British & Foreign Bible Society

2 May, 2002 British & Foreign Bible Society

Machine Assisted Translation Team
1
bsf
The Masoretes and the Punctuation of Biblical Hebrew
Biblical Hebrew was the first classical language to be systematically
punctuated. In comparison with modern languages the system was very
complex: there are more commonly used punctuation marks in Hebrew than
there are letters in the English alphabet, and they are applied not only
between phrases but to virtually every word of the scriptures, indicating very
precisely each words relationship with others in a phrase as well as dividing
each sentence into hierarchically ordered word-groups. This complexity,
coupled with sometimes substantial orthographic variations between printed
editions, has led many students of the language to neglect Hebrew
punctuation entirely.
The advent of computer typesetting, and in particular the inclusion of these
marks in the magnetic text of Biblia Hebraica Stuttgartensia, challenges
scholars in linguistics and computation to cooperate in a more detailed
analysis of this punctuation than has ever before been possible. But even
without further research it provides the opportunity to analyse the logical
structure of the Hebrew text by computer, and to apply the results to Bible
translation and to translation checking. This paper is a preliminary attempt to
stimulate such a dialogue, first by addressing in detail the historical
background which gave rise to the distinctively Hebrew system of
punctuation, and secondly by providing a brief account of the most common
elements of the system.
a. Historical Background
1. The development of punctuation in the West
In general, ancient languages were not punctuated. In some Egyptian poetic texts
line and stanza marks exist as aids to the reader, and some early Greek inscriptions
use a vertical ellipsis as a phrase or sentence divider, but the only punctuation
apparently known to Aristotle was the paragraphos, an underscore to indicate the
beginning of a new topic.
The invention of punctuation to mark the grammatical or logical structure of texts is
generally attributed to Aristophanes of Byzantium. As chief librarian of Alexandria
(c.195 BCE) Aristophanes was responsible for copying and annotating many works
from classical Greek literature. In doing so he added breathings and accents and
devised a simple three part system of punctuation by which a text was divided by
dots placed at the end of words: a short clause (Komma) had the dot in the middle, a
phrase (Klon) at the bottom and a sentence (Periodos) at the top. In spite of its
obvious advantages, this system was scarcely used after Aristophanes except to
clarify a few obscure poetic texts.
Latin inscriptions, unlike Greek ones, marked word divisions by a small dot. At the
beginning of the Common Era these points were also used in books and a paragraph
system akin to that of Greek came into being: a new subject was introduced by
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
2
bsf
starting a new line and changing the indentation, much as in modern European
languages (except that instead of being indented the line was extended into the left
margin as a hanging paragraph). This practice was particularly used in texts which
were composed to be read aloud in public (Cicero used it to mark natural breaks in
his Latin orations as had Demosthenes in Greek) and was comparatively short-lived.
In spite of occasional attempts to revive Aristophanes system for Latin (most notably
by Aelius Donatus who was Jeromes tutor), from the fourth century to the ninth the
only consistently punctuated extant Latin text (and that just by paragraphos) is
Jeromes Vulgate.
It was only under the influence of Charlemagne in the ninth century that punctuation
systems developed in Western Europe. By this time the churches were beginning to
use Neumes as an early musical notation for chanting and some of these, along with
Aristophanes Komma and Periodos, were gradually added to the punctuation
schema, notably the punctus elevatus (!) and the punctus interrogativus (?). These
had a dual function: they indicated both a change of pitch and a grammatical nuance.
Although this system was fully developed by the twelfth century it was largely
confined to chanting, and it was not until after the invention of printing that it was
consistently used. In the work of Caxton there is no very clear relationship between
punctuation and syntax; the consistent modern use of punctuation in the service of
clarity dates back only to the fifteenth century (probably to the Venetian printer Aldus
Manutius, d. 1515 or possibly to his grandson of the same name).
2. Punctuation in Hebrew
Masoretic Hebrew punctuation is similar to that based on the neumes in that it is very
closely related to the musical pitch used in incantation. It is different in two major
respects: on the one hand, it is far more comprehensive, and on the other, it became
a norm for all Jews who read their scriptures as soon as it was introduced. Both of
these differences were caused by political, religious, social and economic forces
which culminated in the appearance of the Masoretic Text in the tenth century.
2a. The Jewish Academies Mishna and Talmud
After the destruction of the Temple Galilee became the administrative centre of
Judaism. The Sanhedrin moved to the new city of Tiberias and the Jewish religious
academies (called yeshivot) moved from Jerusalem to Tiberias and Caesarea. These
academies served two purposes. They were centres for training rabbis in what were
to become known as the Miqra (the written scriptural Law) and the oral teachings of
the great scholars, the Mishna (repeated i.e. from memory). They were also institutes
of advanced scholarship where highly trained rabbis working under the direction of
an officially appointed scholar called the Nasi (literally prince) devoted themselves to
producing a unified and normative interpretation of the oral Law. The first fruits of this
study were the tractates of the Mishna which appeared in the third century CE under
the direction of Juda ha-Nasi. These were almost entirely produced in the academy
at Tiberias.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
3
bsf
Meanwhile, similar academies had been established in Sura and Nehardea in
Babylon. The Babylonian academies were placed under the direct control of the resh
galuta (head of exile or exiliarch), a fact which was later to have serious
repercussions. The quality of scholarship produced by the Tiberian and Babylonian
academies can scarcely be overstated. In them thousands of students were trained
rigorously for many years, and those who graduated and left the academies were
required to return twice a year for a month's refresher course called kalla
(completion of study). There was comparatively free movement, and therefore
exchange of information, between the academies.
Now that Mishna was regarded as virtually canonical, scholars from the academies in
both Palestine and Babylon undertook the task of expanding it to include all the main
teachings of the greatest Jewish scholars a work completed in the fifth century.
This expansion, the Gemara (another word for completion) took different forms in
Tiberias and Sura. Palestine fell under the control of Constantinople, and persecution
under successive Christian emperors severely disrupted Jewish learning there, but
Judaism continued to flourish in Babylon. It is probably for this reason that the
Palestinian gemara is much shorter than the Babylonian one (only about a third of its
length) and does not cover all the tractates of Mishna. The Mishna and the Gemara,
together with some further writings, form the Talmud (Learning). The Babylonian
recension is called the Talmud Bavli or Babylonian Talmud, as opposed to the
inappropriately named Talmud Yerushalmi or Jerusalem Talmud.
It should be stressed that, from a Rabbinic perspective, the oral Law of Mishna and
Gemara, although assembled in the academies over centuries, was no less
authoritative than the Miqra. Like the text of the Torah it had been given as a promise
by Moses. A Rabbinic exegesis of Ex. 24:12 says that the Tables are the Decalogue,
the Law is the Scripture, the Commandment is the Mishna, the that I have written is
the Prophets and Writings and the to teach them is the Gemara ...so telling us
that all of these were given to Moses from Sinai.
2b. From Talmud to Masoretic Text
The Torah or Pentateuch had long been the pivot of the Jewish scriptures and as
such had been transmitted with the utmost care. The remainder of the scriptures, the
prophets and writings, had not been so scrupulously preserved. It was not until the
synod of Jamnia in the last decade of the first century CE that final agreement was
reached on the canon of the Jewish scriptures, and by that time several different
Hebrew (and Aramaic) texts were in circulation. Although the academies had
attempted to produce an agreed text, it was not until a consensus of interpretation
was reached that the rabbis were able to decide on a definitive text of the whole of
scripture. The Talmud was the embodiment of this consensus.
But between the compilation of the Mishna and that of the Gemara the language of
the Jews in Palestine had changed. The Mishna is written in Hebrew: not the Hebrew
of the scriptures but a relatively similar language. It is sometimes asserted that
Hebrew had been replaced by Aramaic as a spoken language before the Common
Era. This is certainly not true: the Jewish revolutionary Bar Kokhba wrote letters to
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
4
bsf
his family and friends, some in Aramaic and some in Hebrew, between the years 132
and 135 CE. There is at least a possibility that even in the third century CE the
language of the Mishna was spoken by ordinary people and not just by Hebrew
scholars. But the Gemara is written in Aramaic. It is clear that by the fifth century both
Biblical and Mishnaic Hebrew were dead languages known only to scholars.
Of course, Hebrew continued to be taught. At a time when only a few highly educated
Europeans could read and write, the Jewish educational system was very advanced.
The education of male children was a religious obligation, and primary schools
existed in all towns and many villages. Boys started learning scripture by six. Several
went on to study Mishna by ten and the most able studied Talmud by fourteen. But of
every thousand boys who started school only one could be expected to become a
proficient scholar. If the scriptures were to survive, new types of help to readers
needed to be devised.
The earliest Hebrew manuscripts, in common with many ancient languages, had no
punctuation system (except for starting a new line to indicate a new topic) and
Hebrew had no vowels in its alphabet. Neither of these omissions was important as
long as Hebrew was a spoken language. Cases where a vowel could readily be
mistaken were to some extent catered for at an early period by inserting consonants
to help the reader: h for an a sound, y for e or i and w for u (although many
ambiguities remained). Full stops were inserted to divide the text into sentences or
verses, possibly as early as the second century BCE, and most verses were marked
to indicate the most important pause to make when reading aloud. Before the Talmud
was completed the rabbis had emended these for consistency and had added other
marks to clarify the text. A vertical bar | was written between words when it was felt
necessary to clarify where one ended and another began (inter-word spaces were
not regularly used), and stress markers were inserted. In Hebrew most words are
stressed on either the last or the last-but-one syllable (a very small number are
stressed on the last-but-two, and several longer words also have a secondary tone),
and it is very important to know where this stress is placed. Pronouncing the accent
on the wrong syllable can drastically change the meaning of a word. These tasks had
been completed and agreed upon by the sixth century. Unfortunately no manuscripts
from this period have survived, but we know from later writers that a Codex Hilleli,
ascribed to Rabbi Hillel ben Moses ben Hillel around the year 600 CE, and a Codex
Muga were regarded as authoritative. After some 500 years of scholarly labour a
definitive text of the Hebrew Bible had at last been produced.
This should have been the end of the Academies work. But the demise of Hebrew as
a spoken language provided a new challenge: without vowels in the text it was very
hard, perhaps impossible, for anyone other than a rabbi to understand the Scripture
or pronounce it correctly. A system of vocalisation had to be made and (since the text
itself was now regarded as definitive) it had to be inserted without making changes to
the letters in the text. To achieve this, the rabbis began to devise a system of dots to
represent the different vowels. The Hebrew word for a dot or spot is niqqud (it is the
word used to describe Josephs speckled cattle in Gen. 30:39). A related and very
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
5
bsf
similar word noqed means a shepherd, and we can imagine a typically rabbinic pun
that the dot is a shepherd to guide the reader.
Because the task of the rabbis from the sixth century onward was no longer to
establish a consensus but to clarify and preserve the traditions which had already
been agreed in the academies, they are usually referred to as Masoretes
(traditionists the English form can equally be written as Massoretes). Pointed texts
(texts with all the vowels represented by dots) first came into existence in Tiberias in
the sixth century, but this earliest form of pointing (sometimes called the proto-
Palestinian niqqud) was complex and took different forms in different academies.
During the sixth and seventh centuries the Babylonian rabbis developed a simpler
and more co-ordinated system, the Niqqud Bavli, with vowel points placed over the
consonants, while those at Tiberias began to produce a different system which
placed the vowels predominantly under the consonants. This Tiberian Niqqud
eventually came to dominate and is the basis of what we now call the Masoretic
pointing. By the eighth or ninth century the aims of Rabbis and Masoretes had
apparently been achieved: there was full consensus among Rabbinic scholars on the
text of all the Scriptures and its correct interpretation
The ninth century, however, saw the rise of several threats to the traditions of
Judaism, threats which substantially changed the perceived role of the Masoretes.
Although at this time Classical Greek literature was unknown in Europe (with perhaps
the sole exception of the Irish monasteries) it had long been kept alive under the
Caliphate. By the ninth century it was being circulated in Arabic translation. Jewish
communities were becoming increasingly urbanised and Arabic was an everyday
language of trade in towns and cities. Tiberias itself, which had fallen under
Byzantine control, was taken by the Arabs in 636. Greek philosophical ideas, in
particular Aristotelian, Platonic and neo-Platonist philosophy, became increasingly
popular and in many cases eroded the ancient Hebrew traditions. Many Jews were
attracted to Gnosticism, and works such as the Sefer Yeira (the Book of Creation
which claimed to be written by Abraham himself) came to be widely circulated,
retelling the Biblical creation accounts in neo-Platonist or Gnostic terms. Among
many of the new urban elite, Aristotelean ethics challenged traditional Jewish beliefs.
A Persian Jew, Hivi al-Balkhi, who had converted to Gnosticism even produced an
expurgated text of Scripture from which, as from Bowdler's Shakespeare, all
offensive material had been removed!
A very different but equally problematic movement from the Rabbinic point of view
was the fast growing Karaite sect. The office of resh galuta (exiliarch) in Babylon had
developed into a hereditary one, but succession to it depended upon the assent of
the Caliph. In the second half of the eighth century the ruling Caliph refused to
confirm the succession of Anan ben David, choosing instead his younger brother.
Anans reaction was to set up a new religion which, he claimed, was a compromise
between Judaism and Islam. His Sefer haMivot (Book of Precepts) written in about
770 asserted the full authority of the Hebrew Bible, but rejected both the Talmud and
the authority of the Rabbinate in the interpretation of scripture, which was held to be
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
6
bsf
self-evident to all readers. In stressing this principle his followers, at first called
Ananites, adopted the name Karaites (properly Qaraites from the Hebrew verb to
read). These Bible readers coined the term Rabbinites for those who followed the
authority of the Talmud. By the mid ninth century the Karaites, under the leadership
of Daniel al-Qumisi, had spread from Babylon to Palestine and attracted a
considerable following there.
Given the very real possibility that large numbers of Jews would read and interpret
the Bible without reference to Rabbinic orthodoxy, one final task remained for the
Rabbinite Masoretes: to restrict the number of ways in which the Bible could be
interpreted by imposing upon it a system of punctuation which was so precise as to
be, in almost all cases, totally unambiguous.
The method they adopted was an ingenious blend of ancient tradition with new
technology. We have noted that major divisions within verses had been established
for at least a thousand years and in general these were followed by the Masoretes,
although some were revised in the interest of greater consistence. A musical tradition
for intoning the scriptures had also been established, although it is not known how
long this had existed in a stable form. In following these two systems the Masoretes
were indeed traditionists. Their innovation was to complete the punctuation of
verses in a rigorously mathematical way.
Although the main task of the Rabbis had always been the preservation and
interpretation of scripture, many of them were also noted for their learning in other
fields. This was particularly true of the Babylonian rabbis. As early as the third
century CE Rabbi Samuel of Nehardea, who contributed substantially to the
Babylonian Talmud, was also renowned for his knowledge of astronomy, medicine
and civil law. He also introduced a rule which was to change the nature of rabbinic
scholarship: that Jews in exile were subject to the civil laws of the state in which they
lived.
By the end of the seventh century the highly advanced mathematics of India had
spread to northern Mesopotamia where it was eagerly embraced by Islamic scholars.
In the early ninth century the House of Wisdom was built in Baghdad, near
Nehardea. This was on the one hand a centre for translation and preservation of
ancient texts (particularly the major Greek mathematical writings) but it was also here
that Islamic mathematics flowered. Perhaps its most influential scholar was al-
Khwarizmi whose name gave us the word algorithm and one of whose works coined
the word algebra (al-jabr, restoring). Among many innovations he developed the
new mathematics of decimal fractions and binary division to calculate the time when
the new moon would be first visible and the direction of Mecca, thus binding Islamic
religion and science inseparably. More significantly for our present purposes, al-
Khwarizmi also established mathematical models for the very complex Muslim rules
of inheritance, which were also binding on Jewish exiles in Islamic territory. This
applied to all the Jewish academies since even Tiberias was now under Islamic
control. Mathematics, to some degree at least, was now an essential tool for both
Jewish and Muslim religious scholars.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
7
bsf
There is no direct evidence that the Masoretes were trained in mathematics, but the
probability seems substantial. The binary division and subdivision of the text of
scripture rapidly became the focus of their punctuation, which combined traditional
musical notation with a consistent binary analysis of the constituents of every verse
of scripture. This task was brought to its spectacular conclusion in two different forms
at Tiberias by Rabbi ben Naftali and Rabbi ben Asher in the first half of the tenth
century. Modern printed texts follow almost exclusively the text of ben Asher.
It is a measure of the rigour of these rabbis work that following the production of the
Masoretic Text all manuscripts copied it exactly, and although the first printed
Hebrew Bible was not produced until 500 years later it is extraordinarily faithful to the
earliest extant manuscripts. In all significant respects the Bible which ben Asher and
ben Naftali completed has remained the only accepted text of the Hebrew Bible for a
thousand years.
b. Hebrew Orthography and the Punctuation System
1. Pre-Masoretic signs
The Hebrew alphabet consists of 23 consonantal signs, five of which take a different
form when they are the last letter of a word:
N 2 . ` * l C ` 2 7 C . C V 3 2 2 !
b g d h v z h t y k l m n s p q r s t
Final: _ C ] ]
Verses are divided by a period or Sof Pasuq :
Words may be joined by a hyphen or Maqqef so, for example, the last three words
of Gen 1.2 are :C`C `.7V...
Several words may be joined in this way: in Gen 25.5 we read `7_2_N72!_N: when
words are joined by Maqqef they effectively become a single word, and all except the
last one lose their accents (but they may retain a secondary tone).
The following table gives the position of these characters in the font SIL Ezra (two
digit hex) and Unicode Standard 3.0 (four digit hex):
N 27 05D0 ` 77 05D5 2 6B 05DB V 76 05E2 2 57 05E9+ ] A4 05DF
2 62 05D1 * 7A 05D6 7 6C 05DC 70 05E4 2 48 05E9+ A5 05E3
. 67 05D2 l 78 05D7 C 6D 05DE 3 63 05E6 ! 74 05EA ] A6 05E5
64 05D3 C 58 05D8 . 6E 05E0 71 05E7 _ A2 05DA : 3A 05C3
68 05D4 ` 79 05D9 C 73 05E1 72 05E8 C A3 05DD 2D 05BE
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
8
bsf
In Unicode the character 2 is a composite of 05E9 + 05C2 and 2 is 05E9 + 05C1.
The neutral form 2 (53 05E9) can also occur as in the name 222` Issachar (Gen
30.18 etc).
One character, the vertical bar | (7C 05C0) serves for both a pre-Masoretic and a
Masoretic function. As a pre-Masoretic sign it is called Paseq and served many
different functions. The most common are:
a. to prevent ambiguity in word division
b. to separate identical letters at the end and beginning of adjacent words
c. to separate words which are identical or very similar
d. to separate words which are contradictory in meaning (e.g. God and sinner)
e. to mark abbreviations.
Its use in the Masoretic accent system will be noted below.
2. Masoretic points and accents
2a. Vocalisation: Hebrew Vowels
The Niqqud dots which the Tiberian Masoretes used to mark vowels are generally
written below the consonant they follow. In what follows they will be placed relative to
the letter C. It should be noted that there are several different ways of transliterating
their names.
_C _C_C _C
Patah (05B7) is a short a. Like most of the vowels it occupies several font positions
to enable it to appear correctly beneath characters with different widths: 61, C2, D1
and DE.
C CC C
Qame (05B8, 41, 6F, C3, C4, D2, D3, DF, E0) is, depending on its context, either a
long a or a short o.
_C _C_C _C
Segol (05B6, 65, C6, D5, E5) is a short e.
_C _C_C _C
ere (05B5, 45, C5, D4, E3) is a long e.
C CC C
Hireq (05B4, 69, C7, D6, E6) is a short i. (It is long if followed by `)
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
9
bsf
C CC C
Holem (05B9, 4F, C8, D7) is a form of long o. It is the only vowel to be written at the
top of a consonant.
C CC C
Qibbu (05BB, 75, C9, D8, E7) is a short u.
C CC C
ewa (05B0, CC, CD, D9, DA, E8, ED, F8, FC) is, depending on its context, a very
light indeterminate vowel or a silent syllable divider.
Other forms of the sounded ewa are:
_N N N N _ N N N N N NN N
These cannot stand under C so here they appear under N. From right to left these are
Hatef Patah (05B2, CF, DC, E1, F0), Hatef Qame (05B3, DD, EC, F1, F3) and
Hatef Segol (05B1, CE, DB, E9, EF).
In SIL Ezra (but not in Unicode) two vowels are made as composites of u and a
niqqud:
` `` `
Holem Vav (F4) is a long o.
` `` `
ureq (FB) is a long u.
The following composite characters are also present in the font:
_ (A7) ] (A8) (A9) _ (AA) _ (AC) _ (AD) (CB) [ (E4) (EB)
(F6)
2b. Other non-punctuating Masoretic signs
Most letters of the Hebrew alphabet could be modified by inserting a dot called dage
in the middle as in 2. This dot had three functions according to its context: in the
letters !2.2 it hardened the sound of the letter: without dage the letter was
aspirated, with dage it was not. When used in this way it is called dage lene. In
other letters except it made the sound more forceful and drawn out, almost
doubling the consonant. This is called dage forte. In (where it is called mappiq) it
tells the reader to give full pronunciation to the h sound rather than just lengthening
the vowel before it. The coding of dage is 05BC, 43, 46, 47, 4A, 4B, 4C, B7, D0.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
10
bsf
Mappiq is also occasionally used with N. When it is used it is positioned in the base
of the letter as N. In Biblia Hebraica Stuttgartensia this occurs only four times: at Gen
43.26, Lev 23.17, Job 33.21 and Ezr 8.18. The font code for this usage is 49: it does
not appear to be included in the Unicode 3.0 specifications.
Rafe is a small bar written above a consonant as 2. This sign was used extensively
by the Masoretes as a check to show that dage lene had been omitted deliberately,
not by scribal error. It is rarely found in printed editions. Its codes are 05BF, 26.
The Masora Circle C (05AF, 7E) is the equivalent of a footnote marker, drawing the
readers attention to a variant reading or other note.
Meteg C (05BD, 95, BD, BE, BF, C0, C1) is used to mark the secondary tone,
reminding the reader to give its vowel full pronunciation. In BHS meteg is also used
sometimes to help the reader distinguish a short o from a long a. An identical mark is
used for the punctuation character Silluq. In fact several of the accents do double
duty in this way, but there is always a clear algorithm to distinguish their use.
2c. Masoretic punctuation signs
There are two distinct systems of punctuation in the Hebrew Bible. One is used only
in the books of Job, Psalms and Proverbs. This system is sometimes referred to as
Poetic punctuation, but the term is misleading insofar as this system is not used
elsewhere in poetry. A better title is the accents of the three books. The initial letters
of Job, Proverbs and Psalms in Hebrew spell the word !CN (emet truth), so the
system is also known as the accents of emet. Although the rest of the Hebrew Bible
consists of thirty-six books in most Christian bibles (more-or-less following the
Septuagint), it makes only twenty-one books in Hebrew: Samuel, Kings and
Chronicles count as one book each, Ezra and Nehemiah form a single book, and the
twelve prophets Hosea to Malachi are also counted as one book. So the other
punctuation system is referred to as the accents of the twenty-one books. The
punctuation system of the three is very much more complex than that of the twenty-
one and is, with one exception, not covered by this paper.
The rabbis divided the punctuation marks of each system into two distinct groups: the
melakhim (kings) and the meshartim (servants or ministers). The Christian
grammarians gave these the Latin names Domini and Servi. The Kings are
disjunctive accents which divide a sentence into phrases and the Servants are
conjunctive: they mark words which are components of phrases to show that they are
connected with each other. Some of these marks are written in various forms in
different editions of the Hebrew text. For simplicity we shall here use the same forms
as Biblia Hebraica Stuttgartensia.
2c1. Disjunctive accents: the melakhim
Following the work of Bohlius in the Lutheran University at Rostock (Scrutinium
Sanct. Scr. ex accentibus, 1636) it became fashionable among Christian hebraists to
divide the kings into a hierarchically structured ruling class consisting of
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
11
bsf
Imperatores, Reges, Duces, Comites etc. Recent years have seen a resurgence of
this approach among both Christian and Jewish scholars. While this system is very
graphic and is a very useful teaching aid, it does not convey what actually happens in
the text: Gesenius-Kautzsch describes it as the source of manifold confusion. In
what follows the accents will be given their traditional Hebrew names.
Most of the punctuating accents serve three purposes: they indicate pitch and
punctuation and they also mark the tone syllable of the word on which they stand. In
some cases this last function is absent. A small number of accents are written after
the end of the word on which they stand: these are called postpositive. Some others,
called prepositive, are written before the beginning of the word.
The term punctuation is perhaps confusing when applied to the Masoretic text. The
system is in fact very similar to that used in marking constituent structure trees in
linguistic analysis and can be illustrated by the same form of tree-diagram as:
Sentence
Noun phrase Verb phrase
Determiner Noun Verb Prepositional phrase
Preposition Noun phrase
Determiner Noun
the cat sat on the mat
The categories of the Masoretic system (except for Sentence) are completely
different from those of phrase structure grammars, but the structure and many of the
mathematical rules for manipulating it are remarkably similar. The structure and
some of the rules are treated in a companion paper, Masoretic Hebrew Cantillation
and Constituent Structure Analysis.
The main melakhim accents are here listed in approximately the descending order of
their disjunctive strength, largely following Wickes
1
.
Silluq
Silluq is the strongest disjunctive accent, the equivalent of a modern full stop. It is
written as a vertical bar under the tone syllable of the last word in a sentence (Gen
1.1 : ]__N_ where N is the accented letter). In appearance it is exactly the same as
meteg. In the vast majority of cases, silluq is written under the word immediately

1
W. Wickes, Taamey 21 sefarim (A Treatise on the Accentuation of the twenty-one so-called Prose Books of the Old
Testament), Oxford 1887
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
12
bsf
before sof passuq (:) so it is usually redundant as a punctuation mark. But the
Masoretes made good use of it in a few cases where they disagreed with the
sentence divisions they had inherited from earlier rabbis. In Gen 35.22, for example,
the end of the verse is doubly accented. The earlier rabbis had not placed a sof
passuq between and Israel heard it and the sons of Jacob were twelve, although
the structure of the narrative clearly requires one it seems likely that this was a
rather delicate means of passing over an unpleasant subject by minimising its
emphasis. The Masoretes were not free to insert a sof passuq, and they obediently
pointed the text in the form they had received it, but also inserted silluq at the end of
and Israel heard it to indicate that there should have been a verse division at that
point. Similar emendations of the traditional verse structure are to be found in Ex
20.2ff and Deut 5.6ff. With these exceptions silluq is always the last accent on a
word. Any mark which appears before it is to be ignored for the purposes of
punctuation.
Ole veyored
This accent (strictly a pair of accents ole and yored ) is only used in the three books,
but is dealt with here because it has a disjunctive force second only to silluq. It is
written as two signs as in C`V2 (Ps 1.1), _`3l (Ps 1.2). The stroke yored V
indicates the beginning of the tone syllable. The accent is written on the last word
(underlined below in English) of the first logical half of the sentence:
Ps 1.2
ole veyored
In the law of the Lord is his delight he meditates day and night on his law
If the word carrying yored is accented on its first syllable, the ole is written on the
beginning of the final syllable of the preceding word `_.N 77CN (Ps 6.3).
The Unicode reference for ole is 05AB, and in SIL Ezra its codes are 86 and BA.
Yored shares 05A5, 42, 9C, F5 with merekha (see Conjunctive accents below).
Atnah
Atnah (also called Etnahta) is the main verse divider for the twenty-one books as ole
veyored is for the three. It is written under the first consonant of the tone syllable as
in C` 7N. It divides a verse into two logical halves as in Gen 1.1:
Atnah
In the beginning created God the heavens and the earth.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
13
bsf
but these logical halves may be quite different in length (Gen 1.7):
Atnah
And God made the firmament and separated and it was so.
the waters beneath it from the waters above it
In the three books Atnah can be the main divider in comparatively short verses (e.g.
Ps 1.4):
Atnah
Not so the wicked they are like chaff which the wind blows away
In verses divided by ole veyored, atnah is used, where necessary, to subdivide the
second half of the verse (e.g. Ps 6.3 English versions 6.2):
ole veyored
Take pity on me Lord, for I am wasting away Atnah
give me strength Lord my body is worn out
The Unicode position of Atnah is 0591, and in SIL Ezra it is A1, F2 and FD.
Segolta & Shalshelet
Segolta is used in the twenty-one books only as the primary subdivider of the first
half-verse (the section ending with Atnah). It is written above and after (to the left of)
the word which carries it: C`7N (it is post-positive), so unlike most punctuation
marks it does not give any indication of the tone syllable. Gen 1.28
Atnah
Segolta
And God blessed them and God said to them rule the fish in the sea and the birds
be fruitful and multiply in the air and everything which lives
fill the earth and control it and moves in the world
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
14
bsf
Shalshelet is a rare variant of Segolta: it is musically, but not grammatically distinct,
and occurs seven times where segolta would appear on the very first word of the
verse. It is made up from two signs, the vertical bar | used for Paseq (b1 above) and
the symbol . So the first word of Gen 19.16 is punctuated |CC!``. The verse is
divided as:
Atnah
Shalshelet
He hesitated so the men took him and his wife and they brought him out and put him
and two daughters by the hand because outside the city
the Lord had mercy on him
Atnah is said to dominate segolta or shalsheleth (the term is used by the rabbis as
well as in modern mathematical descriptions of constituent structures). In structural
terms one node dominates another if there is a path from the first to the second
which only follows the down-ward direction of the arrows.
Segolta is coded as 0592, 82 and B1. Shalshelet is 0593, 52 and 8E.
Zaqef
Zaqef has two forms, Little Zaqef and Great Zaqeph.
Little zaqef is written as a small colon above the beginning of the tone syllable:
C`7N. It may divide either or both halves of a verse (Gen 1.6):
Atnah
Little zaqef Little zaqef
And God said let a firmament come between let it be a divider of waters from waters
the waters
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
15
bsf
Unlike the punctuation dealt with so far, little zaqef can divide a phrase into more
than two sections. in such cases the first dominates the second (Gen 1.7):
Atnah
Segolta
Little zaqef
Little zaqef
And God made the firmament and separated the from the waters above it and it was so
waters beneath it
Here segolta dominates both zaqefs. It immediately dominates the first (is the first
node up from it) and the first zaqef immediately dominates the second.
Great zaqef, in spite of its name, is generally a less forceful disjunctive than little
zaqef (the names refer to musical, not grammatical value), and is frequently used to
mark a sub-phrase which only consists of a single word. It is written as little zaqef
followed immediately by a small bar: C`7N. Gen 1.14 is punctuated:
Atnah
Little zaqef Little zaqef
Great zaqef
And God said Let there to divide day from night and let them be signs for days and years
be lights in the firmament
of heaven
Little zaqef is coded as 0594, 81, AF and great zaqef as 0595, 44, 88. To simplify
tree diagrams we will from now on use the term Zaqef to mean Little zaqef and only
specify when it is Great zaqef.
Tifha
Tifha (also called Tarha) marks the disjunction immediately before silluq or atnah. In
the twenty-one books it is written below the beginning of the tone syllable: !`2N2.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
16
bsf
The full disjunctive pattern of Gen 1.1 is:
Atnah
Tifha Tifha
In the beginning God created the heavens and the earth
In short verses which do not need any strong disjunction Tifha is often the main
punctuation mark. An example is Gen 1.13
Tifha
And there was evening and there was morning day three
The same is sometimes true of longer verses where no strong disjunction is
appropriate, e.g. Gen 3.21:
Tifha
And the Lord God made clothing of skin for Adam and his wife and clothed them
Tifha is coded as 0596, 59, 9B, F7.
Revia
Revia is a single dot placed over the beginning of the tone syllable as in ]N`. It
can be distinguished from the vowel holem as being higher and larger. It is often
printed as a small diamond while holem is circular. Gen 2.17:
Atnah
Zaqef
Tifha Tifha
Revia Revia
And from the tree of knowledge of you will from it because on the day you you will surely die
good and evil not eat eat from it
Revia is coded as 0597, 80, AE, B2.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
17
bsf
Zarqa
Zarqa is postpositive, written above and to the left of the final letter of the word as in
C`7N. Under zaqef above we have given a partial analysis of Gen 1.7. Here is a
fuller (but still incomplete) one:
Atnah
Segolta
Zaqef
Zaqef
Revia
Tifha
Zarqa
And God made the firmament and divided the waters under from the which above the and it was so
the firmament waters firmament
Zarqa is coded as 0598, 51, 56, 8D, 90.
Pashta & Yetiv
Pashta is a postpositive accent which stands over the last consonant of its word:
`N7. Unlike other postpositive accents it does give an indication of where the tone
syllable occurs, because if the tone is not on the last syllable pashta is duplicated
over the tone syllable as in ``!. Since it is only written once on `N7 it follows that
this word must be accented on the final syllable. A musically distinct but
grammatically equivalent variant of pashta is yetiv which is prepositive: 22V. Gen
2.23 contains an example of both pashta and yetiv:
Atnah
Segolta
Zaqef Zaqef
Revia Tifha Tifha
Zarqa Yetiv Pashta
And said the man this at bone from my and flesh from my she will be because she was taken
last bones flesh called woman from man
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
18
bsf
Pashta is coded as 0599, 3D, 66, 89, 8F. Yetiv is 059A, 60, 98.
In modern mathematics (but not in the rabbis) it is customary to refer to a node as the
daughter of the node which immediately dominates it (which is then called its
mother). So in the above tree and said is a daughter of zarqa, which is in turn a
daughter of segolta. Since and said and the man are both daughters of zarqa they
are referred to as sisters. There is a simple rule for daughter association: any
daughter phrase may be directly associated only with the complete set of daughter
phrases which are also dominated by its mother
2
. So bone can only be associated
with from my bones through yetiv. After this association the phrase bone from my
bones can only be associated with this at last through revia, and then the whole
phrase is associated with and flesh from my flesh through tifha and so on. This rule is
specified in greater detail in the accompanying paper Masoretic Hebrew Cantillation
and Constituent Structure Analysis.
Tevir
Tevir is written under the tone syllable and always after its vowel as in `. Gen 1.24
is punctuated:
Atnah
Zaqef
Revia Tifha
Pashta Tevir
and God said let the earth of all cattle and and land of all and it was so
produce life kinds reptiles animals kinds
We may see from this how the punctuation is used to avoid a possible ambiguity:
given the words alone it would be possible to group part of this verse as {cattle and
reptiles}{and land animals of all kinds}. But, given the rule of daughter association, of
all kinds can only be associated with the whole phrase cattle and reptiles and land
animals.
Tevir is coded as 059B, 5F, 99.

2
For further discussion of this see Masoretic Hebrew Cantillation and Constituent Structure Analysis pp 11f.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
19
bsf
Geresh (Teres) & Gerashayim
Geresh, also called Teres, is written above the tone syllable as in C`. Gen 5.22 is
punctuated:
Atnah
Zaqef
Tifha Tifha
Revia
Geresh
And Enoch with God after he begat three years and he had and daughters
walked Methuselah hundred sons
Again we may notice the disambiguation here: this passage is a genealogy of the
male line of descent, and is not concerned with political correctness. The second
tifha clearly forces the interpretation {and he had sons}{(and daughters as well)}
rather than {and he had}{sons and daughters}.
A variant form of Geresh is Gerashayim, double geresh as in Gen 5.5 ````. As
punctuation marks the two are equivalent.
Geresh is coded as 059C, 4D, 8A and gerashayim as 059E, 55, 8B.
Pazer
Pazer is written in several different ways in different editions of the Hebrew bible. Its
form in BHS is 72. A rare variant is Great Pazer, written as ].. This form is also
known as Qarne fara, cow-horns and occurs only sixteen times. A good example of
pazer is Gen 10.13 (which is not divided by atnah):
Tifha
Tevir
Gerashayim
Pazer
And Micraim fathered the Lydians and and the Lehamim and the Naftuhim
the Anamim
Pazer is coded as 05A1, 87, BC and Great Pazer as 059F, 85, B9.
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
20
bsf
Telisha
Telisha has two forms: Great Telisha which is disjunctive and Little Telisha which is
conjunctive and will be discussed below. Great telisha is prepositive and written as
N2 . Gen 6.22 is punctuated:
Atnah
Tifha Tifha
Tevir
Telisha
so did Noah everything which commanded him God so he did
Great telisha is coded as 05A0, 84, B8
Legarmeh
Like Shalshelet, Legarmeh is formed from both an accent and a following bar:
' `l. Also like Shalshelet, if the bar is missing the accent is conjunctive. It does
not have its own coding in Unicode but is a composite of 05A3 and 05C0. In SIL Ezra
it is a composite of 7C and the accents for Munah (for which see below).
Here is the punctuation of Gen 1.21:
Atnah
Zaqef Zaqef
Revia
Pashta
Geresh
Tifha Tifha
Pazer
Legarmeh
God sea- big and every moving the waters of every and every of every and God that it was
made monsters ones living swarm kind bird kind saw good
creature with them
The punctuation here is subtle. In Isa 27.1 we read that the Lord will punish
Leviathan and slay the dragon that lives in the sea. But these are clearly creatures
which the Lord himself created. In the Masoretic punctuation the daughter
association of and God saw that it was good only extends backward to nodes under
the second zaqef, thus avoiding the implication that the sea-monsters were good!
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
21
bsf
We may contrast the punctuation of 1.25 where the unpunctuated verse is
superficially very similar: Atnah
Zaqef
Tifha Tifha
Revia
Pashta
Tevir
Geresh
And God made of every and the of every and every of every and God that it
the land animals kind cattle kind also reptile kind saw was good
Here land animals and cattle are distinguished, but quite closely related to each
other, while reptiles belong to a different category (this is further stressed by the tevir
on and also). Nevertheless, it is true for all subgroups that they were seen to be good
since the daughter association of God saw that it was good covers the whole verse.
Before turning our attention to the conjunctive accents it may be useful to consider
some cases where use of the disjunctive accents can affect the way in which verses
are translated. Without punctuation Gen 1.16 reads: God made the two great lights
the big light to rule the day and the smaller light to rule the night and the stars. It
would seem quite natural to take this as meaning that the smaller light rules both the
night and the stars. But the verse is punctuated as:
Atnah
Zaqef Zaqef
Zaqef
Pashta Pashta
Tifha Tifha
God made the two the great the big to rule and the to rule and the
lights ones light the day smaller light the night stars
If the smaller light ruled both the night and the stars, the punctuation at the end of the
verse would have to be
Tifha
and the to rule the night
smaller light and the stars
The actual schema forces us to divide the final section as {and the smaller light to
rule the night} {and the stars} rather than {and the smaller light}{to rule the night and
the stars}. Thus the King James Version translates ...the lesser light to rule the night:
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
22
bsf
he made the stars also. Most modern translations give a similar rendering. If this is
correct, then ben Ashers punctuation is less than perfect: and the stars has no
possible direct daughter association with either God made or with God made the
two great lights. But if we accept the punctuation as it stands there is a further
possibility: we have noted that while the Masoretes were scrupulous in preserving the
text they had inherited, they were also prepared to use punctuation to challenge what
they considered to be errors
3
. If the final Tifha of the verse is given its regular status
as a binary divider the verse reads:
Atnah
Zaqef Zaqef
Zaqef
Pashta Pashta
Tifha Tifha
God made the two the great the big to rule and the to rule and the
lights ones light the day smaller light the night stars
In this case the rabbis are telling their readers that to rule the night is actually to be
understood twice in this verse: the smaller light to rule the night and the stars to rule
the night. It was a constant rabbinic concern to show that the Hebrew Scriptures
were self-consistent and without contradiction. Read in this way, Gen 1.16 may be
reconciled with Ps 136 7-9, and in particular vs 9: the moon and stars to rule the
night.
4
On any other reading the two passages are in conflict. More interestingly for
our present purposes, the correctional use of punctuation is then fully consistent
with the general scheme of binary division.
Isa 40.3 is also of interest both to Jewish and Christian scholars. The King James
Version translates: The voice of him that crieth in the wilderness, Prepare ye the way
of the Lord, make straight in the desert a highway for our God.
The Masoretic text punctuates the verse as:
Atnah
Zaqef Zaqef
Great zaqef
Tifha Tifha
Pashta
A voice crying in the prepare the way make in the a highway for our God

3
v supra pp 11f. (under Silluq)
4
cf Jer 31.34 (English versions 31.35)
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
23
bsf
wilderness of the Lord straight desert
By the rule of daughter association, in the wilderness must first be associated with
prepare the way of the Lord before that whole phrase can be extended to include A
voice crying. So the verse should be translated something like: A voice crying
Prepare the way of the Lord in the wilderness... Of the four New Testament citations
of part of this verse as a self-description by John the Baptist, only that of the Fourth
Gospel (Jn 1.23) is compatible with this analysis (although few translations indicate
this possibility).
Many modern translations ignore the punctuation of 1 Sam 3.3. Its schema is
Atnah
Zaqef Zaqef
Pashta Tifha Tifha
The lamp (had) not yet and was in the Lord's where the Ark
of God gone out Samuel asleep Sanctuary there was of God
This means that Samuel was asleep in the Lords Sanctuary is not a valid division of
the verse. What the rabbis are stressing is that Samuel was quite naturally asleep
because it was not yet dawn.
We give finally a nice use of punctuation which uses daughter insertion to wryly
humorous effect. Compare Gen 40.13 with 40.19 (we give only the first part verse for
each):
40.13
Atnah
Zaqef ...
Tifha
Revia
Legarmeh
Another three Pharaoh will lift and restore to your
days up your head you position
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
24
bsf
40.19 Atnah
Zaqef ...
Revia Tifha
Pashta
Legarmeh
Another three Pharaoh will lift off your and hang on a
days up your head body you tree
2c2. Conjunctive accents: the meshartim
The conjunctive accents serve to bind words together in much the same way as
maqqef, but less strongly. The rules governing which conjunctive is used to bind a
word to each of the disjunctive accents are complex, but the distinctions are of an
entirely musical nature, so they do not affect the present paper and will not be dealt
with in any detail.
Munah
Munah is written under the tone syllable of a word as N2. It is coded as 05A3, 5A,
9E, F9. It is used, for example, in Gen 1.2 to bind the-Spirit-of with God: C`7N l``.
Mehuppakh
Mehuppakh is written under the tone syllable as C`7N. It is coded as 05A4, 60, 98.
Mehuppakh is often used to bind a word to another which carries pashta as in Gen
1.14: !NC `` joining let-there-be and lights.
Merekha
Merekha is written under the tone syllable as !N. It is coded as 05A5, 42, 9C, F5 (the
same codes as are used for yored). It is often used to bind a word to one which
carries tifha as in Gen 1.3 C`7N C`` said God. A variant form is double merekha or
merekha khefula as in Gen 27.25: ]`` `7 joining wine and for him. Merekha khefula is
coded as 05A6, 54, 9D.
Darga
Darga is written under the tone syllable as in N``. It is coded as 05A7, 5E, 9F, FA.
Azla & Qadma
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
25
bsf
Azla is written above the tone syllable as in N``. It is coded as 05A8, 3D, 66, 89,
8F. In appearance it is identical to pashta, but can readily be distinguished because
the latter is written after its word (and sometimes also on the tone syllable). Azla is
often associated with the disjunctive geresh as in Gen 1.9: C`C ``` let the waters
come together. In such cases it is known as Qadma.
Little Telisha
Little Telisha is written above and after its word (i.e. postpositively) as in 2N. It is
coded as 05A9, 83, B6.
Galgal
Galgal is written under the tone syllable as in V2```. It is coded as 05AA, 6A, 9A. It
is very rarely used (16 times in all, always before great pazer).
Several conjunctive accents can occur consecutively, with or without the hyphen
maqqef, to indicate that all the words so connected are to be treated as a single unit.
In Ezek 48.21 the instructions for tribal allotments have been dealing thus far with
two different areas, the sacred Temple-area and the secular city-area (separated by
disjunctive accents). Now, however, the area belonging to the prince is distinguished
from these areas as a totality, so in this verse they are joined by conjunctives as: the-
holy <maqqef> area <munah> and-the-property <galgal> of-the-city. In Josh 19.51
three parties draw lots at Shiloh to divide the land: the first is Eleazar <munah> the-
priest, the second is Joshua <galgal> son-of <maqqef> Nun and the third is the-
heads-of <munah> the-fathers-of <munah> the-tribes-of <little telisha> the-sons-of
<maqqef> Israel. In this case five Hebrew words are involved, but they are marked
as forming a single semantic unit.
The Masoretic pointing as a whole, and the punctuation in particular, is arguably one
of the greatest literary and linguistic achievments in history. Its development spanned
more than a thousand years and was only possible through the co-operation of
countless forgotten scholars whose dedication to accuracy was without parallel. It
offers to all the people of the Book a detailed explanation of how the great biblical
teachers understood their sacred text. For all its antiquity, it is completely amenable
to computer analysis and its mathematical and semantic structure is available to all
who study or promulgate the scriptures.
David Robinson BFBS MAT Team
Elisabeth Levy Bible Society in Israel
February 2002
British & Foreign Bible Society
2 May, 2002 British & Foreign Bible Society
Machine Assisted Translation Team
26
bsf
Select Bibliography
The classic account of the accentuation system was given by
Rabbi David Qimhi (died ca 1235)
Sefer Miklol (The Book of completeness).
This was translated into English by William Chamaky 1933.
The earliest non-rabbinic treatment was by
S. Bohlius
Scrutinium S.S. ex accentibus 1636
The most comprehensive study in modern times is
W. Wickes
Accents of the Poetic Books Taamey emet 1881
Accents of the Twenty-one Books Taamey 21 sefarim 1887
A recent study which takes a very different approach from this paper is
M. Broyer
Taamey hammiqra be-21 sefarim uvesifrey emet 1981
(Biblical accentuation in the twenty-one books and the poetic books)
Brief surveys of the system are hard to find. Probably the best is in
W. Gesenius ed. E. Kautzsch tr. A.E. Cowley
Hebrew Grammar 1813, 28th edition 1910
We understand that a computer-based concordance to the cantillation, Tanach
Cantillation Concordance, is now available, published by Davka 1996, but have not
been able to consult it.