Cantillation marks (also known as "taamey ha-miqra", "teamim", "trope", "neginot", "accents") are diacritic symbols annotating the Hebrew Bible text for the purpose of cantillation, similar to neumes. They were introduced by the Masoretes around the end of the first millenium of the common era. They can serve three different purposes:
They give instructions for cantillation, which is their primary purpose at least today, and some of their features are plausible only in this context. It is not clear whether cantillation has been the historically first purpose of the marks, or whether they first served only to depict which parts of a sentence belong logically together for easier reading. This article is not concerned with cantillation. These questions are explained in more detail in the Wikipedia entry on cantillation and the Web sites linked to from there.
They depict the syntactic structure of the underlying text. This will be explained in the next section.
As nearly all of them are placed on or below the consonant of the stressed syllable, they generally indicate word stress. See the tables and their legend for details of how to exploit this feature.
The tune to use for cantillation and the distribution of pauses depend on the syntactic structure of the text. It is therefore not surprising that the cantillation marks give information on the syntactic structure. The way this happens has some resemblance with the usage of punctuation marks in English. There, after each word, we have a punctuation mark telling what syntactic entity is terminated with this word: a full stop, question mark, or exclamation mark terminates a sentence; a colon terminates a partial sentence; semicolons, dashes, and commas terminate smaller units; a blank terminates a word; a hyphen terminates a part of a compound word. This is also how the cantillation marks work: they, just like English punctuation marks, indicate that the word where they are placed terminates a smaller or larger syntactic unit. The equivalent of the full stop is the Sof Pasuq, the end of the Bible verse, which was also the basis of verse numbering (so verse division is a Jewish invention, while verse numbering and chapter division is a later Christian invention).
Each cantillation mark belongs to a class indicating its dividing power. The word where it is placed terminates the syntactic entity consisting of this word and all preceding words where the dividing power of the respective mark is smaller. Of two marks with equal power with no intervening stronger mark, the earlier one is considered stronger for reasons to be explained below.
The cantillation marks in the class with the least dividing power, that is, those that will always terminate syntactic units consisting of only one word, do in fact indicate that the word where they are placed is closer connected to the following word than to the preceding one. It is therefore reasonable to describe their role as conjunctive. The other marks are called distinctive.
The analogy between English punctuation marks and Hebrew cantillation marks must not be taken too far. There are also important differences:
The English sentence structure, at least as far as revealed by punctuation marks, is rather flat: most words have only a blank attached (which ordinary people would not call a punctuation mark at all). The sentence structure of a Bible verse, however, is a real tree with some depth.
Never forget that the primary purpose of cantillation marks is cantillation, not syntactic decomposition. When the tune requires that each verse be composed of two halves divided by a caesura, then this will happen, irrespective of whether the caesura falls right in the middle of a sentence in the usual sense of the word or between sentences. Indeed, virtually every Bible verse is divided in this way by a mark called Atnach.
| In the beginning¹ created God° the heavens¹ and the earth.° Now the earth² was formless² and empty¹ and darkness¹ [was] over the abyss° and the Spirit of God¹ [was] hovering¹ over the waters.° |
These are the first two verses of the Bible with all distinctive marks given as superscript numbers where a low number stands for a high dividing power of the mark. The binary decomposition of the verse can quite well be seen by the indentation. The Sof Pasuq is the superscript 0 with a preceding full stop, the Atnach is the superscript 0 at the remaining places. In the first verse, the Atnach is in the middle of a sentence between verb/subject and object, in the second, it is between the second and the third of three complete sentences. - How the position of the marks yields the binary decomposition (and hence the indentation to illustrate it) is explained later. |
One could thus say that at least the distinctive marks tell the distance to the entire verse (the root of the syntax tree), and not to the words (its leaves) as do English punctuation marks.
For the same reason, to wit their musical significance, cantillation marks are generally less predictable than punctuation marks when the text including its syntactic structure is known. There are, however, quite a lot of rules governing the placement of the marks and thus restricting the freedom of choosing a particular musical realisation. After decision on the syntactic decomposition of the Bible verse, there are at each word seldom more than one or two marks that could be possible. An entire chapter of this Web page is devoted to these syntax rules.
It happens relatively infrequently that the meaning of a text is affected by its syntactic decomposition, so that the marks disambiguate the text. One example is:
A voice of one calling: "In the desert prepare the way for the LORD; make straight in the wilderness a highway for our God. . . ." (Is.40:3, NIV)
where, just as in English, the same sequence of words could also mean:
A voice of one calling in the desert: "Prepare the way for the LORD; make straight in the wilderness a highway for our God. . . ." (Is.40:3 footnote, NIV)
Another example is the list of names in Is.9:5-6 where it is not obvious whether the eight names form four compounds of two each.
Three books in the Bible, Psalms (Tehillim), Proverbs (Mishley), and Job (Iyov) are often printed differently from the other books: they use separate lines for verses as is customary for poems in English. The same books use also a different system of cantillation marks which is therefore called the poetic system as distinct from the prosaic system for the other books. As the other books contain also some poetry, and some are even entirely poetic, such as the Song of Songs and the Lamentations, we shall use the more neutral terms the 3 books and the 21 books instead. If you wonder where the 15 books went that are missing to a total of 39 books in the Hebrew Bible: here the two-volume books (Samuel, Kings, Chronicles, and Ezra/Nehemiah) are counted as one book each, and the 12 shorter prophetic writs (Hosea through Maleachi) are combined to form a single book.
The differences between the two systems are:
Some marks exist only in one of the two systems.
Some marks have different names: the conjunctive marks Mahpakh²¹=Mehuppakh³, Qadma²¹=Azla³, Galgal³=Yerach Ben Yomo²¹, and the distinctive mark Zarqa²¹=Tsinnor³. With the exception of the last pair, the "wrong" names are often synonyms of the "right" ones, so that the difference is not grave. To avoid confusion, we use consistently the names left of the equals signs above, and we avoid using the name "Azla" which sometimes is a synonym of "Qadma" (especially in the term "Azla Legarmeh") and sometimes of "Geresh" (especially in the term "Qadma we-Azla").
Some marks are similar but have entirely different significance: the same sign that is distinctive Tipcha²¹ is conjunctive Tarkha³, and conjunctive Merkha (occurs in all books) looks like Yored³, a component of the strongly distinctive Ole We-Yored.
The top-level decomposition is a bit more complicated in the 3 books than in the 21 books. There is not a single mark, Atnach, to divide each entire verse into halves, rather, this can be done by any of Ole We-Yored, Atnach, or Revia. The latter can also play many other roles in this system which, with several marks showing such overlap in significance, appears to be rather enigmatic when compared with the more lucid system in the 21 books.
For the placement of cantillation marks, a number of general rules apply:
The overwhelming majority of words carries one and only one cantillation mark.
A Maqqef (hyphen) connects words close enough that, as a rule, only the last of them carries a cantillation mark. Note that if you regard the Maqqef as a conjunctive mark belonging to the first of the two words it connects, rule 1 is still obeyed in this case.
A Paseq (vertical bar) can sometimes be regarded as part of a mark and renders an otherwise conjunctive mark distinctive (see tables); sometimes it indicates a division that is independent of the structure represented by the other cantillation marks.
A Meteg (secondary stress) does not count as a cantillation mark; it appears together with another mark or a Maqqef at the same word. A Silluq (which looks the same as a Meteg), however, does count as a cantillation mark; only in quite rare cases it appears with another mark in the same word or hyphenated ("maqqefated"?) chain.
While there is an abundance of how a Bible verse may be structured, there is one general pattern: Virtually all verses consist of two halves with an Atnach between. In the 3 books, there can be a stronger division than Atnach by Ole We-Yored, for instance when a short headline of a Psalm (e.g. "A psalm of David") is part of a longer Bible verse: then this headline is typically separated by an Ole We-Yored from the remaining portion of the verse which then usually has an Atnach to divide it.
Unfortunately, for each of these rules there are numerous places in the Hebrew Bible where they are violated.
As described above, the cantillation marks belong to different classes describing their dividing power. These classes carry the titles of rulers: a qeysar (Caesar, emperor) terminates an entire Bible verse and "reigns" it; a melekh (king) divides the realm of an emperor and reigns the first half which it terminates while the other half is still under the reign of the emperor. Likewise, analogous rules apply to the lower ranks of rulers: a mishne (duke) divides the realm of a king and reigns the first half which it terminates; a shalish (officer) divides the realm of a duke and reigns the first half which it terminates. Lowest in rank is a mesharet (servant), that is, a conjunctive mark. Only the term for a distinctive mark in general, mafsiq (divider), does not fit into this imagery of rulers.
In the 21 books, the very first division of each entire Bible verse is done by an emperor, Atnach, which divides the realm of another emperor, Sof Pasuq, thus violating the general rule that always the lower rank divides the realm of the immediately higher. However, the preceding paragraph remains valid even in this case when we regard the half-verse, terminated by either Sof Pasuq or Atnach, as the top level of the decomposition, of course keeping in mind that the Sof Pasuq divides stronger than the Atnach. In the 3 books, there is no such special rule: there, the Sof Pasuq is the only emperor, but two of the kings can appear at most once in a verse and only before the other kings so that in effect a similar division is obtained.
Now, let us look at such a decomposition. The lines are to be read from left to right.
| We start with an entire half-verse, terminated by an emperor: | ||||||||
| . . . emperor | ||||||||
| The first step of division is obvious: | ||||||||
| . . . king | . . . emperor | |||||||
| It is also obvious that the first half has to be divided by a duke. It is, however, not immediately clear whether the second half should be divided by a king (as lower by one rank than the emperor at the end of this section), or by a duke (thus reflecting the real nesting level of the newly created section which is only 1/4 of the text at the outset). Actually, the rule is to proceed in the former way: | ||||||||
| . . . duke | . . . king | . . . king | . . . emperor | |||||
| Now the next step of decomposition follows the same rule again: | ||||||||
| . . . officer | . . . duke | . . . duke | . . . king | . . . duke | . . . king | . . . king | . . . emperor | |
From this sketch one sees immediately why, of any two marks with equal rank with no intervening stronger mark, the earlier one is considered stronger: the later one divides not the entire realm of its superior but only a portion of it, only its "title" profits from its superior's high rank. Note that the binary decomposition of the text is still unique. We assert the following statements about the resulting pattern, leaving the proof as an exercise for the reader:
If in a correctly marked text a distinctive mark M2 follows a distinctive mark M1 with no or only conjunctive marks between, then the rank of M2 may be arbitrary low compared to the rank of M1, but it can at most be higher by one. The latter case occurs if and only if M1 belongs to the realm of M2.
The converse is also true: If in a marked text no distinctive mark is higher by two or more ranks than the preceding distinctive mark, and the last mark is the only emperor, then the sequence of the marks is a possible pattern in a complete half-verse, according to the above rules.
The real nesting levels (0 for top-level, 1 for halves of top-level, 2 for quarters etc.) of the marks can be recovered from the ranks:
For an emperor (or king, or duke, or officer) that has no mark of the same or a higher rank ahead of it, the nesting level is 0 (or 1, or 2, or 3, resp.).
For a mark that is lower in rank than its predecessor by k ranks or equal in rank (that is: lower by k=0 ranks), the nesting level is k+1 more than the nesting level of the predecessor.
For a mark that is higher in rank than its predecessor, the nesting level is the same as it would have been, had all marks of lower rank been removed.
In the Genesis 1 example, the indentation of each line reflected the nesting level of the mark that terminated the preceding line.
While the real nesting levels stay nowhere constant and can jump up and down by arbitrary amounts, the first of these observation shows that the ranks of the marks have a smoother behaviour, without any loss in precision. Therefore the ranks are better suited for cantillation where the marks correspond to pieces of music which must fit together and end in a contiguous tune. A caesura is the more adequate, the less the text before the caesura is continued after it. As we have seen, it plays a role for this distinction whether a mark of some rank is the last in a chain of equals (then it is dominated by the next mark that is not inferior) or whether others will follow in the chain (then it dominates them). Consequently, there are non-final and final marks. Also, which of several possible marks is selected may depend on which mark it "reports" to as its superior. These details are not explained here but only reflected in the chapter on syntax rules. There, one finds also which concrete marks have which ranks.