ugaritic numbers

Upload: alpeters8703

Post on 07-Apr-2018

228 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 Ugaritic Numbers

    1/40

    1

    Al Peters 15/05/2011

    The Phonology and Morphology of the Cardinal and Ordinal Numerals in Ugaritic

    Introduction1

    The subject of this paper is the phonology and morphology of numerals in Ugaritic.2

    Because the

    primary focus of this paper is phonological/ morphological, the terms noun and nominal are

    used for all units that take nominal inflection, regardless of their syntactic function. First, the

    cardinals of all the independent numerals (i.e., the basic components from which compound

    numerals are formed) will be discussed individually. Then, all of the ordinals numberswith the

    exception of the ordinal 1st, which presents its own peculiaritieswill be discussed together,

    according to the understanding that this category is more morphologically homogeneous and

    systematic than cardinals.3

    The syntax of numerals, including the morpho-syntactical issue of

    how compound numerals are formed (e.g., 246 from 200, 40, and 6), though a

    1 The following abbreviations are used for conciseness: CS = Central Semitic, NWS = Northwest Semitic, PS =

    proto-Semitic, WS = West Semitic, OSA = Old South Arabian, MSA = Modern South Arabian. The term absolute

    case is used to refer to that inflection (far more limited in WS than Akkadian) referred to by Assyriologists as the

    status absolutus. This is to distinguish it from the absolute or unbound state (i.e., where a noun is not bound to

    another in a genitival construct chain). ll garitic passages are cited according to Dietrich, Loretz, and anmartn ,

    1976.

    2The subject under consideration is not Ugaritic phonology generally, but only in specific relation to the numerals,

    so the ramifications of the Ugaritic reflexes of PS consonants (for which, see Tropper, 1994, 345ff.) are not

    addressed, but rather considered a given of the discussion.

    3 Multiplicatives (e.g., {ld} three times) cannot be discussed here for space reasons. Furthermore, insofar as they

    are construed as a numeral in conjunction with a particle, they should really be treated separately in a study of this

    construction in its own right, with reference to its phonotactics, the use and behaviour particle outside of the

    numerical context, and so forth. While a study such as the current undertaking should ideally also include a

    discussion of fractions, this has regrettably not been possible here owing to space restrictions.

  • 8/4/2019 Ugaritic Numbers

    2/40

    2

    dimension in which Ugaritic is distinct from other Semitic languages, and a subject worthy of

    attention, will only be addressed inasmuch as it is relevant to their morphology.4

    This paper will make extensive reference to other Semitic languages, with specific focus on

    what the data tell us about the relationship of Ugaritic to these, one the one hand, and about what

    it tells us about the historical development of Ugaritic specifically, on the other. The absence of

    internal evidence has resulted in uncertainty regarding the morphology of some numerals (for

    example, ordinals and small details of some cardinals). Given, then, that one has to rely on

    comparative evidence, ones opinion regarding the relationship of garitic to other (emitic)

    languages is inevitably a major influence on the reconstruction of its phonology and morphology.

    The basic genetic classification of the Semitic languages adhered to here is a modified version of

    Hetzrons model, as presented by Faber.5 With regard to Ugaritic specifically, notwithstanding

    the attitudes expressed in the rest of his paper, Blaus statement, Ginsberg has proven his point

    that Ugaritic is more closely related to Canaanite than to Aramaic. On the other hand, in my

    opinion, Phoenician and Hebraic belong together (Canaanite), as opposed to garitic, strikes

    me as basically sound.6

    The exact position of Ugaritic within NWS, though, is not undisputed. This should be

    recognised in order to avoid circular arguments based solely on cognate evidence, which do not

    benefit scholarship, but instead merely entrench their proponents in (possibly correct, but

    crucially unfalsifiable) speculations. Furthermore, the role of areal diffusion of features among

    4For example, Ugaritic has a method of forming compound numerals, unparalleled elsewhere in emitic, using the

    preposition {l} (Loewenstamm, 1980, 3131ff.). Elsewhere, while a breakdown of chiastic concord has occurred in

    several modern emitic languages, particularly in popular vernaculars (for which, see Lipiski , 1997, 25.6), its

    attestation in Ugaritic is virtually unique amongst the classical Semitic languages (Tropper, 2000, 69.133.1).

    5Faber, 1997.

    6Blau, 1978, 36; for arguments given in support of his classification, see 6.2ff. The same basic situation is

    expressedby Troppers (1994, 343353) identification of garitic as North Canaanite.

  • 8/4/2019 Ugaritic Numbers

    3/40

    3

    languages (particularly within NWS, but also as a general principle of historical linguistics) must

    not be ignored, provided one uses rigorous methodology to determine which common features

    may be ascribed to such a process, rather than being shared innovations or retentions, or

    independent parallel developments.7

    Preliminaries

    Firstly, it is noted that logographic representation of numbers is occasionally attested in Ugaritic,

    though seldom in alphabetic texts.8

    This matter is not further addressed here as it does not inform

    directly about linguistic matters. While the observation may be made that it can help us to

    understand cultural relationships with other (specifically Mesopotamian and Anatolian)

    traditions, I am wary of inferring from it that Ugaritic numerals necessarily have any particular

    affinity to their Akkadian counterparts. Considering that logographic representations are rare in

    alphabetic texts, I feel that their use in Ugaritic generally is rather an indication of the extent of

    Akkadian influence in the arena of scribal practices specifically.9

    It is perhaps a simplification, but in my opinion a useful and broadly speaking correct one, to

    characterise Ugaritic as using a base-ten numerical system. This is quite common in the Semitic

    languages, and in the world generally (likely a reflection of human anatomy, namely the ten

    fingers of the hands).10 Although evidence that an alternative, base-sixty numerical system (and

    7Some authors, in order to deal with problematic datasuch as the language of the Deir lla inscription, or

    dialectal variation within Phoenician and Old Aramaichave championed areal diffusion of linguistic features as

    the primary criterion according to which one should view language relationships, specifically within a dialect-

    continuum model (see, e.g., Garr, 1985). In my opinion, though, areal diffusion is better understood as a secondary

    phenomenon, modifying existing relationships established primarily by genetic divisions.

    8Tropper, 2000, 61.4.

    9Tropper, 2000, 61.4.

    10 For a discussion of numerical and measuring systems in relation to human anatomy, see Blaek, 1999, 325.

  • 8/4/2019 Ugaritic Numbers

    4/40

    4

    possibly other systems besides) was used in Semitic, the origin and status of this alternative

    system are a matter of some debate, and one beyond the scope of the current paper.11 Suffice it

    here to identify the primary numerical system of Ugaritic as base ten, and recognise that using a

    variety of numerical systems is not uncommon.12

    Tropper does well to remind us that [d]ie bildung von Zahlwrtern unterscheidet sich im

    Prizip nicht von der anderer Nomina und bedarf deshalb keiner gesonderten Behandlung.13 I

    agree entirely with this sentiment and would further add that many basic numerals (certainly 3

    10 and 100) should be reconstructed to PS as primary nouns.14

    s an extension of Troppers

    statement, numerals should be treated as any other nouns are in terms of inflectional as well as

    derivational morphology. In this regard, Sivan states that we have no way of knowing whether

    the forms had case endings (as in rabic) or whether they lacked them as in kkadian.15 This is

    not entirely true; if one knows where to look, there is at least some, admittedly subtle, evidence

    11 There are those who see the sexagesimal system as having entered Semitic from Sumerian, via Akkadian, though

    others argue for the opposite direction of transmission (for discussion, see Dombrowski and Dombrowski, 1991,

    344345). Tropper (2000, 61.3) thinks this sexagesimal system may underpin the use of certain collectives.

    Loewenstamm (1980, 310311) further identifies remnants of a further alternative primitive numerical system in

    garitic: Besides the normal numbers, which belong to the decimal system, Ugaritic preserves a few remnants of a

    more primitive modethe abnormal dual forms t [2x3=6] andt [2x6=12] should be regarded as the remnants

    of an old and archaic arrangement.12

    In English, for example, traces of a base-twelve system are reflected in the base numerals dozen (12) and

    gross (12x12 =144), and the practice of selling certain basic commoditieseggs and bottled beer, for example

    by the six, twelve, and twenty-four. As late as the 1970s, the United Kingdom used a partially duodecimal monetary

    system (1 = 240d; 1s = 12d).

    13Tropper, 2000, 61.2.

    14Greenberg (1950, 176) brings a genial proof of this for 9 specifically: A PS root-consonant restriction (**R1=

    dental, R2= sibilant) resulted in their metathesis. This started in the prefix conjugation, in which these two were in

    contact and subsequently spread through the paradigm. The lexeme *tis was immune to this process, since, as it was

    a primary noun, R1 was never in contact with R2.

    15Sivan, 2001, 86.

  • 8/4/2019 Ugaritic Numbers

    5/40

    5

    for the former, which is the position adopted by this paper. Firstly, the syllabic transcription [a-

    ]a-du bears a case vowel even though syllabic orthography is perfectly capable of expressing a

    form without it (as *a-a-ad, for example).16 Secondly, the alphabetic form {mit}, with {i}

    representing syllable-closing //, when compared to the corresponding kkadian form, argues

    against absolute-case numerals in Ugaritic (the argument is explained in detail below). This is

    not to mention the broad methodological stance that, considering the use of the absolute case in

    Ugaritic, while imperfectly understood, is apparently far more restricted there than in Akkadian,

    and bearing in mind the genetic classification of Ugaritic as WS, it is unreasonably

    Akkadocentric to presume that the Ugaritic numeral system was similar in this respect.

    17

    Thebest counterevidence from the numeralsthe absence of {m} in {n}/{t} 2 and the {h} of

    {rh}is discussed below.18

    1: {ad}, [a-]a-du = /aad-/, {at} = /aatt-/

    The attestation of related forms from I =yfor example, the adjective {yd} /yad/ alone

    and the (adverbially used) noun {yd} /ya(a)da/together, the latter to be considered, at some

    level, a by-form of the numeral {ad}19shows that whatever the origins of this lexeme, it is

    considered to have an entirely triconsonantal, albeit variable, root by the proto-Ugaritic stage.20

    16Huehnergard, 1987, 293.

    17This statement is hardly subject to the criticism I have made of circular arguments based on preconceptions about

    language classifications, which really relates more specifically to the position of Ugaritic within NWS, and, at the

    most, in relation to Arabic. Anyway, this there is evidence from the Ugaritic numerals specifically, discussed below,

    to suggest they behave like those of other WS languages rather than Akkadian.

    18 For discussion see Tropper, 2000, 54.5ff.

    19Del Olmo Lete and anmartn, 2003, 960.

    20In my opinion, forms from R1 = wbeing attested in kkadian as well as W (rabic; > R1=y in NWS)

    suggest that if indeed this lexeme was ever biradical, it had already been expanded by PS. Accordingly, Aramaic

  • 8/4/2019 Ugaritic Numbers

    6/40

    6

    Comparing the relative frequency of forms from dor ydin a concordance of Ugaritic

    clearly shows that the former is primary in Ugaritic.21 In this respect, it resembles Hebrew and

    Aramaic (on the assumption that the aphaeresis of // is more likely than that of /y/), in contrast

    to Arabic.22

    s to the morphology of the garitic numeral 1 from d, the presence of {a} in the

    orthography not only confirms that aphaeresis did not occur, as it does in Aramaic and some

    dialectal Arabic, but also specifies the vowel quality of the initial syllable as low, which is in

    total accordance with comparative evidence from cognate languages. There is no reason to

    believe that this vowel was long, as in Arabic wid, since no other NWS exhibits a similar

    form, and Arabic anyway possesses a form aad. Historical gemination in the proto-Hebrew

    reflex of this lexeme is indicated by the vowel following // in Tiberian vocalisations /_GG23) and a< *aadt(since *a in an open syllable would

    lower to ). Considering Aramaic forms, this cannot have been the case for proto-NWS. Thus, if

    one were to posit gemination of // in the Ugaritic forms, it would necessarily suggest that

    Ugaritic had a closer relationship with the predecessor of Hebrew than with that of Aramaic,

    a/, and likewise forms with no consonant preceding // in the language of Deir lla, perhaps mmonite

    (contra Garr, 1985, 5052), some Arabic dialects (e.g., Damascene in the expression dno one, for which

    see Cowell, 2005, 285), and several Ethiopic languages (Leslau, 1991, 12), result from aphaeresis, rather than

    retaining an original biradical form (contra Lipiski, 1997, 35.3a).21

    Cunchillos, Vita, and Zamora, 2003, 5075, 1388. Naturally, in using a concordance of this kind, one must

    individually assess each entry to ascertain that it is, in fact, a lemma from the relevant root, taking care not to

    include, for example, instances of the form {d} that are 1st person singular prefix conjugation forms from dy.22 It does not matter in this regard that Hebrew and Aramaic have verbs from ydand rabic from both ydand

    d, since, as a logical corollary of the aforementioned fact that numerals are primary nouns in Semitic, the verbal

    roots must be denominative, and therefore secondary to the roots of the numerals themselves.

    23The pausal feminine form (Gen 11:1; Exod 36:10) likewise results from this phonological process.

  • 8/4/2019 Ugaritic Numbers

    7/40

    7

    whether in terms of genetic subgrouping, or areal diffusion (assuming that parallel independent

    development of such a phenomenon would be unexpected).

    Although this is an entirely reasonable attitude, in the absence of evidence from other

    Canaanite languagesabjadic orthographies do not indicate gemination, while El Amarna

    Canaanite uses iten rather than a lexeme from dthis criterion cannot contribute further to

    any discussion regarding their subclassification; it is possible that gemination of // was a

    Hebrew-specific development, a common innovation of Troppers outh Canaanite group, or

    even of Canaanite as a whole (according to Troppers terminology), including garitic.24 My

    instinctive inclination is that it may well have been a Hebrew-specific development, particularly

    considering there exists in Hebrew an inflected form of this lexeme in which the proto-form

    cannot have had gemination of //.25 All that remains to be noted, in this regard, is the slim piece

    of evidence Huehnergard identifies from the syllabic form [a-]a-du (g. 5 138 1) to suggest

    Ugaritic did not possess the gemination of //: Although the first sign is broken, there is only

    room for A, not for A. Double consonants are often written singly in the syllabic transcriptions,

    as is generally true in kk. texts, but an initial v sign where vCis possible is not common.26

    24See Tropper (1994) for this use of the terminology.

    25The proto-Hebrew base ofi (Gen 11:1; 27:44; 29:20; Dan 11:20) cannot have had a geminated // since

    the vowel in its initial (closed) syllable would have been incapable of reduction. The semantics of this form are

    somewhat problematic: in the latter three examples, i qualifies the nounymimin the sense a few days; in

    the firstin which the very use of a plural form is likely a kind of wordplay, or some other rhetorical device,

    resulting in unusual, highly stylised languageit qualifies drim, apparently meaning something like one [kind

    of] words. In any case, there is no reason to suspect that these forms are anything other than the lexeme *aad

    inflected for masculine gender and plural number. Nor am I aware of any process that would allow one to posit

    proto-Hebrew *aadand account for the loss of gemination of //. Recourse to suppletion to account for this data

    is an undesirable solution and anyway recognises the existence of a stem *aadin proto-Hebrew, and therefore

    quite possibly elsewhere in NWS.

    26Huehnergard, 1987, 67n22.

  • 8/4/2019 Ugaritic Numbers

    8/40

    8

    Finally, a note is in order on the feminine form of this numeral. There is entirely reasonable

    consensus amongst scholars of Ugaritic that the orthography {at} of this numeral as it appears

    when qualifying feminine entities represents the masculine stem, reconstructed here as /aad/, to

    which the allomorph [t] of the feminine morpheme (rather than the allomorph [at]) is suffixed.

    This assumes, then, that a regressive assimilation (*dt> tt) has taken place.27

    This is not

    restricted to Ugaritic, but also occurs in Hebrew and Phoenician inscriptions, not to mention the

    consonantal text of the Hebrew Bible, which exhibit the same orthography.28

    This use of the [t]

    allomorph will come up again in connection with other numerals and may be a useful criterion

    for analysis.

    1, also the unit in 11: {t(y)} /at(+ya)/ a in a closed unaccented syllable in the

    27See, e.g., Tropper, 2000, 35.115.11.

    28Hoftijzer and Jongeling, 1995, 32. pace does not permit full discussion of the form { tt} (KAI 110:3), but the

    weight of evidence is in favour of a scribal error, as Hoftijzer and Jongeling propose.

    29Dombrowski and Dombrowski, 1991, 348. For a thorough discussion of this lexeme, see Dombrowski and

    Dombrowski, 1991, 347ff. I am not qualified to determine the extent to which data from El Amarna (for which, see

    Rainey, 1996, I:183184) reflect Canaanite influence, so they have not been included in the discussion.

  • 8/4/2019 Ugaritic Numbers

    9/40

    9

    environment of // to explain the Hebrew data), to which various suffixes are attached.30Gelbs

    preferable proposal, which is cautiously adopted here, is a base atiy with the suffix n in the

    masculine, and atin the feminine, and it more satisfactorily explains several Akkadian variants

    (notably i-ti-a-num), as well as the Hebrew /e/ (resulting from triphthong collapse).31

    Dombrowski and Dombrowski bring an impressive amount of evidence in favour of an extra-

    Semitic (specifically Hurro-Urartian) origin for the lexeme, where it had a deictic function.32

    Fascinating though this might be, it cannot be further treated here, other than to the extent that it

    informs the discussion on language classification by disproving the previously popular view that

    the Hebrew reflex (and by extension, that of Ugaritic) was a loan from Akkadian.

    33

    This is

    further confirmed by the preservation of // in NW.34

    The presence of {y} in the unique orthography {ty} (1.161:27) when the numeral occurs

    independentlyin the compound numeral 11, one finds the orthography {t.r(t)}35is

    30 Lipiski, 1997, 35.3b. The origin of Hebrew /e/ would thus be an archaic plural morpheme /ay/. Accordingly,

    Akkadian itn alongside itn would speak to the archaic nature of the form, because later the dipthong /ay/ reduced

    to //, not //. The /n/ element of kkadian (and O) is still more difficult to explain. If, as Lipiski would have it ,

    this represents a non-boundedness morpheme, then it would have to either be attached to a dual form (which would

    be extremely peculiar), or date to a very early stage of Akkadian (i.e., pre-Sargonic), before it lost non-boundedness

    morphemes on plural forms. Furthermore, unbound forms like itenu(and O {s1tnm}) would necessarily result

    from reanalysis. For the dating of Akkadian processes described, see Hasselbach, 2005, 179180. For Akkadian and

    OSA forms, see Oppenheim (ed.), 1960, 278, and Hfner, 1943, 110, respectively.

    31Gelb, cited in Dombrowski and Dombrowski, 1991, 350351. It should not be considered problematic that

    reconstructing the PS vowel *a in the initial syllable necessitates reading ein kkadian, even when an I sign is

    used, but rather reflects the idiosyncrasies of Mesopotamian syllabary orthography. The main criticism of this

    position is its failure to explain the absence of /n/ in NWS reflexes of this word.

    32Dombrowski and Dombrowski, 1991, 351ff.

    33Joon, 1991, 100f: These formsare considered by most authorities to be derived from the kkadian.34

    Loewenstamm, 1980, 316. Hetzrons (1977, 178) counterarguments are unconvincing and stem largely from the

    fact that he was unaware of the single independent use of {ty} (1.161:27) in Ugaritic.35

    Tropper, 2000, 62.112.

  • 8/4/2019 Ugaritic Numbers

    10/40

    10

    intriguing and, according to Troppers analysis of the behaviour of triphthongs in Ugaritic, lies

    behind his vocalisation /aty-/ or alternatively /aty-/, even though no comparative evidence

    supports this.36

    The crucial problem here is how one can reconcile the presence of {y} in

    1.116:27 and its absence in all other forms; one possibility is that in its independent use, the

    word bears a primary stress, while when it occurs as part of the compound expression {t.r(t)},

    it does not, resulting in different behaviour of the triphthong in the two instances, being

    preserved in the former, and collapsing in the latter.37

    There are, however, alternative explanations. Taking the numerals in 1.161:2730 as

    functioning adverbially better explains the use of cardinal numbers (i.e., {} not {d} in l. 29)

    than does Troppers Kardinalia fr Tageszhlung.38 However, thisproposal of Pardees has a

    further advantage he does not mention: it allows one to speculate that the first in the sequence

    may have been doubly marked for adverbial usage, both by accusative case and by use of enclitic

    {y}, in which case the morphology/at-ya/

  • 8/4/2019 Ugaritic Numbers

    11/40

    11

    this elsewheresuch as {mn} 8 /amnv /

  • 8/4/2019 Ugaritic Numbers

    12/40

    12

    possesses {i}.44 Nor does it behave like ramaic, in which /n/ becomes /r/. The extent to which

    the garitic lexeme {n} 2 is similar to its Hebrew reflex will now be examined, with the

    proviso that the latter is itself far from problematic.

    It is the feminine form that is most informative about the development of this lexeme in

    Hebrew. To explain the unique word-initial consonant clusterthe absence of any vowel at all

    following // is confirmed by the plosive pronunciation of /t/Tropper suggests the development

    t < *itt (vgl. samaritanisch-he.ittm).45 However, the reduction of a vowel in a closed

    syllable is problematic, and it may be preferable to allow for occasional instances of vocalic

    consonants as late as the proto-Hebrew stage (i.e., a developmenttyi /in-/; in the case of the feminine form

    {t}, where the sonorant /n/ has apparently assimilated to the follow /t/, assuming that the form is

    not like that of Hebrew (i.e., /t-/rather than /itt-/ or something similar) is only motivated by an

    unwillingness to reconstruct word-initial consonant clusters in Ugariticit should be borne in

    mind that this is the only instance of this phenomenon in Hebrew. I have mentioned that the

    absence of {n} in {t} indicates assimilation of /n/ to the feminine morpheme, which means that

    the allomorph [t] was used.46

    This is also the case in Akkadian, Aramaic, Hebrew, and even

    Arabic (in the by-formintni), a language in which the [t] allomorph is extremely rare.47

    44Sivan, 2001, 128.

    45Tropper, 2000, 62.121.

    46 Tropper, 2000, 62.121. possible exception is {bnt.rt} (4.146:8), but Cunchillos, Vita, and Zamora (2003,

    3156) consider a scribal error for {mnt}.47

    Moscati et al., 1964, 14.1. It is noteworthy that in Akkadianitta the [t] allomorph is used, even though this

    requires the helping vowel -a, because the absolute-case marker - more normally used with numbers would result

    in a prohibited word-final consonant cluster.

  • 8/4/2019 Ugaritic Numbers

    13/40

    13

    With regard to what follows the stem {n}, comparative evidencethat is, from all Semitic

    languages except Akkadiansuggests that ifcase vowels follow, then they should be those of

    the dual (nom. , obl. ). However, Ugaritic is remarkable insofar as the non-boundedness

    marker {m} never appears on {n}/{t}, even where the syntactic context demands the absolute

    state. Tropper gives two possible explanations for this, of which he favours the latter: (1) the

    numeral 2 usually appears in the construct state, and this has become seen as the normal form,

    and has been extended to the instances of the absolute state; (2) this represents the use of the

    absolute case with numbers in Ugaritic.48

    Arguing against the former is the absence of a similar development in any other language.

    However, one could equally argue that, apart from Amorite, the absolute state marked with the

    endings ora is quite rare in WS (some might say it only exists in remnants).49 The latter

    solution therefore relies either on positing an affinity between Ugaritic and Akkadian, as opposed

    to other WS languages, which may be methodologically suspect, or otherwise implies that, like

    Amorite, Ugaritic is more archaic than other WS with respect to the preservation of an earlier

    case system.50 Furthermore, if one is to posit the use of the absolute case in the Ugaritic numeral

    2, then one must then either explain why it would be restricted to this numeral or analyse the

    entire numerical system as functioning in the absolute case. On balance, then, however the

    48Tropper, 2000, 62.121.

    49For an up-to-date discussion of the development of the absolute case, see Tropper, 1999, and Hasselbach,

    forthcoming.

    50Troppers (2000, 62.121) suggestion that the first component of Hebrew n randt re is possibly

    absolute case rather than merely in construct to the word for 10 may be rejected considering that the Arabic reflex

    is inflected according to casenom. ()itn aara/()itnat arata; obl. ()itnay aara/()itnatay aratacontra

    Tropper, 1999, 190: die klassisch-arabische Kardinalzahl fr zwlf der Form (i)n aara, in welcher der

    Auslautvokal - von (i)n [sic.] offensichtlich die Endung des Absolutivkasus m.du. ist.

  • 8/4/2019 Ugaritic Numbers

    14/40

    14

    absence of the non-boundedness marker in {n}/{t} is to be explained, it is hardly certain proof

    of the absolute-case use with Ugaritic numerals.

    2: {klat} /kilat-/

    One lacuna in Troppers otherwise remarkably comprehensive treatment of garitic numerals is

    the treatment of the lexeme {klat}. It does not matter that this numeral, in contrast to {n}, which

    is used more generally, is used to qualify 2 of something specifically in a pairin fact, in all

    but one of the six attestations to qualify hands51and therefore ought better to be understood

    as both than as 2; insofar as numerals are being treated as nouns (in the broad sense), any

    noun qualifying another by determining a specific number of it should be considered a

    numeral.52 It is just a coincidence that Ugaritic only possesses forms with a feminine morpheme,

    and hardly a surprising one considering that the most common pairs (body parts) are feminine.53

    One might reasonably suppose that the [at] allomorph occurs by virtue of the prohibition against

    suffixing the [t] allomorph to a *qitl base.54

    Comparative evidence suggests that, as with {n}/{t} 2, the dual case vowels (nom.: , obl.:

    < *ay) should be suffixed that to the stem, which is entirely natural. 55 The form {klatnm} (1.14

    II:15, III:57) is variously explained, but crucially both Tropper and Pardee reconstruct the dual

    51Cunchillos, Vita, and Zamora, 2003, 1642.

    52For example, {kl} all is not a numeral since it does not qualify in terms of a specific number, but if one includes

    {rbt} in a treatment of numerals, considering the uncertainty as to whether this really indicates 10,000 of

    something, rather than just a loton this lexeme, see belowone should certainly include {klat}. Cf. the Geez

    cognate ke, which functions as the normal numeral 2.53 Cf. the other lexeme for 2, {}/{t}, wobei zwischen Zahlwort und Gezhlten Genuskongruenz (Tropper,

    2000, 69.121).54

    It is noteworthy that the Arabic cognate kit (Q 18:33)exceptionally uses the [t] allomorph, which has been

    allowed by the elision of consonantal //.55 Lipiski, 1997, 35.4b.

  • 8/4/2019 Ugaritic Numbers

    15/40

    15

    oblique case vowel and non-boundedness marker.56

    On the one hand, this is an argument against

    an absolute-case use of numerals in Ugaritic; on the other, the one instance of the form {klat} of

    this lexeme not in construct to hands, namely {klat.tqtnn} (1.23:57) both [women] squatted

    down, does not bear the expectedness non-boundedness marker. This, though, is not conclusive

    evidence in favour of absolute-case use of the numerals in Ugaritic; it could be an aberration, or

    otherwise formed by analogy to the other lexeme for 2 (whatever the reason forthe absence of

    {m} there is). The only observation that may be made regarding the classification of Ugaritic on

    the basis of the form {klat(nm)} is that, like the reflexes in the rest of the WS languages, the

    Ugaritic lexeme is formed from k, in contrast to Akkadian kian, which is from kll.57

    3: {l} = /al/, {lt} /alat-/

    While no internal evidence exists for the vocalisation of this numeral, comparative evidence is

    unanimous in suggesting the PS base *qat.58 With regard to the root, reflexes of P *lare

    found in Geez, OSA, and MSA (other root by-forms are also attested in the latter two).

    Reconstructing this as the original PS root, rather than an innovation of southern Semitic (this

    term, like northern below, is used in a strictly geographical sense), is confirmed by the Old

    Akkadian orthography sa-li-i-tim.59 Nothing may be inferred about the classification of Ugaritic

    56I certainly agree with Pardee (2003/2004, 8384), that T.s preferred explanation of the /n/ in klatnm as the

    consonantal element of the dual morpheme m which has undergone dissimilation from the following /m/ is forced

    in the extreme, and it should better be interpreted as the derivational morpheme /n/. 57

    Oppenheim, 1971, 353.

    58Moscati et al., 1964, 14.2.

    59 Lipiski, 1997, 35.8.

  • 8/4/2019 Ugaritic Numbers

    16/40

    16

    from the fact that it has taken part in the progressive distant assimilation >(which probably

    occurred through areal diffusion), since it is so widespread in northern Semitic.60

    It seems unlikely that the Ugaritic form of this numeral with the feminine morpheme used the

    allomorph [t] rather than [at], as in the Hebrew construct state. If this were the case, one

    might expect at least occasional instances of assimilation of R3 to the feminine morpheme, giving

    the unattested sequence **{lt}, hence my reconstruction /alat/.61 All that may be said with

    certainty is that if the [t] allomorph was used, the prohibited syllable structure **/CvC/ would

    have to be avoided, most likely through shortening of, as in Geezalastv.62

    4: {arb} /arba-/, {arbt} /arba(a)t-/

    Internal evidence for this numeral is limited, telling us only the quality (low) of the first vowel.

    However, as with 3, the cognate evidence is unanimous, and a form /arba/ may confidently

    be reconstructed back to PS.63

    While I agree that this lexeme clearly possesses a trilateral root in

    Ugaritic as in other languagesnamely, rb, as ordinals, discussed below, confirmthe

    prothetic // in the cardinal dates back at least as far as PS, and those forms without it in the

    60Tropper, 2000, 33.114.2.

    61Cunchillos, Vita, and Zamora, 2003, 3087. The two consonants are close, though not identical, in terms of

    location of articulation, making assimilation not implausible (cf. the similar assimilation, albeit with concurrent

    progressive voicing assimilation, iddaara < itaara in Arabic). The sound change *t> ttis not, however, among

    those listed by Tropper (2000, 33.113ff.) as occurring in Ugaritic.

    62Moscati et al., 1964, 14.1; the cited Hebrew(e.g., Gen 30:36) is in this respect aberrant, reflecting a

    proto-Hebrew form *atv with the theoretically impermissible sequence CvC (cf. b< btv [?]).63

    Moscati et al., 1964, 14.1.

  • 8/4/2019 Ugaritic Numbers

    17/40

    17

    descendent languages (e.g., Mehri and Harsusi) result from aphaeresis rather than retention of the

    original form.64

    Pardee criticises Tropper for not allowing the possibility that {arbt} represents /arbat-/,

    since, as was noted in relation to the numeral 3, the Hebrew construct form arba(e.g.,

    Num 7:7) uses the allomorph [t].65

    Because the assimilation *t > ttis not identified elsewhere, I

    agree in principle that {t} may indeed express /t/ rather than /at/; in practice, a decision must be

    made largely on the basis of personal methodology, specifically to what extent, and in what

    circumstances, the allomorph [t] was used in Ugaritic.66

    Analysis and conclusions on the

    variation between [t] and [at] in the numerals as a whole are presented below. It is to be noted

    that the suffixing of [t] to the stem /arba/ presupposes the use of case vowels on numerals, since

    otherwise the impermissible sequence **/CvCC/ (i.e., /bat/) would obtain.

    5 {m} /am-/, {mt} /amat-/

    There is no way of knowing whether the Ugaritic numeral 5 has a *qatil base, as Hebrew

    e, a *qattil base, like Aramaic a()e, or a *qatl base, like Arabic *ams. To assert

    64 Contra Lipiski, 1997, 35.7. The objection here is basically the same as was made above to his statements

    regarding Aramaic aand cognate forms. Despite masterly control of the data, Lipiskis stance the Proto-

    emitic root morphemes of the numerals three to ten can be established as follows: at-, rba-, ha-, sidt-,

    ab-, tn-, ti-, arwhich is questionable also with regard to the numerals 3 and 8, results from his

    overriding desire to reconstruct an unnecessarily (and unnaturally) symmetrical PS paradigm with monosyllabic

    bases for all the numerals.

    65Pardee, 2003/2004, 199.

    66Tropper, 2000, 52.213: Die Morphemvariante /-t/ ist im Ug. mit Sicherheit hufiger bezeugt als etwa im Ar., wo

    beinahe nur /-at/ gebraucht wird, anderseits aber gewi seltener als im Akk., wo /-t/ deutlich berweigtDer

    sukundre Ersatz von /-at/ durch /-t/ lt sich im g. meist auf konkrete phonet. Regeln zurckfhren. What he

    fails to note, however, is that of the extremely rare cases of the [t] allomorph in Classical Arabic, half are in

    numerals (namely, the words for 2, discussed above), the only other two of which I am aware being the primitive

    kinship terms utsister and bintdaughter.

  • 8/4/2019 Ugaritic Numbers

    18/40

    18

    otherwise would be to fall into the aforementioned pitfall of circular argument based on assumed

    classification of Ugaritic. It is therefore somewhat short-sighted of Tropper to reconstruct this

    numeral as /ami/, am(i)at/, with no discussion of possible alternatives for the base; this is

    presumably based on the genetic classification of garitic as North Canaanite, though I am

    anyway aware of no specific evidence indicating a *qatilbase for 5 in otherCanaanite

    languages. Moscatis proposals regarding Hebrew and Aramaic forms are interesting, but

    ultimately do not help to clarify the Ugaritic situation.67

    With regard to the form with the

    feminine morpheme, given that one cannot determine the stem with any certainty, all that may be

    said is that no forms {mt} (indicating assimilation of // to the allomorph [t]) are attested,though that does not rule out use of the allomorph [t], as in the Hebrew construct (e.g.,

    Num 18:16), and that use of the [t] allomorph would require a vowel (whether original or

    epenthetic) preceding // to avoid the prohibited sequence **/CCC/ (i.e., /mt/).

    6 {} /i-/, {t} /iat-/

    Assuming that OSA {s1d} reflects the original PS *sid, then the form of the numeral 6 in

    Ugaritic (and indeed in all the Semitic languages, including variant OSA forms) may be

    explained by a series of assimilations. This much is uncontroversial. The complication lies in the

    order in which these processes took place, and which languages shared in which changes. If, as

    Tropper would have it, and it is my intuition that he is correct, the development was * sid> *id

    >i, then there are ramifications for the position of NWS.68 This ordering would necessarily

    group Ugaritic closer to Canaanite than Aramaic, since Biblical Aramaici(Dan 3:1), for

    67Moscati et al., 1964, 14.2: Hebrew five is formed on the analogy of six (*a > i on the analogy of

    i); yriac five is formed on the analogy of four (* e > aeon the analogy ofarba).68

    Tropper, 2000, 62.161.

  • 8/4/2019 Ugaritic Numbers

    19/40

    19

    example, cannot have undergone the first of these assimilations.69

    If these assimilations occurred

    in the opposite order, it is technically possible that the second change only occurred in Ugaritic;

    asis the Canaanite reflex of PSand s alike, proto-Canaanite *ican be accounted for by the

    first assimilation alone.70Whatever the case may be, Loewenstamms suggestion that {d} (1.14

    II:31) is an archaic cardinal without assimilation is to be rejected as unnecessarily complicating

    the situation, and it results from a fundamental misunderstanding of the use of ordinals.71

    With regard to the form of the numeral 6 with the feminine morpheme, I opine that {t}

    probably represents /iatv/, because the [t] allomorph would result in the prohibited sequence

    **/CCC/ (i.e., /t/). True, the Hebrew construct form(Exod 16:26; cf. absolutei [Gen

    30:20]) simplifies the geminate cluster in order to avoid this, but there is no reason to believe that

    this necessarily happened in Ugaritic. Moreover, as mentioned in relation to the numeral 3,

    though there is no decisive evidence for the assimilation *t> ttin Ugaritic, on strictly phonetic

    grounds, it is not hard to imagine that it occurred, and there are no instances of the numeral 6

    written {t}.

    7 {b} /ab-/, {bt} /abat/

    Despite a total absence of internal evidence, the reconstruction of a *qatl base is supported by

    evidence from almost all Semitic languages.72 Moreover, the [t] allomorph of the feminine

    69This statement is not meant to be understood in terms of a genetic, family-tree model, but, since sound changes

    like those under discussion are the kind of features particularly prone to areal diffusion, instead refers to proximity

    within a wave model, which is proximity nonetheless, and a valid means of subclassification.

    70 For further ramifications of this order, as proposed by Lipiski (1997, 35.11), see the discussion of the numerals.71

    Loewenstamm, 1980, 312.

    72Moscati et al., 1964, 14.1. Only the slightly troubling Hebrewi and the corresponding decadeii (see

    Gen 4:24 for both) have apparent *qitl stems. However, these probably merely reflect peculiar Hebrew-specific

  • 8/4/2019 Ugaritic Numbers

    20/40

    20

    morpheme can be ruled out with reasonable confidence since, in Hebrew, whose construct forms

    of numerals with this morpheme use the allomorph [t] where possibleand in the case of 3,

    discussed above, even where theoretically not possiblewe find the formiat- (e.g., Gen

    8:10).

    8 {mn} /amn/, {mnt} /amn?t-/

    s for the vocalisation of the garitic form of the cardinal 8 without the feminine morpheme,

    considering the evidence of the other CS languagesArabic (nom./acc.)anin/aniyan,

    Aram tmne, and Hebrewonthe reconstruction of a development /anv1/ */aniyv1/in proto-Ugaritic, with various triphthong-collapse developments occurring

    independently in each language, is quite uncontroversial.73 This is Troppers position, even

    though, by his own admission, it contravenes his understanding of the behaviour of triphthongs

    in Ugaritic, in which he argues that the sequence /iyv/ is regularly preserved (as it apparently is,

    for instance, in the numeral {mnym} = /amniyma/ 80).74

    Complications begin to arise when one wishes to address the reconstruction of the form of this

    numeral bearing the feminine morpheme, {mnt}. Because both the proto-forms of both Hebrew

    (abs./cstr.)mon/monatand Aramaic tny must have used the allomorph [at] rather than

    [t] (i.e., *aniyat), to suggest that Ugaritic did otherwise would implicitly infer some

    behaviour in *qvtlatnouns (cf. k, ki, kalamb. For discussion, see Joon, 1991, 88Ca*, 100d, who

    sees no problem in identifying a *qatlbase for 7).73

    Moscati et al., 1964, 14.1. In fact, the data from the other Semitic languages do not contradict such a proposal,

    but for simplicitys sake, I exclude them from the discussion here.74 Tropper, 2000, 33.322.3d, 62.181. Pardee (2003/2004, 106n394) states, T. indicates both /-iyt-/ and /-iyat-/ as

    becoming /-t-/ in gariticNowhere in the section on triphthongs, but at various points further on in the grammar.

    This is best taken as an example of the inconsistency sometimes found in Troppers grammar, and more generally as

    evidence that the Ugaritic triphthong situation is not yet fully understood and requires further study.

  • 8/4/2019 Ugaritic Numbers

    21/40

    21

    difference in behaviour of Ugaritic (specifically that the [t] allomorph was more common here,

    an idea addressed in the conclusions section), and perhaps, by extension, in its subclassification.

    Therefore, in order to arrive at the vocalisation /amnt-/ from a proto-form */amniyat/,

    Tropper posits an intermediate stage (namely *aaniyt), where syncope of the a of the feminine

    morpheme has been occasioned by a rule CvCvCvCvCv > CvCvCvCCv.75

    It is true that the

    aforementioned rule is undesirably specificcan it in fact be applied to explain any other

    data?so Pardee is justified in his (implicit) query whether Ugaritic may have undergone a

    development similar to Hebrew (which apparently underwent a contraction *iya > a in these

    forms).

    76

    It must be acknowledged, however, that the sole basis for reconstructing *aniyat>

    anat(rather than a more cross-linguistically expected >ant, for example) is a presumed

    affinity between Hebrew and Ugaritic, rather than any internal evidence.

    9 {t} /ti-/, {tt} /tiat-/

    I consider this the least problematic of all the cardinal numerals. There is no internal evidence to

    aid reconstruction, but the comparative evidence points unanimously to a *qitl base. The same

    observations regarding the improbability of the allomorph [t] having been used that were made

    above in reference to the numeral 7 also apply here (cf. Hebrew tia- [Num 34:13]).

    10 {r} /ar-/, {rt} /aar(a)t/, {rh}

    Let us begin with the less problematic matter first (another instance in which we have no internal

    evidence), that of the base of this numeral in Ugaritic, and Semitic more generally. Joon

    concisely sums up the distributionof these two bases in the stems of Hebrew forms[t]he form

    75Tropper, 2000, 33.243.12.

    76Pardee, 2003/2004, 106.

  • 8/4/2019 Ugaritic Numbers

    22/40

    22

    of the masc. is a qatl while the form of the fem. is a qatal.77 Since the same distribution obtains

    in Arabic, and the Aramaic data reflect the expected development according to the normal

    vowel-reduction and epenthesis rules of that language, there is no reason to suspect that it did not

    in Ugaritic as well.78 With regard to the feminine morpheme {t}, we have a situation similar to

    that described for other numerals; while there is no internal evidence for this numeral

    specifically, considering Hebrew construct forms like r qlim (Lev 27:5), it would be

    short-sighted to rule out the possibility of the use of the [t] allomorph.

    Less straightforward is the garitic form {rh} of the cardinal 10. Various proposals made

    to explain its etymology will now be addressed. These have a direct bearing on the classification

    of Ugaritic insofar as they address the relation to Hebrew and Aramaic reboth spelt {rh}

    in the earliest instances, ramaic developing to {sry} and {sr} over timeas part of the teen

    compound numeral qualifying feminine nouns, or to the Akkadian absolute case of numerals. In

    assessing the value of the different explanations, a few things should be borne in mind: (1) this

    form is used only in nonliterary texts in Ugaritic,79 (2) like Hebrew re, with the possible

    exception of {rh.snm} (4.174:2), it is only used in compound teen numerals, although unlike

    Hebrew, it may (rarely) be used to qualify a masculine noun.80

    77Joon, 1991, 100d. The issue of chiastic concord confuses issues somewhat, so it is well to illustrate with

    examples. The numeral qualifying feminine nouns has a stem with the base *qatl, as r nim (Gen 5:14) or

    e re n(Gen 5:10); that qualifying the masculine noun a *qatal base, as r gmallim (Gen 24:10) and

    ir nim (Exod 26:25). This principle apparently holds, then, regardless of whether the numeral is

    independent or occurs as part of a compound numeral.

    78Hetzron, 1977, 185; Moscati et al., 1964, 14.1.

    79Hetzron, 1977, 183.

    80Tropper, 2000, 62.201.

  • 8/4/2019 Ugaritic Numbers

    23/40

    23

    It is because of the first of these considerations that Blau suggests that this form may simply be

    substandard orthography, exceptionally using {h} as a mater lectionis for // (< */ayv/).81 Such

    an explanation could also account for the Hebrew and Aramaic forms, though the

    aforementioned distribution, restricted to compound numerals, would remain a mystery, and the

    origin of the *ay morpheme would still require explanation. Hetzron posits Hebraic intrusion into

    nonliterary language: [I]t could also be the case that this form was actually borrowed from a

    proto-Hebrew-type dialect that still pronounced the final -h of the feminine ending (-t > -h > ),

    but whose writing system perpetuated the second stage.82 This explanation is unsatisfactory

    because the necessary implication that the {h} of the Hebrew and Aramaic feminine morpheme

    was ever anything other than a mater lectionis is, in my opinion, highly suspect. Moreover, it

    cannot account for the final vowel /e/ of Hebrew and Aramaic.

    Both of these explanations require that the Ugaritic form is cognate to those cited from Hebrew

    and Aramaic. Pardee is adamant that that is the caseclaiming an irregular form in Ugaritic

    that does not correspond to the Hebrew form is hardly a viable solution83and feels that all the

    data may be explained if one takes {h} as cognate to the Akkadian terminative ending i. On the

    one hand, among the advantages of this proposal are that it does not ignore the distribution of the

    morpheme, which other explanations are bound to do: [T]he exclusive use ofrh with a

    number noun in the digits to express a number in the teens reminds one of the use of the

    preposition l in compound numbers.84 On the other, if one is to apply this explanation to

    Hebrew reas well as garitic {rh}, it has the serious drawback that it precludes identifying

    81 Hetzron, 1977, 183184.82

    Hetzron, 1977, 184.

    83Pardee, 2003/2004, 82.

    84Pardee, 2003/2004, 82.

  • 8/4/2019 Ugaritic Numbers

    24/40

    24

    the so-called he localis of Hebrew with the Akkadian terminative; it is not possible to explain

    how the same morpheme would develop sometimes into -, and elsewhere into .

    Finally, Tropper makes two other proposals that do not rely on identifying Hebrew and

    Aramaic re as etymologically related to garitic {rh}. The first, that {h} steht fr ein

    spezifisches, vom Morphem /-(a)t/ zu unterscheidendes Femininmorphem, he rightly dispenses

    with on the grounds that a similar morpheme occurs neither in other Ugaritic numerals, nor

    elsewhere in Semitic.85

    He ultimately decides that {h} in {rh} is a phonetic variant of the

    feminine morpheme, resulting from aspiration (Verhauchung) of word-final /t/ ([arah] [t]. Lipiskis (1997, 35.20) linking

    the Ugaritic form with Mari Amorite me-etand Late Babylonian (t), and subsequently vocalising it as eetor

    t, deserves no credence and does not conform to either the phonological system (since he posits a short vowel /e/)

    or alif-orthography practices (he otherwise suggests {i} is a mater lectionis for // resulting from quiescence of

    syllable-closing //) best reconstructed for garitic.96

    Huehnergard, 1987, 144; contra Tropper, 2000, 62.51.

  • 8/4/2019 Ugaritic Numbers

    28/40

    28

    Akkadian, than a helping vowel (e.g., *ita or something similar) in order to avoid the

    prohibited sequence **/CvCC/ (i.e, **it).

    1000: {alp} = /alp-/

    Reconstructing a *qatl base from lp meaning 1000 at least as far back as C is entirely

    unproblematic.97

    Ugaritic has then preserved this form. The only observation its presence in

    Ugaritic allows in relation to the classification of this language is to align it with WS languages,

    as expected, in contrast to Akkadian, which uses the lexeme .

    10,000 {rbt} = /ribbat-/

    Insofar as the Ugaritic lexeme {rbt} is abgesehen von einer formelhaften Wendung in 5,9 I:5

    nur in der Poesie bezeugt,98frequently in parallel to {alp(m)} thousand(s), one may wonder

    to what extent it is, properly speaking, a numeral referring to 10,000 specifically rather than

    just a word meaning a great number. Whatever the case may be, two observations are in order

    regarding the morphology of this form. Firstly, comparative evidenceAramaic(ising) ribbo

    (Dan 7:10) and Hebrew r (Jdg 20:10), while Akkadian has both ribbatu and rabbatu99

    suggests the proto-vowel *i in the first syllable.100

    Secondly, it seems that proto-NWS may have

    had by-form bases *qillatand *qilalat.101In this case, the plural {rbbt} = /rababt/ (e.g., 1.4 I:28)

    97Moscati et al., 1964, 14.6.

    98Tropper, 2000, 62.9.

    99Reiner and Roth, 1999, 14, 314.

    100Contra Tropper (2000, 62.914), who vocalises /rabbat-/.

    101 Pardee, 2003/2004, 103: One must query whetherrbtis simply a secondary form ofrbbt; since Hebrew has the

    same two forms, they would appear at least to go back to proto-West emitic. Tropper (2000, 62.914) is even

    more confidentes gibt keinen Zweifel da rbbtdie typologisch ltere Form darstellt. I am doubtful as to whether

    the development was necessarily in this direction, because generally *a does not syncopate between R2 and R3 in

  • 8/4/2019 Ugaritic Numbers

    29/40

    29

    is merely the plural of the form *rababat, no longer preserved in Ugaritic, rather than the plural

    of a *qillat-based singular with dissociation of R2 and R3. Drawing conclusions about the

    classification of Ugaritic from his numeral, though, is difficult; it is very likely just a coincidence

    that Ugaritic has preserved the monosyllabic by-form like Aramaic, while Hebrew has preserved

    the disyllabic one.102

    Ordinals, 1st

    The ordinal 1st exhibits remarkable heterogeneity in the emitic languages. The various

    languages use different forms from sundry verbal roots or primary nounsunsurprisingly, these

    usually relate to primacy, precedence, or similar concepts and in several cases are terms for body

    parts, namely head and faceand among the Semitic languages, only Akkadian and OSA

    form this numeral from the corresponding cardinal.103

    I am in agreement with Loewenstamm that

    the explicit expression of the ordinal 1st was not a feature of P;104 indeed, considering

    Hebrew rion and Aramaic qady, it is clear that no single lexeme may even be reconstructed

    as far back as proto-NWS. This should not be particularly surprising, though, since in the most

    forms from II=III in W, and in the comparable by -form *libb~*libab, I feel confident that the second form is the

    neological development.

    102Even the second part of this claim is not beyond doubt and relies on identifying Hebrew ribbo (e.g., Dan 11:12)

    as an Aramaism.

    103Dombrowski and Dombrowski (1991, 370) summarise the situation regarding the various Semitic means of

    expressing 1st. It is noted that the derivation of this ordinal (in contrast to other ordinals) from a source other than

    the corresponding cardinal is a situation not unfamiliar in European languages (e.g., French un~premierversus

    deux~deuxieme and Dutch een~eerstversus twee-twede).

    104Loewenstamm, 1980, 311.

  • 8/4/2019 Ugaritic Numbers

    30/40

    30

    common context in which the notion of 1st featuresin a list of ordinals, or at least in contrast

    to another ordinalthe idea is implicit in the noun itself.105

    Further, it has been suggested that the ordinal 1st in Ugaritic may be expressed simply with the

    cardinal 1, Sivan citing only {ym ad} (1.115:14), Tropper also {lb.ad} (4.146:1).106 Pardee

    queries whether the latter example is really an ordinal usage, pointing out that the word for

    one is often used in lists that otherwise consist of ordinals (as in the Biblical-Hebrew creation

    story: day one, the second day, etc.).107 I would take this further and say that in English it is

    entirely natural to use cardinal numerals when enumerating items in a list, without any loss of a

    sense of ordinal continuity.

    108

    Therefore, one might wonder to what extent such use is

    characteristic of Ugaritic (and Hebrew apparently), rather than being a linguistic universal.

    Finally, on the basis of the unique {ym pr} first day (4.279:1), it seems that the lexeme

    {pr} may have been emerging as an ordinal 1st in garitic. This could reflect the incipience

    of a semantic bleaching from the meaning first fruits or otherwise result from the meaning

    precedenceof the root pr(which conforms to the formation of the ordinal 1st in other

    Semitic languages, as mentioned above).109 In any case, this use is unique, thus calling {pr} the

    Ugaritic for the ordinal 1st would be to identify a productivity that simply does not exist. By

    105 For this idea, see ivan, 2001, 93; Tropper, 2000, 63.311. ee, for example, {ym.wn.l.rb.ym.y !m.d.ym} a

    [first] day and a second, a third, fourth day, a fifth, a sixth day (1.14 IV:4445), for which the inclusion first in an

    English translation is admittedly more natural sounding if the noun day is translated as definite. 106

    Sivan, 2001, 93; Tropper, 2000, 368369.107

    Pardee, 2003/2004, 205.

    108For example, if one were to say, step one: beat the eggs; step two: mix in the flour; step three: place in a pre-

    heated oven, there would be no doubt that the order of the steps is integral to the meaning. One might argue that the

    idea of ordinal procession lies in the noun step, but this cannot be maintained of strike in the expression strike

    3, which is, of necessity, the third in a sequence of strikes. 109 For the first possibility, see ivan, 2001, 93; for the latter, Tropper, 2000, 63.312. For the lexeme, see del Olmo

    Lete and anmartn, 2003, 679.

  • 8/4/2019 Ugaritic Numbers

    31/40

    31

    Troppers own admission, {bym.d} means am Neumondstag, and so citing it as an ordinal

    erster Tag is somewhat disingenuous; one could only reasonably say it bore the specific

    meaning of1st if it was attested with such a meaningcases like {tr.d} (1.14 II:48), in

    which {tr} means new, recent, that is, newly-wed, cannot be considered ordinal usesin a

    context unrelated to the word new-moon.110 In summary, the absence of an ordinal 1st

    reveals little about the classification of Ugaritic, other than confirming a degree of separation

    from any other NWS language, specifically characterising it as more archaic than its NWS

    neighbours and Akkadian, inasmuch as it has not developed one.

    2nd and higher

    In Ugaritic, as elsewhere in emitic, above 10th, cardinal numerals can perform the function of

    ordinals.111

    It is clear, however, from the forms {rb} 4thand {d} 6th, which differ in

    orthography (and therefore necessarily in phonology) from their respective cardinals, that there is

    a distinct set of ordinals at least from 2 to 10, as in other Semitic languages. 112 (The sequence of

    {m}-preformative forms in 1.14 I:1620 are not, strictly speaking, ordinals, but rather passive

    participles from a derived verbal stem [D-stem], i.e., {mrbt} =/murabba(a)t-/, lit. the one made

    fourth.) Pardee concisely sums up the problems with reconstructing their vocalisation:

    T. cites Akkadian, Arabic, and Ethiopic as a basis for reconstructing the ordinals numbers as

    formed on a /qtil/ base, then concludes Das g. unterscheidet sich somit vom kan. und aram.

    Befund This positive presentation does not adequately convey to the reader the simple fact

    110Tropper, 2000, 63.313; cf. ides, which would never be used in a context other than referring to a day of the

    month (specifically, that of the full moon), so listing it as a lexeme with the meaning 13th or 15th (which is the

    day of the month on which the ides falls) would be unreasonable.

    111Dombrowski and Dombrowski, 1991, 1; Sivan, 2001, 94.

    112 We may leave aside the questionable instances of {nt}, the inclusion of which may jeopardise the analysis.

  • 8/4/2019 Ugaritic Numbers

    32/40

    32

    that we have no way of knowing whether Ugaritic had the more broadly attested form or the

    Northwest emitic /qatl/ form, or, for that matter, yet another.113

    There is, however, a modest amount of evidence that, while it may not decisively answer the

    question, it is certainly pertinent when considering what the Ugaritic ordinals may tell us about

    the classification of this language.

    It is clear because of the absence of {y} in their orthography that Ugaritic numerals, unlike

    their Hebrew and Aramaic counterparts, do not bear the gentilic ending; this does not, of course,

    rule out that they may have had the same *qatstem, althoughassuming the use of the gentilic

    is an innovation, and probably not an independent oneit would make Ugaritic more archaic

    than the rest of NWS in this respect, and to some extent differentiated from Canaanite and

    Aramaic.114 Certainly the base of the numerals must have been disyllabic to avoid assimilation in

    6th as occurs in the cardinal, and Pardee further observes that, for the same reason, it may be

    advantageous to reconstruct a long vowel in at least one syllable. By comparing the various

    forms, it may be possible to further narrow down which vowels these are.

    I begin with the assumption that in order to form the Ugaritic ordinals, a triradical root must be

    used, even where the cardinals do not strictly adhere to this. Thus, from {arb(t)} 4, one

    extracts rb and forms {rb}4th. (Whence dof {d} 6th is derived is discussed below.)

    The numeral {mn} 8th does not differ orthographically from its cardinal, but that is not to say

    it has not undergone a similar tri-radicalising process, which cognate evidence suggests it

    has.115 This means the form {n} 2nd should result from triphthong collapse of R3 =y (i.e.,

    */vnvyv/ > /vnv/). ccording to Troppers understanding of triphthong behaviour in garitic,

    113 Pardee, 2003/2004, 202.

    114Tropper, 2000, 63.111.

    115Here I disagree with Pardee (2003/2004, 202203n766), who argues that the garitic ordinal {mn} 8th may

    have retained its quadriconsonantal character (i.e., /amn/ < */amnyv/).

  • 8/4/2019 Ugaritic Numbers

    33/40

    33

    neither of the sequences /iyv/ nor /yv/ ordinarily collapses in Ugaritic.116 It seems likely that the

    triphthong with the long vowel would be less likely to collapse, which means, given a choice

    between the bases *qtiand *qatfor the formation of Ugaritic ordinals, the former is to be

    preferred.117

    Finally, as mentioned above, Lipiski claims that in gariticthe formfirst implies the

    regressive assimilation d>and then a second regressive assimilation>.118 However,

    the existence of the garitic ordinal {d} argues against this since there would be no stage at

    which a speaker could analyse the root of the cardinal numerals 6 as dand form the

    corresponding ordinal accordingly. We are therefore obliged to reconstruct the assimilations

    resulting in the attested form {} of the cardinal as taking place in the other order: *sid > *id>

    i.119 This suggests that the garitic ordinal form {d(t)} 6th (discussed below) is more

    archaic than those of Hebrewii, since unlike them, it was formed before the latter assimilation

    took place (i.e., in the second of the diachronic stages listed).120

    Such a position contains the

    116 Tropper (2000, 33.322.2, 33.322.42) in fact identifies {n} as an exception to the preservation of /iyv/.117

    That Ugaritic has a unique stem from which it forms its numeralsunlike either of those used in southern

    Semitic and Akkadian, on the one hand, or in the rest of NWS, on the otherstrikes me as fairly unlikely.118 Lipiski (1997, 35.11) argues that the widespread assimilation of R2 to R3in the cardinal 6 in various emitic

    languages supports this assimilation preceding that of R1 to R3 in Ugaritic. However, such an assimilation is an

    entirely natural occurrence anywhere these two phonemes are in contact (cf. Geez, where the form sssu without a

    feminine suffix may undergo this assimilation while that with [sdstu] does not), and no realistic objection can be

    raised to it being an independent parallel development in several of these (groups of) languages.

    119 Cf. Tropper, 2000, .62.161. The same applies to the D-stem from which {mdt} (1.14 I:19) is formed.120

    Cf. Arabic, where the Classical ordinalsdis6th shows a stage in which R3 had assimilated to R1, while the

    neological (low-prestige) Damascene counterpartstit(Cowell, 2005, 316) is formed after mutual total-contact

    assimilation of R2 and R3 had occurred in the cardinal (i.e., *sids > sitt, with regressive devoicing and progressive

    assimilation in manner of articulation). Aramaic, unlike Hebrew, cannot have undergone assimilation of R1 to R3;

    the ordinal iy is instead developmentally similar to the aforementioned Damascene form.

  • 8/4/2019 Ugaritic Numbers

    34/40

    34

    implicit suggestion that the Ugaritic ordinals most likely possess a *qtibase (like Classical

    Arabicsdis, Old Assyrianditu, and Geezsds), preserved from PS.

    Conclusion

    In conclusion, although often the unvocalised nature of alphabetic Ugaritic and the lack of

    syllabic evidence leave us in the dark, there is a modest amount of internal evidence, often

    indirect and the result of several stages of argument, that allows some analysis of the phonology

    and morphology of the cardinal and ordinal numerals in Ugaritic. The contribution this analysis

    may make to a discussion on the classification of Ugaritic will now be summarised. It ought to

    be reiterated first, though, that such a classification is not exclusively genetic, but must

    incorporate elements of a wave model of areal diffusion of features.

    The cardinal {ad} 1, not exhibiting aphaeresis, is more similar to its Canaanite reflex than

    its Aramaic one. It probably did not have a geminated // (i.e., {ad} = /aad-/); whether this is

    an isogloss separating Ugaritic from all other Canaanite languages, or gemination here is instead

    a phenomenon restricted to Hebrew, cannot be determined, though my instinct would favour the

    latter option. The morphology of the other word for 1, {t(y)}, is far less straightforward,

    although I have suggested the possibility that {y} in the single attestation of {ty} 1.161:27 may

    not be part of the lexeme, thus obviating the need for contradictory explanations of the

    morphological development of {t} and {ty}. t the very least, thislike other instances in

    the morphological analysis of numerals, specifically 2 and 8highlights the difficulty of

    understanding the behaviour of triphthongs in Ugaritic, and of explaining it in terms of a regular,

    systematic set of sound changes. Whatever the case, the morphology of {t(y)} is similar to

    Hebrew (and Aramaic), in distinction to Akkadian and OSA, appearing as it does with no token

  • 8/4/2019 Ugaritic Numbers

    35/40

    35

    of an /n/, whose loss may be considered an innovation of NWS. In usage, too, Hebrew and

    Ugaritic share a development that neither Akkadian nor OSA do, a movement away from

    independent use of this lexeme towards restriction to the context of the compound numeral 11.

    It is not possible to tell exactly how similar the phonology of {n}/{t} 2 is to its Hebrew

    counterpart, which behaves most peculiarly, although there is a clear divide from the other main

    NWS language, Aramaic, as well as from Arabic. The absence of a non-boundedness marker on

    this numeral, regardless of syntactic context, is certainly an isogloss separating Ugaritic from all

    other WS languages, but I consider it most unwise to assume it constitutes a Ugaritic-Akkadian

    isogloss, namely, the use of the absolute case for numerals. The other piece of evidence cited in

    support of this idea, the form {rh}, is also equivocal in this respect. What scant alphabetic

    evidence we have supports the notion that Ugaritic numbers took case vowel like any other

    nouns, as they do in Arabic, for example (and indeed all WS languages prior to the breakdown of

    the case system). This fits perfectly within the undisputed classification of Ugaritic as a WS

    language and should be seen within the context of a broader methodological rejection of

    Akkadocentricism in the study of Ugaritic.

    The ordinal {d} 6th allows us to order the assimilations that have resulted in the cardinal

    {(t)} 6 in garitic: *sid > *id > i. Aramaic cannot have undergone the first of these

    assimilations, while in Canaanite the merger of PS *s and *would effectively have the same

    result as it; considering how infrequent distal assimilation is, I do not find it implausible that the

    Ugaritic assimilation may have been triggered by contact with Canaanite languages, in which R1

    and R3 were the same phoneme.121 garitic {d} 6th also suggests that the formation of

    121One could, of course, posit the counterargument that Arabic has a similar, but inverse, assimilation of R3 to R1 (as

    evidenced in the ordinalsdis6th). Moreover, there is admittedly no evidence of a stage where the Canaanite root

    from which numerals were formed was d.

  • 8/4/2019 Ugaritic Numbers

    36/40

    36

    ordinals was of a more archaic nature than that of Canaanite and ramaic. This, in concert with

    the absence of {y} in the ordinal {n} 2nd, provides what I believe is a persuasive argument

    that Ugaritic retained an archaic base *qtifor ordinals, in contrast to the rest of NWS, which

    innovated a form *qat+ gentilic morpheme in this function.

    With regard to the decades, their vocalisation cannot be reconstructed based on internal

    evidence alone, though comparative evidence suggests they are plural type formations of their

    unit counterparts. The most noteworthy observation, though, is the isogloss shared by Ugaritic

    with all the other NWS languages, in contradistinction to Arabic, that they always retain their

    non-boundedness morpheme, even when preceding the qualified noun (a context in which one

    might imagine a construct form occurring). This is best understood as some kind of reanalysis

    process, whereby /m/ or /n/ is not viewed as a plural non-boundedness morpheme, but rather as

    part of the lexeme itself. Such an interpretation offers the advantage of explaining why decades

    whose unit counterparts are segholate forms possess monosyllabic stems in Hebrew (and

    possibly Ugaritic); they are not seen as plurals, but as lexemes in their own right.

    The final point to be made does not really relate to the position of Ugaritic, but is rather an

    interesting observation that has arisen in the course of this study regarding the distribution of the

    [at] and [t] allomorphs of the feminine morpheme in numerals. Ugaritic is the only Semitic

    language to use it in the numeral 100 ({mit} = /mit-/). Admittedly, here, it is an invariable

    part of the lexeme, rather than the feminine morpheme suffixed onto a stem, so this may not be

    entirely pertinent. However, {at} = /aatt-/ 1 and {t} = /itt-/ 2 can only be explained as

    assimilation of the final consonant of the stem to the [t] allomorph, while forms like {arbt} are

    ambiguous, and could represent either /arbaat-/or /arbat-/. Semitic languages show a

    tendency, where possible, to use the [t] allomorph; Hebrew generally uses the [t] allomorph on

  • 8/4/2019 Ugaritic Numbers

    37/40

    37

    numerals in the construct state (and, like ramaic, always on 1 and 2), while rabic uses the

    [t] allomorph on the numerals 2 and both (optionally in the first case), even though this

    allomorph is otherwise all but absent from Arabic.122

    Considering the form {klat}, the third sign

    of which confirms the use of the [at] allomorph, it is perhaps best to assume that in Ugaritic, the

    allomorph [t] was used with any stems with which it would not create a prohibited syllable

    structure: 1, 2, 4, 8, possibly 5 (depending on the vocalisation of the stem, which

    cannot be determined), and 10 (assuming in garitic, as elsewhere in W, the stem to which

    the feminine morpheme was attached was disyllabic). The allomorph [at], then, was used to

    avoid the sequences **/CvC/ and /CCC/ in 3, 6, 7, and 9. ltimately, though, this is farfrom certain, because in those languages where the distribution of these allomorphs is generally

    phonologically motivatedAkkadian, Aramaic, and Ethiopicthe same principles do not

    always apply with the numerals. Moreover, the Hebrew formmust have developed from

    *atv, whose sequence /CvC/ (i.e., //) is theoretically impermissible in proto-Canaanite.

    To conclude, the analysis of the phonology and morphology of Ugaritic cardinal and ordinal

    numerals confirms the North Canaanite position of Ugaritic, placing it within NWS, and

    further, to a certain extent, aligning it closer to Hebrew (or better, Canaanite) than Aramaic, not

    necessarily (only) in genetic, but likely also (or rather) in areal terms. Ugaritic numerals possess

    certain archaic features (such as the form of the ordinals), but this is in no way to say that they

    have preserved the absolute-case use of numerals, likely a remnant of PS, still found in

    Akkadian. Instead, though it may sound trite, the best broad analysis of the Ugaritic numeral

    system is that it is wholly WS in character, having undergone, or at least being in the incipiency

    of, certain NWS innovations more typical to Canaanite than Aramaic.

    122Its retention here, and otherwise only in utsister and bintdaughter, girl, is likely due to the primary,

    archaic nature of these words.

  • 8/4/2019 Ugaritic Numbers

    38/40

    38

    Bibliography

    Blaek, V. 1999.Numerals: Comparative-Etymological Analyses of Numeral Systems and their

    Implications. Brno: Masaryk University.

    Blau, J. 1978. Hebrew and North West emitic: Reflections on the Classification of the emitic

    Languages.Hebrew Annual Review 2: 2144.

    Brockelmann, C. 1908. Grundriss der vergleichenden Grammatik der semitischen Sprachen,

    volume 1:Laut- und Formenlehre. Berlin: Reuther and Reichard.

    Cowell, M. 2005.A Reference Grammar of Syrian Arabic. Washington DC: Georgetown

    University Press.

    Cunchillos, J.-L., J.-P. Vita, and J.-. Zamora. 2003.A Concordance of Ugaritic Words. Madrid:

    Laboratorio Hermeneumtica.

    Dietrich, M., O. Loretz, and J. anmartn. 1976.Die keilalphabetischen Texte aus Ugarit.

    Kevelaer: Butzon and Bercker.

    Dombrowski, F., and Dombrowski, B. 1991. Numerals and Numeral ystems in the Hamito-

    emitic and other Language Groups. In Semitic Studies in Honor of Wolf Leslau on the

    Occasion of His Eighty-fifth Birthday, volume 1, edited by A. Kaye, 240281. Wiesbaden:

    Harrassowitz.

    Faber, . 1997. Genetic ubgrouping of emitic Languages. In The Semitic Languages,edited

    by R. Hetzron, 315. London: Routledge.

    Garr, W. 1985.Dialect Geography of Syria-Palestine, 1000-586 B.C.E. Philadelphia: University

    of Pennsylvania Press.

    Greenberg, J. 1950. Root Patterning in emitic. Word2: 162181.

  • 8/4/2019 Ugaritic Numbers

    39/40

    39

    Hasselbach, R. 2005. Sargonic Akkadian: A Historical and Comparative Study of the Syllabic

    Texts. Wiesbaden: Harrassowitz.

    Hasselbach, R. Forthcoming. Grammatical Roles and Relations in emitic. Chicago: The

    Oriental Institute.

    Hetzron, R. 1977. Innovations in the emitic Numeral ystem.Journal of Semitic Studies

    22/2: 167201.

    Hfner, M. 1943.Altsdarabische Grammatik. Wiesbaden: Harrassowitz.

    Hoftijzer, J., and K. Jongeling. 1995.Dictionary of North-West Semitic Inscriptions. 2 volumes.

    Leiden: Brill.

    Huehnergard, J. 1987. Ugaritic Vocabulary in Syllabic Transcription. tlanta: cholars Press.

    Joon, P. 1991.A Grammar of Biblical Hebrew. 2 volumes. Revised and translated by T.

    Muraoka. Rome: Editrice Pontificio Istituto Biblico.

    Lipiski, E. 1997. Semitic Languages: Outline of a Comparative Grammar. Leuven: Peeters.

    Loewenstamm, . 1980. The Numerals in garitic. In Comparative Studies in Biblical and

    Ancient Oriental Literatures, edited by K. Berghof et al., 310319. Kevelaer: Butzon and

    Bercker.

    Moscati, S., A. Spitaler, E. Ullendorff, and W. von Soden. 1964.An Introduction to the

    Comparative Grammar of the Semitic Languages: Phonology and Morphology. Wiesbaden:

    Harrassowitz.

    Del Olmo Lete, G., and J. anmartn. 2003.A Dictionary of the Ugaritic Language in the

    Alphabetic Tradition. 2 volumes. Leiden: Brill.

    Oppenheim, A (ed.). 1960. The Assyrian Dictionary of the Oriental Institute of the University of

    Chicago.I/J, volume 7. Chicago: The Oriental Institute.

  • 8/4/2019 Ugaritic Numbers

    40/40

    Oppenheim, A (ed.). 1971. The Assyrian Dictionary of the Oriental Institute of the University of

    Chicago.K, volume 8. Chicago: The Oriental Institute.

    Pardee, D. 2003/2004. Review of Josef Tropper. garitische Grammatik.Archiv fr

    Orientforschung 50: 1404.

    Rainey, A. 1996. Canaanite in the Amarna Tablets. 4 volumes. Leiden: Brill.

    Reiner, E., and Roth, M. 1999. The Assyrian Dictionary of the Oriental Institute of the University

    of Chicago.R, volume 14. Chicago: The Oriental Institute.

    Simeone-Senelle, M.-C., 1997. The Modern South Arabian Languages. In The Semitic

    Languages, edited by R. Hetzron, 378423. London: Routledge.

    Sivan, D. 2001.A Grammar of the Ugaritic Language. Atlanta: Society of Biblical Literature.

    Soden, W. von. 1995. Grundriss der Akkadischen Grammatik. 3rd edition. Rome: Editrice

    Pontificio Istituto Biblico.

    Testen, D. 1995. econdary Vowels in emitic and the Plural Pronominal Endings. In

    Analecta Indo-europaea Cracoviensia, vol. 2:Kuryowicz Meoria Voue, edited by W.

    Smoczyfiski, 543551. Krakow: In Officina Cuius Nomen Universitas.

    Tropper, J. 1994. Is garitic a Canaanite Language? In Ugarit and the Bible, edited by G.

    Brooke, A. Curtis, and J. Healey, 343353. Mnster: Ugarit Verlag.

    Tropper, J. 1999. Die Endungen der semitischen uffixkonjugation und der bsolutivkasus.

    Journal of Semitic Studies 44/2: 175193.

    Tropper, J. 2000. Ugaritische Grammatik. Mnster: Ugarit Verlag.