ENCYCLOPEDIA of HEBREW LANGUAGE and LINGUISTICS Volume 3 P–Z

ENCYCLOPEDIA OF HEBREW LANGUAGE AND LINGUISTICS Volume 3 P–Z

General Editor Geoffrey Khan

Associate Editors Shmuel Bolokzy Steven E. Fassberg Gary A. Rendsburg Aaron D. Rubin Ora R. Schwarzwald Tamar Zewi

Volume One

Introduction ...... vii List of Contributors ...... ix Transcription Tables ...... xiii Articles A-F ...... 1

Volume Two

Transcription Tables ...... vii Articles G-O ...... 1

Volume Three

Transcription Tables ...... vii Articles P-Z ...... 1

Volume Four

Transcription Tables ...... vii Index ...... 1

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 100 phonology: biblical hebrew the basis of the children’s productions, there is Echols, C., and E. Newport. 1992. “The role of evidence that it also affects their perception. stress and position in determining first words”. Language Acquisition 2:189–220. Gerken, L.-A. 1994. “A metrical template of chil- References dren’s weak syllable omission from multisyllabic Adam, Galit. 2002. “From variable to optimal gram- words”. Journal of Child Language 21:565–584. mar: Evidence from language acquisition and Grunwell, Pamela. 1982. Clinical phonology. Lon- language change”. PhD dissertation, Tel-Aviv don: Croom Helm. University. de Lacy, Paul. 2006. Markedness: Reduction and Adam, Galit and Outi Bat-El. 2008a. “The trochaic preservation in phonology. Cambridge: Cambridge bias is universal: Evidence from Hebrew”. Lan- University Press. guage Acquisition and Development: Proceedings of Lindblom, B. 1986. “Phonetic universals in vowel GALA 2007, ed. by A. Gavarró and M. J. Freitas, systems”. Experimental Phonology, ed. by J. Ohala 12–24. Newcastle: Cambridge Scholars. and J. Jager, 13–44. Orlando, Florida: Academic. ——. 2008b. “Segmental effects on syllable selection: Nespor, Marina, Marcela Peña, and Jacques Mehler. Evidence from Hebrew”. Language Acquisition 2003. “On the different roles of vowels and con- and Development: Proceedings of GALA 2007, sonants in speech processing and language acquisi- ed. by A. Gavarró and M. J. Freitas, 1–11. New- tion”. Lingue e Linguaggio 2:203–229. castle: Cambridge Scholars. Schwartz, R. 1988. “Phonological factors in early ——. 2009. “When do universal preferences emerge lexical acquisition”. The Emergent Lexicon: The in language development? The acquisition of Child’s Development of a Linguistic Vocabulary, Hebrew stress”. Brill’s Annual of Afroasiatic Lan- ed. by Michael Smith and John Locke, 185–222. guages and Linguistics 1:1–28. San Diego: Academic. Adi-Bensaid, Limor. 2006. “The Prosodic Develop- Tubul-Lavy, Gila. 2005. “The phonology of Hebrew ment of Hebrew-Speaking Hearing Impaired Chil- speaking dyspraxic children” [in Hebrew]. PhD dren”. PhD dissertation, Tel-Aviv University. dissertation, Tel-Aviv University. Adi-Bensaid, Limor, and Outi Bat-El. 2004. “The development of the prosodic word in the speech of Outi Bat-El a hearing impaired child with a cochlear implant (Tel-Aviv University) device”. Journal of Multilingual Communication Disorders 2:187–206. Adi-Bensaid, Limor, and Gila Tubul-Lavy. 2009. Phonology: Biblical Hebrew “Consonant-free words: Evidence from Hebrew speaking children with cochlear implants”. Clini- cal Linguistics and Phonetics 23:122–132. Introduction Bat-El, Outi. 2005. “The emergence of the trochaic foot in Hebrew hypocoristics”. Phonology 22:1–29. This entry treats the phonology of Biblical ——. 2009. “Harmonic domains and synchro- nization in typically and atypically developing Hebrew, though on occasion we will refer to Hebrew-speaking children”. Language Sciences data from beyond the domain of BH per se. The 31:117–135. methodology utilized here is that of historical lin- Ben-David, Avivit. 2001. “Language acquisition and guistics, especially since the relevant information phonological theory: Universal and variable pro- cesses across children and cross languages” (in covers more than a thousand years (for an earlier Hebrew), PhD dissertation, Tel-Aviv University. treatment, on which the current essay is largely Ben-David, Avivit, and Ruth Berman. 2007. “Israeli based, see Rendsburg 1997; for amplification of Hebrew speech acquisition”. The International Guide to Speech Acquisition, ed. by Sharynee some of the topics treated herein, see Kutscher McLeod, 437–456. 1982:12–30; for theoretical approaches to the Clements, G. 1990. “The role of the sonority cycle subject Phonology, Generative and Phonol- in core syllabification”. Papers in Laboratory Pho- ogy, Optimality Theory: Biblical Hebrew; for a nology I: Between the Grammar and Physics of Speech, ed. by John Kingston and Mary Beck- synchronic description of the Tiberian tradition man, 283–333. Cambridge: Cambridge University of Hebrew on the basis of medieval sources Press. Tiberian Reading Tradition). Demuth, Katherine. 1996. “The prosodic structure The subject of Biblical Hebrew phonology of early words”. Signal to syntax: Bootstrapping from speech to grammar in early acquisition, ed. is complicated by the fact that ancient Hebrew by J. Morgan and K. Demuth, 171–184. Mahwah, was written with a 22–consonant alphabet— New Jersey: Lawrence Erlbaum Associates. though as we shall see, Hebrew possessed more Demuth, Katherine, and Jane Fee. 1995. Mini- than 22 consonantal phonemes, so that some mal words in early phonological development. Manuscript, Brown University and Dalhhousie of the graphemes (letters) served double duty. University. Moreover, vowels were not represented in the

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 phonology: biblical hebrew 101 writing system (except to some extent via the the ancient Hebrew period; (c) certain sound use of matres lectionis), though, presumably, shifts may be dated to the post-biblical period, any reader of an ancient Hebrew composition as can be determined from the transcriptions would have known how to recite the text on of Hebrew into Greek and Latin; and (d) one the basis of an oral reading tradition passed cannot be certain that the Greek and Latin from tradent to tradent (especially for literary transcriptions of Hebrew represent the same works, such as those which eventually entered reading tradition as the one underlying the the biblical canon). Tiberian Masora. The first three of these issues Eventually, Jewish scholars, known as the are addressed below, while the fourth one .mås< òrå< , should be kept in mind when relevant ָמ ָסוֹרה Masoretes (from the word meaning either ‘tradition’ or ‘counting’ [see The picture presented here is further compli- Dotan 2007:614–615]), created a system of cated by the fact that several reading traditions vowel (along with accent and punctuation) of Biblical Hebrew besides the Tiberian one notation during the 8th–9th centuries C.E. In existed at later periods, including the Babylo- fact, there were three main systems of vocaliza- nian ( Vocalization, Babylonian), Palestinian tion, though in this article we will for the most ( Vocalization, Palestinian) and Samaritan part focus on the Tiberian system, which in ( Samaritan Hebrew: Biblical Hebrew). The time emerged as the dominant one in Jewish Samaritans, who developed as an offshoot of society. Judaism c. 500 B.C.E., also possess the first five The question remains as to how accurately books of the Hebrew Bible (the Torah or Penta- the oral reading tradition of the biblical text teuch) as canonical. They have an independent and the Masoretic transcription thereof reflects reading tradition for their Scripture, though in ancient Hebrew. That is to say, the Masoretic this entry we refrain from entering into these Text (i.e., the traditional text of the Bible) differences. dates to c. 850 C.E. and reflects the manner in which Biblical Hebrew was pronounced at that 1. Consonants time. But how conservative, i.e., how ancient, was the reading tradition of the readers for the At least 29 consonantal phonemes are trace- centuries before c. 850 C.E.? In other words, able to Proto-Semitic (comparison with other does the Masoretic Text reflect Hebrew as it families in the Afroasiatic phylum suggests was pronounced five hundred years earlier, one the possibility of still other phonemes). The thousand years earlier, even fifteen hundred most ancient Hebrew attested retained 25 of years earlier? In some cases, we can answer this these; one local variety of Israelian Hebrew question, but no definitive conclusion can be retained one other phoneme; and the remaining reached. Nevertheless, we will base ourselves three phonemes merged with other phonemes on the assumption that the readers of the 1st (though one cannot discount the possibility millennium C.E. were extremely conservative that any or all of these three may have been in their biblical reading tradition, so that the retained in some restricted geographical locale, Masoretic Text more or less accurately reflects lack of evidence notwithstanding; see below the pronunciation (or at least one pronuncia- §1.13). tion) of ancient Hebrew in the 1st millennium As noted above, the Hebrew (Phoenician) B.C.E., i.e., the time of the composition of the alphabet has only 22 signs, so the recovery of biblical books (see Morag 1974). I say “more the additional three or four phonemes requires or less” because, among other points, (a) in special comment (see below for the individual some instances we know that the Masoretes cases). Below I list the consonantal phonemes no longer recognized consonantal phonemes of ancient Hebrew, grouped according to place which were distinguished in ancient Hebrew, and/or manner of articulation. Transliteration but which merged at a later date; (b) the system is based on the standard system utilized in of vowels according to the Masoretic notation Semitics. Where the IPA symbol differs, it also has a large number of allophones, some or is noted. I also present the letter of the alphabet many of which may have developed only after used to render each phoneme.

1.1. Bilabial plosives /†/—a voiceless emphatic dental plosive, indi- ’On the nature of the ‘emphatics .ט cated by פ—/p/ see §1.14. ב—/b/ 1.4. Nasals 1.2. Interdentals מ—/m/ /μ/ (IPA [θ]). In virtually all dialects of Hebrew, נ—/n/ .(שׁ) ש this phoneme shifted to /š/, indicated by In some Israelian (northern) Hebrew texts, the 1.5. Rolled -may be observed (Rends (ת) /shift of /μ/ > /t burg 2003:10–11). However, in the Hebrew /r/—either a rolled dental or a rolled uvular of Transjordan (specifically Gilead), as well (its exact articulation in ancient Hebrew is .ר as in the neighboring Canaanite dialect of unknown), indicated by Ammonite, this phoneme was retained. The evidence for this comes from the famous pas- 1.6. Sibilants sage in Judg. 12.6, known as the ‘shibboleth ס—/s/ incident’ ( Shibboleth). The story relates how ז—/z/ the Gileadites controlled the fords of the Jordan / /—a voiceless emphatic sibilant (according to River. When retreating Ephraimites (from Cis- ß most opinions it is a fricative [cf., e.g., LipiÐski jordan) sought to cross, the guards at the fords 1997:122]; others hold it to be an affricate [see asked them to pronounce the word μibbòlÆt most importantly Steiner 1982]), indicated by [μibbòlÆt], which in Hebrew means ‘stream, tor- .On the nature of the ‘emphatics’ see §1.14 .צ rent’, a fitting password for the crossing of the Since this letter represented .ש—([∑] š/ (IPA/ Jordan River. Since most Israelites did not pos- more than one sound relatively late in the his- sess this sound in their phonetic inventory, the tory of Hebrew, a diacritical mark was added Ephraimites would say [sibbòlÆt], thus reveal- by the Masoretes on the right side to produce ing the fact that they were not Gileadites (cp. the grapheme שׁ; see further §2.1. the manner in which various foreign speakers of English [Germans, for example] pronounce 1.7. Laterals English /θ/ as [s], or the manner in which Per- ל—/sians and other non-Arab Muslims pronounce /l -For a thorough survey con .ש—([Arabic /θ/ as [s].) Since standard Hebrew (and /«/ (IPA [ the dialect of Canaanite for which the alphabet cerning this phoneme, see Steiner 1977. Since represented more than one sound ש was invented) did not possess this phoneme, the letter there was no special grapheme for representing relatively late in the history of Hebrew, a dia- this sound. In the passage just mentioned, Judg. critical mark was added by the Masoretes on see ;שׂ š> is used. the left side to produce the grapheme> =(שׁ) ש the letter ,12.6 For the secondary development of /μ/ = [θ] as further §2.1. the fricativized form of /t/, see §2.4. (Note, inci- On the one remaining lateral of Proto- dentally, that in my above transcriptions of the Semitic, see §1.13. Hebrew words underlying the ‘shibboleth incident’, I do not reflect this phonetic shift, which 1.8. Velar plosives occurred at a later date, since in this instance I כ—/k/ have attempted to replicate the pronunciation ג—/g/ of Hebrew during the Early Iron Age, the pur- /q/—a voiceless emphatic velar plosive, indi- ported setting of Judg. 12.) ’On the nature of the ‘emphatics .ק cated by On the two remaining interdentals of Proto- see §1.14. Semitic, see §1.13. 1.9. Velar fricatives 1.3. Dental plosives This sign was also used to .ח—([IPA [x) / / • ת—/t/ represent / /. We are able to postulate the exis- ™ ד—/d/

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 phonology: biblical hebrew 103 tence of both phonemes in the ancient period 1.11. Laryngeals on the basis of transcriptions of Hebrew words א—([IPA [π) /± / (mainly proper names) in the Septuagint (the ה—/h/ ancient Greek translation of the Bible) of the Pentateuch (c. 250 B.C.E.; see Blau 1982). 1.12. Glides (semi-vowels) When Proto-Semitic comparisons indicate that ו—/the consonant /•/ is present in the Hebrew /w י—([word, the Septuagint transcription uses χ (see /y/ (IPA [j §1.10 for the practice of transcribing /™/). For the eventual merger of /•/ and /™/ see §2.2. For 1.13. The remaining Proto-Semitic phonemes the secondary development of / / = [x] as the ú There are three remaining traceable Proto- fricativized form of /k/ see §2.4. Semitic phonemes: /≈/ (IPA [ð]), /Ω/ (IPA [ð’]), This sign was also used to .ע—([IPA [y) /// and /∂/ [IPA [ ’]). There is no evidence for represent / /. We are able to postulate the exis- ≠ the preservation of these sounds in ancient tence of both phonemes in the ancient period Hebrew. Instead, in most regional dialects of on the basis of transcriptions of Hebrew words ancient Hebrew, /≈/ shifted to /z/ (in some (mainly proper names) in the Septuagint of Israelian dialects it shifted to /d/); and both /Ω/ the Pentateuch (c. 250 B.C.E.; see Blau 1982). and /∂/ shifted to /ß/ (in some Israelian dialects When Proto-Semitic comparisons indicate that the former shifted to /†/ and the latter shifted the consonant /// is present in the Hebrew to /q/ or later to / ≠/). At the same time, it must word, the Septuagint transcription uses γ (see be admitted that any one, two, or three of §1.10 for the practice of transcribing / /). For ≠ these phonemes may have been preserved in the eventual merger of / / and / / see §2.2. For / ≠ some locales. But since the Hebrew alphabet the secondary development of /: / = [ ] as the g y lacks special signs to represent these sounds, fricativized form of /g/ see §2.4. it is difﬁcult to ascertain if and where such phonemes may have been retained. Were it not 1.10. Pharyngeal fricatives for the story in Judg. 12.6 (see §1.2), we would This sign was also used to not know that Gileadite Hebrew retained the .ח—([IPA [Ó) /™/ represent /•/. We are able to postulate the exis- voiceless interdental /μ/, so it is conceivable that tence of both phonemes in the ancient period elsewhere in ancient Hebrew /≈/, /Ω/ and /∂/ per- on the basis of transcriptions of Hebrew words sisted. Finally, note that Huehnergard (2003) (mainly proper names) in the Septuagint of has posited the existence of yet another proto- the Pentateuch (c. 250 B.C.E.; see Blau 1982). Semitic consonant, namely, an emphatic velar When Proto-Semitic comparisons indicate that fricative /!x/ (IPA /x’/), though to be sure neither the consonant /™/ is present in the Hebrew Hebrew nor any other known Semitic language word, the Septuagint transcription shows no actually attests to this proto-phoneme. consonant (see §1.9 for the practice of transcribing /•/). For the eventual merger of /™/ and 1.14. The nature of the emphatics / / see §2.2. • The exact nature of the emphatic consonants This sign was also used to .ע—([∏] IPA) /≠ / /†/, /ß/, and /q/ cannot be determined. Among represent / /. We are able to postulate the exis- / Semitic languages still spoken today, the corre- tence of both phonemes in the ancient period sponding consonants in Ethiopian and Modern on the basis of transcriptions of Hebrew words South Arabian are glottalized, while in Arabic (mainly proper names) in the Septuagint of they are velarized or pharyngealized. Most likely the Pentateuch (c. 250 B.C.E.; see Blau 1982). the glottalization is the original Proto-Semitic When Proto-Semitic comparisons indicate that manner of articulation (see Aro 1977), so that the consonant / / is present in the Hebrew ≠ this can be postulated for ancient Hebrew (for word, the Septuagint transcription shows no the opposite view see LipiÐski 1997:105–106; consonant (see §1.9 for the practice of tran- see also Steiner 1982; Emphatic Consonants; scribing / /). For the eventual merger of / / and / ≠ Affrication). /// see §2.2.

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 104 phonology: biblical hebrew 2. Historical Changes in the Bible, which was accomplished several decades Consonantal Phonology or perhaps even a century later, this consistency disappears (see Blau 1982). Accordingly, we The consonantal phonology described above can confidently fix this phonological develop- is correct for Hebrew in its earliest attested ment to c. 200 B.C.E. phase. But already by the biblical period there is evidence of various changes, and in the post- 2.3. Weakening of the pharyngeals and biblical period still more changes are evident. laryngeals These historical developments will be presented In the preceding paragraph we observed that here. c. 200 B.C.E. the velar fricatives /•/ and /// merged with the corresponding pharyngeals /™/ 2.1. The shift of /«/ to /s/ and / ≠/. As time passed, there is evidence for an In the course of time the voiceless lateral frica- overall weakening of the pronunciation of the tive /«/ shifted to a sibilant and merged with /s/. pharyngeals and laryngeals, especially in Qum- This is indicated by the numerous interchanges ran Hebrew and Rabbinic Hebrew ( Guttural in the spelling of ancient Consonants). On the one hand, the tradition ס and (שׂ) ש between Hebrew (see Blau 1970:23–24, 114–125). This that yielded the Tiberian Masorah preserved tendency is less acute in the pre-exilic (pre-586 the articulation of these consonants; on the B.C.E.) books of the Bible, but becomes more other hand, certain effects of their weakening common in the exilic and post-exilic (post-586 are discernible in the Masoretic vocalization B.C.E.) books. Thus, we may conclude that the system. These include: (a) the consonants /™/, merger of /«/ and /s/ occurred in Late Biblical / ≠/, /h/, and /±/ cannot be geminated (this holds Hebrew and continued in still later phases of for /r/ as well); (b) these consonants cannot the language. This shift may be the result of be vocalized with simple shewa, but instead Aramaic influence. require an auxiliary vowel; and (c) in final posi- In the centuries after the merger occurred, tion an anaptyctic vowel is required after a long ַרוּח < /™copyists of the Bible remained faithful to the high vowel (for all except / ±/), e.g., /rù received text. Accordingly, even though his- rùa™ ‘wind’. torical /«/ now was pronounced the same as /s/, in the great majority of cases the biblical 2.4. Fricativization (spirantization) of manuscripts continue to represent this sound non-emphatic plosives When the Masoretes devised their .(שׂ) ש with At some point in ancient Hebrew, the six system of marking all phonetic distinctions in non-emphatic plosives, /p/, /b/, /t/, /d/, /k/, /g/, the received text, diacritic marks were invented developed a twofold realization. In post-vocalic to distinguish the two sounds represented by position they came to be pronounced as frica- With the dot placed over the upper left hand .ש tives (spirants); otherwise they retained their corner, the grapheme שׂ represented the former original plosive character. The correspond- lateral fricative /«/, now pronounced [s]. With ing fricative (spirantized) pronunciations are, the dot placed over the upper right hand cor- respectively: /ƒ/, /v/, /μ/ (IPA [θ]), /≈/ (IPA [ð]), ner, the grapheme שׁ represented /š/. /ú/ (IPA [x]), /g: / (IPA [y]). Almost without exception, these sounds are allophones. Only 2.2. Merger of /•/ and /™/ and merger of /// in rare instances, due to other factors, did and / ≠/ phonemic differences arise. Exactly when the In c. 200 B.C.E., the phoneme /•/ merged with fricativization of the non-emphatic plosives in the phoneme /™/, and the phoneme /// merged post-vocalic position occurred cannot be deter- with the phoneme / ≠/. This can be determined mined, though most scholars date the phenom- from the following. In the Septuagint of the enon to the 5th century B.C.E., perhaps under Pentateuch, accomplished c. 250 B.C.E., these Aramaic influence. individual phonemes are represented differently Several of these allophones are equivalent to in the Greek transcription of proper names and other phonemes in the language. For example, occasional common nouns (see §1.9, §1.10). /ú/ is the same as /•/ (both IPA [x]), and /g: / is But in the Septuagint of the other books of the the same as /// (both IPA [y]). Assuming, as © 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 phonology: biblical hebrew 105 most scholars do, that the fricativization of ized/pharyngealized as well. This pronunciation /k/ to /ú/ [x] and of /g/ to /g: / [y] occurred c. remains current among the Jewish communities 400 B.C.E., and that /•/ [x] and /// [y] were of North Africa and the Middle East. Jews in still articulated as late as c. 200 B.C.E. (see Europe, on the other hand, lost the ability to §2.2), then we may posit the coexistence of pronounce the emphatic consonants altogether. t], so that it merged] < (ט) /†/ ,two phonemes and their respective phonetically Thus, in time k], so that it merged with] < (ק) /q/ ;(ת) /identical (or almost identical) allophones for with /t ts], a phoneme common] < (צ) /and /ß ;(כ) /about two centuries (Blau 1982:74–75). Simi- /k larly, the fricativization of /t/ to /μ/ may have in many European languages, e.g., German resulted in another such case, if we assume that (again, see Morag 2007; cf. Steiner 1982; at the same time at least one Hebrew dialect Affrication). retained the original phoneme /μ/ (see §1.2). Yet there is hardly any confusion between the 2.6. Shift of initial w- > y- respective phones (one a phoneme, the other an A standard historical phonological rule is the allophone), thus demonstrating that speakers shift of initial w- > y-, not only in Hebrew, of a language may possess abstract representa- but throughout Northwest Semitic. Thus, for tions of the relevant sounds beneath the level of ,’yåša< ∫ ‘he sat, he dwelt ָי ַשׁב < example, *wašab surface phonetics (Blau 1982:75). though in non-initial position (as expected Clearly the six fricatives under discussion based on the rule) the /w/ remains, as in here were pronounced by all (?) Jews c. 800 ’hòša∫tì ‘I caused to dwell ַ֙הוֹשׁ ְב ִ ֙תּי < hawšabti* C.E., when the Tiberian system of the Masora > ‘I settled’ (Lev. 23.43), with the requisite was developed. In time, however, the ability to monophthongization aw > ò (see below, §3.3). pronounce some of these sounds was lost by The main exception to this rule is the ubiqui- various Jewish communities, especially those w- (vocalized in different ו- tous conjunction in Europe. ways) ‘and’. The three sounds which remained most stable were /v/, / /, and /ƒ/. Among most European ú 3. Vowels Jews, however, /μ/ was realized as [s] (compare the ‘shibboleth incident’ described in §1.2, The exact pronunciation of the vowels of though there is no direct connection between ancient Hebrew cannot be recovered. How- the two phenomena). In the two remaining ever, we may assume that the classical pattern cases, /: / and / /, fricativization disappeared g ≈ of Semitic (illustrated best in Classical Arabic) and /g/ and /d/ were pronounced as [g] and [d] was operative in Hebrew in its earliest histori- in all environments. On the other hand, some cal period. Thus we can reconstruct three basic Jewish communities in Arab lands retained vowels, short and long: /a/, /i/, /u/, /à/, /ì/, /ù/. most, if not all, of the fricativized allophones The Masoretic notation system, as noted above, into the 20th century. The Jews of Yemen dates to the 8th–9th centuries C.E., and most are the best-known example of a community accurately reflects the pronunciation of Hebrew whose pronunciation of Hebrew includes the in the early medieval period. By this time, the proper realization of all six allophones. For classic three-vowel (short or long) system was further discussion see Morag 2007. no longer operative, as many allophones had developed, based on a complex system of syl- 2.5. Velarization of the emphatics labification and accentuation. Again, exactly Above (§1.14) we discussed the nature of the when the shift from a quantitative system of the emphatics, with the supposition that they were basic three vowels (short or long) to the quali- most likely originally glottalized. Because the tative system to be described below occurred corresponding consonants in Arabic are velar- is unknown. But it is apposite to quote the ized or pharyngealized, and because the major- view of Jerome (c. 400 C.E.): “It is of no con- ity of Jews in the world c. 1000 C.E. lived in an sequence whether [the word Shalem] is pro- Arabic-speaking milieu and themselves spoke nounced Salem or Salim, because Hebrew very Arabic as their native language, in time the rarely uses vowel letters in the course of words, emphatic consonants in Hebrew became velar- and according to the discretion of readers

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 106 phonology: biblical hebrew ( ַ ) ™and the different regions the same word is /a/ > [a] pata ( ִ ) pronounced with different sounds and accents” /i/ > [i] ™iriq (Letter ad Evangelum, no. 73 ed. Migne). In or ( ֶ ) other words, there was much local variation /i/ > [Æ] segol ( ֻ ) in the realization of the vowels. One may wish /u/ > [u] šureq to compare the situation in colloquial Arabic, or (where slight changes in vowels are noticeable /u/ > [å] qameß ( ָ ) (typically called qameß qa†an in its various dialects (for example, the definite article can be [al], [el], [il], [ël], or [l]). Different environments usually determine Below we present the vowel system accord- whether /i/ > [Æ] as opposed to remaining [i], ing to the Tiberian Masoretic system. We begin and whether /u/ > [å] as opposed to remain- with the long vowels, which are far simpler in ing [u]. For example, if the vowel is followed their historical development, then move to the by a geminated consonant, one can expect /i/ ,[libbì ‘my heart’, and /u/ > [u ִל ִבּי ,.short vowels, and conclude with a treatment of > [i], e.g kullåm< ‘all of them’ respectively; by ֻכּ ָלּם ,.the diphthongs. e.g lÆ∫-yåm< ‘heart ֶל ָב־ים contrast witness /i/ > [Æ] in kål-±ìš ‘every ָכּ ִל־אישׁ Long vowels of the sea’, and /u/ > [å] in .3.1 man’. Typically, the Proto-Semitic long vowels retain If the short vowel occurs in an open syllable their basic pronunciation in all environments. more than one syllable before the accent, then Thus, /ì/ is almost always [ì], and /ù/ is almost .( ְ the vowel is reduced to shewa [ë] (noted by always [ù]. The only area of real fluctuation is If, however, the consonant involved is a pha- with /à/. When Semitic cognates indicate /à/, ryngeal or a laryngeal, then an auxiliary vowel the Hebrew reflex is /ò/ ( Canaanite Shift). is necessary (often called ‘compound vowel’ ל ֹא Thus, for example, Arabic là = Hebrew due to its orthographic representation in the šål< òm ָשׁלוֹם lò ‘no’; Arabic salàm = Hebrew Masoretic system) (see §2.3). The auxiliary ‘peace’; etc. vowel is halfway between a true shewa and the corresponding short vowel. Thus one finds the 3.2. Short vowels following: The above discussion (§3) about the numerous ( ֲ) [≤] < /vowel allophones refers most importantly to /a ( ֱ) [ °the short vowels. The Tiberian Masoretic nota- /i/ > [ł ( ֳ) [tion system reflects different realizations of the /u/ > [å three original vowels /a/, /i/, and /u/, depending on both the accent and the nature of the syl- We illustrate this whole process with one exam- lable in which the vowel occurs. ple. The proto-Hebrew word for ‘word’ is If the short vowel occurs in an accented *dabár (with short /a/ vowels in both syllables, syllable, or in an unaccented open syllable and with the accent mark indicating the stress) då< ∫år< . The first å< occurs ָדּ ָבר immediately preceding the accent, the following > Masoretic developments occur (I include here the name because it appears in an unaccented open syl- of the Hebrew vowel, along with its Tiberian lable immediately preceding the accent; while symbol in parentheses): the second å< occurs because it appears in an ְדּ ַבר־יהוה accented syllable. In the expression : a/ > [å] qameß ( ָ) dë∫ar-yhwh ‘the word of YHWH’, the two words/ i/ > [è] ßere ( ֵ ) together have only a single accent, at the end/ u/ > [ò] ™olem ( )ׂ of the divine name. The first /a/ vowel now/ appears in an unaccented open syllable more If the short vowel occurs in an unaccented closed than one syllable before the accent, and thus it syllable, typically the original pronunciation is is reduced to shewa. The second /a/ vowel now not affected, though with two of the vowels appears in an unaccented closed syllable and there is the possibility of an allophone. Thus: thus is realized as [a].

Note that one Hebrew vowel sign, the qameß 3.3. Diphthongs is transliterated as å< when it derives from ,( ָ ) Two diphthongs are reconstructed for ancient an /a/ vowel, but is transliterated as å when it Hebrew (as in general Semitic) in its earliest derives from an /u/ vowel. In essence, however, stage: aw and ay. In some cases, mainly in final the Masoretic notation clearly demonstrates a position, these diphthongs remain unchanged, single pronunciation for this vowel, as an open ay ‘alive’ (though with™ ַחי ,’qåw< ‘line ָקו ,.e.g mid back vowel = IPA [ ]. This shows that /a/, – the former note again the raising of the vowel when it was accented and when it appeared to [–]). Typically, however, one of two changes in an open syllable immediately preceding the occurs. The first option is the insertion of an accent (see above), was raised to a quality anaptyctic vowel to form two syllables, thus, approaching /o/. Such a process is clearly indi- ַבּ ִית < måw< Æμ ‘death’, *bayt ָמֶות < e.g., *mawt cated for Phoenician, and was no doubt true bayiμ ‘house’ (again note the raising of the of ancient Hebrew as well, at least in the vowel in the former example), in effect creating pronunciation tradition which was transmitted something close to a triphthong in each case. by the Tiberian Masoretes. As such, this shift The second option is monophthongization, parallels the case of long /à/ > /ò/ (see §3.1); which in Hebrew almost always means aw > ò and thus we may wish to postulate a general šòr ‘bull’, *bay∂a שׁוֹר < and ay > è, e.g., *μawr drift in this direction in ancient Hebrew and bèßå< ‘egg’. Though in a small number of ֵבּ ָיצה < Phoenician. instances, these two diphthongs monophthon- It is important to note that the above chart- ,ån< ‘(to) where± ָאן < gize to å< = [–], e.g., *±ayn ing of rules governing the short vowels is not until when’ (1 Sam. 10.14; Job 8.2). Examples to be taken as hard and fast. As in most lan- of this latter process may be localized to two guages, also in Hebrew, /a/ is the most stable geographical regions in Israel: the northern part vowel. When an /i/ vowel or an /u/ vowel is of the country (Galilee) and a small pocket in present, the above rules are often violated. southern Judah (northern Negev) (see Rends- b r š ְבּרוֹשׁ < For example, *bur š > *bur š à ò ë ò burg 1990). ‘juniper, cypress’ shows reduction of the /u/ ֲחמוֹר < vowel to shewa, and * im r > * im r ™ à ™ ò 3.4. Historical changes concerning the vowels ™≥mòr ‘donkey’ shows reduction of the /i/ vowel to composite shewa (due to the presence of the 3.4.1. /i/ > /a/ in an originally closed accented pharyngeal /™/ ), even though in both cases syllable the open syllable in which these vowels occur This law is known as Philippi’s Law. An origi- immediately precedes the accented syllable. By nal /i/ vowel shifts to /a/ in an originally closed contrast, of similar nominal pattern is *šalòš > accented syllable (that is, a syllable that was šål< òš ‘three’, with the /a/ vowel retaining closed even in its proto-form [as opposed to ָשׁלוֹשׁ its character (though with slight raising to [–], a closed syllable brought about by some other as discussed in the preceding paragraph). historical development]) (for further details Similarly, auxiliary vowels can arise after Philippi’s Law). Thus, for example, Proto- consonants which are not pharyngeals or laryn- Semitic *gint > *gitt (via assimilation, see §4.2) geals. For example, /u/ does not reduce to > *git (with surrendering of word-final gemi- gaμ ‘winepress, olivepress’. In ַגּת < (hag-går≥ ån< òμ ‘the nation ַה ֳגָּרנוֹת shewa in the word threshing floors’ (1 Sam. 23.1; Joel 2.24); Akkadian transcriptions of the city in Canaan rather it appears as /å≥ / = IPA / ≥– /. This is due by this name (English ‘Gath’), dating to as late to the circumstance of back vowel /u/ follow- as c. 720 B.C.E., the form is still Gint (or Gimt ing the velar consonant /g/. Instead of reducing [with partial dissimilation]). In the Greek trans- fully to shewa, as normally would be expected lation of the historical and prophetic books c. in the case of an unaccented open syllable more 200 B.C.E., the form is Γεθ (GÆμ), and in the than one syllable before the accent, /u/ retains Masoretic text the pronunciation is (as noted part of its original quality (i.e., as a back vowel) above) gaμ. Accordingly, we are able to trace following a consonant pronounced in the back the historical development of this shift, even of the mouth (i.e., the velar /g/) (see Garr though the Greek transcription is too equivo- 1990:59). cal to allow us to pinpoint the exact century

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 108 phonology: biblical hebrew in which Philippi’s Law was operative. Further /l/, a common trigger for metathesis in world complicating the matter is the fact that the languages. Septuagint and Hexapla reflect a lowering of /i / > /e/ generally in contexts other than those 4.2. Assimilation where the Masoretic pronunciation reflects the Anticipatory (so-called ‘regressive’) assimila- /i/ > /a/ shift in accordance with Philippi’s Law tion occurs with unvocalized /n/, except before ( Transcription into Greek and Latin Script: pharyngeals and laryngeals ( Assimilation: Pre-Masoretic Period). Pre-Modern). Thus, for example, to use an item noted earlier (see §3.4.1), *gint > *gitt (eventu- 3.4.2. /a/ > /i/ in an originally closed unaccent- .(’ga ‘winepress, olivepress ַגּת ally shifting to ed syllable μ ִידֺּר Similarly, *yandur eventually emerges as This law does not have an official name, but it yidd r ‘he vows’. Note also the same phenom- may be called the corollary to Philippi’s Law ò enon with vowelless /l/ in various forms of (others prefer the term ‘attenuation of /a/ ’). ִי ַקּח < l-q- ‘take’ (e.g., *yilqa לק"ח the verb An original /a/ vowel shifts to /i/ in an origi- ™ ™ yiqqa ‘he takes’), though this particular case nally closed unaccented syllable (again, that ™ is due to analogy with the antonymic verb is, a syllable that was closed even in its proto- yitt n ‘he ִי ֵתּן < n-t-n ‘give’ (e.g., *yintin נת"ן form [as opposed to a closed syllable due to è gives’). some other historical development]). Thus, for Other examples of anticipatory assimilation mig: dål< ‘tower’ (also ִמ ְג ָדּל < example, *magdal involve the dental consonants, e.g., when vow- šimšòn ִשׁ ְמשׁוֹן < a toponym ‘Migdal’); *šamšòn elless /d/ precedes its voiceless counterpart /t/. ‘Samson’; etc. In the Septuagint (3rd century ’a a ‘one ַא ַחת < A regular example is * a adt B.C.E.) and the New Testament (1st century ± ™ ± ™ μ (f.). A unique example occurs in 1 Sam. 4.19: C.E.), the Greek renderings of proper names *lalidt ‘to give birth’ > *laladt (via Philippi’s reflect the original /a/ vowel (witness the Eng- -låla< (with surrender ָל ַלת < Law) > *lalatt lish forms: Samson, Mary Magdalene, etc.). μ ing of final gemination); the normal form is Jerome (c. 400 C.E.) still has Magdal in his lål< ‘to give birth’, arrived at through ָל ֶל ֶדת Latin translation of the Bible. The Masoretic Æ≈Æμ different means. In like fashion, vowelless /t/ text reflects the shift to /i/ at some point within assimilates to its voiced counterpart /d/ in vari- the following four and a half centuries. Thus, ִמ ַדּ ֵבּר < ous Hitpa≠el forms, e.g., *mitdabber we may date this shift to sometime between middabb r ‘conversing’. 400 C.E. and 850 C.E. è Partial lag (so-called ‘progressive’) assimilation occurs in the Hitpa≠el form of the verb, 4. Sound Changes when the first root consonant is /z/ or /ß/ and it precedes /t/ (see above §4.1). Examples with /ß/ 4.1. Metathesis include the following: *nitßaddaq > *nißtaddaq niß†addåq< ‘(how) shall we justify ִנ ְצ ַט ָ ֑דּק < The most consistent case of metathesis occurs ourselves’ (Gen. 44.16); and *hitßayyadn ù > in the Hitpa≠el form of the verb, when the first ’hiß†ayya≈nù ‘we provisioned ourselves ִה ְצ ַטַיּ֤ ְ דנוּ root consonant is any of the sibilants, /s/, /z/, (Josh. 9.12). In both of these, the characteristic /š/, /ß/, or the lateral fricative /«/. In such cases, /t/ of the Hitpa≠el stem shifts to /†/ because of the /t/, which forms part of the morphology the preceding /ß/. No examples with /z/ occur of this verbal stem and which normally pre- in the Bible, but from post-biblical Hebrew cedes the first root consonant, interchanges one may cite forms such as *mitzayyip > *miz- with the above consonants, e.g., *wa-±itšammir mizdayyep< ‘is forged’ (Tosefta ִמ ְז ַדּ ֵיּיף < tayyip .wå-< ±Æštammèr ‘I guarded myself’ (Ps ָ֝ו ֶא ְשׁ ַתּ ֵ֗מּר < ≠Avoda Zara 4.12), in which /t/ shifts to /d/ 18.24). because of the preceding /z/. Other examples of metathesis are the word k « ‘sheep’, and ֶכּ ֶשׂב ~ » k ֶכּ ֶבשׂ pairs Æ∫Æ Æ Æ∫ 4.3. Elision ,’almå< ‘article of clothing» ַשׂ ְל ָמה ~ >imlå» ִשׂ ְמ ָלה both of which interestingly contain the lateral Intervocalic /h/ and /y/ frequently elide in Bibli- fricative /«/, with the second set also including cal Hebrew, especially in certain morphological

© 2013 Koninklijke Brill NV ISBN 978-90-04-17642-3 phonology: biblical hebrew 109 / >bå< ±å ָ֗בּ ָאה .environments. Examples include: *bètahum > ticiple, accented on the ultima) vs bèμåm< ‘their (mpl) house’ (with the 3mpl. ‘she came’ (3fs suffix-conjugation, accented on ֵבּ ָיתם bab- the penult), both of which are predicated of ַבּ ַבּ ִית < pronominal suffix); *bë-hab-bayit bayiμ ‘in the house’ (with the definite article); Rachel in the same story (Gen. 29.6, 9, respec- bån< ù\ ‘they built’ Gen. 11.5 (3cpl ָבּ ֖נוּ (yaq†ìl (paradigmatic form of tively); b ַי ְק ִטיל < yëhaq†ìl* (suffix-conjugation, accented on the ultima ָבּנוּ < the Hiph≠il prefix-conjugation); *banayù / bån< ù ‘over us’ Gen. 37.8 (preposition ָ ֑בּנוּ .bån< ù ‘they built’ (with the 3mpl suffix-conju- vs verbs); etc. It should be noted, bë- + 1cpl suffix pronoun -nù, accented on the ל"י gation of wë-±åmart< Û ‘and I shall ְו ָא ַמְר ִ ֣תּי (though, that instances of the non-elision of /h/ penult); and c or /y/ are attested in Israelian Hebrew (Rends- say’ Gen. 24.43 (1cs wëqa†al form, with the burg 2003). accentual shift from the penult to the ultima) wë-±åmárt< ì ‘and I said’ Qoh. 9.16 ְו ָא ַ ֣מְר ִתּי .vs 4.4. Anaptyxis (1cs suffix-conjugation, preceded by conjunc- tive wë-, with no accentual shift). The presence of anaptyctic vowels has been noted on several occasions above (see §2.3, References §3.3). One further example occurs in the cre- Aro, Jussi. 1977. “Pronunciation of the ‘emphatic’ ation of the segholate nouns, e.g., *dalt ‘door’ consonants in Semitic languages”. Studia Orien- .dål< Æμ (attested in Hebrew in sentence talia 47:5–18 ָדּ ֶלת < Blau, Joshua. 1970. On pseudo-corrections in some positions requiring a pause, e.g., at the end Semitic languages. Jerusalem: Israel Academy of .dÆlÆμ (reflecting vowel har- Sciences and Humanities ֶדּ ֶלת < (of a verse mony) ( Segholates: Pre-Modern Hebrew). ——. 1982. On polyphony in Biblical Hebrew (Pro- Greek and Latin transliterations of such words ceedings of the Israel Academy of Sciences and Humanities 6/2). Jerusalem: Israel Academy of tend to show the forms without anaptyxis, Sciences and Humanities. though they do so inconsistently ( Transcrip- Dotan, Aron. 2007. “Masorah”. Encyclopaedia tion into Greek and Latin Script: Pre-Masoretic Judaica, 2nd edition, vol. 13, 603–656. Period). In any case, this development most Garr, W. Randall. 1990. “Interpreting orthography”. The Hebrew Bible and Its Interpreters, ed. by likely occurred in the 1st millennium C.E. William H. Propp, Baruch Halpern, and David Noel Freedman, 53–80. Winona Lake, Indiana: 4.5. Prosthetic vowel Eisenbrauns. Huehnergard, John 2003. “Akkadian • and West Semitic *™”. Studia Semitica (Orientalia: Papers The pronunciation of (potential) initial conso- of the Oriental Institute 3; Alexander Militarev nant clusters is assisted by the placement of a volume), ed. by Leonid Kogan, 102–119. Moscow: prosthetic vowel. The best example is the attes- Russian State University for the Humanities. Kutscher, Eduard Yechezkel. 1982. A history of the ֶא ְז ַרוֹע zëròa≠ (passim) and ְז ַרוֹע tation of both Hebrew language, ed. by Raphael Kutscher. Jeru- ±Æzròa≠ ‘arm’ (Jer. 32.21; Job 31.22), though salem: Magnes / Leiden: Brill. the latter may be limited to specific regional LipiÐski, Edward. 1997. Semitic languages: Outline dialects and/or should be deemed an Arama- of a comparative grammar. Leuven: Peeters. ba ‘finger’, Morag, Shelomo. 1994. “On the historical validity of ֶא ְצ ַבּע ism. Another example is ±Æß ≠ the vocalization of the Hebrew Bible”. Journal of built from an earlier (albeit unattested) form the American Oriental Society 94:307–315. without initial ±Æ-, as demonstrable from both ——. 2007. “Pronunciations of Hebrew”. Encyclo- cognate evidence (cf. Egyptian ≈b≠) and inter- paedia Judaica, 2nd ed., vol. 16, 547–562. nal Hebrew evidence (note the denominative Rendsburg, Gary. A. 1990. “Monophthongization of aw/ay > à in Eblaite and in Northwest Semitic”. .haßbì≠ù ‘raise a finger’ [> ‘point’] Eblaitica 2:91–126 ַה ְצ ִבּיעוּ verb [Mishna Yoma 2:1]). The cardinal numeral ——. 1997. “Ancient Hebrew phonology”. Phonolo- ,arba≠ ‘four’ (f) also contains a pros- gies of Asia and Africa ed. by Alan S. Kaye, vol. 1± ַאְר ַבּע pp. 65–83. Winona Lake, Indiana: Eisenbrauns. rë∫ì≠ìμ ——. 2003. “A comprehensive guide to Israelian ְר ִב ִיﬠית thetic vowel (cf. the ordinal form ‘fourth’ [fem.]), though this is true for all the Hebrew: Grammar and lexicon”. Orient 38:5–35. Semitic languages (cf. Akkadian erba, Ugaritic Steiner, Richard C. 1977. The case for fricative-lat- ±rb≠, Sabaic ±rb≠, Arabic ±arba≠, etc.). erals in Proto-Semitic. New Haven, Connecticut: American Oriental Society. ——. 1982. Affricated ßade in the Semitic Lan- 5. Stress guages. New York: American Academy for Jewish Research.