitic Languages and Hebrew”. In: EHLL. Ed. byGeorge Khan. Leiden/Boston: Brill, pp. 852–863

2013b – “Wilhelm Gesenius als Semitist: Das “Lehrgebaude” in seinemWissenschaftsgeschichtlichen Kontext”. In: Biblische Exegese undHebraische Lexikographie: Das “Hebraisch-DeutscheHandworterbuch” von Wilhelm Gesenius als Spiegel und QuelleAlttestamentlicher und Hebraischer Forschung, 200 Jahre nach seinerErsten Auflage. Ed. by Stefan Storch and Ernst-Joachim Waschke.Vol. 427. BZAW. Berlin/Boston: De Gruyter, pp. 184–208

2014 – “Language and Script”. In: The Arameans in Ancient Syria. Ed. byHerbert Niehr. Leiden/Boston: Brill. Chap. 4, pp. 71–107

2015 – A Cultural History of Aramaic: From the Beginnings to the Advent ofthe Islam. Leiden/Boston: Brill

Hackett, Jo-Ann and Pat-El, Na’ama2010 – “On Canaanite and Historical Linguistics: A Rejoinder to Anson

Rainey”. In: Maarav 17.2, pp. 173–188Harris, Zellig S.

1939 – Development of the Canaanite Dialects: An Investigation inLinguistic History. New Haven: American Oriental Society

Hasselbach, Rebecca2004 – “Final Vowels of Pronominal Suffixes and Independent Personal

Pronouns in Semitic”. In: JSS 49.1, pp. 1–20Hess, Richard S.

2007 – “Arrowheads from Iron Age I: Personal Names and Authenticity”. In:Ugarit at Seventy-Five. Ed. by K. Lawson Younger Jr. Winona Lake:Eisenbrauns, pp. 111–129

Hesse, Kristina J.2008 – Hesse, Kristina J. “Contacts and Trade at Late Bronze Age Hazor:

Aspects of Intercultural Relationships and Identity in the EasternMediterranean”. PhD thesis. Umea University

Hetzron, Robert1972 – Ethiopian Semitic: Studies in Classification. Manchester: The

University Press1974 – “La Division des Langues Semitiques”. In: Actes du Premier Congres

International de Linguistique Semitique et Chamito-Semitique. Ed. byAndre Caquot and David Cohen. The Hague: Mouton, pp. 181–194

58

1976 – “Two Principles of Genetic Reconstruction”. In: Lingua 38.2,pp. 89–108

1997 – The Semitic Languages. Ed. by Robert Hetzron. New York:Routledge

Hoch, James E.1994 – Semitic Words in Egyptian Texts of the New Kingdom and Third

Intermediate Period. Princeton: Princeton University PressHorowitz, Wayne, Oshima, Takayoshi, and Sanders, Seth

2006 – Cuneiform in Canaan: Cuneiform Sources from the Land of Israel inAncient Times. Jerusalem: Israel Exploration Society and TheHebrew University Jerusalem

Huehnergard, John1987a – “Northwest Semitic Vocabulary in Akkadian Texts (Review of Sivan,

1984)”. In: JAOS 107.4, pp. 713–7251987b – “Review of Garr (1985)”. In: JBL 106.3, pp. 529–5331987c – “Three Notes on Akkadian Morphology”. In: Working with No Data:

Semitic and Egyptian Studies Presented to Thomas O. Lambdin.Ed. by David M. Golomb and Susan Tower Hollis. Winona Lake:Eisenbrauns, pp. 181–194

1987d – Ugaritic Vocabulary in Syllabic Transcription. Vol. 32. HSS. Atlanta:Scholars Press

1991 – “Remarks on the Classification of the Northwest Semitic Languages”.In: The Balaam Text from Deir ‘Alla Re-evaluated. Ed. byJ. Hoftijzer and G. van der Kooij. Leiden/Boston: Brill, pp. 282–293

1992a – “Historical Phonology and the Hebrew Piel”. In: Linguistics andBiblical Hebrew. Ed. by Walter R. Bodine. Winona Lake:Eisenbrauns, pp. 209–229

1992b – “Languages (Introductory)”. In: The Anchor Bible Dictionary.Ed. by David Noel Freedman. Vol. 4 (K-N). New York: Doubleday,pp. 155–170

1995 – “What is Aramaic?” In: ARAM 7, pp. 261–2821998 – “A Grammar of Amarna Canaanite (review of Rainey, 1996)”. In:

BASOR 310, pp. 59–772002 – “Comparative Semitic Linguistics”. In: IOS 20, pp. 119–1502003 – “Akkadian h

˘and West Semitic *h. ”. In: Studia Semitica. Ed. by

Leonid Kogan. Vol. 3 (Fs. Alexander Militarev. Orientalia: Papers ofthe Oriental Institute. Moscow: Russian State University for theHumanities, pp. 102–119

2005 – “Features of Central Semitic”. In: Biblical and Oriental Essays inMemory of William L. Moran. Ed. by Agustinus Gianto. Roma:Editrice Pontificio Istituto Biblico, pp. 155–203

2006 – “Proto-Semitic and Proto-Akkadian”. In: The Akkadian Language inits Semitic Context: Studies in the Akkadian of the Third and SecondMillennium BC. Ed. by G. Deutscher and N. J. C. Kouwenberg.Leiden: Nederlands Instituut voor het Nabije Oosten, pp. 1–23

2013 – “Canaanite Shift”. In: EHLL. Ed. by Geoffrey Khan. Leiden/Boston:Brill

Huehnergard, John and Hasselbach, Rebecca2007 – “Northwest Semitic Languages”. In: EALL. Ed. by K. Versteeg.

Vol. 3. Leiden/Boston: Brill, pp. 408–422

59

Huehnergard, John and Rubin, Aaron D.2011 – “Phyla and Waves”. In: The Semitic Languages: An International

Handbook. Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach-und Kommunikationswissenschaft. Berlin: De Gruyter-Mouton.Chap. 9, pp. 259–278

Huffmon, Herbert B.1965 – Amorite Personal Names in the Mari Texts: A Structural and Lexical

Study. Baltimore: Johns Hopkins PressIzre’el, Shlomo

1978 – “The Gezer Letters of the el-Amarna Archive: Linguistic Analysis”.In: IOS 8, pp. 13–90

2003 – “Vocalized Canaanite: Cuneiform-Written Canaanite Words in theAmarna Letters: Some Methodological Remarks”. In: DS 5.1-2,pp. 13–34

2007 – Canaano-Akkadian: Some Methodological Requisites for the Study ofthe Amarna Letters from Canaan. Unpublished paper, accessible viaacademia.edu/229790/

Al-Jallad, Ahmad2014 – “Final Short Vowels in G@‘@z, Hebrew »atta, and the Anceps

Paradox”. In: JSS 59.1, pp. 315–327Jouon, Paul

2006 – A Grammar of Biblical Hebrew. 2nd edn. Roma: Gregorian & BiblicalPress

Killebrew, Ann E.2005 – Biblical Peoples and Ethnicity: An Archaeological Study of Egyptians,

Canaanites, Philistines, and Early Israel, 1300-1100 b.c.e. Vol. 9.Archaeology and Biblical Studies. Atlanta: Society of BiblicalLiterature

Kitchen, Kenneth A.1969 – “Review of Edel (1966)”. In: BiOr 26, pp. 198–202

Knauf, Ernst Axel and Ma’ani, Sultan A.1987 – “On the Phonemes of Fringe Canaanite: The Cases of Zerah-Ud

¯ruh.

and “Kamash¯alta””. In: UF. Ed. by K. Bergerhof, M. Dietrich, and

O. Loretz. Vol. 19, pp. 91–94Knudsen, E. E.

2004 – “Amorite Vocabulary: A Comparative Statement”. In: Assyria andBeyond: Studies Presented to Mogens Trolle Larsen. Ed. byJ. G. Dercksen. Leiden: Nederlands Instituut voor het Nabije Oosten,pp. 317–331

Kogan, Leonid2011a – “Proto-Semitic Lexicon”. In: The Semitic Languages: An

International Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 8, pp. 179–258

2011b – “Proto-Semitic Phonetics and Phonology”. In: The SemiticLanguages: An International Handbook. Ed. by Stefan Weninger.Vol. 36. Handbucher zur Sprach- und Kommunikationswissenschaft.Berlin: De Gruyter-Mouton. Chap. 6, pp. 54–151

Krahmalkov, Charles R.2001 – A Phoenician-Punic Grammar. Leiden/Boston/Koln: Brill

60

Krebernik, Manfred– “The Linguistic Classification of Eblaite: Methods, Problems, and

Results”. In: The Study of the Ancient Near East in the Twenty-FirstCentury. Ed. by J. S. Cooper and G. M. Schwartz. Winona Lake:Eisenbrauns, pp. 233–249

Lane, Edward W.1863 – An Arabic-English Lexicon. London: Williams & Norgate. (= Lane)

Lawson Younger Jr., K.2007 – “The Late Bronze Age / Iron Age Transition and the Origins of the

Arameans”. In: Ugarit at Seventy-Five. Ed. byK. Lawson Younger Jr. Winona Lake: Eisenbrauns, pp. 131–174

Leslau, Wolf1991 – Comparative Dictionary of Ge–ez (Classical Ethiopic). Wiesbaden:

Harrassowitz. (= CDG)Lipinski, Edward

1997 – Semitic Languages: Outline of a Comparative Grammar. Leuven:Peeters

Loprieno, Antonio and Muller, Matthias2012 – “Ancient Egyptian and Coptic”. In: The Afroasiatic Languages.

Ed. by Zygmunt Frajzyngier and Erin Shay. Cambridge: CambridgeUniversity Press. Chap. 3, pp. 102–144

Luffin, Xavier2011 – “Arabic-based Pidgins and Creoles”. In: The Semitic Languages: An

International Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 58, pp. 990–1000

Mallory, James P.1989 – In Search of the Indo-Europeans: Language, Archaeology and Myth.

London: Thames & HudsonMeyer, Ronny

2011 – “Amharic”. In: The Semitic Languages: An International Handbook.Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach- undKommunikationswissenschaft. Berlin: De Gruyter-Mouton. Chap. 70,pp. 1178–1212

Millard, Alan2012 – “The Alphabet”. In: Languages from the World of the Bible. Ed. by

Holger Gzella. Berlin: De Gruyter, pp. 14–27Moran, William L.

1961 – “The Hebrew Language in its Northwest Semitic Background”. In:The Bible and the Ancient Near East: ESsays in Honor of WilliamFoxwell Albright. Ed. by G. Ernest Wright. London:Routledge/Kegan Paul, pp. 54–72

1992 – The Amarna Letters. Baltimore/London: The John HopkinsUniversity Press

Moscati, Sabatino1964 – An Introduction to the Comparative Grammar of the Semitic

Languages: Phonology and Morphology. Vol. VI. Porta LinguarumOrientalium. Wiesbaden: Harrassowitz

Mugnaioni, Remo

61

2000 – “Note pour servir a une approche de l’ Amorrite”. In: La Semitologie,Aujourd’hui. Ed. by Philippe Cassuto and Pierre Larcher. Vol. 16.Cercle Linguistique d’Aix-en-Provence, Travaux. Aix-en-Provence:Publications de l’Universite de Provence, pp. 57–68

Na’aman, Nadav1981 – “Economic Aspects of the Egyptian Occupation of Canaan”. In: IEJ

31, pp. 172–1851994 – “The Hurrians and the End of the Middle Bronze Age in Palestine”.

In: Levant 26, pp. 175–1872000 – “Rubutu/Aruboth”. In: vol. 32. UF, pp. 373–384

Noorlander, Paul2012 – “Sam’alian in its Northwest Semitic Setting: A

Historical-Comparative Approach”. In: Orientalia 81.3, pp. 202–238Oppenheim, Leo, Reiner, Erica, and Roth, Martha T.

2010 – Chicago Assyrian Dictionary. Chicago: The Oriental Institute. (=CAD)

Pardee, Dennis1999 – “Review of Rainey (1996)”. In: JNES 58.4, pp. 313–3172011 – “Ugaritic”. In: The Semitic Languages: An International Handbook.

Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach- undKommunikationswissenschaft. Berlin: De Gruyter-Mouton. Chap. 20,pp. 460–472

2012 – The Ugaritic Texts and the Origins of West-Semitic LiteraryComposition: Schweich Lectures of the British Academy 2007.Oxford: Oxford University Press

Payne Smith, Robert1903 – A Compendious Syriac Dictionary. Oxford: Clarendon Press. (=

CSD)Pitard, Wayne T.

1987 – Ancient Damascus: A Historical Study of the Syrian City-State fromEarliest Times until its Fall to the Assyrians in 732 b.c.e. WinonaLake: Eisenbrauns

Posener, Georges1940 – Princes et Pays d’Asie et de Nubie: Textes Hieratiques sur des

Figurines d’Envoutement du Moyen Empire. Brussels: FondationEgyptologique Reine Elisabeth

Pritchard, James B.1969 – Ancient Near Eastern Texts Relating to the Old Testament. 3rd edn.

Princeton: Princeton University PressRainey, Anson F.

1972 – “The World of Sinuhe”. In: IOS 2, pp. 369–4081975 – “Toponymic Problems”. In: TA 2, pp. 13–161982 – “Linguistic Notes on Thutmose III’s Topographical List”. In:

Egyptological Studies. Ed. by Sarah Israelit-Groll. Vol. XXVIII. SH.Jerusalem: The Magnes Press, pp. 335–359

1996 – Canaanite in the Amarna Tablets: A Linguistic Analysis of the MixedDialect Used by the Scribes from Canaan (4 Vols.) Leiden/NewYork/Koln: Brill

1998 – “Syntax, Hermeneutics and History”. In: IEJ 48.3-4, pp. 239–2512015 – The El-Amarna Correspondence. Leiden/Boston: Brill

62

Rosenthal, Franz1983 – A Grammar of Biblical Aramaic. Vol. V. Porta Linguarum

Orientalum. Wiesbaden: HarrassowitzRubin, Aaron D.

2008 – “The Subgrouping of the Semitic Languages”. In: LLC 2.1, pp. 61–84Saenz-Badillos, Angel

1993 – A History of the Hebrew Language. Trans. by John Elwolde.Cambridge: Cambridge University Press

Sasson, Jack M.1984 – “Zimri-Lim Takes the Grand Tour”. In: BA 47.4, pp. 246–251

Sethe, Kurt H.1926 – Die Achtung Feindlicher Fursten, Volker und Dinge auf

Altagyptischen Tongefassscherben des Mittleren Reiches. Vol. 5.Abhandlungen der Preussischen Akademie der Wissenschaften -Philosophisch-Historische Klasse. Berlin: Verlag der Akademie derWissenschaften

Sethe, Kurt H. and Helck, Wolfgang1927-1958 – Ukrunden der 18. Dynastie. Leipzig/Berlin: J.C.

Hinrichs/Akademie-VerlagSimeone-Senelle, Marte-Claude

2011 – “Modern South Arabian”. In: The Semitic Languages: AnInternational Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 64, pp. 1073–1113

Simons, Jan Jozef1937 – Handbook for the Study of Egyptian Topographical Lists Relating to

Western Asia. Leiden: BrillSivan, Daniel

1984 – Grammatical Analysis and Glossary of the Northwest SemiticVocables in Akkadian Texts of the 15th-13th C.B.C. from Canaan andSyria. Vol. 214. AOAT. Neukirchen-Vluyn: NeukirchenerVerlag/Verlag Butzon & Bercker Kevelaer

2000 – “The Status of Ugaritic among the Northwest Semitic Languages inthe Wake of New Research”. In: UF. Ed. by M. Dietrich K. Bergerhofand O. Loretz. Vol. 32. Munster: Ugarit-Verlag, pp. 531–541

Sivan, Daniel and Cochavi-Rainey, Zipora1992 – West Semitic Vocabulary in Egyptian Script. Beer-Sheva. Beersheva:

Ben-Gurion University of the Negev PressStreck, Michael P.1998a – “Name, Namengebung. E. Amurritisch”. In: RlA. Vol. 9, pp. 127–1311998b – “Name, Namengebung. F. Westsemitisch in Keilschrifttexten des I.

Jt.” In: RlA. Vol. 9, pp. 131–1342000 – Das amurritische Onomastikon der altbabylonischen Zeit. Band I:

Die Amurriter: Die onomastische Forschung; Orthographie undPhonologie; Nominalphonologie. Munster: Ugarit-Verlag

2011a – “Eblaite and Old Akkadian”. In: The Semitic Languages: AnInternational Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 13, pp. 340–359

63

2011b – “Amorite”. In: The Semitic Languages: An International Handbook.Ed. by Stefan Weninger. Vol. 36. Handbucher zur Sprach- undKommunikationswissenschaft. Berlin: De Gruyter-Mouton. Chap. 19,pp. 452–459

Suchard, Benjamin2013 – “Regularizing the Canaanite Shift”. MA thesis. Leiden University

Trask, R. Larry and McColl Millar, Robert2007 – Trask’s Historical Linguistics. 2nd edn. London: Hodder Education

Tropper, Josef1990 – Der Ugaritische Kausativstamm un die Kausativbildungen des

Semitischen: Eine Morphologisch-Semantische Untersuchung zumS-Stamm und zu den Umstrittenen NichtsibilantischenKausativstamme des Ugaritischen. Munster: Ugarit-Verlag

1994 – “Is Ugaritic a Canaanite Language?” In: Ugarit and the Bible:Proceedings of the International Symposium on Ugarit and the Bible,Manchester, September 1992. Ed. by George J. Brooke,Adrian H.W. Curtis, and John F. Healey. Vol. 11. UBL. Munster:Ugarit-Verlag

2000 – Ugaritische Grammatik. Vol. 273. AOAT. Munster: Ugarit-Verlag2002 – Altathiopisch: Grammatik des Ge‘ez mit Ubungstexten und Glossar.

Munster: Ugarit-Verlagvan Soldt, Wilfred H.

2008 – “Why did they write? On Empires and Vassals in Syria and Palestinein the Late Bronze Age”. In: Theory and Practice of KnowledgeTransfer: Studies in School Education in the Ancient Near East andBeyond. Ed. by Wolfert S. van Egmond and Wilfred H. van Soldt.Leiden: Nederlands Instituut voor het Nabije Oosten, pp. 103–114

2010 – “Ugarit as a Hittite Vassal State”. In: AF 37.2, pp. 198–2072013 – “The Extent of Literacy in Syria and Palestine during the Second

Millennium b.c.e.” In: Time and History in the Ancient Near East:Proceedings of the 56th Rencontre Assyriologique Internationale atBarcelona. Ed. by LLuıs Feliu et al. Winona Lake, Indiana:Eisenbrauns, pp. 19–31

Vanderhooft, David S.1995 – “The Edomite Dialect and Script: A Review of the Evidence”. In:

You Shall Not Abhor an Edomite for he is your Brother: Edom andSeir in History and Tradition. Ed. by Diana Vikander Edelman.Atlanta: Scholars Press, pp. 137–157

von Soden, Wolfram1995 – Grundriss der Akkadischen Grammatik. 3rd edn. Vol. 33. AO. Roma:

Pontificio Istituto BiblicoWeninger, Stefan2011a – The Semitic Languages: An International Handbook. Ed. by

Stefan Weninger. Vol. 36. Handbucher zur Sprach- undKommunikationswissenschaft. Berlin: De Gruyter-Mouton

2011b – “Reconstructive Morphology”. In: The Semitic Languages: AnInternational Handbook. Ed. by Stefan Weninger. Vol. 36.Handbucher zur Sprach- und Kommunikationswissenschaft. Berlin:De Gruyter-Mouton. Chap. 7, pp. 151–178

Wilhelm, Gernot

64

1995 – “The Kingdom of Mitanni in Second-Millennium UpperMesopotamia”. In: Civilizations of the Ancient Near East. Ed. byJack Sasson. Vol. II, pp. 1243–1254

Wright, William1896 – A Grammar of the Arabic Language. 3rd edn. Cambridge: Cambrdige

University Press

65