Phonetics of Ancient Hebrew

Alexander Liss


The Task *

Slow Speech *

Specifics of Ancient Hebrew Notation *

Block, Emphasis and Shewa *

Assimilation, Exchange and End Forms *

Notation *


We rely in this work on methodology, which we worked out in the "Written Languages of Arts". One can find it in the same site.

The Task

Here we attempt the restoration of phonetics of the ancient Hebrew speech. This work is interesting by itself, because it allows to get a glimpse at ancient poetry. Also, it can be used as a tool of creation of new poetry and new songs.

We have a rich material for such work - ancient religious texts are well preserved through generations, and a tradition of reading them gives us a firm foundation for such work.

We make a few assumptions in our work, which we spell out here clearly.

First, we assume that notation of ancient texts closely reflects pronunciation at that time. In other words, we assume that in the time of writing the language was not developed to the degree where the speech and written language went there separate ways of development.

We know from ancient texts that there were a few dialects in use. Hence, writing should reflect the dominant form of pronunciation. From the description in ancient texts of the differences between dialects, we learn specifics of the pronunciation.

Second, we assume that the ancient speech was slow.

We know that, for example, old English pronunciation was slower than current one. It affected phonetics. We can see the difference between old and current phonetics in the difference between written and oral forms of the English language. It is reasonable to assume that ancient Hebrew was adjusted to slow speech.

This is an important assumption, because slow speech can use tools unavailable to fast one. We will explore these tools in a moment.

Third, there are specifics of ancient grammar - the way the text was written, the way prefixes and infixes are affected (transformed) by preceding or following phonemes, etc., which we assume all are caused by the natural transformation of sounds, which follow each other. In other words, we assume that these specifics of grammar can be explained by specifics of phonetics only.

Phonetic system, which is acceptable as a foundation of the language used in everyday life, has to be stable. It is remarkable that with these obvious assumptions we arrived to the stable phonetic system, which does not have many variations. This is not a proof that we restored ancient phonetic system (no one can provide such proof with the limited amount of information we have), but it is a hope that we are close.

It is still quite possible that what is written in ancient books reflects special, ritual pronunciation and singing. However, even in this case the restoration of this cultural layer of ancient life is interesting.


Slow Speech

Slow speech can use elements of phonetic structure, which are rarely used currently, when we only have examples of fast speech.

Fast speech is structured by the

Slow speech cannot use these tools; it uses its own structuring tools.

Its rhythm is slow, related to the natural rhythm of breathing, not heartbeat. The moments of air intake should be incorporated naturally in such speech as breaks between phrases, for example. The excess of air, which has to be exhaled before an air intake in such break, could be used for some energetic emphasis in the end of a phrase - a special ending intonation.

Slow speech uses special markers - hard and short breaks of the stream of air. These breaks can be produced by lips, a tongue, a throat and muscles closing nasal cavity. This kind of interrupter was denoted in writing in ancient Hebrew with a "light" dagesh.

The rest of phonetic states, which interrupt the stream of air are either "open" - they allow the continuing exhaling of air, albeit constrained, or they present a soft brief break of the stream.

This second type of interrupter is rarely used in fast speech - this dynamic interrupter needs a relatively long period to be executed. Some remnant of it in fast speech is denoted in writing with double consonant.

However, it was widely used in ancient Hebrew. It was denoted in writing with a "heavy" dagesh. This kind of interrupt requires a vowel before and after it.

Slow speech allows deployment of a throat.

In fast speech, sometimes we find deployment of a glottal stop.

There are other tools of throat also - constraining the stream of air with or without vocal support and producing the vibration of the stream, similar to vibration produced by a tongue (consonant r in some variants). Ancient Hebrew deployed all four tools.

The movement of muscles of throat is slow, and these tools of language are lost, when speech speeds up.

The other specific type of an interrupt is produced by very back of a tongue. It is slow and cannot be used in fast speech, but it was used in ancient Hebrew - kof.


Specifics of Ancient Hebrew Notation

We analyze vowels first.

In our phonetic presentation, a vowel consists of two transitions-"graces" with the special state between them. This state has to constrain the stream of air in a minimal way and has to be supported by vocal cord. This is vague definition, but it is sufficient for our goals here.

Usually, in Hebrew notation, vowels are denoted with nikudot - special signs above and below letters. Some of them carry modifiers, which denote a short form of a vowel. The distinction between a short and a long form of a vowel can be easily picked up in slow speech; it is a valid tool of slow speech. In contrast, in fast speech, what was a long vowel became louder vowel or a vowel, where the tone is changing.

Usually, in Hebrew notation, each letter, except the last in a word, has accompanying nikudot, which correspond to vowels. Only special vowels formed with lips (o and u) are denoted with a special letter instead of nikudot. Sometimes, however, there are letters, which do not carry nikudot and they are not followed by these special vowels. Only a few letters can have such distinction. Ancient Hebrew writing is very compressed, nikudot are used only in small set of texts used for special purposes. It is unlikely, that these special letters are used to distinguish similar sounding words, most likely that the words, with these letters present, sounded differently from similar words, without these letters in them. Hence, these special letters were used together with nikudot to denote special vowels.

This special letters jod, he and alef should represent "modifiers" of vowel, constraining the stream of air with a tongue, a throat and vibration of a throat correspondingly.

Letter he in the end of a word is usually a modifier of a vowel, in this case absence of nikudot with it is properly interpreted; but sometimes it should denote an interrupt, in this case it is supplied with mappik.

The collision of the speech elements produced with a throat and specially "modified" vowels manifests itself in some words, which end with the throat interrupter, which is denoted with letters ain (glottal stop), het (without vocal support) and he with mappik (with vocal support). They cannot follow a vowel, which is a variant modified with tong or formed with lips (o and u). To make such combination possible in some form, the language consistently inserts between them an interrupt, which corresponds to the way the vowel constraints the stream (for tongue it is j for lips it is w) with following vowel a. In writing this trick of language is presented as nikudot patah associated with the last letter.


Block, Emphasis and Shewa

The stream of slow speech is naturally divided on blocks by

One such block can include a few vowels. In ancient Hebrew one such block usually contained not more than one long vowel.

Basic state allows flow of air, it produces a soft boundary. States-breaks and graces without vocal support, which correspond to a quick change of states where air stream is constrained, produce a hard boundary.

In ancient Hebrew, ware two types of emphasis - primary, closely related to the semantic division of a stream of speech, and secondary, mostly phonetic.

Vowels, which are additionally constrained, cannot precede a hard boundary, unless they get a boost from a primary emphasis.

Similarly, a secondary emphasis cannot precede a hard boundary.

Shewa is a hint of a vowel. It is vocalized and short. It cannot carry emphasis.

In the Hebrew notation graces without vocal support inside the word are marked with the special symbol, which is the same as a symbol of shewa.

The fact that the same symbol is used to denote two different phenomena, when it is so easy to introduce additional nikudot, means that the logic of their differentiation is simple and clear.

For example:

shewa cannot precede hard boundary,

first grace of a block cannot be voiceless.


Assimilation, Exchange and End Forms

From the semantic analysis we can find in interesting cases, when instead of states, called by the language patterns, we find other states in the word, caused by phonetics. One such case when two states kind of fuse together, and the other when special infix tav (in mitpael form of verbs) appears in a "wrong" place modified and sometimes fused with previous state. This gave us insight into states tav, dalet, tet, sameh, tsade, zain, and shin.

It is unlikely that a few letters, which have "end forms", have these forms only for semantic reasons. Most likely they were phonetically deferent, but we do not know where to place them.



In our phonetic notation we use symbols of a computer keyboard - Latin letters and special symbols. However we follow the classification of states, which is present in ancient Hebrew notation.

Potentially, states-interrupters of the same class can have three forms

We denote


There are four groups of triads of states open-closed-dynamic (12).

v - \v - /v (bet)

all thee states are made with lips only, with voice support: v sounds as v and \v sounds as b


f - \f - /f (pe)

counterpart to bet without voice support, also made with lips only: f sounds as f and \f sounds as p


g - \g - /g (gimel)

made with tongue, with voice support: \g sounds as g and g as a soft, open form of it

q - \q - /q (kaf)

counterpart of gimel made with tongue, without voice support: \q sounds as k and q as a soft, open form of it

This group is complemented by two pairs of open - dynamic forms:

m - /m (mem)

as m

k - /k (kof)

state similar to k, but produced by the very back of a tongue (cannot be used in fast speech)


The triad

y - \y - /y (tav)

is produced with teeth and the tip of a tongue: y sounds as in three and \y sounds similar to t, but produced with tongue-to-teeth

This triad is in a close relationship to the following group of states (18)


n - /n (nun)

n sounds as n

l - /l (lamed)

l sounds as l



t - ~t (tet)

t sounds as t, made with the tip of a tongue, and ~t is its more constrained form made with the other part of a tongue, closer to middle

d - ~d (dalet)

d sounds as d (vocalized), made with the tip of a tongue, and ~d is its more constrained form made with the other part of a tongue, closer to middle



x - /x (tsade)

x sounds similar to ts and /x sounds as "t-ts"

z - /z (zain)

z sounds as z (vocalized) and /z sounds as dz

A special pair of pairs of states shin, They are produced with a tongue, one pair with its front and the other with its middle (softer):

#s - /#s

#s sounds as sh and /#s sounds as tsh

*s - /*s

*s sounds as a soft sh and /*s sounds as a soft dsh

(reduced to sound s in current pronunciation)


s - /s (sameh)

s sounds as s

Special states (4):

j - ~j (jod)

j souns as j, ~j its more constrained form

w - ~w (wav)

w sounds as w, ~w its more constrained form


States produced by a throat (4):

' - ain - glottal stop

h - het - without vocal support

: - he - vocalized

; - alef - series of interrupts, vocalized


r - resh

it produces series of interrupts with a tongue, vocalized.


States leading to vowels are (19):

& - shewa, a hint of vowel

first group (8):

a - open a

o - produced with the back of a tongue o

&a - short a

&o - short o

#a - constrained by throat a

#o - constrained by throat o

~a - vibrating by throat a

~o - vibrating by throat o

second group (6):

e - made with throat e

~e - constrained with a middle of tongue e

#e - further constrained with a tongue e

&e - short e

i - with the middle of a tongue i

~i - constrained with a middle of tongue i


third group (4):

~o - made with lips o

*o - further constrained with lips o

~u - made with lips u

*u - further constrained with lips u

All together, we have here 66 states (with basic state), which produce a palette of about 4,000 graces.