Cantillation: Some Observations – Part 1

Print Friendly, PDF & Email

William Gewirtz


My hope is that this brief paper will contain something of value even for those familiar with cantillation, henceforth referred to as trop, and is not too cursory for those with only limited familiarity. In this essay, trop will be briefly introduced, followed by a look at its importance in a local context, structuring a phrase within a pasuk. It concludes with an unrelated topic: some signs of trop’s rabbinic origin. A follow-up essay will look at trop in its global context, structuring the whole pasuk. The process by which the trop operates on a pasuk demonstrates its surprisingly recursive nature, providing the first such example in a musical context of which I am aware.

Before going any further, it is critical to recognize the role of trop in providing (only) syntax as opposed to semantics. Semantics specifies the meaning of a word, phrase, sentence, etc., something that Onkelos and other interpreters do; syntax provides only the structure. A given syntax can rule out a specific semantic interpretation, remaining consistent only with other interpretations. A semantic interpretation will normally imply a specific syntax and invalidate (some) other syntactic alternatives. Multiple examples in the next section will illustrate.

Both trop and Onkelos are accorded an ancient origin in the Talmud.[2] However, on arguably well over fifty occasions, they differ with respect to the meaning of verses in the Torah.

Despite a dispute between traditional and academic scholars over the identity and dating of Onkelos, many contemporary scholars date Onkelos’ commentary to the later part of the 4th century CE to the early part of the 5th century,[3] during the period of the amoraim. Not surprisingly, at times Onkelos differs from the view of the bavli. However, while the trop of various pesukim was still unsettled in Talmudic times,[4] the trop does not differ from halakhic conclusions of the bavli in any critical instance of which I am aware. While there are minor differences in the trop currently in use, our system of trop correlates with the Aleppo codex. No version of trop in our possession predates the end of the period of the geonim.

The trop – a simplified overview[5]

Trop contains 4 levels of separators (mafsikim) and a single set of connectors/servants (meshartim). The first level separators (often referred to as keisarim, Caesars) are the sentence ending sof pasuk, and the etnahta, which identifies the midpoint of the sentence, a semi-colon of sorts. Both parts of the sentence, before and after the etnahta, are treated identically by the syntax defining rules of trop.[6] The second level of separators (often referred to as melakhim, kings), the zakeif katon, zakeif gadol, segol, shalshelet and tipha, define the major structure of the pasuk. Pashta, revii, and tevir are common third level separators, while darga, pazeir and telisha gedolah are common fourth level separators.

Munah, merha, mahapah, and kadma are common connectors; there should not be an apparent pause between the reading of words where they appear and the following word.

Trop structures both at a global / macro level (the entire pasuk and its two major components) and at a local / micro level (each individual phrase).

Some examples of trop’s importance locally

To begin examining trop in a localized context, let’s look at the significant impact that can be drawn from the placement of the tipha, a second level separator, versus the munah and merha, connectors that almost always occur prior to the sof pasuk and the etnahta. The examples below further illustrate the difference between semantics and syntax. Two phrases from Az Yashir, מַרְכְּבֹ֥ת פַּרְעֹ֛ה וְחֵיל֖וֹ יָרָ֣ה בַיָּ֑ם (Shemot 15:4) and צָֽלֲלוּ֙ כַּֽעוֹפֶ֔רֶת בְּמַ֖יִם אַדִּירִֽים (Shemot 15:10), illustrate this difference. Both phrases have two different interpretations; in each case, the first interpretation, coming from Onkelos and based purely on semantics, is inconsistent with the trop; the trop is, however, consistent with the second interpretation.

יָרָ֣ה בַיָּ֑ם can mean either:

  1. The army was shot while at sea or
  2. The army drowned in the sea.

There is a major difference between

  1. being shot at the sea, where the sea does not play a participating role but is simply the location where the shooting occurs, and
  2. being drowned in the sea, where the sea is an indispensable part of the event.

By use of a connector, the munah, linking יָרָ֣ה to בַיָּ֑ם, the trop implies a significant link between the sea and the event. The trop is identical to that on רָמָ֥ה בַיָּֽם, slightly earlier in אָ֣ז יָשִֽׁיר, again indicating a significant connection between the sea and the action. The connecting munah in the word יָרָ֣ה is consistent with being cast into the sea (and dying as a result of of drowning in the sea) as opposed to simply being shot while at sea. Onkelos’ translation, shedi ba’yamoh, unquestionably means shot at sea.

Similarly, בְּמַ֖יִם אַדִּירִֽים can mean either:

  1. The Egyptians sank in the mighty waters or
  2. The mighty (Egyptians) sank in the water.

There is a major difference between

  1. the mighty waters, where mighty is an adjective describing the waters, and
  2. the mighty (Egyptians) being drowned in the sea, where the two words are an independent noun and verb, and the noun appears awkwardly, alone, at the end of the verse.

By (strongly) separating בְּמַ֖יִם and אַדִּירִֽים, the trop is consistent only with the second interpretation where the words are independent, telling us who, the אַדִּירִֽים, and where (they drowned), בְּמַ֖יִם. Onkelos, on the other hand, translates the phrase as be’mayin takifin, the mighty waters.

When a tipha and a munah or merha are interchanged, as in the above two examples, the impact on the semantics must be carefully examined. In the above examples, even not following the trop results in a different but still very plausible reading.

Most often, however, an incorrect reading has no coherent interpretation. Another phrase from Az Yashir, נִצְּב֥וּ כְמוֹ־נֵ֖ד נֹזְלִ֑ים, is illustrative. The phrase has a tipha on the third word, separating the first three words from the fourth, which supports a meaning like “the waters formed a heap.” However, erroneously reading the tipha on the first word and connecting the second, third and fourth word would support a farfetched and rather unlikely meaning, which alleges that God formed a leaky heap.

A second localized area of oft overlooked significance concerns pausing in a manner consistent with the trop’s four levels of separators. Pausing properly for the four levels requires a full stop, one-half stop, one-quarter stop and one-eighth stop respectively. Stopping is rarely explained or practiced; when the proper length of stops is violated the resulting error has varying consequences.

An amusing example involves a tevir, requiring a one-quarter stop, followed a word or two later by a tipha, requiring a recognizably longer one-half stop.

Correctly read, וּמְקַלֵּ֥ל אָבִ֛יו וְאִמּ֖וֹ מ֥וֹת יוּמָֽת׃ (Shemot 21:17) means

And one who curses his father or mother is put to death.

With improper pausing, a longer pause after the tevir on the word אָבִ֛יו than after the tipha on the word וְאִמּוֹ, the sentence can be misinterpreted to mean:

And if one who curses his father, then his mother is put to death.

Equally shocking is the second half of Shemot 31:15, which states כָּל־הָעֹשֶׂ֧ה מְלָאכָ֛ה בְּי֥וֹם הַשַּׁבָּ֖ת מ֥וֹת יוּמָֽת:

He who works on the Shabbat is executed.

With improper pausing exactly as above, it might be misinterpreted to mean:[7]

He who works is executed on the Shabbat.

While humorous examples like the two above are rare,[8] there are typically one or two such examples in every week’s Torah reading. Some are of minimal consequence at best; the trop repeated 12 times in parshat Nasso ending the sacrifice of each tribe’s head is a good example.[9] However, many cases of improper pausing work at cross purposes with the trop, modifying associations that the trop intends. For example, the trop on the pasuk פַּ֣ר אֶחָ֞ד בֶּן־בָּקָ֗ר אַ֧יִל אֶחָ֛ד כֶּֽבֶשׂ־אֶחָ֥ד בֶּן־שְׁנָת֖וֹ לְעֹלָֽה׃ implies that the bull, the ram and the lamb in its first year are all sacrificed as olot. Improper pausing could imply that only the lamb is sacrificed as an olah. Such examples abound; three more examples are given in the footnote below.[10]

A famous example is the pasuk in Ha’azinu:

                                    שִׁחֵ֥ת ל֛וֹ לֹ֖א בָּנָ֣יו מוּמָ֑ם דּ֥וֹר עִקֵּ֖שׁ וּפְתַלְתֹּֽל

The pasuk has multiple interpretations; most fundamental is the decision whether to connect the word לֹ֖א with the next word/phrase, בָּנָ֣יו מוּמָ֑ם, or (as the trop does) with the prior word/phrase, שִׁחֵ֥ת ל֛וֹ. The former would refer to a group characterized as not His children; the latter a negative response to either a quizzical or an assertive assignment of responsibility for destruction to God.

Trop is Rabbinic:

There has been reference to the Karaite leanings[11] of (some of) those involved in the transmission or transcription of trop. Other sefarim, first among them Ve’yavinu Ba’mikra[12] by R. Naftali Tzvi Yehudah Gettinger, try to explain idiosyncrasies in the trop by citing a rabbinic view, halakhic or midrashic, that might explain an otherwise (often mildly) troublesome sequence of trop. While many of these explanations are plausible or even brilliant, they are not always entirely convincing.[13] I will illustrate that surprisingly, non-literal, rabbinic interpretations, as opposed to ones that adhere more closely to the text, are supported by the trop in its most critical decision, the placement of the etnahta.

Such examples are not common; in most cases rabbinic interpretations:

  • augment the text, providing missing context but leaving the text itself unchanged; or
  • modify the semantics in a way that does not impact the syntax.

The rabbinic constraints on parameters surrounding a ben sorer u’moreh are a classic example of the former; the additional constraints are derived from the words in the text without changing their inter-relationship and hence their trop. Similarly, the rabbinic implications drawn from lo ba’shamayim hi or treating lex talionis as requiring monetary compensation illustrates the latter; both the literal and the rabbinic interpretation would suggest similar syntax and trop.

The first example of a non-literal Rabbinic reading being supported by the trop is Shemot (20:20):

לֹ֥א תַעֲשׂ֖וּן אִתִּ֑י אֱלֹ֤הֵי כֶ֙סֶף֙ וֵאלֹהֵ֣י זָהָ֔ב לֹ֥א תַעֲשׂ֖וּ לָכֶֽם׃

The most literal interpretation would divide the sentence into two parts, the first ending with the word כֶ֙סֶף֙; the pasuk prohibits graven images of both silver and gold, using a chiastic structure. However rabbinic interpretation lists three prohibitions:

  1. לֹ֥א תַעֲשׂ֖וּן אִתִּ֑יforbidding making images of my celestial beings.
  2. אֱלֹ֤הֵי כֶ֙סֶף֙ (do not make)[14] the keruvim from silver, as opposed to gold.
  3. וֵאלֹהֵ֣י זָהָ֔ב לֹ֥א תַעֲשׂ֖וּ לָכֶֽם other than the keruvim, make no other images of gold.

A rather idiosyncratic trop, with an otherwise inexplicable etnahta on אִתִּ֑י, is in complete alignment with rabbinic interpretation.

A second verse will illustrate the challenges that are associated with the methodology. Consider Shemot (22:12):

אִם־טָרֹ֥ף יִטָּרֵ֖ף יְבִאֵ֣הוּ עֵ֑ד הַטְּרֵפָ֖ה לֹ֥א יְשַׁלֵּֽם׃

As written, the etnahta divides the pasuk at the word עֵ֑ד. However, the literal interpretation embraced by many rishonim interprets the pasuk to mean that the body of the animal is brought as witness to its having been devoured by a wild animal. This explanation would place the etnahta on the word הַטְּרֵפָ֖ה, one word beyond עֵ֑ד.

The rabbis interpret eid,[15] a witness, to mean two eidim, or witnesses, despite the word being written in the singular, and explain that the witnesses tell of the occurrence, as opposed to bringing the physical carcass as evidence. Their interpretation is therefore consistent with an etnahta on the word עֵ֑ד, as occurs in the trop. This proof was convincing, until modern scholars gave two alternate readings that would also imply the same trop as in the rabbinic interpretation (the first ironically providing added rationale / support for the reading in the Talmud):

  1. Witnesses do not have to bring physical evidence; their word is adequate.
  2. The word יְבִאֵ֣הוּ, the word preceding עֵ֑ד, already refers to the carcass,. The Pasuk is saying implicitly that it is to be brought as a witness.

As well, the halakhah may also favor the trop over Onkelos. Consider the oft repeated phrase throughout selihot:

וַיִּקְרָ֥א בְשֵׁ֖ם יְקוָֽק

Whereas Onkelos’s translation places a dalet in front of יְקוָֽק, connecting the word to בְשֵׁ֖ם, meaning “that we call in the name of God,” the trop separates the word בְשֵׁ֖ם from יְקוָֽק, which would support several alternative meanings, including “we call to God by His Name.”

Ashkenazic practice when reciting selihot follows the trop.[16]

There are also many instances where the trop follows a midrashic interpretation, as for example in Bereishit (13:13) וְאַנְשֵׁ֣י סְדֹ֔ם רָעִ֖ים וְחַטָּאִ֑ים לַיקוָ֖ק מְאֹֽד. A possible translation given by JPS reads: “Now the inhabitants of Sodom were very wicked sinners against the LORD.” This and other translations would not comport with the presence of an etnahta on the word וְחַטָּאִ֑ים. Other interpretations, like: “Now the inhabitants of Sodom were very wicked and sinners against the LORD,” might move the etnahta one word forward. The trop seems to support various midrashic interpretation that lists specific sins (blasphemy, idolatry, sexual promiscuity, etc.) associated with both the words לַיקוָ֖ק and מְאֹֽד.[17]

This topic has other examples, almost always involving second level separators.[18] Clearly providing examples from only 4 of over 4000 meaningful sentences in the Torah, some of which are potentially arguable, do not constitute proof of a Rabbinic origin for trop; influence, undoubtedly, but determining origin requires more extensive analysis.

In summary, almost every instance that attempts to demonstrate trop’s rabbinic origin may be disputable. However, the existing evidence and the absence of any contradictory indication supporting a non-rabbinic reading makes a Karaite one unlikely. In fact, over the last 1000 years we do not have examples where the trop was determined to be in such significant opposition to rabbinic interpretation to result in raising fundamental questions.[19]


The Halakhah requires that we correct errors that impact meaning during the (public) reading of the Torah.[20] Increased awareness of trop’s implications may require halakhists to create additional guidelines with respect to trop implementing that rule more precisely. While I do not feel it is my place to shout out corrections, I have on occasion told the reader afterwards what interpretations his reading might suggest. On rare occasions, I have also told a reader that his reading was consistent with Onkelos’s interpretation as opposed to the trop.

It has been jokingly remarked that the full understanding of trop is an example of something lost in the transmission of Torah from Moses to Joshua. My goal was to illustrate some remarkable features of trop, which might increase the level of interest in trop’s essential role.

[1] This essay is dedicated in honor of my father’s 21st yahrzeit. My father died peacefully on Shabbat after davening, telling my sister to go home to hear kiddush from her husband, telling the nurse to say goodbye to his wife, and then settling into bed. He died on the 21st of Elul, which fell on Shabbat parshat Ki Tavo, as occurred this past year and this year as well. My father was an expert’s expert ba’al keriah to whom I asked too few questions.

[2] Nedarim 37b and Megillah 3a. It is not clear if trop denoted the same system throughout its history; in fact, the trop now in use is assumed to be post-talmudic. How it might relate to earlier such systems is unknown.

[3] See for example:

[4] The bavli in Yumah 52a and 52b lists 5 places where there is uncertainty over the placement of the etnahta, the middle of the pasuk, the most important decision made by the trop. All 5 examples involve a dispute concerning the literal as opposed to rabbinic interpretation.

[5] A comprehensive review of trop is provided in Ta’amai Ha’mikrah by Rav Mordechai Breuer, who also authored a much shorter overview included in the first volume of Daat Mikrah, Bereishit. Also, Joshua Jacobsen has authored both an abbreviated and a comprehensive version of “Chanting the Hebrew Bible.”

[6] The equal treatment of both parts of a pasuk, covered in Part 2, is fundamental to how trop operates.

[7] Shlomo Zuckier pointed out that the erroneous meaning associated with improper pausing is not correct grammatically. While certainly in this instance and in several others that is true, one cannot assume such knowledge of grammar among all listeners.

[8] Another arguably bizarre example is Vayikra (14:7) where improper pausing would / might imply that diseased scale, as opposed to water, is to be sprinkled on the person being purified.

[9] The result of improper pausing creates a tighter connection between one’s name and one’s father’s name than the trop correctly read would suggest; the trop makes a tighter connection between the name and the korban.

[10] There are numerous examples that the reader can examine. See for example Vayikra (11:31), which requires a longer pause after the tipha than the tevir; improper pausing might imply that if you are in contact with someone while they are still alive, you become impure after their death. Other clear examples are in Bamidbar (10:29), identifying Moshe’s father-in-law, and Bamidbar (16:27) identifying from whose tents to separate.


[12] The classification of trop in some of these sefarim, including Ve’yavinu Ba’mikre, differs from that described, more in details than in fundamentals. Interestingly, the recursion to described in Part 2 is asserted only for the system of trop described by R. Breuer.

[13] We are not talking about a strict proof; that would be impossible. Instead, a more intuitive (and biased) sense of the likelihood that it was the rabbinic interpretation that drove the trop’s composition. Devarim (28:22), in parshat Ki Tavo, is a perfect example of the literal explanation given by Rashi being reflected in the trop. Provable, never; very likely, yes.

[14] This second phrase borrows לֹ֥א תַעֲשׂ֖וּ לָכֶֽם from the end of the third phrase.

[15] There is a dispute about whether the carcass or witnesses are brought, (see Baba Kama 11a and Meḥilta, which quote the opinion of Abba Shaul, a mid-2nd century tanna, who supports bringing the carcass) the (uncontested) conclusion of the gemara, (Sotah 2a) brings proof from Devarim (19:15) that qualifies eid with ehad, implying eid without qualification, although singular (can) mean a pair of witnesses.

[16] See Avudraham in the Laws of Fasts where in alignment with the trop he suggests pausing after בְשֵׁ֖ם. The phrase בְשֵׁ֖ם יְהוָֽה occurs in multiple locations throughout the Torah. At times the trop and Uneklos have the same disagreement as they do in this example, at times they reverse positions, and at times they agree.

[17] See for example Sanhedrin 109a, Tosefta Sanhedrin chapter 13, and Torat Kohanim Be’Ḥukotai, parsha 2.

[18] A good example is found throughout the beginning of parshat Tzav where the different types of korbanot are preceded by the phrase “zot torat ha- While Onkelos separates the word zot from the word torat, consistent with the assumed pshat, the trop links them. See Titein Emet le’Yaakov by R. Yaakov Kaminetsky for various rabbinic interpretations the trop supports.

[19] As Shlomo Zuckier noted it is still possible that Karaites created much of the trop, which the Rabbinites modified in several places. One might, however, argue that it is unlikely that subsequent Karaite generations involved in trop’s transmission would not restore the Kariate version of the trop. I would very much appreciate being e-mailed halakhic examples consistent (or inconsistent) with the trop.

[20] Rambam, Hilkhot Tefillah (12:6) and Rabbi Yosef Karo in Shulhan Arukh (Orah Hayyim 142:1). Some, including Kaf Hahayyim to Orah Hayyim 142, paragraphs 1-12, extend this to the trop as well.