A Concise Historical Morphology of Biblical Hebrew

appendix A Concise Historical Morphology of Biblical Hebrew

We have considered the most important sound changes from Proto-Northwest-Semitic to Biblical Hebrew. This appendix reviews the consequences of these developments for the historical morphology of the inflected word classes of Biblical Hebrew: pronouns, nouns and adjectives, numerals, and verbs. Frequent reference is made to the chapters discussing individual sound changes.

1 Pronouns

The personal pronouns, both independent and suffixed, are discussed in detail in Chap- ter 8. :derives from the old genitive *ḏī הֶז The masculine near demonstrative pronoun word-final stressed *-ī > *-ē and word-final *-ē > *-ɛ̄. The associated nominative *ḏū and the accusative *ḏā developed to the וּז regularly yields the rare relative pronoun ,is difficult to reconstruct הֶלֵּ֫א due to the Canaanite Shift.The plural וֹז / הֹז demonstrative with different Semitic languages showing irreconcilable forms.The Hebrew form seems to go back to *ʔillay: the unstressed diphthong contracted to *-ē and then participated in the shift to *-ɛ̄ in word-final position. , הֶז < does not show the same shift as *ḏī יִמ As the animate interrogative pronoun it should not be reconstructed with word-final *-ī, but as *mīya. The inanimate inter- with gemination of the following consonant) goes ־הַמ normally) הָמ rogative pronoun back to *mah.

2 Nouns and Adjectives1

2.1 Roots with One or Two Radicals also used before , יִפּ mouth’ has only one radical. The construct state‘ הֶפּ ,Synchronically suffixes, derives from *pī, originally the construct state genitive.2 This was either generalized to the absolute state, where stressed word-final *-ī shifted to *-ē and further to *-ɛ̄, or the absolute state regularly developed from *pīm > *pím > *pém > *pḗ. man’ (only attested in the plural) should‘ * תֵמ bottom’, and‘ תֵשׁ ,’name‘ םֵשׁ ,’son‘ ןֵבּ be reconstructed as biradical *qṭum nouns without a vowel between the first and second radical (Testen 1985, Blau 2006).3 The loss of mimation and the case vowel in the singular was probably analogous to their deletion in all other nominals. In the unsuffixed singular, an epenthetic vowel *e was inserted to break up the consonant cluster. his‘ וֹמְשׁ ,’my son‘ יִנְבּ The lack of an original vowel can still be seen before suffixes, as in -forms the irreg ןֵבּ .men’, all without pretonic lengthening‘ םיִתְמ name’, and in the plural banīma, a form which is at least Proto-West-Semitic (cf. its reflexes* > םיִנָבּ ular plural has extended the singular stem to the םֵשׁ in Arabic and Modern South Arabian), and . תוֹמֵשׁ ,plural Nouns with a short vowel between their first and second radical undergo tonic and הֶשׂ < hands’. *śayum‘ םִיַ֫דָי < hand’, *yadayma‘ דָי < pretonic lengthening, e.g. *yadum ‘sheep’ shows triphthong contraction and the subsequent development of word-final ,’father‘ בָא ē > *-ɛ̄. No unambiguous *quṭum nouns are attested. The irregular nouns-* father-in-law’ extend their stem with -i in the construct state and‘ םָח brother’, and‘ חָא ’brothers‘ םיִחַא before suffixes, originally a lengthened genitive case vowel. The plural goes back to *ʔaḫīma > *ʔaḫḫīma with pretonic gemination (Bergsträsser 1918, 139–149); his brothers’.4‘ ויָחֶא < * ויָחַא before å in the suffix, the a regularly assimilates to ɛ, as in

1 In this section, I will make frequent reference to the recent overview of Hebrew nominal morphology by Huehnergard (2015), where the reader will also find many references to the extensive literature on this subject. I have generally not indicated the many points on which we are in agreement. 2 Huehnergard (2015, 31) reconstructs the absolute state stem as *piy-, while Militarev and Kogan (2005, 195–197) reconstruct *pay- or *pVw-. According to the rules for triphthong con- although *piy- is , הֶפּ traction given in Chapter 5, all of these reconstructions would also yield incompatible with the Classical Arabic accusative fā (construct), not **fiya. 3 Many other scholars reconstruct a short *i or *u in these nouns, e.g. Huehnergard (2015, 30– 31). דַחַא one (f.) (pause)’ but no change in‘ תָ֑חֶא < * תָ֑חַא ,’(one (m.) (absolute‘ דָחֶא < * דָחַא .Cf 4 lying‘ םיִשָׁחֶכּ < * םיִשָׁחַכּ ;’charcoal‘ םָחֶפּ < * םָחַפּ ;’(one (f.) (context‘ תַחַא ,’(one (m.) (construct‘ ;’(the sword (pause‘ בֶרָ֑חֶה < * בֶרָ֑חַה ;’(he will comfort himself (pause‘ םָ֑חֶנְתִי < * םָ֑חַנְתִי ;’(.m.pl) etc. No sequences of aHå (with H representing any guttural) are retained.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 233

If the stress is farther removed from the first syllable, the original single *ḫ was pre- םֶכיֵחֲא < brothers (construct)’ (Num 27:10), *ʔaḫaykimma‘ יֵחֲא < served, as in *ʔaḫay ‘your (m.pl.) brothers’ (Hos 2:3).5 Nouns with *ī and *ū retain their original vowel, while those with *ā change it to *ō > o due to the Canaanite Shift. תֶלֶ֫דּ .Some feminine biradical nouns have incorporated a third radical in the plural ’maidservant‘ הָמָא ; תוֹתָלְדּ door’ was reanalyzed as a segolate, giving rise to the plural‘ תוֹמָח sister’ and‘ תוֹחָא The feminine nouns . תוֹהָמֲא ,has a third radical *h in the plural ‘mother-in-law’ were originally formed by adding the feminine suffix *-at- to their associated masculines; due to the pre-Proto-Semitic loss of their third radical, presumably *w, *-at- was lengthened to *-āt-, which became *-ōt- due to the Canaanite Shift. This parallels the lengthening of the case vowels still seen in the masculine construct states .(father-in-law’ (only attested before a suffix‘ * יִמֲח brother’ and‘ יִחֲא

2.2 *qVṭl(at)um (segolates) After the loss of the case endings, *qVṭlum nouns contained a word-final consonant cluster in the singular absolute state. In nouns from strong roots, this was resolved at a relatively late point in the history of Hebrew by the insertion of an epenthetic vowel, usually *ɛ, but *i after *y and usually *a after gutturals.6 If the third radical was -path’ (Bauer and Lean‘ חַרֹ֫א < h, ḥ, or ʕ, *ɛ regularly assimilated to *a, as in *ʔórɛḥ רֶדֶ֫ח room’ (absolute‘ רַדֲח der 1922, 569). In a few words, a separate construct state like is attested. With Steiner (1976), we may attribute this to early insertion of ( רֶדֵ֫ח and an epenthetic vowel before resonants; the further development of these forms, presumably to be reconstructed like *ḥadr > *ḥader, is then like that of construct states of *qaṭilum nouns, discussed below. The plural forms of these segolates generally developed in the same way: in the abso- -with reduction of the first vowel and pretonic length , םיִלָטְק < lute state, *qVṭalīma7 ening of the second one. *quṭlum nouns seem to have preserved short *o in the first when secondarily accented, this vowel is ;( ןֶרֹ֫גּ .threshing floors’ (sg‘ תוֹנָרֳגּ syllable, as in In the construct state and before heavy .( שֶׁדֹ֫ק .holies’ (sg‘ םיִשָׁדָֽק lengthened to ɔ, as in

5 It thus appears that the gemination in the other forms of the plural is a (pre-)Hebrew-internal development, making the connection with Akkadian /aḫḫū/ ‘brothers’ more tenuous (pace Huehnergard 2015, 35). ןֶהֹ֫בּ bride-price’ (and also‘ רַהֹ֫מ tent’ and‘ לֶהֹ֫א Based on cognate evidence, it seems likely that 6 ‘thumb/big toe’?) were originally *qaṭlum nouns (Huehnergard 2015, 40, n. 37). Similarly to head’, they would then have lost their *h with compensatory‘ שׁאֹר < II-ʔ forms like *raʔsum lengthening of *a and a subsequent change to *ō under the Canaanite Shift, e.g. *ʔahl- > *ʔāl- > *ʔōl-. The consonantal h would then have been reintroduced based on the spelling, as may also have happened in some II-ʔ nouns (see below). 7 With the insertion of *a in the plural stem, possibly an Afroasiatic feature (cf. Greenberg 1955).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 234 appendix suffixes, *qVṭalay yielded *qəṭəlē. The first vowel developed to i or a, depending on the surrounding consonants and was frequently analogically influenced by the singular (see Chapter 7); *quṭlum again preserved ɔ. The second šwå was deleted, as it occurred in an open syllable. In the singular, we find different developments of the vowel in the first syllable. in context, with assimilation of *a to the epenthetic לֶטֶ֫ק qaṭlum regularly developed to* seed’ shows that‘ עַרֶ֫ז < ɛ. The development of words like *ḏarʕum > *zárɛʕ > *zɛ́rɛʕ* this assimilation took place before that of the second *ɛ to *a before gutturals. In major Before . לֶטָ֑ק pause, *a was lengthened to *ā, which did not assimilate, resulting in pausal suffixes, *a was largely preserved, but sometimes analogically replaced by *i due to influence from the *qiṭlum paradigm. *qiṭlum regularly merged with *qaṭlum in the context form of the absolute state due to Philippi and Blau’s Laws (see Chapter 6). Thus, the regular Biblical Hebrew -righteousness’. As Blau’s Law post‘ קֶדֶ֫צ < as in *tṣidqum , לֶטֶ֫ק outcome of *qiṭlum is Before suffixes, *i . לֶטֶ֑ק dated major pausal lengthening, the regular pausal form is also the , לֶטֶ֫ק < was regularly preserved. Since *qaṭlum and *qiṭlum partly merged in *qaṭl two paradigms exerted a large amount of analogical influence on each other, resulting his grave’ for regular‘ וֹרְבִק milkahu,8* > ** וֹכְּלִמ his king’ for regular‘ וֹכְּלַמ in forms like -riglum. *e < *i was also fre* > ** לֶגֶ֑ר foot’ for regular‘ לֶגָ֑ר qabrahu, and pausal* > ** וֹרְבַק quently restored in the absolute state of *qiṭlum nouns, based on analogies like *kaspṓ ‘his silver’ : *kásp ‘silver’ = *ʕeglṓ ‘his calf’ : *ʕégl ‘calf’. These analogically restored *qeṭl . לֶטֵ֫ק nouns then developed into

2.2.1 Weak Roots In an early change, syllable-final *ʔ was lost, with compensatory lengthening of the preceding vowel. In *qiṭlum and *quṭlum nouns, this regularly resulted in qeṭ and qoṭ, respectively, but where the *ʔ was maintained in spelling, it was usually secondarily muʔdam* > דֹאְמ ,’biʔrum ‘well* > רֵאְבּ reintroduced, yielding such artificial forms as ‘much’. The *ā that resulted from this loss changed to *ō due to the Canaanite Shift, as -head’. In the plural, intervocalic *ʔ was lost with contrac‘ שׁאֹר < in *raʔsum > *rāsum heads’ shows that this development‘ םיִשׁאָר < tion of the surrounding vowels; *raʔasīma postdated the Canaanite Shift. *qaṭlum nouns with *y as their second radical insert i instead of ɛ in the absolute house’. In other forms of the singular, *ay is contracted to *ē > e. In‘ תִיַ֫בּ singular, as in *qiṭlum nouns (and II-y *quṭlum nouns, if they existed), *iy (and *uy) was contracted city’. Different plural formations are attested: the normal‘ ריִﬠ < to *ī > i, as in *ʕiyrum

8 Fassberg (2002) argues for a reconstruction as *malk- for this word and explains the then ְךֶלֹ֫מ as due to avoidance of the similar-sounding ,** ְךֶלָ֑מ not , ְךֶלֶ֑מ unexpected pausal form ‘Moloch’.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 235 segolate plural with a-insertion can undergo triphthong contraction, as in *ʕiyarīma > .wild asses’ (sg‘ םיִרָיֲﬠ cities’;9 the consonantal *y can be analogically restored, as in‘ םיִרָﬠ or the stem of the singular can be extended to the plural and undergo diphthong ;( רִיַ֫ﬠ .( תִיַ֫ז .olive trees’ (sg‘ םיִתיֵז contraction, as in -cf. Aramaic bāttīn; Ugaritic ⟨bhtm⟩ /bahat) םיִתָּֽבּ house’ has the irregular plural‘ תִיַ֫בּ V̄ma?/ may reflect an attempt to make the biconsonantal singular ⟨bt⟩ /bēt-/ < *baytum conform to the /qaṭalV̄ma/ plural pattern, or the h may simply serve to make the root , הָנָּֽא triconsonantal again; Pardee 2003–2004, 45). Together with the precative particle this is the only word in which å (normally reflecting *ā) occurs in a closed, unstressed syllable. In both cases, å is followed by a geminate; we may assume that, like Classical Arabic, Hebrew did not shorten long vowels in this position, contrary to their behaviour in other closed syllables. The *ā may result from contraction of a triphthong, leading as *bayattīma. This is similar to the regular a-insertion found in םיִתָּבּ us to reconstruct the plural of segolates (cf. the singular *baytum), differing only in the gemination of the third radical. This so-called pretonic gemination is also found in other words, e.g. mVrḥaqum, and has* > קָחְרֶמ mVrḥaqqīma ‘distances’ besides the singular* > םיִקַּחְרֶמ not yet been convincingly explained; given the existence of this phenomenon and its .though, the reconstruction of *bayattīma seems certain , םיִתָּבּ regular outcome as Only a few *qaṭlum nouns with *w as their second radical maintain it, inserting ɛ in most cases, *aw is contracted to *ō > o, as in ; תֶוָ֫מ and assimilating their *a > å, as in day’ (see Chapter 5). *quṭlum (and II-w *qiṭlum) nouns contract *uw‘ םוֹי < yawmum* םיִקָוְשׁ market’. Plurals can be analogically restored, like‘ קוּשׁ < ū > u, as in *suwqum* < ,days’ < *yamīma‘ םיִמָי ;( םוֹצ .fasts’ (sg‘ תוֹמוֹצ markets’, or based on the singular stem, like‘ .is derived from a different stem than the singular , יֵמְי cf. the construct state Two *qiṭlum nouns with *n as their second radical, *bintum and *gintum (< *gim- תַבּ tum), have undergone Philippi’s Law and subsequent n-assimilation, resulting in winepress’, respectively. The *n is preserved in the irregular plural‘ תַגּ daughter’ and‘ . תוֹנָבּ < banātum* *qVṭlum nominals from geminate roots did not undergo segolization, probably because the two adjacent identical radicals were realized as a single, long consonant. As they also did not participate in tonic lengthening or Philippi’s Law, their original , טָק ,and, in some prosodic contexts) טַק vocalism is largely preserved. *qaṭṭum yields The singular stem has . טֹק and *quṭṭum yields , טֵק see Section 4.3.4), *qiṭṭum yields been generalized to the plural in most cases; alternatively, the originally inserted *a may have been deleted due to a haplological sound change that also affects geminate Plurals .( םָﬠ / םַﬠ .peoples’ (sg‘ םיִמַּﬠ < roots in other forms, as in *ʕamamīma > *ʕammīma cities‘ יֵרָﬠ Huehnergard (2015, 32, n. 22) notes the retention of the vowel in the construct state 9 (construct)’, which I attribute to the length of the vowel resulting from triphthong contraction; Huehnergard instead suggests that this derives from a suppletive plural stem, *ʕarr-.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 236 appendix also occur. Although םיִמָמֲﬠ < with analogically restored a-insertion like *ʕamamīma Proto-Semitic did not originally have any *qVṭlum adjectives (Fox 2003, 107), the sound change mentioned above also deleted the second vowel in *qaṭVlum adjectives from עָר / עַר unblemished (m.sg.)’ and‘ םָתּ adjectives like טָק / טַק geminate roots, resulting in ‘evil (m.sg.)’. In *qaṭlum nouns with *y as their third radical, this was vocalized to *i in word-final position. In context, the stress shifted forward to this vowel and the preceding vowel kid’. *qaṭlum nouns partially assimilated *a to *ɛ due‘ יִדְגּ < was reduced, as in *gadyum Similar changes affected . יִד ֶ֑גּ to the following *y; this vowel was preserved in pause, as in *qaṭlum nouns with *w as their third radical: *a assimilated to å10 and *w was vocalized -waste, void’.The plural reg‘ וּהֹ֫תּ to *u, but the stress did not shift, resulting in words like but ,( יִלְכּ .vessels’ (sg‘ םיִלֵכּ < ularly underwent triphthong contraction, as in *kilayīma .’kids‘ םיִיָדְגּ y was usually restored, as in*

2.2.2 With a Feminine Suffix The development of *qVṭlatum nouns is largely similar to that of their masculine counterparts, with the important difference that the second and third radical were not sepa- see Chapter) הָלְטִק rated by an epenthetic vowel. *qaṭlatum was regularly attenuated to 7), but a was often restored through analogy with the masculine. In the feminine, too, the stem was replaced by *qVṭal- in the plural. In the absolute state, this resulted in .in the construct and before suffixes תוֹלְטַק or תוֹלְטִק while we find , תוֹלָטְק

2.3 *qV̆ṭV̆l((a)t)um In the singular absolute state, *qV̆ṭalum nominals undergo both tonic and pretonic hair’; no clear *quṭalum‘ רָﬠֵשׂ < word’, *śiʕarum‘ רָבָדּ < lengthening, as in *dabarum ,Plural forms . לַטְק nominals are attested (Fox 2003, 221). The construct state is regularly deriving from *qV̆ṭalīma (absolute) and *qV̆ṭalay (construct), develop like the formally identical plurals of *qVṭlum nouns (see above). in the absolute state. In the construct state, many לֵטָק qaṭilum regularly yields* forms underwent Blau’s Law (see Chapter 6), developing like *ḏaqinu > *zaqen > old (m.sg.)’.11 In other cases, e was analogically extended to the construct‘ ןַקְז < zəqɛn* state. Before suffixes and in the absolute plural, *i was originally deleted before non-low

Nile reeds’, but this is a loan from Egyptian and‘ וּחָ֫א At least, this may have happened in 10 thus the original shape of the word is unclear. No assimilation has taken place in the struc- .’and he prostrated himself‘ וּחַ֫תְּשִׁיַּו < turally similar verbal form *wayyistaḥwi 11 Huehnergard (2015, 43) follows Lambdin (p.c. to H.) in attributing this to analogy with constructs of *qaṭalum nominals after the construct state plural forms, *qaṭalay and לַטְק This is an attractive solution, but it does not explain the . יֵלְטִק qaṭilay, had merged into* place to lie‘ ץַבְּרִמ < change of *i to a in construct forms of other patterns, e.g. *marbiṯụ (construct)’ (Ezek 25:5).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 237 vowels, but it was analogically restored and subsequently underwent pretonic length- ,old (m.pl.)’ (see Chapter 4). The construct plural, *qaṭilay‘ םיִנֵקְז < ening as in *ḏaqinīma behaves like *qaṭalay, discussed above. *qaṭulum nominals developed similarly to *qaṭilum forms, the main difference קֹמָﬠ < being the different outcome of tonic and pretonic lengthening (as in *ʕamuqum idem (m.pl.)’) and the non-operation of Blau’s Law‘ םיִקֻּמֲﬠ < deep (m.sg.)’, *ʕamuqīma‘ in the construct singular. *u does shift to a in the construct state before word-final gut- .’(.high (m.sg‘ הַּבְגּ besides construct ַהֹּבָגּ turals, leading to pairs like absolute dream’ have‘ םוֹלֲח < quṭulum nouns are rare; the few possible cases like *ḥulumum* completely merged with the *qiṭālum paradigm (Fox 2003, 205). Besides the regular construct forms given above, a number of *qV̆ṭV̆lum nominals דֶבֶ֫כּ besides construct דֵבָכּ form constructs from another stem, *qV̆ṭlum, e.g. absolute rib’. The vowel in‘ עַלֵ֫צ besides construct עָלֵצ heavy (m.sg.)’, absolute‘ ( דַבְכּ and regular) the first syllable is always identical in both forms, suggesting that this alternation is the result of syncope of the second vowel in the construct state, a development which is also reflected outside of Northwest Semitic and which must therefore be early (Steiner 2012).

2.3.1 Weak Roots If the second radical was *w or *y, *qV̆ṭV̆lum nominals underwent triphthong contrac- if it לֵק ,(’cloud‘ בָﬠ < if the second vowel was *a (as in *ʕayabum לָק tion, resulting in םיִשׁוֹבּ < if it was *u (as in *bawuṯīma לֹק sojourner’), and‘ רֵגּ < was *i (as in *gawirum ‘ashamed (m.pl.)’; see Chapter 5). As this contraction resulted in a long vowel, the stem remains unchanged in all forms of the paradigm. Active qal participles of II-wy verbs etc.12 , םָק < have the same stem as the perfect, yielding *qawamum In *qV̆ṭV̆lum nominals from geminate roots, the vowel in the second syllable was deleted in a Proto-Semitic sound change, merging these words with the *qVṭlum paradigm. הֶדָשׂ < qV̆ṭV̆lum nominals with *w or *y as their third radical, e.g. *śadayum* ‘field’, underwent triphthong contraction. As this was preceded by the dissimilation of unvoiced *u to *i next to bilabial consonants (discussed in Chapter 3), this resulted in a paradigm with forms like *śadḗm in both the nominative (< *śadayum) and the genitive (< *śadayim) and *śadā́m in the accusative; the nominative-genitive form was generalized throughout the whole paradigm. Mimation was deleted with

12 Huehnergard (2015, 6, n. 7) rather reconstructs these participles with a short *a (analogically shortened from older *ā), as in *qamum. The preservation of the stem’s vowel in -standing (m.pl. construct)’,must then be ana‘ יֵמָק unstressed, non-pretonic position, as in nouns and adjectives, which Huehnergard reconstructs לֵק logical.The same goes for II-wy as *qilum (p. 31).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 238 appendix compensatory lengthening. Together with pretonic lengthening, this resulted in the -with the regular change of word-final *-ē > *-ɛ̄; in the (pro , הֶדָשׂ attested forms like In . הֵדְשׂ clitic) construct state, this change did not take place, resulting in forms like the plural, where the second syllable was followed by a long vowel, contraction yielded forms where the second radical was immediately followed by the plural ending, as in .’fields‘ תוֹדָשׂ

2.3.2 With a Feminine Suffix Tonic and pretonic lengthening operated on *qV̆ṭalatum, *qV̆ṭilatum, and *qV̆ṭulatum .respectively , הָלֻּטְק and , הָלֵטְק , הָלָטְק nominals, which are reflected in Biblical Hebrew as -The abso . תַלְטִק < In the construct state, *qV̆ṭV̆latu generally developed to *qəṭəlat are , תוֹלְטִק as well as the construct plural , תוֹלֻּטְק and , תוֹלֵטְק , תוֹלָטְק lute plural forms explained in the same way. Only few *qV̆ṭV̆ltum nominals are attested, if any. Notably, the possible examples stone wall’, if from *gadirtum) do not show pretonic lengthening of the‘ תֶרֶ֫דְגּ like) vowel in the first syllable. It seems likely that the *-t- form of the feminine suffix was originally associated with non-absolute forms of the noun. These forms would then be secondary, only extended to the absolute state after the operation of pretonic length- truth’, if this‘ תֶמֱא ening. This would also explain the non-operation of Philippi’s Law in stem derives from the non-absolute form *ʔamint- associated with a hypothetical absolute state *ʔaminatum.

2.4 *qV̆ṭV̄l((a)t)um The development of *qaṭV̄lum nominals is fairly straightforward: the long vowel in the second syllable remains unchanged in all forms, surfacing in Biblical Hebrew as i < *ī, u < *ū, or o < *ā (with the Canaanite Shift), while the short *a in the first syllable is pretonically lengthened to å in the absolute singular and reduced in all other forms. *ō < *ā regularly shifted to *ū in unstressed non-word-initial syllables, as in *matāqīma > .sweet (m.pl.)’, but o was analogically restored in most cases‘ םיִקוּתְמ According to Fox (2003), the only *qV̆ṭV̄lum pattern with two high vowels that is reconstructable for Proto-Semitic is *quṭūlum. In Hebrew, the first *u was reduced due This form of the stem is unchanged . לוּטְק to the following non-a vowel, resulting in throughout the paradigm. As *quṭīlum and *qiṭīlum are not attested in other Semitic -nouns (and one adjective) as loan ליִטְק languages, Fox explains the Biblical Hebrew words or back formations, based on forms of *qaṭīlum nouns or languages where the *a was reduced (pp. 193–194). *qiṭālum undergoes the same reduction as *quṭūlum, as well as the Canaanite Shift, While this is also sometimes seen as the outcome . לוֹטְק yielding an unchanging stem of *quṭālum (e.g. Fox 2003, 234), it is argued in Chapter 3 that the Canaanite Shift did not operate after rounded vowels in the previous syllable. Thus, *quṭālum should have

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 239

-no unambiguous examples of this development are attested in the singu ;* לָטְק yielded shins’. A small number of other‘ םִיַ֫ﬠָרְכּ < lar, the evidence being limited to *kurāʕayma *quṭālum nouns occur in Biblical Hebrew, but as their *u is adjacent to a bilabial con- .e.g , לוֹטְק < sonant in each case, it dissimilated to *i, merging these words with *qiṭālum .’juniper‘ שׁוֹרְבּ < burāṯum > *birāṯum*

2.4.1 Weak Roots Some words with *ʔ as their first radical have e in the first syllable instead of an expected ʔidzārum ‘loincloth’. In Section 4.4, it is argued that ṣere was* > רוֹזֵא ḥåṭep̄ vowel, like originally in free variation with the ḥåṭep̄ vowels in this position, which explains its .’(ʔināsum ‘man(kind* > שׁוֹנֱא absence in similar words like

2.4.2 With a Feminine Suffix When the feminine suffix *-at- is added, the vowel in the first syllable is reduced in all In words with . הָלוֹטְק and , הָלוּטְק , הָליִטְק cases, while the long vowel remains, resulting in the feminine suffix *-t-, the long vowel, which is now in a closed syllable, is shortened, and the *qV̆ṭV̄ltum noun merges with the corresponding *qV̆ṭV̆ltum pattern. As in the bronze’, if from‘ תֶשֹׁ֫חְנ latter paradigm, pretonic lengthening does not take place, as in This may indicate the .( שׁוּחָנ and הָשׁוּחְנ naḥustum < *naḥūstum (cf. the rare synonyms* secondary origin of these nouns.

2.5 *qVṭṭVl((a)t)um According to Fox (2003, 280), the reconstructible nominal patterns with a short vowel following a geminate second radical are *qaṭṭalum, *qaṭṭilum, *qaṭṭulum, and *quṭ- ṭulum. The first vowel undergoes a qualitative change in all patterns but *qaṭṭalum, .with tonic lengthening in the absolute singular , לָטַּק which is regularly reflected by As is argued in Chapter 7, unstressed *a before a geminate assimilated in quality to a following stressed short vowel. Thus, *qaṭṭilum became *qiṭṭilum, while *qaṭṭulum as the second *i follows ; לֵטִּק merged with *quṭṭulum. *qiṭṭilum regularly developed to a heavy syllable, it did not undergo pretonic lengthening and is reduced in the plural quṭṭulum seems* .( םֵלִּא .dumb (m.pl.)’ (sg‘ םיִמְלִּא < and before suffixes, as in *ʔillimīma -as in one of the very few identifiable nouns of that pat , לֹטִּק to yield Biblical Hebrew bird’. The shift of *u to *i, generally seen as a form of dissimilation, has been‘ רוֹפִּצ ,tern explained in different ways: as the result of a general dissimilation of rounded vowels to *i before other rounded vowels (Huehnergard 1992, 222) or resulting from the same reduction of *i and *u to *ə before non-low vowels seen in the *qV̆ṭVlum patterns, with a subsequent change of *ə > *i in closed, pretonic syllables (Garr 1985b). The opera- ,(/chair’ (cf. Akkadian /kussûm‘ אֵסִּכּ < -tion of this change in the loanword *kutstsiʔ where it cannot be dissimilatory, supports Garr’s solution. This change is limited to vowels preceding geminates, while *u before two different consonants is unchanged,

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 240 appendix sacrifice’.In many of the possible *quṭṭulum nouns, the change‘ ןָבְּרָק < as in *qurbānum may also be due to the dissimilation of unstressed *u > *i next to bilabials proposed in Chapter 3. The restrictions on vowel quality for *qVṭṭV̄lum nominals given by Fox (2003) are the same as those for the *qVṭṭV̆lum patterns discussed above. *qaṭṭīlum and *qaṭṭūlum respectively; these stems are unchanging throughout the , לוּטַּק and ליִטַּק regularly yield paradigm. No secure *qaṭṭālum nominals are attested in Hebrew, although this is a very frequent pattern in other Semitic languages. Perhaps these words were morpho- nouns may be לוּטִּק logically transferred to *qaṭṭalum. The rather common category of reconstructed as *quṭṭūlum, with the same fronting in the first syllable seen in *quṭ- . לֹטִּק < ṭulum Pace Fox (2003), a number of Biblical Hebrew nouns do seem to reflex a *qiṭṭālum or *quṭṭālum pattern. The high vowel in the first syllable surfaces as i in Hebrew, while .’(.drunk (m.sg‘ רוֹכִּשׁ the long *ā undergoes the Canaanite Shift, resulting in forms like

2.5.1 Weak Roots Words with a guttural second radical show degemination, either with compensatory lengthening of the preceding short vowel or without (virtual gemination). In *qaṭṭalum nominals with a virtually geminated second radical, *a has assimilated to ɛ before the charcoal’.13‘ םָחֶפּ < following å, as in *paḥḥamum Instead of *qVṭṭVlum nominals with a geminated second radical, some II-w roots form *qVṭlVlum nominals, reduplicating their third radical. Thus, we find *ṯawbabum .(to turn back’ (Barth 1897‘ בושׁ from the root ** בָוַּשׁ backsliding’ instead of‘ בָבוֹשׁ < The same development in *qiṭṭālum nominals resulted in forms like *niwwāḫum > .’to rest‘ חונ appeasing (m.sg.)’ from‘ ַחֹחיִנ < niwḫāḫum > *nīḫāḫum*

2.5.2 With a Feminine Suffix *qaṭṭVltum nouns do not show the same assimilation of *a to the following vowel as their masculine counterparts. As with the apparent non-operation of pretonic lengthening in *qaṭVltum nouns, this may be due to the original non-absolute function of the *qaṭṭVlt- form of the stem. As the stress would not have fallen on the second vowel in non-absolute forms of the noun, the assimilation did not take place. *qaṭṭalatum yields .’doe‘ הָלָיַּא < as in *ʔayyalatum , הָלָטַּק

2.6 *mVqṭVl((a)t)um Of the various noun patterns with an m-prefix, special developments occur in *maq- ṭalum and *maqṭilum nouns. In a process often called the Law of Attenuation (see

13 See note 4.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 241

Chapter 7), the first *a of *maqṭalum nouns dissimilated to *e > i in a late change, limited to the Tiberian reading tradition, as in *magdalum > *maḡdål > *meḡdål > tower’. This change was blocked before geminates, gutturals, and in a few other‘ לָדְּגִמ environments listed in Chapter 7. As this development greatly increased the number of nouns with a mi-prefix, this may also have spread to other nouns where *ma- -psalm’ (earlier *mazmōr, as in Hexaplaric μαζ‘ רוֹמְזִמ should have been preserved, as in μωρ). with tonic , לֵטְקַמ The absolute state of *maqṭilum nouns regularly developed to lengthening of the *i. In the construct state, this vowel developed to a through Blau’s Law (see Chapter 6 and the discussion of *qaṭilum nominals above), triggering the ץַבְּרִמ < same dissimilation of *a > i seen in *maqṭalum nouns, as in *marbiṯụ > *marbaṣ ‘lying place (construct)’.

2.6.1 Weak Roots The most important deviation from the strong pattern is found with II-wy roots, which lost their second radical with lengthening of the following vowel in a pre-Proto-Semitic sound change. If this resulted in *ā, it regularly underwent the Canaanite Shift (see ,place’. Note that the *a is pretonically lengthened‘ םוֹקָמ < Chapter 3), as in *maqāmum as it is in an open syllable. In unstressed non-word-initial syllables, *ā > *ō regularly his net’, but o is analogically restored in most‘ וֹדוּצְמ < shifted to *ū, as in *matṣādahu cases.

2.6.2 With a Feminine Suffix Nouns with suffixed *-at- and *-t- behave as expected, with the minor point that flight’ besides‘ הָסוּנְמ < unstressed *ō < *ā regularly develops to *ū, as in *manātsatum ,refuge’; in some words, o was analogically restored‘ סוֹנָמ < the masculine *manātsum .’idem‘ דוֹצָמ net’ based on‘ ( הָדוּצְמ also) הָדוֹצְמ as in

2.7 Miscellaneous Of the noun patterns with other prefixes than *mV-, only those with *ta- are somewhat frequent. These behave like the corresponding *maqṭVlum patterns. -in most cases (see Chap - ןוֹThe frequent nominal suffix *-ān- has regularly shifted to offering’, the Canaanite Shift was blocked due to the‘ ןָבְּרָק ter 3). In a few words like preceding rounded vowel; from forms like these, the -ån form of the suffix was analogically extended to some other forms. A unique interchange between geminated and ungeminated consonants is found -idem (con‘ ןוֹרְכִז remembrance (absolute)’ besides‘ ןוֹרָכִּז .noun pattern, e.g ןוֹלָטִּק in the your (f.sg.) remembrance’, etc. These words are usually reconstructed‘ ְךֵנוֹרְכִז ,’(struct as *qVṭalānum, with irregular gemination of the second radical in the absolute singular (Blau 2010, 275). This gemination is ad hoc, however, and contradicted by many

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 242 appendix counterexamples. Pace Blau, it is easier to reconstruct this pattern as *qiṭṭalānum, with degemination occurring in forms where the non-pretonic *a was reduced.14 In Rab- –Hurvitz 1968; Samet 2016, 1071) ןוֹלְטִק binic Hebrew, this pattern is largely replaced by ןוֹלָטִּק we might also derive the construct and suffixed stem of Biblical Hebrew ;(1073 your‘ ְךִיַ֫נוֹבְזִﬠ nouns from *qiṭlānum, although the spirantization of the third radical in (f.sg.) merchandise’ (six times in Ezek 27) points towards a reduced vowel between the second and third radical in this pattern. The adjectivizing i-suffix is also common. The vocalization of the plural, -iyyim (also -im with haplology) and -iyyoṯ, suggests a reconstruction as *-iyy-, but it should rather be reconstructed as *-īy-, which is more easily connected to the deviant form found in Aramaic, -āy. *-īyV is regularly reflected by -iyyV in Biblical Hebrew, as can be seen -innocent’. The same gemination took place before the femi‘ םיִיִּקְנ < from *naqīyīma nine suffix *-at-, as in *-īyatum > -iyyå, while the form with the shorter feminine suffix, *-īytum, shortened its vowel in a closed syllable, developing to *-iytum > *-ītum > - iṯ. Finally, it should be noted that many reduplicated nouns do not undergo tonic lengthening or the Law of Attenuation, which would disrupt the symmetry of the redu- .** רָדְּרִדּ thistles’ for regular‘ רַדְּרַדּ ,plicated parts; thus, for instance

3 Numerals

Table 45 lists the Biblical Hebrew absolute states of the cardinal numerals 1–10 and their Proto-Northwest-Semitic reconstruction. All pnws forms are given in the nominative, but note that the Biblical Hebrew forms reflect the original genitive. For the reconstruction of an initial cluster in *ṯnāna ‘two (m.)’, cf. Classical Arabic iṯnāni with with the change of *n > r , ןיֵרְתּ loss of the i after preceding vowels and Biblical Aramaic limited to word-initial clusters (Testen 1985), besides the lack of pretonic lengthening -cf. Blau 2006). Original *d has assimilated to a following coro) םִיַ֫נְשׂ in Biblical Hebrew nal in *ʔaḥadtum > *ʔaḥattum and *sidṯ(at)um > *siṯṯ(at)um; in the latter case, the *d is still attested in forms like the Classical Arabic cardinal sādisun (with dissimilation of *ṯ > s).

14 Blau rejects this possibility because unemphatic plosives tend to retain their gemination in all circumstances, but there are a few other cases of this degemination (Bergsträsser 1918, 142). Assuming degemination before šwå was regular, analogical restoration of the gemination would be easy and desirable in most morphological classes, especially with the unemphatic plosives, where a postvocalic singleton would have undergone spirantization. In this noun pattern, however, the degemination seems to have been generalized as a morphological rule.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 243 table 45 Cardinal numerals 1–10

Meaning Masculine Feminine

pnws bh pnws bh

תַחַא ʔaḥattum* דָחֶא ʔaḥadum* 1 םִיַ֫תְּשׁ ṯintāna* םִיַ֫נְשׁ ṯnāna* 2 שֹׁלָשׁ ṯalāṯum* הָשֹׁלְשׁ ṯalāṯatum* 3 עַבְּרַא ʔarbaʕum* הָﬠָבְּרַא ʔarbaʕatum* 4 שֵׁמָח ḫamisum* הָשִּׁמֲח ḫamisatum* 5 שֵׁשׁ siṯṯum* הָשִּׁשׁ siṯṯatum* 6 עַבֶ֫שׁ sabʕum* הָﬠְבִשׁ sabʕatum* 7 הֶנֹמְשׁ ṯamāniyum* הָנֹמְשׁ ṯamāniyatum* 8 עַשֵׁ֫תּ tisʕum* הָﬠְשִׁתּ tisʕatum* 9 רֶשֶׂ֫ﬠ ʕaśrum* הָרָשֲׂﬠ ʕaśaratum* 10

The ḥ in ‘1’ has undergone pretonic gemination (Bergsträsser 1918, 139–140), hence the lack of pretonic lengthening of *a; the original stem is preserved in the plural, the same (m.pl.)’. In the masculine, *a has changed to ɛ through‘ םיִדָחֲא < ʔaḥadīma* height assimilation to the following å.15 The best explanation for the unique initial with a plosive t,16 is given by Bravmann (1952), who believes that, as , םִיַ֫תְּשׁ cluster in in Syriac *šittā > štā ‘six (m.)’, the sequence *šitt- was reduced to št- due to the phonetic -the expected second vowel **å, with tonic length , עַבְּרַא similarity between š and i. In is הָשִּׁמֲח ening, is assimilated to the following guttural (Blau 1983). The gemination in Brockelmann 1908, 486). Both forms of) הָשִּׁשׁ based on analogy or contamination with ‘eight’ show the regular contraction of a triphthong. all show the תֶרֶ֫שֲׂﬠ and , תֶשֵׁ֫שׁ , תֶשֵׁ֫מֲח , תַﬠַ֫בְּרַא , תֶשֹׁ֫לְשׁ The masculine construct states *-t- allomorph of the feminine suffix, rather than the *-at- found in the absolute state. This shorter allomorph is also associated with the construct state in some nouns. The but given the , תֶשֵׁ֫מֲח is usually attributed to influence from תֶשֵׁ֫שׁ lack of gemination in ,’construct) ‘woman) תֶשֵׁ֫א /(absolute) הָשִּׁא close parallel in the interchange between and עַבְשׁ this may be the result of sound change. In the feminine, the construct states are vocalized as so-called reverse segolates; given the high sonority of their final עַשְׁתּ

15 See Note 4. see Hoberman (1989), who also gives an , תְּשׁ- For this analysis of the ambiguous spelling 16 alternative, analogical explanation for the origin of this cluster.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 244 appendix consonant, this is probably due to early insertion of an epenthetic vowel, also found in .(Steiner 1976) עַרֶ֫ז seed (construct)’ besides the absolute state‘ עַרְז forms like head’, with dissimilation of‘ שׁאֹר first’ is derived from‘ ןוֹשׁאִר The ordinal numeral *ō > *ī before *ō (Bauer and Leander 1922, 628) or contamination with *tiwkānum > middle (m.sg.)’. The other ordinals all have the adjectivizing suffix -i < *-īyum‘ ןוֹכיִתּ ליִטְק.- < -sixth’, the base is consistently *qaṭīl‘ יִשִּׁשׁ second’ and‘ יִנֵשׁ attached. Except in Note that the word for ‘second’ shows a pretonically lengthened e; if the reconstruction of the cardinals given here is correct, this *i may have been taken over from the feminine form at some point after the reduction of *i in open syllables before non-low may have been motivated by the dissimilarity יִשִּׁשׁ vowels. The secondary creation of between the ordinal *siṯṯum, with assimilation of *-dṯ- > *-ṯṯ-, and the regular ordinal *sadīṯīyum. The numbers 11–19 are formed by combining a (frequently reduced) form of the unit with feminine הֵרְשֶׂﬠ with masculine nouns and רָשָׂﬠ ,’and a word expressing the ‘-teen nouns. The masculine ‘-teen’ can probably be reconstructed as *ʕaśarVm; the Classical Arabic cognate ʕašara does not show nunation, but this may be due to its different syn- .suggests that this form was mimated רָשָׂﬠ tactic behavior, as the tonic lengthening in Given the existence of Ugaritic ⟨ʕšrh⟩, the most straight-forward reconstruction for the feminine is *ʕiśrVhV, with yet another form of the stem and an otherwise unknown suffix. The higher numerals all regularly derive from their Proto-Northwest-Semitic recon- seventy’ have been‘ םיִﬠְבִשׁ fifty’ and‘ םיִשִּׁמֲח :structions, with some minor exceptions seven (m.)’, respectively. The‘ הָﬠְבִשׁ five (m.)’ and‘ הָשִּׁמֲח analogically remodeled after other decades are regularly formed by attaching the masculine plural endings to the derived from the root of , םיִרְשֶׂﬠ < associated numeral, with the special form *ʕiśr-īma -one hundred’ with pretonic lengthen‘ הָאֵמ ten’, for ‘twenty’. *miʔatum regularly yields‘ םיִשׁאָר < as in *raʔasīma ; םִיַ֫תאָמ < ing.The dual is retained for ‘two hundred’,*miʔatayma ‘heads’, intervocalic *ʔ has been lost before an unstressed vowel, resulting in long *ā. -one thousand’ with segolization. Finally, the most com‘ ףֶלֶ֫א ʔalpum regularly yields* while its Late , הָבָבְר < mon word for ‘ten thousand, myriad’ appears as *rVbabatum .is most likely an Aramaic loanword וֹבִּר Biblical Hebrew synonym

4 Verbs

Given the close similarity between the Classical Arabic and Ugaritic verbal paradigms, we may securely base the reconstruction of the Proto-Northwest-Semitic verb on these forms. The development of the perfect endings is discussed in Chapter 8. One development that was not mentioned there was the change of word-final *-at to *-ā, which affected the third person feminine singular ending. This took place after triphthong

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 245 contraction, as *banayat becomes *banāt with preserved *t (see the section on III- wy verbs below), and before tonic lengthening, as the newly open syllable *-ā blocks . הָלְטָק < lengthening of the preceding vowel as in *qaṭálā

4.1 The Strong Verb 4.1.1 Fientive qal As the unsuffixed verbal forms all ended in a long or short vowel, tonic lengthening did not apply to them. Either the short vowel was lost in the first apocope, closing or ,( לַטָק < the syllable before tonic lengthening could operate (e.g. *qaṭala > *qaṭál the long vowel remained, blocking the operation of tonic lengthening (e.g. *qaṭalū ,Pretonic lengthening did take place in all open, originally pretonic syllables .( וּלְטָק < . לַטָק < i.e. in all forms but the second person plural; thus, for instance, *qaṭál > *qāṭál At some point after the operation of pretonic lengthening, the third person feminine singular and third person plural participated in the stress shift from their light penultimate syllable to the ultimate, followed by reduction of the short vowel: *qāṭálā > In pause, the originally stressed vowel was . וּלְטָק < qāṭálū > *qāṭalū́* , הָלְטָק < qāṭalā́* For the perfect forms with suffixes, see Chapter . וּלָ֑טָק lengthened and preserved, as in 8. In the imperfect, the *a of the prefix shifted to *e and, much later, to i, resulting in see Chapter 7). Due to a stress shift (discussed) לֹטְקִי < forms like *yaqṭulu > *yeqṭol merged לֹטְקִי < in Chapter 4), the unsuffixed forms of the jussive like *yáqṭul > *yaqṭúl with the corresponding imperfect forms. This led to the extension of the original jussive forms in *-ū and *-ī to the imperfect, causing the reflexes of *yaqṭulū, *taqṭulū, and *taqṭulī to largely replace those of *yaqṭulūna, *taqṭulūna, and *taqṭulīna. These forms with a final long vowel underwent the same stress shift and reduction as some forms -The second and third person fem . יִלְטְקִתּ and , וּלְטְקִתּ , וּלְטְקִי of the perfect, resulting in inine, *taqṭulna, acquired an additional word-final *-h through contamination with . הָנְלֹ֫טְקִתּ < the independent personal pronouns (see Chapter 8), resulting in *taqṭulnah Pronominal suffixes on the imperfect either attach to the original energic suffix, *-Vn-, or to a historically long e, which originated in the III-wy paradigm. The second person ;is exceptionally not connected to a linking vowel , ָךְלָטְקִי masculine singular suffix, as in this may be due to the secondary form of this suffix. The imperative can originally be reconstructed as the second person jussive without ,Suchard 2017). Before vocalic suffixes) לֹטְק < a prefix, as in the masculine singular *qṭul the stem vowel was reduced after losing the stress; as can be seen in the masculine plu- The epenthetic schwa (also phonetically present in the . וּלְטִק < ral, *qṭúlū > *qəṭəlū́ singular) develops to i, as in the construct plural noun forms like *dabaray > dəbərē > This . יִנֵ֫לְטָק Notably, we find a different form of the imperative before suffixes, as in . יֵרְבִדּ allomorphy may be attributed to the different original position of the stress: *qṭúlū > This suggests a regular metathesis in this environment, *#CCoCV́ . יִנֵ֫לְטָק < qṭulḗnī* , וּלְטִק

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 246 appendix

Since imperatives with an a-vowel lengthen . יִנֵ֫לְטָק < CoCCV́: thus, *qṭolḗnī > *qoṭlḗnī# < call (m.sg.) me’, this sound change may be considered‘ יִנֵ֫אָרְק .it in this environment, e.g a special case of pretonic lengthening. The reconstruction of the infinitive construct is somewhat uncertain. It is closely connected to the imperfect stem in all paradigms, but does undergo tonic lengthening in many cases; this suggests a reconstruction as *qṭulum, in which case the forms without tonic lengthening were analogically rederived from the prefix conjugation stem. The same metathesis seen in the imperative takes place before suffixes, although the vs. the verbal suffixes , וֹלְטָק infinitive construct mostly takes nominal suffixes, as in this offers some confirmation for the sound law ; וּהֵ֫לְטָק found on the imperative, as in regularly reflects , לוֹטָק ,and reconstruction suggested above. The infinitive absolute *qaṭālum, cf. Akkadian /qaṭālum/. The active participle, *qāṭilum in the masculine singular, undergoes the Canaan- ,In the feminine singular . לֵטֹק ite Shift and tonic lengthening, yielding Biblical Hebrew with Philippi’s Law and segolization affecting , תֶלֶ֫טֹק < the regular forms are *qāṭiltum -with non-operation of pretonic lengthen , הָלְטֹק < the second syllable, and *qaṭilatum is analogically modeled after the הָלֵטֹק ing after a heavy syllable; the alternate form f.), pretonic lengthening does) תוֹלְטֹק m.) and) םיִלְטֹק ,masculine. In the plural forms not apply either. One of several Proto-Northwest-Semitic qal passive participles, *qaṭ- like other לוּטָק ūlum, became the productive form in Hebrew, regularly developing to nominals of the same pattern.

4.1.2 Passive qal Like the other internal passive stems, the passive qal is largely formed by changing the vowels of the corresponding active stem to *-u-a-. In the perfect, pretonic lengthening Otherwise, the passive qal perfect . לַטֻּק < geminated the second radical, as in *quṭála and imperfect were affected by the same sound changes as their active counterparts. As in the stative qal and nip̄ʕal, the passive qal participle has the same stem as the . לָטֻּק < associated perfect; this develops regularly like *quṭalum

4.1.3 Stative qal respectively, both with , לֹטָק and לֵטָק In the perfect, *qaṭila and *qaṭula developed to pretonic lengthening. *qaṭil- first and second person perfect forms underwent Philippi and Blau’s Laws (see Chapter 6), changing the vowel in the second syllable to a as in .the corresponding forms from *qaṭul- verbs regularly retain their o ; ָתְּלַ֫טָק < qaṭilta* This partial merger of *qaṭila and *qaṭala led to the transfer of some verbs from the former category to the latter. -without any lengthened vow , לַטְקִי In the imperfect, *yiqṭalu regularly developed to els. Stress-bearing endings in nun caused pretonic lengthening, also in context, as in .(you (f.sg.) will stick’ (Ru 2:8, 21‘ ןיִקָבְּדִתּ < tadbaqīna*

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 247

The development of the participle is similar to that of the perfect, with the difference that as a nominal form, it undergoes tonic lengthening: contrast perfect *gabuha .’(.high (m.sg‘ ַהֹּבָגּ < it was high’ with the participle, *gabuhum‘ הַּבָגּ <

4.1.4 Nip̄ʕal -in closed sylla ִנ- The prefix of the nip̄ʕal perfect developed from *na- to *ne- and later bles (see Chapter 7). The second syllable of *naqṭal- developed like that of the fientive qal perfect, *qaṭal-. -the *n-prefix assimilated to the following rad , לֵטָקִּי < In the imperfect, *yinqaṭilu ical. The *a was pretonically lengthened. In the second person feminine singular and masculine plural, the stress shifted to the word-final vowel, with reduction of the penultimate vowel. Philippi’s Law affected the second and third person feminine plural, The merger of the jussive with the imperfect was not caused by . הָנְלַ֫טָקִּתּ < tinqatilna* any identifiable sound change in the nip̄ʕal, but is rather analogical to the same merger in the qal. in the imperative is probably secondary, analogically taken from the ִה- The prefixed hip̄ʕil, where the imperative stem is also synchronically formed by changing the prefix ,The same goes, mutatis mutandis .( לֵטְקַה to h (as in ( לֵטְקַתּ .consonant of the jussive (e.g is derived לוֹטְקִנ ,Of the two infinitive absolute forms . לֵטָקִּה ,for the infinitive construct -from the prefix conjugation. As in the stative qal, the par לוֹטָקִּה from the perfect and ticiple is identical to the perfect stem, only with tonic lengthening.

4.1.5 Piʕel, puʕal, and hiṯpaʕel The change in the piʕel perfect of Proto-Northwest-Semitic *qaṭṭil- to Proto-Canaanite *qiṭṭil- is best explained as the result of an assimilation rule: *#CaC1C1V̆́1 > *#CV1C1C1V̆́1 (see Chapter 7). Before consonants, the second *i of the perfect stem was affected by This form of the stem was extended to the . ָתְּלַ֫טִּק Philippi’s Law, resulting in forms like second person plural, even though the *i was unstressed there, and sometimes to the . לֵטִּק to compete with לַטִּק third person masculine singular, causing No special developments affected the imperfect. The second and third person feminine plural should have undergone Philippi’s Law, but *e was analogically restored. As in the other derived stems (except for the nip̄ʕal), the participle is formed by adding a prefix *mu- to the imperfect stem. In Hebrew, the *u of this prefix dissimilated to *i (Suchard 2016), but it was later lost in most forms. , לַטֻּק < The development of the puʕal forms is straightforward: perfect *quṭṭala . לָטֻּקְמ < participle *muquṭṭalum , לַטֻּקְי < imperfect *yuquṭṭalu The stem of the hiṯpaʕel, originally *-tqaṭṭal-, has changed to *-tqaṭṭil- in Hebrew, although some forms with *a between the second and the third radical remain. This may be due to contamination with the original Gt-stem, which was lost in Hebrew and seems to have merged with the tD-stem in some cases (Arnold 2005, 48–50). In that

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 248 appendix case, the Hebrew hiṯpaʕel would be something of a hybrid, combining features of the was prefixed to ִה- ,Gt-stem, *-qtaṭil-, as well as the tD-stem, *-tqaṭṭal-. As in the nip̄ʕal the stem in word-initial position based on analogy with the hip̄ʕil.

4.1.6 Hip̄ʕil and hɔp̄ʕal In the strong verb, the hip̄ʕil contains the reflex of a long *ī between the second and third radical in all originally open syllables. This replaced the original short *i through analogy with the II-wy verb (see below): jussive *yaqim : imperfect *yaqīmu : perfect *hiqīma = jussive *yaqṭil : imperfect *yaqṭīlu : perfect *hiqtīla. This change in vowel length contributed to the preservation of the imperfect/jussive distinction in the hip̄ʕil. As the stressed syllable contained a long vowel in the perfect and imperfect, the stress did not shift to the ultimate in context, but rather stayed in place, yielding forms like . וּליִ֫טְקִה In the perfect, original *ha- was probably replaced by *hi- through analogy with the piʕel (Huehnergard 1992). As in other stems with *i before the third radical, the first and second persons were affected by Philippi’s Law. In the jussive, imperfect, and participle, the intervocalic *h was deleted together and ליִטְקַי < with the preceding vowel, resulting in forms like *yahaqṭilu > *yaqṭilu . ליִטְקַמ < muhaqṭilum > *maqṭilum* The development of the hɔp̄ʕal parallels that of the puʕal, with the same elision of intervocalic *h seen in the hip̄ʕil.

4.2 Guttural and Weak Verbs 4.2.1 Guttural Verbs In all verbs with one or more guttural radical(s), reduced vowels usually become a ḥåṭep̄ vowel instead of the normal šwå; šwå that does not reflect a historical vowel is also often replaced by a ḥåṭep̄ vowel in verbs with a guttural first radical. If this ḥåṭep̄ vowel would they‘ וּדְבַﬠַי occur in a closed syllable, it is written with the corresponding full vowel, as in . וּלְטְקִי parallel to non-guttural ,** וּדְבֲﬠַי m.) will serve’ instead of) The main difference between non-guttural verbs and those with a guttural first radical is that in the prefix conjugation of the fientive qal, the prefix vowel *a did not shift to *e > i. Thus, the Barth-Ginsberg law still applies, as seen in the opposition between -he will be rich’. The *a in the nip̄ʕal perfect pre‘ רַשְׁﬠֶי he will serve’ and‘ דֹבֲﬠַי forms like he was forsaken’, which shows that this is a‘ בַזֱﬠֶנ < fix did shift to *e > ɛ, as in *naʕzaba different sound change than that affecting the qal prefixes (Lambdin 1985, 144). In the nip̄ʕal prefix conjugation, the gemination of the first radical was given up with compen- דֵבָﬠֵי < satory lengthening of the preceding *e, e.g. *yinʕabidu > *yeʕʕābed > *yēʕābed ‘it will be cultivated’. The same change also affects verbs with reš as their first radical. In verbs with a guttural second radical, short vowels following it are changed to a ( לַטָק ) in forms that also occur with a for morphological reasons, like the qal perfect

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 249

-It is unclear whether this is a morpho .( לַטִּק ) and the piʕel perfect ( לַטְקִי ) and imperfect logical change, that only affected those verb forms where a could plausibly occur, or a sound change that was analogically canceled in those forms where the non-guttural The verb stems with a .( לֵטַּקְי verb only has non-a vowels, like the piʕel imperfect (only geminated second radical lost this gemination with compensatory lengthening if the radical was guttural; depending on when this happened, the reflexes of *i and *u in this position are either e and o or i and u. In verbs with a guttural third radical, short *e and *o have assimilated to a before the guttural in word-final position. In the infinitive, participle, hip̄ʕil, and pausal forms, páṯaḥ furtivum is inserted between a long, non-a vowel and a word-final guttural.

4.2.2 I-ʔ While most verbs with *ʔ as their first radical behave like other I-guttural roots, a few frequent verbs have retained forms that are the regular outcome of sound change. In the qal prefix conjugation, *aʔC > *āC > *ōC (Canaanite Shift), e.g. *yaʔkulu > *yōkol. Verbs with *u as their theme vowel underwent dissimilation: after *ō in the preceding ,he will eat’, while pausal, long *ō shifts to *ē‘ לַכאֹי syllable, short *o becomes *a, as in you (m.sg.) will eat’ (see Chapter 6). The consecutive imperfect *wayyáʔkul‘ לֵ֑כאֹתּ as in and he ate’ preserves penultimate stress in context, as the a-vowel in the final‘ לַכאֹ֫יַּו < syllable is secondary, deriving from dissimilation from the preceding rounded vowel, and therefore did not attract the stress. In the imperative and infinitive construct, *u was retained, as there was no prefix to cause dissimilation. Hence, we find forms and he‘ זֶחאֹ֫יַּו I will seize’ and‘ זֵחֹא to eat’. The forms like‘ לֹכֲא eat (m.sg.)’ and‘ לֹכֱא like originally had an *i theme vowel, rather than *u, although זחא seized’ may suggest that this is hard to square with both cognate evidence (Classical Arabic yaʔḫuḏu) and the but the latter may reflect the inflection as a strong verb ; זֹחֱא imperative forms like Judg 16:3, 1Kgs 6:10). The development of the) זֹחֱאֶיַּו also attested in prefixed forms like to say’ is further complicated by the dissimilation of *u > *i‘ רמא highly frequent verb in unstressed syllables, caused by the preceding *m (see Chapter 3). Thus, while we -in the imperfect, the consecutive imper לכא find the same development as that of ,’and he said‘ רֶמאֹ֫יַּו < fect developed from *wayyáʔmur > *wayyáʔmir > *wayyṓmer with the pausal stress shift and Blau’s Law (also discussed in Chapter רַ֑מאֹיַּו pausal 6).

4.2.3 I-wy Fientive verbs with *w as their first radical share certain weak features in all classical Semitic languages and are probably partially biconsonantal in origin.This situation was complicated in Northwest Semitic by the sound change of word-initial *w- > *y-, which affected the perfect of these verbs in the qal, piʕel and puʕal; *y was later analogically extended to the prefix conjugation of these stems, as well as the hiṯpaʕel.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 250 appendix

As reflected by Classical Arabic and Gəʕəz, the fientive qal prefix conjugation stem was originally only based on the second and third radical, as in *ya-ṯib-u ‘he will sit’ from *yṯb (< *wṯb). In Hebrew, the first radical *y was analogically introduced into these The normally defective spelling of the prefix vowel . בֵשֵׁי < forms, resulting in *yayṯibu e may indicate that the expected archaic forms like *yāšeb were still in use when the orthography of the Hebrew Bible was fixed. This may be contrasted with the stative qal, where the historically present *y is regularly spelled, as in (pre-Proto-Northwest- -he will inherit’. In the consecutive imperfect, fien‘ שַׁריִי < Semitic *yiwraṯu >) *yiyraṯu , בֶשֵׁ֫יַּו tive forms maintained the original penultimate stress in unsuffixed forms, like .(see Section 4.2) שַׁ֫ריִיַּו while the sonorous *a attracted the stress in stative forms like The qal imperative and infinitive construct are biradical, an inherited feature of these verbs.The imperative is formed from the old stem of the prefix conjugation, without the analogically added *y, while the infinitive construct adds a feminine suffix *-t- (rarely *-at-). In these latter forms, Philippi’s Law was operative (see Chapter 6), the to sit’; the‘ תֶבֶ֫שׁ < most common form developing like *sibtum > *šebt > *šɛbt > šabt .’your (m.sg.) sitting‘ ָךְתְּבִשׁ original *i is retained in suffixed forms like In the nip̄ʕal, *w was preserved due to its word-internal position. In the perfect, the unstressed diphthong *aw was contracted to ō, presumably before the change of *a > *e it was left’. In the prefix conjugation, n-assimilation‘ רַתוֹנ < in the prefix, as in *nawtara .’it will be left‘ רֵתָוִּי < resulted in a geminate *ww, as in *yinwatiru he‘ ביִשׁוֹה < The hip̄ʕil forms are similar to those of the nip̄ʕal perfect, e.g. *hawṯiba -he will settle’. A few verbs with original *y as their first rad‘ ביִשׁוֹי < settled’, *yahawṯibu הָקיִ֫ניֵה < ical are still distinguishable in the hip̄ʕil, where they develop like *hayniqat ‘she suckled’. In the hɔp̄ʕal, verbs with original *w contract the diphthong in their first .’he was settled‘ בַשׁוּה < syllable to *ū, as in *huwṯaba

4.2.4 I-n If the first radical of a verb is *n, it regularly assimilates to a following consonant, as in he will fall’, although it is analogically restored or failed to assimilate in‘ לֹפִּי < yanpulu* verbs with a guttural second radical. Most I-n verbs lose the first radical in the imperative and infinitive construct. These verbs, however, are only weak in languages that undergo n-assimilation, but not in other languages, like Arabic or Gəʕəz. This biradical formation of the imperative and the infinitive construct should therefore be attributed to the assimilation of the first radical in the prefix conjugation, which was analogically extended to the imperative: jussive *teqṭol : imperative *qṭol = jussive *tetten ‘may you (m.sg.) give’ : imperative *tten (or perhaps *n also simply assimilated in the -give (m.sg.)’ with degemination of the word‘ ןֵתּ < (expected imperative, *nten > *tten initial geminate. These imperative forms then gave rise to biradical infinitive construct forms through analogy with the I-w verbs (see above).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 251

4.2.5 II = III17 Most of the weak features of the geminate verbs are inherited from earlier stages of the language. They show the result of two sound changes: the deletion of a short vowel he turned’, and the‘ בַס < between two identical consonants, as in *tsababa > *tsabba metathesis of a short vowel in order to produce a geminate, as in *yatsbubu > *yatsubbu he will turn’. In the perfect of the qal, nip̄ʕal, hip̄ʕil, and hɔp̄ʕal, a linking vowel *ā‘ בֹסָי < occurs between the stem and suffixes starting with a consonant; this *ā is affected by I turned’. Imperfect feminine plural forms‘ יִתוֹ֫בַּס < the Canaanite Shift, as in *tsabbātu of these stems insert a linking vowel -ɛ- after the stem through analogy with the III- wy verbs (see below): *yebnū ‘they (m.) will build’ : *tebnēnā ‘they (f.) will build’ = *yasobbū ‘they (m.) will turn’ : *tasobbēnā ‘they (f.) will turn’. In the qal imperfect, nip̄ʕal perfect, and hip̄ʕil, the prefix vowel is in an open syllable, unlike in the strong verb. It is accordingly lengthened in pretonic position and reduced when further away from the stress. The qal and hip̄ʕil consecutive imperfect maintain the original penulti- and he turned’ (see Section 4.2). So-called Aramaising‘ בָסָ֫יַּו mate stress in context, as in ןֵתּ :are based on analogy with the I-n verbs , בֹסִּי forms with a geminated first radical, like he will turn’.18 Finally, the geminate‘ בֹסִּי : ’(.turn (m.sg‘ בֹס = ’he will give‘ ןֵתִּי : ’(.give (m.sg‘ verbs can either form a regular piʕel, with gemination of the second radical, or a polel, This is an analogy . בֵבוֹס < with *w inserted before the second radical, as in *tsawbiba with the II-w verbs (see below): what was originally the second radical in these forms in a *qaṭlila form has been reanalysed as an infix, allowing the pattern to be extended to the geminate verbs as *qawlila. The presence of a long *ū in the hɔp̄ʕal prefix, as in is also taken over from the II-wy verbs, while the a vowel in some hip̄ʕil perfects , בַסוּה .(he broke (pause)’ is due to dissimilation from the preceding e (see Chapter 6‘ רַפֵה like

4.2.6 II-wy The hollow verb is another category that inherited some weak features from Proto- Semitic. Wherever the first and second radical came into contact, the second radical םוּקָי < was lost with lengthening of the following vowel, as in *yaqwumu > *yaqūmu ‘he will stand up’. When this resulted in *ā, as in a few stative qal imperfects and he will‘ 19 שׁוֹבֵי < the nip̄ʕal, this vowel underwent the Canaanite Shift, as in *yibāṯu it is established’. This ō frequently surfaces as u in‘ ןוֹכָנ < be ashamed’ and *nakāna unstressed position. As in the geminate verbs, prefix vowels occur in open syllables, resulting in pretonic lengthening or reduction.

17 The ideas expressed in this and the next section are argued more fully in Suchard (2016). 18 The same analogy is posited as the origin of Aramaic I-wy forms with geminated second radicals by Huehnergard (1995, 272). 19 With analogical restoration of the prefix vowel, which should be lost in this position (see Section 4.4).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access 252 appendix

The qal perfect underwent triphthong contraction, resulting in a long *ā in the fientive verb and *ē or *ō in the stative verb (see Chapter 5). In closed syllables, this vowel you (m.sg.) stood up’. The‘ ָתְּמַ֫ק < was shortened, as in *qawamta > *qāmta > *qamta same contraction affected the qal participle, which has an identical stem to that of the standing (m.sg.)’. The imperfect feminine plural forms insert -ɛ- before‘ םָק .perfect, e.g their suffix, like the II = III verbs. The imperative has secondarily been adapted to the while the infinitive , םוּק < by *qūm ** םֹק < imperfect stem, replacing the expected *qum construct may simply have preserved the non-word-final form of the jussive stem, as in As in some other categories of weak verbs, the consecutive imperfect . םוּק < qūmum* The infinitive absolute regularly . םָקָ֫יַּו preserves the penultimate stress in context, as in . םוֹק < contracts its triphthong, resulting in *qawāmum > *qawōmum > *qōmum The same linking vowel *ā > -o- found in the geminate paradigm occurs in the nip̄ʕal and usually hip̄ʕil perfect of hollow verbs. In analogy with the geminate verb, the nip̄ʕal בַסִּי imperfect : בַסָנ has extended its perfect stem to the prefix conjugation: perfect An analogy with the I-y verbs is found in the causative . ןוֹכִּי imperfect : ןוֹכָנ perfect = stems, where the similarity between II-wy forms like *hiqīma > *hēqīm ‘he erected’ and I-y forms like *hayṭiba > *hēṭīb ‘he did well’ caused the I-wy form of the hɔp̄ʕal Another peculiarity of these verbs is . םַקוּה to be extended to the II-wy paradigm, as in The historically short vowel in the first syllable . םיִקֵמ their hip̄ʕil participle, formed like results from loss of the original intervocalic *h with preservation of the preceding vowel in an open syllable: *muhaqīmum > *mihaqīmum (with dissimilation, see Chapter 3) . םיִקֵמ < miqīmum* < Instead of stems with a geminated second radical, II-wy verbs usually form a polel, polal, and hiṯpolel.20 The o in these forms goes back to a contracted diphthong *aw, ;* םֵמוֹק < where *w was originally the second radical of the II-w verbs, as in *qawmima *qawmima, in turn, regularly goes back to the expected D-stem, *qawwima (Barth 1897).

4.2.7 III-wy21 Most forms of verbs with a glide as their third radical were still strong in Proto- Northwest-Semitic. Triphthong contraction (see Chapter 5) affects the endings of the third person perfect, most imperfect forms and the active participle of verbs with *w or *y as a third radical in all stems. The regular outcome of this contraction is -å for -he built’; -ɛ for forms ending in *-u, as in *yab‘ הָנָבּ < forms ending in *-a, as in *banaya u for- ;( הֶנֹבּ .he will build’ (including the masculine active participle, e.g‘ הֶנְבִי < niyu

20 Cf. the discussion in Pardee (2003–2004, 279–280). 21 Some of the reconstructions in this section were first presented at the 12. Mainz Interna- tional Colloquium on Ancient Hebrew, held 30 October through 1 November 2015 at the Johannes Gutenberg Universität Mainz. This paper has now appeared as Suchard (2017).

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access a concise historical morphology of biblical hebrew 253

-they built’; and -i in the second person femi‘ וּנָבּ < forms ending in *-ū, as in *banayū you (f.sg.) will build’. The merger of‘ יִנְבִתּ < nine imperfect, ending in *-ī, as in *tabniyī the stative and fientive paradigms in most forms prompted the generalization of the you (m.sg.) built’ also‘ ָתיִ֫נָבּ < stative paradigm, as can be seen from forms like *baniyta occurring in historically fientive verbs. The third person feminine perfect first regularly developed to *banayat > *banāt > *banat and then added an additional feminine if not , הָניֶ֫נְבִתּ The feminine plural imperfect form . הָתְנָבּ < ending, yielding *banatat originally stative, is due to analogy with the strong verb, where the feminine plural is .to the singular - הָנformed by adding In the jussive, as in Classical Arabic, the third radical was absent in unsuffixed forms, as in *yabni. The word-final vowel was regularly lost. The resulting word-final conso- and he drank’, and‘ ְתְּשֵׁיַּו < nant cluster is retained if it ends in a plosive, as in *yasti -and he built’.The mascu‘ ןֶבִ֫יַּו < otherwise resolved by an epenthetic vowel, as in *yabni line singular imperative, which similarly goes back to a form with a lost third radical, is due ,** יוֹנָבּ for expected הֹנָבּ ,The infinitive absolute . הֵנְבּ < regularly develops like *bni . הֹנָבּ infinitive : הָנָבּ perfect = לוֹטָק infinitive : לַטָק to analogy with the strong verb: perfect The infinitive construct regularly develops from Proto-Northwest-Semitic *binātum > ,see Chapter 4 for the reduction of *i); this was originally a *qiṭlatum verbal noun) תוֹנְבּ with a pre-Proto-Semitic sound change turning *binyatum into *binātum. The passive . יוּנָבּ < participle regularly retains its glide, as in *banūyum In the nip̄ʕal, the first and second person perfect forms do not have -i- < *-iy- after the ,’and it was revealed‘ לָגִּתַּו second radical, but -e- < *-ay-. The consecutive imperfect, like it will be‘ הֶלָגִּתּ is analogically formed by simply removing the ending of the imperfect revealed’, based on the model of the qal. Similarly, the word-final vowel is reintroduced be revealed (m.sg.)’, through analogy with the interchange‘ הֵלָגִּה ,in the imperative between ɛ and e in the qal imperfect and imperative. The same happens in the other -command (m.sg.)’ are also pre‘ וַצ < derived stems, although regular forms like *tṣawwi served. The occasional first person singular perfect forms with e in the piʕel and hip̄ʕil, are best explained with Rubin (2001) as resulting , יִתיִ֫לִּגּ I revealed’ for normal‘ יִתיֵ֫לִּגּ like from a dissimilatory sound change with a very limited conditioning.

4.2.8 III-ʔ The peculiarities of verbs with *ʔ as their third radical are due to a fairly late elision of that consonant in syllable-final position (but before segolization), with compensatory I‘ יִתאָ֫רָק < he called’, *qāraʔtī‘ אָרָק < lengthening of the preceding vowel, e.g. *qāraʔ in ; שׁאֹר < called’. This is a later sound change than that affecting words like *raʔsum these verbs, *ʔ was analogically restored and maintained until much later. Other devi- ations from the strong paradigm are due to analogy with the III-wy paradigm based on the model of *bānā : x = *qārā : y.

Benjamin D. Suchard - 9789004390263 Downloaded from Brill.com09/30/2021 07:59:23AM via free access