Biblical Hebrew phonology

The phonology as reconstructed for Biblical Hebrew is as follows:

Consonants
Consonants lost and gained during the lifetime of Biblical Hebrew are color-coded respectively.

The phonetic nature of some Biblical Hebrew consonants is disputed. The so-called "emphatics" were likely ejective, but possibly pharyngealized or velarized. Some argue that were affricated.

Originally, the Hebrew letters ח and ע each had two possible phonemes, uvular and pharyngeal, unmarked in Hebrew orthography. However the uvular phonemes ח and  ע merged with their pharyngeal ones  ח and  ע respectively c. 200 BCE.

This is observed by noting that these phonemes are distinguished consistently in the Septuagint of the Pentateuch (e.g. Isaac יצחק = Ἰσαάκ versus Rachel רחל = Ῥαχήλ), but this becomes more sporadic in later books and is generally absent in Ezra and Nehemiah.

The phoneme, is also not directly indicated by Hebrew orthography but is clearly attested by later developments: It is written with $\langleש\rangle$ (also used for ) but later merged with (normally indicated with $\langleס\rangle$). As a result, three etymologically distinct phonemes can be distinguished through a combination of spelling and pronunciation: written $\langleס\rangle$,  written $\langleש\rangle$, and  (pronounced  but written $\langleש\rangle$). The specific pronunciation of as  is based on comparative evidence ( is the corresponding Proto-Semitic phoneme and still attested in Modern South Arabian dialects) as well as early borrowings (e.g. balsam < Greek balsamon < Hebrew baśam). began merging with in Late Biblical Hebrew, as indicated by interchange of orthographic $\langleש\rangle$ and $\langleס\rangle$, possibly under the influence of Aramaic, and this became the rule in Mishnaic Hebrew. In all Jewish reading traditions and  have merged completely; however in Samaritan Hebrew  has instead merged with.

Allophonic spirantization of to  (known as begadkefat spirantization) developed sometime during the lifetime of Biblical Hebrew under the influence of Aramaic. This probably happened after the original Old Aramaic phonemes disappeared in the 7th century BCE, and most likely occurred after the loss of Hebrew  c. 200 BCE. It is known to have occurred in Hebrew by the 2nd century CE. After a certain point this alternation became contrastive in word-medial and final position (though bearing low functional load), but in word-initial position they remained allophonic. This is evidenced both by the Tiberian vocalization's consistent use of word-initial spirants after a vowel in sandhi, as well as Rabbi Saadia Gaon's attestation to the use of this alternation in Tiberian Aramaic at the beginning of the 10th century CE.

The Dead Sea scrolls show evidence of confusion of the phonemes, e.g. חמר ħmr for Masoretic אָמַר 'he said'. However the testimony of Jerome indicates that this was a regionalism and not universal. Confusion of gutturals was also attested in later Mishnaic Hebrew and Aramaic (see Eruvin 53b). In Samaritan Hebrew, have generally all merged, either into, a glide  or , or by vanishing completely (often creating a long vowel), except that original  sometimes have reflex  before.

Geminate consonants are phonemically contrastive in Biblical Hebrew. In the Secunda are never geminate. In the Tiberian tradition cannot be geminate; historically first  degeminated, followed by, , and finally , as evidenced by changes in the quality of the preceding vowel.

Vowels
The vowel system of Biblical Hebrew has changed considerably over time. The following vowels are those reconstructed for the earliest stage of Hebrew, those attested by the Secunda, those of the various vocalization traditions (Tiberian and varieties of Babylonian and Palestinian), and those of the Samaritan tradition, with vowels absent in some traditions color-coded.

Sound changes
The following sections present the vowel changes that Biblical Hebrew underwent, in approximate chronological order.

Proto-Central-Semitic
Proto-Semitic is the ancestral language of all the Semitic languages, and in traditional reconstructions possessed 29 consonants; 6 monophthong vowels, consisting of three qualities and two lengths, i.e., where the long vowels only occurred in open syllables; and two diphthongs. The stress system of Proto-Semitic is unknown but it is commonly described as being much like the system of Classical Latin or the modern pronunciation of Classical Arabic: If the penultimate (second-from-last) syllable is light (has a short vowel followed by a single consonant), stress goes on the antepenultimate (third-from-last), else on the penultimate.

Various changes, mostly in morphology, took place between Proto-Semitic and Proto-Central-Semitic, the language at the root of the Central Semitic languages. The phonemic system was inherited essentially unchanged, but the emphatic consonants may have changed their realization in Central Semitic from ejectives to pharyngealized consonants.

The morphology of Proto-Central-Semitic shows significant changes compared with Proto-Semitic, especially in its verbs, and is much like in Classical Arabic. Nouns in the singular were usually declined in three cases: /-u/ (nominative), /-a/ (accusative) or /-i/ (genitive). In some circumstances (but never in the construct state), nouns also took a final nasal after the case ending: nunation (final /-n/) occurred in some languages, mimation (final /-m/) in others. The original meaning of this marker is uncertain. In Classical Arabic, final /-n/ on nouns indicates indefiniteness, and disappears when the noun is preceded by a definite article or otherwise becomes definite in meaning. In other languages, final /-n/ may be present whenever a noun isn't in the construct state. Very Early Biblical Hebrew (pre-1500 BCE) had mimation of uncertain meaning in an occurrence of the word urušalemim (Jerusalem).

Broken plural forms in Arabic are declined like singulars, and often take singular agreement as well. Dual and "strong plural" forms use endings with a long vowel or diphthong, declined in only two cases: nominative and objective (combination accusative/genitive), with the objective form often becoming the default one after the loss of case endings. Both Hebrew and Arabic had a special form of nunation/mimation that co-occurred with the dual and masculine sound plural endings whenever the noun wasn't in the construct state. These endings were evidently felt as an inherent part of the ending, and as a result have survived to the present day. Examples are Arabic strong masculine plural -ūna (nominative), -īna (objective), and dual endings -āni (nominative), -ayni (objective); corresponding construct-state endings are -ū, -ī (strong masculine plural), -ā, -ay (dual). (The strong feminine endings in Classical Arabic are -ātu nominative, -āti objective, marked with a singular-style -n nunation in the indefinite state only.)

Hebrew has almost lost the broken plural (if it ever had it), and any vestigial forms that may remain have been extended with the strong plural endings. The dual and strong plural endings were likely much like the Arabic forms given above at one point, with only the objective-case forms ultimately surviving. For example, dual -ayim is probably from *-aymi with an extended mimation ending (cf. Arabic -ayni above), while dual construct -ē is from *-ay without mimation. Similarly, -īm < *-īma, -ōt < *-āti. (Note that expected plural construct state *-ī was replaced by dual -ē.)

Feminine nouns at this point ended in a suffix /-at-/ or /-t-/ and took normal case endings. (When this ending /-at-/ became final due to loss or non-presence of the case ending, both Hebrew and Arabic show a later shift /-at/ > /-ah/ > .)

Canaanite shift
Hebrew shows the Canaanite shift whereby often shifted to ; the conditions of this shift are disputed. This shift had occurred by the 14th century BCE, as demonstrated by its presence in the Amarna letters (c. 1365 BCE).

Proto-Hebrew
As a result of the Canaanite shift, the Proto-Hebrew vowel system is reconstructed as (and possibly rare ). Furthermore, stress at this point appears to have shifted so that it was consistently on the penultimate (next to last) syllable, and was still non-phonemic. The predominant final stress of Biblical Hebrew was a result of loss of final unstressed vowels and a shift away from remaining open syllables (see below).

Loss of final unstressed vowels
Final unstressed short vowels dropped out in most words, making it possible for long vowels to occur in closed syllables. This appears to have proceeded in two steps: Vowel lengthening in stressed, open syllables occurred between the two steps, with the result that short vowels at the beginning of a -VCV ending lengthened in nouns but not verbs. This is most noticeable with short /a/: e.g. *kataba "he wrote" > but *dabara "word (acc.)" >.
 * 1) Final short mood, etc. markers dropped in verbal forms.
 * 2) Final short case markers dropped in nominal forms.

The dropping of final short vowels in verb forms tended to erase mood distinctions, but also some gender distinctions; however, unexpected vowel lengthening occurred in many situations to preserve the distinctions. For example, in the suffix conjugation, first-singular *-tu appears to have been remade into *-tī already by Proto-Hebrew on the basis of possessive -ī (likewise first singular personal pronoun *ʔana became *ʔanī).

Similarly, in the second-singular, inherited *-ta -ti competed with lengthened *-tā -tī for masculine and feminine forms. The expected result would be -t or -tā for masculine, -t or -tī for feminine, and in fact both variants of both forms are found in the Bible (with -h marking the long -ā and -y marking the long -ī). The situation appears to have been quite fluid for several centuries, with -t and -tā/tī forms found in competition both in writing and in speech (cf. the Secunda (Hexapla) of Origen, which records both pronunciations, although quite often in disagreement with the written form as passed down to us). Ultimately, writing stabilized on the shorter -t for both genders, while speech choose feminine -t but masculine tā. This is the reason for the unexpected qamatz vowel written under the final letter of such words.

The exact same process affected possessive *-ka ("your (masc. sing.)") and *-ki ("your (fem. sing.)"), and personal pronouns *ʔanta, *ʔanti, with the same split into shorter and longer forms and the same ultimate resolution.

Short vowel lengthening (esp. pretonic), lowering
The short vowels tended to lengthen in various positions. In the process of lengthening, the high vowels were lowered. In the Secunda, the lengthened reflexes of are ; when kept short they generally have reflexes.
 * First, short vowels lengthened in an open syllable in pretonic position (i.e. directly before the stressed syllable).
 * Later, short vowels lengthened in stressed open syllables.

Reduction of short open stressed syllables
Stressed open syllables with a short vowel (i.e. syllables consisting of a short vowel followed by a consonant and another vowel) had the vowel reduced to and the stressed moved one syllable later in the word (usually to the last syllable of the word). Stress was originally penultimate and loss of final short vowels made many words have final stress. However, words whose final syllable had a long vowel or ended with a consonant were unaffected and still had penultimate stress at this point. This change did not happen in pausal position, where the penultimate stress is preserved, and vowel lengthening rather than reduction occurs.

The previous three changes occurred in a complex, interlocking fashion:


 * 1) Shift of stress to be universally penultimate.
 * 2) Loss of final short vowels in verbs, pre-stress lengthening in open syllables. Pre-stress lengthening/lowering becomes a surface filter that remains as a rule in the language, automatically affected any new short vowels in open syllables as they appear (but ultra-short vowels are unaffected).
 * 3) Stress movement from light syllable to following heavy syllable when not in pausa, with newly unstressed light syllable reducing the schwa.
 * 4) Tonic lengthening/lowering in open syllables.
 * 5) Loss of final short vowels in nouns.

Examples:

Note that many, perhaps most, Hebrew words with a schwa directly before a final stress are due to this stress shift.

This sound change shifted many more originally penultimate-stressed words to have final stress. The above changes can be seen to divide words into a number of main classes based on stress and syllable properties:


 * 1) Proto-Hebrew words with an open penult and short-vowel ending: Become final-stressed (e.g. "he killed" <  PHeb. ).
 * 2) Proto-Hebrew words with an closed penult and short-vowel ending: Become penultimate due to segholate rule (e.g. "king" < ).
 * 3) Proto-Hebrew words with an open short penult and longer ending: Become final-stressed due to sress shift (e.g. "they killed" <  PHeb. ).
 * 4) Proto-Hebrew words with a closed penult and longer ending: Remain penultimate (e.g. "I killed" <  PHeb. ).
 * 5) Proto-Hebrew words with an open long penult and longer ending: ???

Pre-stress reduction of short vowel
were reduced to in the second syllable before the stress, and occasionally reduced rather than lengthened in pretonic position, especially when initial (e.g. σεμω = שמו  "his name"). Thus the vowel system of the Secunda was.

Later developments
The later Jewish traditions (Tiberian, Babylonian, Palestinian) show similar vowel developments. By the Tiberian time, all short vowels in stressed syllables and open pretonic lengthened, making vowel length allophonic. Vowels in open or stressed syllables had allophonic length (e.g. /a/ in יְרַחֵם  "he will have mercy" < previously short  <  by Tiberian degemination of  < PSem ). The Babylonian and Palestinian vocalizations systems also do not mark vowel length. In the Tiberian and Babylonian systems, and lengthened  become the back vowel. In unaccented closed syllables, become  (Tiberian),  (Babylonian), or  (Palestinian) – generally becoming the second vowel before geminates (e.g. לִבִּי) and the first otherwise. In the Tiberian tradition pretonic vowels are reduced more commonly than in the Secunda. It does not occur for, but is occasional for (e.g. מסמְרים  'nails' < ), and is common for  (e.g. רְחוֹב  'open place' < ). In Tiberian Hebrew pretonic is most commonly preserved by geminating the following consonant, e.g. אדֻמּים  'red (pl.)' (cf.  'red (sg.)'); this pretonic gemination is also found in some forms with other vowels like אַסִּיר~אָסִיר  'prisoner'.

The Babylonian and Palestinian systems have only one reduced vowel phoneme like the Secunda, though in Palestinian Hebrew it developed the pronunciation. However the Tiberian tradition possesses three reduced vowels of which  has questionable phonemicity. under a non-guttural letter was pronounced as an ultrashort copy of the following vowel before a guttural, e.g. וּבָקְעָה, and as preceding , e.g. תְדֵמְּיוּ֫נִי , but  was always pronounced as  under gutturals, e.g. שָחֲחו, חֲיִי. When reduced, etymological become  under gutturals (e.g. אֲמרתם 'you (mp.) said" cf. אָמר 'he said'), and generally  under non-gutturals, but  >  (and rarely  > ) may still occur, especially after stops (or their spirantized counterparts) and  (e.g. דֳּמִי /dɔ̆ˈmi/).  Samaritan and Qumran Hebrew have full vowels in place of the reduced vowels of Tiberian Hebrew.

Samaritan Hebrew also does not reflect etymological vowel length; however the elision of guttural consonants has created new phonemic vowel length, e.g. רב 'great' vs. רחב 'wide'. Samaritan Hebrew vowels are allophonically lengthened (to a lesser degree) in open syllables, e.g. המצרי, היא , though this is less strong in post-tonic vowels. Pretonic gemination is also found in Samaritan Hebrew, but not always in the same locations as in Tiberian Hebrew, e.g. גמלים TH SH ; שלמים TH  SH. While Proto-Hebrew long vowels usually retain their vowel quality in the later traditions of Hebrew, in Samaritan Hebrew  may have reflex  in closed stressed syllables, e.g. דין /,  may become either  or, and  >. The reduced vowels of the other traditions appear as full vowels, though there may be evidence that Samaritan Hebrew once had similar vowel reduction. Samaritan results from the neutralization of the distinction between  and  in closed post-tonic syllables, e.g.  בית 'house'  הבית 'the house'  גר  הגר.

Various more specific conditioned shifts of vowel quality have also occurred. Diphthongs were frequently monopthongized, but the scope and results of this shift varied among dialects. In particular, the Samaria ostraca show <  <  for Southern  'wine', and Samaritan Hebrew shows instead the shift  >. Original tended to shift to  (e.g. אֹמֶר and אִמְרָה 'word'; חוץ 'outside' and חיצון 'outer') beginning in the second half of the second millennium BC. This was carried through completely in Samaritan Hebrew but met more resistance in other traditions such as the Babylonian and Qumran traditions. Philippi's law is the process by which original in closed stressed syllables shifts to  (e,g,  > בַּת  'daughter'), or sometimes in the Tiberian tradition  (e.g.  > אֱמֶת  'truth'). This is absent in the transcriptions of the Secunda, but there is evidence that the law's onset predates the Secunda. In the Samaritan tradition Philippi's law is applied consistently, e.g. >  'heart'. In some traditions the short vowel tended to shift to  in unstressed closed syllables: this is known as the law of attenuation. It is common in the Tiberian tradition, e.g. > Tiberian שִבְעָה  'seven', but exceptions are frequent. It is less common in the Babylonian vocalization, e.g. 'seven', and differences in Greek and Latin transcriptions demonstrate that it began quite late. Attenuation generally did not occur before, e.g. Tiberian מַפְתֵּחַ 'key' versus מִפְתַּח  'opening (construct)', and often was blocked before a geminate, e.g. מתנה 'gift'. Attenuation is rarely present in Samaritan Hebrew, e.g. מקדש. In the Tiberian tradition take offglide  before. This is absent in the Secunda and in Samaritan Hebrew but present in the transcriptions of Jerome. In the Tiberian tradition an ultrashort echo vowel is sometimes added to clusters where the first element is a guttural, e.g. יַאֲזִין 'he will listen' פָּעֳלוֹ  'his work' but יַאְדִּיר  'he will make glorious' רָחְבּוֹ  'its breadth'.

The following charts summarize the most common reflexes of the Proto-Semitic vowels in the various stages of Hebrew:
 * 1) Samaritan vowels may be lengthened in the presence of etymological guttural consonants. results from both  and  in closed post-tonic syllables.
 * 2) under the conditions of the law of attenuation
 * 3) under the conditions of Phillipi's law
 * 4) Samaritan are nearly in complementary distribution ( in open syllables,  in closed syllables)
 * 5) lengthening occurs in some open pretonic syllables and some stressed syllables; precise conditions depend on the vowel and on the tradition
 * 6) reduction occurs in the open syllables two syllables away from the stress and sometimes also in pretonic and stressed open syllables
 * 7) effectively in most closed syllables
 * 8) more common before geminate consonants
 * 9) Samaritan Hebrew has full vowels when the other traditions have reduced vowels, but these do not always correlate with their Proto-Hebrew ancestors

Stress
Proto-Hebrew generally had penultimate stress. The ultimate stress of later traditions of Hebrew usually resulted from the loss of final vowels in many words, preserving the location of proto-Semitic stress. Tiberian Hebrew has phonemic stress, e.g. בָּנוּ֫ 'they built' vs. בָּ֫נוּ  'in us'; stress is most commonly ultimate, less commonly penultimate, and antipenultimate stress exists marginally, e.g. הָאֹ֫הֱלָה  'into the tent'. There does not seem to be evidence for stress in the Secunda varying from that of the Tiberian tradition. Despite sharing the loss of final vowels with Tiberian Hebrew, Samaritan Hebrew has generally not preserved Proto-Semitic stress, and has predominantly penultimate stress, with occasional ultimate stress. There is evidence that Qumran Hebrew had a similar stress pattern to Samaritan Hebrew.