♫ Cantillizer ש֕ - (Mis)encoding Issues

1. Data Processing
   1. Coding Standard
   2. Initial Data Corrections
      1. Sign Inversion
      2. Azla & Pashta
   3. Data Retention
2. Initial Data Properties
   1. Defining Digraphs & Homographs
   2. Defective Digraphs
   3. Legarmeh Digraphs
      1. Prosodic Legarmeh Rule & Exceptions
      2. Psalmodic Pasek List
3. Initial Database Queries
   1. Disjunctive Signs
   2. Conjunctive Signs
4. Misencoding
   1. Misalignment
   2. Refinements
5. GUI
   1. Find Verses Dialog Box
   2. Find Disjunctive Signs Dialog Box
   3. Find Disjunctive Environment Dialog Box
   4. Find Disjunctive Patterns Dialog Box
   5. Find Conjunctive Signs Dialog Box
   6. Find Conjunctive Patterns Dialog Box
   7. Find Any Signs Dialog Box
   8. Word Count Dialog Box

Data Processing

Coding Standard

v 2.0 of the Ezra SIL fonts requires data to be encoded in the following order:

Unicode assigns to cantillation marks the decimal characters 1425-1441, 1443-1454, 1469-1470, 1472, 1475, to vowel points 1456-1465 (1466 in some code), 1467-1469, to diacritics 1468, 1471, 1473-1474, and to the letters of the Hebrew alphabet 1488-1514. It should be noted that 1469 is either the tonic cantillation mark silluk or the pretonic vowel qualifier meteg, and that 1468 is either the vowel shuruk on וּ (if no other vowel is written, under most conditions) or the diacritic dagesh on almost any letter.

If Hebrew language text is maintained, one of two proposed versions of the cantillation virtual keyboard will facilitate searches, assuming performance issues can be resolved. The latter could be targeted (morphological, syntactic, cantillation) by limiting criteria among the following: letters, diacritics, vowels, cantillation marks. Hebrew language text should be searched for any other characters. None should occur, except space character between words and 0160 (nbsp) before pasek.

Initial Data Corrections

Sign Inversion

The coding standard is not uniformly followed in the source, as in this example: נִטְמְאִ֤֨ים (Ezekiel 20:31) encoded in Aleppo אִ֤֨ non-standard but consistent with the purposes of syntactic configuration, while in Leningrad it is ֤֨ standard but inconsistent with the goals of syntactic configuration. All examples of two signs falling on the same letter (consecutive signs in unprocessed source code) should be visually checked and semi-manually corrected, if necessary. When available, same-word occurrences are the best test case for same-letter occurrences. Since many cases of two signs on the same letter (particularly those of one sublinear and one superlinear) may be controversial, all should be flagged in the database.

The coding standard explains the misrepresentation of sublinear prepositive signs: dekhi & yetiv (when falling on the same letter as meteg or munakh) and ole veyored (ole + merekha, when falling on the same letter), but not revia mugrash (geresh mukdam + revia, when falling on the same letter), inexplicably misencoded in the source. These errors are apparent only in Unicode, as the alignment property of the signs themselves guarantees their correct rendering in a graphical browser irrespective of their order. Full examples (with letter & vowels) of misencoding:

Psalms 137:6, meteg + dekhi = אַֽ֭ = אַֽ֭ (in Leningrad Psalms 73 occurrences, 2:3; 5:6; 9:8 etc. Munakh + dekhi in Job 32:13)

Merekha + ole, example not found, may not be attested in source, as ole is pretonic and may be removed to previous word. Variant reading of Job 8:6, אָ֥֫ = אָ֥֫

Psalms 137:6, revia + geresh mukdam = עַ֗֝ = עַ֗֝ (not misencoded in Leningrad)

Algorithm: if meteg/munakh + dekhi/yetiv (with no intervening character), invert; if merekha + ole (with no intervening character), invert; if revia + geresh mukdam (with no intervening character), invert.

All changes to data source must be checked visually beforehand and duly noted by verse (retrieved from file name & named anchors) in documentation.

Azla & Pashta

Data Retention

If Hebrew language text is not maintained, the following information is retained from the data source: book (derived from file name), chapter (derived from file name), verse (derived from named anchors), and Unicode decimal characters: 1425, 1426, 1427, 1428, 1429, 1430, 1431, 1432, 1433, 1434, 1435, 1436, 1437, 1438, 1439, 1440, 1441, 1443, 1444, 1445, 1446, 1447, 1448, 1449, 1450, 1451, 1452, 1453, 1454, 1469, 1470 (makef), 1472 (pasek), 1475 (sof pasuk), and space character, non-breaking space. The thirty-three Unicode characters could be transliterated, in some random or less-than-random order, to SignID 0-9/a-w, with x-y serving as space character and non-breaking space, if necessary. The z character could be used as needed. If for some reason more characters proved to be necessary, A-Z would still be available, assuming case sensitivity in the database. This is probably the most efficient use of resources, with all other distinctions (digraphs, homographs, etc.) being drawn in the sign properties of the configuration. This allows for competing theories of cantillation to be entertained without prejudice to the database.

All other data is deleted. In addition, the following data is deleted, all variants and annotations: (text), [text], and {text}. (text) readings, while statistically insignificant, seem to be preferable, but might prove technically infeasible. On the other hand, the received readings, which seem to lack signs, would return anomalous results. Perhaps batch processing could solve the problem.

Initial Data Properties

Limited or console visualization is required to perform testing, data verification, and correction of encoding errors. The following minimal properties must be informed: configuration (syntactic), book, type (Pros/Pslm), chapter, verse, SignID and Pros/PslmSynx (Conj/Disj). The following properties will be informed by the initial queries: ProsRank, ProsValu and PslmRank, PslmValu. (May need to assume initial values in tables.)

Defining Digraphs & Homographs

The following syntactic digraphs must be defined (in configuration/sign properties or as distinct signs): azla legarmeh, great shalshalet, mehupakh legarmeh, munakh legarmeh, ole veyored, revia mugrash, and reduplicated pashta. Use lack of word break to detect reduplicated signs. The following homographs must be defined (in configuration/sign properties or as distinct signs): silluk/gaya, tifkha/tarkha, and zarka/tsinor.

Defective Digraphs

In psalmody the distinctive, pretonic element (ole or geresh mukdam) of the digraphs ole veyored and revia mugrash sometimes fails, e.g. Psalms 68:20, in which case the string of what otherwise appear to be conjunctive signs indicates the anomaly: י֤וֹם ׀ י֥וֹם יַֽעֲמָס־לָ֗נוּ (cf. pasek list below). If the tonic element (merekha or revia) fails, it is far more easily detected, since the pretonic elements never occur in isolation otherwise. Richter claims that defective ole veyored always follows tsinor, except for Job 8:6. Similarly, defective revia mugrash should always immediately precede silluk. James Price, author of a Concordance of cantillation marks, has kindly provided a list of 57 cases where merekha stands for ole veyored, 213 cases where revia stands for revia mugrash (most likely in verses lacking atnakh), and 24 cases where geresh mukdam stands for revia mugrash. Circumspection is warranted. If rigorous analysis of the environment is conclusive, then the records of such defective digraphs must be flagged individually in the database. All decisions should be duly noted by verse. This action may be taken after, and be informed by, the initial database queries.

Legarmeh Digraphs

The environment of legarmeh digraphs (azla, mehupakh, and munakh legarmeh) must be checked for anomalies. Nothing else distinguishes these constructions from random occurrences of conjunctive sign + pasek. Dr. Price has also provided a list of 1,167 (munakh), 310 (azla), and 258 (mehupakh) legarmeh digraphs. The records of legarmeh digraphs must be flagged individually in the database. Freefind dialog boxes (listing geresh mukdam and ole separately) and alternate configurations may choose to ignore the flags. All decisions should be duly noted by verse.

Prosodic Legarmeh Rule & Exceptions

In prosody munakh legarmeh (but never munakh + pasek) always precedes revia or itself with sixteen exceptions according to Wickes, citing unspecified Masorah notes to Leviticus 10:6; Jeremiah 4:19. All cases of munakh + pasek (+ conjunctives) + revia should be flagged as legarmeh digraphs, taking into account the following:

Geresh (11) follows Munakh Legarmeh in Genesis 28:9; 1 Samuel 14:3, 47; 2 Samuel 13:32; 2 Kings 18:17; Jeremiah 4:19 (variant reading); 38:11; 40:11; Ezekiel 9:2; Haggai 2:12; 2 Chronicles 26:15.

Pashta (3) follows Munakh Legarmeh in Leviticus 10:6; 21:10; Ruth 1:2.

Tevir (1) follows Munakh Legarmeh in Isaiah 36:2. Wickes cites Rashi and David Kimhi (cf. 2 Kings 18:17). He further claims that Darga in this verse is Geresh transformed for musical reasons in the interest of a “lighter melody” (Mrka MnLg Azla Drga Tvir Mrka Tfka Mnkh Atnk).

Revia (1) follows Munakh + Pasek in Isaiah 42:5 below left, pasek פסק visible in the left margin keyed to הָאֵ֣ל ׀֯ יְהוָ֗ה by the Masorah circle. Pasek also separates these two names of God in Psalms 85:9.

Wickes quibbles (prose chapter XIII, note 6) with the following exceptions taken from the Masorah notes of the Leningrad Codex (Samuel ben Jacob, c. 1009).

Little Pazer (3) follows Munakh Legarmeh in Daniel 3:2 below center, the abbreviation of legarmeh לגרמׄ in the left margin keyed to לְמִכְֿ֯נַ֣שׁ ׀; Nehemiah 8:7 (bis) below right, the synonymous abbreviation לגרׄ in the right margin keyed to וְשֵׁרֵ֥בְֿיָ֣֯הֿ ׀ and cf. שַׁבְּתַֿ֣י ׀ at the end (left) of the column.

Isaiah 42:5 page 476 column 3 line 19

Daniel 3:2 page 883 column 3 line 2

Nehemiah 8:7 page 921 column 3 line 13

The last three instances of munakh legarmeh share the same environment, verse initial little pazer twice preceding and twice following before pashta in an enumeration of high functionaries.

Daniel 3:2 [Aramaic] Then Nebuchadnezzar the king sent to gather together the princes, the governors, and the captains, the judges, the treasurers, the counsellors, the sheriffs, and all the rulers of the provinces [...]

Mnkh LtPz LtPz MnLg LtPz Mnkh LtPz LtTl Mhpk Psta LtZk Tfka Mnkh Atnk

Nehemiah 8:7 Also Jeshua, and Bani, and Sherebiah, Jamin, Akkub, Shabbethai, Hodijah, Maaseiah, Kelita, Azariah, Jozabad, Hanan, Pelaiah, and the Levites, caused the people to understand the law [...]

LtPz LtPz MnLg LtPz LtPz MnLg LtPz LtPz Mnkh LtTl Azla Mhpk Psta LtZk Mrkh Tfka Atnk

Critically conflating a variety of sources, Wickes gives a pasek list of 416 entries. Controversies in prosody may also be compared with Verboomen.

Psalmodic Pasek List

In psalmody Wickes counts fifty-seven occurrences of pasek. All occurrences of azla/mehupakh + pasek should be flagged as legarmeh digraphs, and then the following 57 unflagged. In at least one case (Psalms 137:7) there is a flagged AzLg and an unflagged mehupakh + pasek in the same verse.

Psalms (43) 5:2*, 5, 7*; 10:3*, 13*; 18:50*; 20:7*; 35:21†; 40:16†; 41:14†; 44:24*; 57:10*; 58:7*; 59:2*; 61:9†; 66:8*, 18*; 67:4*, 6*; 68:20†, 21*; 70:4†; 72:19†; 74:18*; 77:8*; 78:65*; 85:9*; 86:8*; 89:9*, 50*, 52*, 53†; 94:3*; 108:4*; 113:4*; 116:1*; 119:52*, 156*; 137:7 (MhLg per Price)†; 139:19*, 21*; 141:4; 143:9*.

Job (6) 27:9*, 13*; 35:13*; 38:1; 40:6, 9*.

Proverbs (8) 1:22; 6:3, 9; 8:21, 30 (MhLg per Price)†, 34 (MhLg per Price)†; 15:25*; 30:15†.

* 37 occurrences of pasek precede and/or follow one of the names of God apparently for reasons of decorum, in order to avoid the sullying, obscuring, profanation, or anthropomorphism of the divine being, e.g. Psalms 85:9 הָאֵ֪ל ׀ יְה֫וָ֥ה.

† 12 pairs of doublets are distinctively separated by pasek, e.g. Psalms 68:20 י֤וֹם ׀ י֥וֹם. Wickes, however, points out that pasek fails to separate the doublets in Psalms 22:2 אֵלִ֣י אֵ֭לִי, which breaks the preceding rule as well; 68:13; Proverbs 20:14, and many prosodic verses.

Initial Database Queries

Disjunctive Signs

Initial database queries provide ProsSynx and PslmSynx values, the syntactic hierarchy of the signs: n (all disjunctive signs) precedes n (all disjunctive signs). Work backwards from atnakh/silluk (prosody) and ole veyored/atnakh/silluk. Return number of occurrences. Taking Job 42:7-17 (the prosodic epilogue) as an example:

Disjunctive Sign Hierarchy

y precedes x	LtPz	GtTl	MnLg	DbGr	Grsh	Tvir	Zrka	Ytiv	Psta	Rvia	Tfka	GtZk	LtZk	Sgol	Atnk	Sluk	SfPs	Total
Little Pazer					2													2
Great Telisha					1	2												3
Munakh Legarmeh										1								1
Double Geresh									1									1
Geresh			1			1				4								6
Tevir											9							9
Zarka														1				1
Yetiv													3					3
Pashta													9					9
Revia				1		2	1	2	5									11
Tifkha															9	11		20
Great Zakef											1							1
Little Zakef		2				1		1			8	1	2					15
Segolta									1									1
Atnakh	1				1	2				3	2							9
Silluk																	11	11
Sof Pasuk	1	1			2	1			2	3			1					11
Total	2	3	1	1	6	9	1	3	9	11	20	1	15	1	9	11	11

Prosody and psalmody must be treated separately. For another example of the statistical analysis of sign order, see 2 Samuel 12.

Conjunctive Signs

Create similar table for all disjunctive signs: n (disjunctive) follows n (conjunctive). Return number of occurrences. Taking Job 42:7-17 (the prosodic epilogue) as an example:

Conjunctive Sign Subordination

y follows x	Total	None	1st	2nd	3rd	4th	5th	6th
Little Pazer	2	2
Great Telisha	3	2	1 Mnkh
Munakh Legarmeh	1	1
Double Geresh	1	1
Geresh	6	0	6 Azla	3 LtTl	1 Mnkh
Tevir	9	3	3 Drga 3 Mrka	2 Azla 2 Mnkh
Zarka	1	0	1 Mnkh	1 Azla
Pashta	9	2	7 Mhpk	1 Azla
Revia	11	6	5 Mnkh	1 Drga
Tifkha	20	9	11 Mrka
Little Zakef	15	9	6 Mnkh
Segolta	1	1
Atnakh	9	8	1 Mnkh
Silluk	11	3	8 Mrka

Instances of different signs in the first position (e.g. tevir above and c.f. below), which are common, branch out in order to be differentiated in the ensuing positions, so that each individual sequence is properly attested. Wickes claims that in some cases darga is geresh transformed with azla as the leftover first conjunctive of geresh. Prosody and psalmody must be treated separately. In the former Richter observes:

• Some distinctive marks have their own conjunctive marks serving them: Mahpakh serves Pashta, Darga serves Tvir, Qadma [azla] serves Geresh, Galgal serves Qarney Para [great pazer].

• The remaining distinctive marks, and Pashta and Tvir in many cases as well, are served by two “generic” conjunctive marks: Munnach serves Zarqa, Gershayim [double geresh], and all non-final marks [atnakh, segolta, little zakef, revia, little pazer, telisha]; Merkha serves the final marks [silluk, tifkha, legarmeh] except Zarqa and Gershayim.
  

This distinction (final/non-final disjunctive sign) is probably not necessary, as the Disjunctive Sign Hierarchy table will likely show this, as geresh, tevir, zarka, yetiv, pashta, and tifkha above, all precede signs of a higher rank. In psalmody this distinction seems confused.

Misencoding

All changes to data source must be checked visually beforehand and duly noted by verse in documentation. No data (except initial inversion correction) shall be emended until visualization and syntactic queries (n precedes/follows n) are supported independently for both prosodic and psalmodic books, conjunctive and disjunctive signs.

Misalignment

Possible misalignment leading to homographic ambiguity.

Check positive signs for prepositive environment*:

geresh (pros - disj) > geresh mukdam (pslm - NA)
mehupakh (pros/pslm - conj) > yetiv (pros - conj)
tifkha (pros - disj) / tarkha (pslm - conj) > dekhi (pslm - disj)

Check positive signs for postpositive environment:

tsinorit (pslm - conj) > zarka (pros - disj) / tsinor (pslm - disj)
azla (pros/pslm - conj) > pashta (pros - disj)

Check prepositive signs for positive environment:

geresh mukdam (pslm - NA) > geresh (pros - disj)
dekhi (pslm - disj) > tifkha (pros - disj) / tarkha (pslm - conj)
yetiv (pros - conj) > mehupakh (pros/pslm - conj)

Check postpositive signs for positive environment:

pashta (pros - disj) > azla (pros/pslm - conj)
zarka (pros - disj) / tsinor (pslm - disj) > tsinorit (pslm - conj)

* Pros/Pslm + n + conjunctives/disjunctive (for conjunctive signs); Pros/Pslm + n + disjunctive (for disjunctive signs); n + pasek + disjunctive (for azla & mehupakh legarmeh). Use lack of word break to detect reduplicated pashta.

Refinements

After data is emended, check and if necessary update definition of digraph reduplicated pashta. If geresh > geresh mukdam correction is made in psalmody, update definition of digraph revia mugrash.

Update syntactic hierarchy and conjunctive subordination tables.

In prosody define stich break as following atnakh. Short verses lacking atnakh may contain no stich break. In psalmody visible caesura (space characters/non-breaking spaces) is retained. (Line wrap may cause problems.) In distichs atnakh generally falls before the caesura, sometimes ole veyored. In tristichs the caesuræ seldom correspond to both ole veyored and atnakh, but are often removed to revia/revia mugrash. More research is needed on this point. Problems may be foreseen. It may turn out that the concepts of caesura and stich break are in fact distinct.

GUI

The following forms illustrate the basic functions of Cantillizer (but are not intended to represent, or in any way restrict, the recommended or actual GUI, which will be determined at a later date based on criteria yet to be determined).

Find Verses Dialog Box

Cantillizer displays selected verses or one full chapter of the Bible. Chapter and verse list boxes are context sensitive to the number of chapters in the selected book and the number of verses in the selected chapter respectively. Multiselect verses by means of Shift and Control keys.

Job 42 verses 1-6 are psalmodic, while the epilogue (verses 7-17) is prosodic, as is the prologue (verses 1:1-3:1). (For reference, here is the entire WTC Leningrad Codex book of Job and the same file in slightly processed form.) Price lists 42:1 as an instance of the defective digraph revia mugrash standing in place of atnakh, which never falls on either of the last two words before silluk in psalmody. Atnakh often fails in similar first chapter verses of the dialogue that amount to speech prefixes: “Then X answered, and said.” Wickes deems 42:2 an instance of revia mugrash transformed into the conjunctive sign munakh before silluk for lack of sufficient intervening word length, as measured in syllable count and vowel quantity/quality, and he further claims that, but for transformation, silluk is never preceded by more than one conjunctive sign. In the very short prosodic verses 42:13 and 42:17 atnakh fails. The full chapter could be displayed in the Cantillizer workspace thus (Colors might be deemed too gaudy, but serve a useful visual-aid purpose. Might be too hard to implement):

Find Disjunctive Signs Dialog Box

Cantillizer displays all verses of the selected books that fit the selected criteria and boolean operators. Either prosodic or psalmodic books may be selected, but not both. The three menu items for Job divide that book into prosodic prologue, psalmodic body, prosodic epilogue. Multiselect books by means of Shift and Control keys. Disjunctive sign list boxes are context sensitive (not implemented in order to show full menu in the mockup below) to the signs that occur in the selected books, so (depending on the books selected) atnakh, double geresh, geresh, great pazer, great shalshelet, great telisha, great zakef, little pazer, little zakef, munakh legarmeh, pashta, revia, segolta, silluk, tevir, tifkha, yetiv, zarka (in all 21 prosodic books) and atnakh, azla legarmeh, dekhi, great shalshelet, little pazer, mehupakh legarmeh, ole veyored, revia, revia mugrash, silluk, tsinor (in all 3 psalmodic books). If possible, in order to avoid futile searches, each subsequent disjunctive sign list box should be context sensitive to the signs that occur in the same verse as the signs selected in the previous list boxes in the books selected. There is no reason for the eight-sign limit on queries. It should be user-defined, if possible, or not be fewer than six list boxes.

Results could be displayed thus:

Find Disjunctive Environment Dialog Box

Cantillizer displays distinctive environments (not individual verses) for the selected sign in descending order of frequency. The disjunctive sign list box is context sensitive (not implemented in order to show full menu in the mockup below) to the signs that occur in the selected books.

Following signs could be displayed for thus:

♫ Cantillizer ש֕ - Job Epilogue: 11 instances of 5 signs in 11 verses

You searched for Disjunctive signs following Revia

ש֙	ש֗	5
ש֚	ש֗	2
ש֛	ש֗	2
ש֮	ש֗	1
ש֞	ש֗	1

Preceding signs could be displayed for thus:

♫ Cantillizer ש֕ - Job Epilogue: 11 instances of 4 signs in 11 verses

You searched for Disjunctive signs preceding Revia

ש֗	ש֜	4
ש֗	ש׃	3
ש֗	ש֑	3
ש֗	ש֣ ׀	1

(Sof pasuk stands for the beginning or end of the verse.) Note that within a given frequency the patterns are sorted by the disjunctive hierarchy of the preceding/following sign.

Find Disjunctive Patterns Dialog Box

For the moment enough is known about prosody alone to map the syntax of the signs. Only nine of the eighteen disjunctives may be said to govern a subordinate clause. Great shalshelet, great zakef, and yetiv stand alone in their clause, while the earls (double geresh, geresh, great pazer, great telisha, little pazer, munakh legameh) sit at the foot of cantillation hierarchy. Some authors (Wickes, Jacobson) claim that there is a fifth level to the hierarchy, in which earls (little pazer, great telisha) toil under earls (geresh, little pazer, great telisha), but the evidence seems sketchy. Cantillizer hopes to contribute to research in this field. The initial, working assumption is that strings of earls, like strings of emperors, kings and dukes, obey the rule of in-rank precedence.

End of previous clause*	Subordinates	Governor of clause	Remarks
Atnakh or Verse break	Kings, Dukes & Earls	Silluk	In fact silluk governs the whole verse, but in searching for patterns it is also useful to compare stichs, i.e. silluk’s immediate clause to that of atnakh.
Verse break	Kings, Dukes & Earls	Atnakh
King, Atnakh or Verse break	Dukes & Earls	Little Zakef or Tifkha
Verse break	Dukes & Earls	Segolta	Neither king nor atnakh precedes segolta.
Duke or Verse break	Earls	Zarka	Zarka aways occurs immediately preceding segolta.
Duke, King, Atnakh or Verse break	Earls	Revia, Pashta or Tevir

* Only the nearest preceding instance must be sought, duke, king, atnakh, or the beginning of the verse.

Cantillizer displays distinctive patterns (not individual verses) for the selected sign or signs in descending order of frequency. Disjunctive checkboxes are context sensitive (not implemented in the mockup below) to the signs that govern a subordinate clause (i.e. follow a sign of lower rank) in the selected books. A naming convention lacks. Silluk governs the verse, atnakh and silluk (the emperors) govern the stichs or empires, kings govern hemistichs or realms, dukes govern quarterstichs or duchies, earls govern feet or earldoms. Perhaps a better terminological standard could be found.

Only the Compare Verses button is active for psalmody. More research is needed, particularly data collected from the Find Disjunctive Signs dialog box and input into the Disjunctive Sign Hierarchy table. Comparing the clauses of ole veyored (characterized by tsinor), atnakh (characterized by dekhi), and silluk (characterized by revia mugrash) does not seem promising. The four-to-six sign disjunctive hierarchy, governed by one or two kings and an emperor, may not yield fruitful results. Richter claims that azla legarmeh may replace ole veyored and that revia or little pazer may replace atnakh as kings.

(Data are not especially meaningful without a sample size in the hundreds, if not thousands, of verses.) Disjunctive patterns could be displayed for verses thus:

♫ Cantillizer ש֕ - Job Epilogue: 11 instances of 11 patterns in 11 verses

You searched Disjunctive signs for Verse patterns of Silluk

שֽ

ש֖

ש֑

ש֖

ש֔

ש֙

1

שֽ

ש֖

ש֑

ש֖

ש֛

ש֔

ש֔

ש֙

ש֗

ש֜

1

שֽ

ש֖

ש֔

ש֙

ש֗

ש֑

ש֖

ש֔

ש֙

1

שֽ

ש֖

ש֔

ש֙

ש֗

ש֜

ש֡

ש֑

ש֖

ש֛

ש֗

1

שֽ

ש֖

ש֔

1

שֽ

ש֖

ש֕

ש֔

ש֚

ש֗

ש֑

ש֖

ש֔

ש֚

ש֔

ש֙

ש֒

ש֮

ש֗

ש֜

ש֠

1

שֽ

ש֖

ש֛

ש֑

ש֖

ש֔

ש֚

ש֗

1

שֽ

ש֖

ש֛

ש֑

ש֖

ש֛

ש֜

1

שֽ

ש֖

ש֛

ש֠

ש֔

ש֙

ש֗

ש֜

ש֑

ש֖

ש֛

ש֗

1

שֽ

ש֖

ש֛

ש֠

ש֔

ש֙

ש֞

ש֗

ש֑

ש֖

ש֔

ש֔

ש֙

ש֗

ש֣ ׀

ש֜

ש֡

1

שֽ

ש֖

ש֛

1

Disjunctive patterns could be displayed for stichs thus:

♫ Cantillizer ש֕ - Job Epilogue: 20 instances of 15 patterns in 11 verses

You searched Disjunctive signs for Stich patterns of Silluk & Atnakh

שֽ	ש֖	ש֛									3
ש֑	ש֖	ש֔	ש֙								2
ש֑	ש֖	ש֛	ש֗								2
שֽ	ש֖										2
ש֑	ש֖	ש֔	ש֔	ש֙	ש֗	ש֣ ׀	ש֜	ש֡			1
שֽ	ש֖	ש֔	ש֙	ש֗	ש֜	ש֡					1
שֽ	ש֖	ש֔	ש֙	ש֗							1
ש֑	ש֖	ש֔	ש֚	ש֔	ש֙	ש֒	ש֮	ש֗	ש֜	ש֠	1
ש֑	ש֖	ש֔	ש֚	ש֗							1
שֽ	ש֖	ש֔									1
שֽ	ש֖	ש֕	ש֔	ש֚	ש֗						1
ש֑	ש֖	ש֛	ש֔	ש֔	ש֙	ש֗	ש֜				1
ש֑	ש֖	ש֛	ש֜								1
שֽ	ש֖	ש֛	ש֠	ש֔	ש֙	ש֗	ש֜				1
שֽ	ש֖	ש֛	ש֠	ש֔	ש֙	ש֞	ש֗				1

Note that within a given frequency the patterns are sorted by the disjunctive hierarchy of the first governed sign (second from left), then the second, third, etc. and the disjunctive hierarchy of the governor is used only as the tiebreaker of last resort. Depending upon how the ProsValu and PslmValu properties are handled, certain signs could be considered equivalent: segolta/great shalshelet (not statistically significant), little/great zakef, pashta/yetiv, geresh/double geresh, little/great pazer (not statistically significant), and (in psalmody) azla/mehupakh legarmeh. They are essentially contextual variants or allographs subject to no difference in disjunctive hierarchy.

Disjunctive patterns could be displayed for kings thus:

♫ Cantillizer ש֕ - Job Epilogue: 36 instances of 15 patterns in 11 verses

You searched Disjunctive signs for Royal patterns of Segolta & Little Zakef & Tifkha

ש֖						11
ש֖	ש֛					4
ש֔	ש֙					3
ש֔						3
ש֔	ש֙	ש֗	ש֜			2
ש֔	ש֚	ש֗				2
ש֖	ש֛	ש֗				2
ש֖	ש֛	ש֠				2
ש֔	ש֙	ש֗	ש֜	ש֡		1
ש֔	ש֙	ש֗	ש֣ ׀	ש֜	ש֡	1
ש֔	ש֙	ש֗				1
ש֔	ש֙	ש֞	ש֗			1
ש֔	ש֚					1
ש֒	ש֮	ש֗	ש֜	ש֠		1
ש֖	ש֛	ש֜				1

Disjunctive patterns could be displayed for dukes thus:

♫ Cantillizer ש֕ - Job Epilogue: 29 instances of 10 patterns in 11 verses

You searched Disjunctive signs for Ducal patterns of Revia & Pashta & Tevir

ש֙				8
ש֗				6
ש֛				6
ש֗	ש֜			2
ש֛	ש֠			2
ש֗	ש֜	ש֠		1
ש֗	ש֜	ש֡		1
ש֛	ש֜			1
ש֙	ש֞			1
ש֗	ש֣ ׀	ש֜	ש֡	1

Find Conjunctive Signs Dialog Box

Cantillizer displays all verses of the selected prosodic or psalmodic books that fit the selected criteria. The disjunctive sign list box is context sensitive (not implemented in order to show full menu in the mockup below) to the disjunctive signs that occur following a conjunctive sign in the selected books. It should be remembered that a few disjunctives never follow a conjunctive, e.g. great zakef and yetiv. The number of conjunctive sign list boxes never exceeds six and is context sensitive to the number of conjunctive signs that may precede the selected disjunctive sign in the selected books. Conjunctive sign list boxes are context sensitive to the conjunctive signs that occur in the selected position preceding the selected disjunctive sign in the selected books, so (depending on the disjunctive selected in the books selected) azla, darga, double merekha, galgal, little telisha, mehupakh, merekha, munakh (in all 21 prosodic books) and azla, galgal, illuy, little shalshelet, mehupakh, merekha, munakh, tarkha, tsinorit (in all 3 psalmodic books). Many of the conjunctive sign list boxes are inactive or contain only one item. It should be noted that dekhi occasionally precedes its conjunctive munakh in the same word.

Results could be displayed thus:

Find Conjunctive Patterns Dialog Box

Cantillizer displays distinctive patterns (not individual verses) for the selected sign in descending order of frequency. The disjunctive sign list box is context sensitive (not implemented in order to show full menu in the mockup below) to the signs that follow a conjunctive sign in the selected books (thus for example the absence of great zakef and yetiv, but the presence of great shalshelet, which may follow a conjunctive sign in the psalmodic books). As in the table above, the longest pattern consists of a disjunctive sign preceded by six conjunctive signs. It should be noted that dekhi occasionally precedes its conjunctive munakh in the same word.

Conjunctive patterns could be displayed thus:

♫ Cantillizer ש֕ - Job Epilogue: 9 instances of 5 patterns in 11 verses

You searched Conjunctive signs for patterns of Tevir

Disj.	1st	2nd	3rd	4th	5th	6th	Freq.
ש֛							3
ש֛	ש֧	ש֨					2
ש֛	ש֥	ש֣					2
ש֛	ש֧						1
ש֛	ש֥						1

There are no especially relevant criteria for sorting patterns within a given frequency. The sign that occurs more frequently in the first position altogether should precede. For ease of visibility, table should be sortable by sign per position column. It should be noted that dekhi occasionally precedes its conjunctive munakh in the same word. These instances must be flagged in the database.

Find Any Signs Dialog Box

Cantillizer displays all verses of the selected books that fit the selected criteria without distinction between disjunctive and conjunctive signs. Sign list boxes are context sensitive (not implemented in order to show full menu in the mockup below) to the signs that occur in the selected books. If possible, in order to avoid futile searches, each subsequent sign list box should be context sensitive to the signs that occur in the same verse as the signs selected in the previous list boxes in the books selected.

Results could be displayed thus:

Word Count Dialog Box

Cantillizer displays all verses of the selected prosodic or psalmodic books that fit the selected criteria. The first sign list box is context sensitive to the signs that occur in the selected books.

Whether a conjunctive or disjunctive sign is selected in the first sign list box, both the Same Letter and the Same Word menu items are potentially active in the Select an Interval list box. If either the Same Letter or the Same Word menu item is selected, then all signs in the second sign list box are potentially active.

If a disjunctive sign is selected in the first sign list box, then the 1-nth Word menu items are also potentially active in the Select an Interval list box. If one of the 1-nth Word menu items is selected, then only disjunctive signs one rank below (and revia mugrash preceding silluk in psalmody) the disjunctive sign selected in the first sign list box are potentially active in the second sign list box. (A few other combinations are also irrelevant, e.g. segolta/great shalshelet before silluk [prosodic] and tsinor before atnakh [psalmodic]. Perhaps a general rule could be established defining clauses based on descending sign hierarchy.)

If possible, in order to avoid futile searches, the second sign list box should be context sensitive (not implemented in order to show full menu in the mockup below) to the relevant signs that occur in the selected position preceding the selected sign in the selected books. This dialog box is complex, but it alone allows the user to search for signs on the same letter or in the same word, as well as to check Wickes’ widely received theory of word count determining the choice of signs. It is worth implementing, if possible.

If not too numerous, list all occurrences of Same Letter and Same Word in sign notes. If too numerous, list number of occurrences. Results could be displayed thus:

This search may be further refined in order to isolate the major dichotomy (note the absence of Job 42:7-9, 12, 14-15, and especially 17 in the results below) as follows:

In the last sign list box the user could restrict as follows (only one menu item active per situation):

Last Sign	Previous Sign	Not Preceded By	Notes
Silluk	Atnakh	Ole Veyored	Psalmodic
Silluk	Revia Mugrash	Ole Veyored or Atnakh	Psalmodic
Ole Veyored	Tsinor or Revia *	Revia	Psalmodic. * so-called Little Revia
Atnakh	Revia  or Dekhi	Revia **	Psalmodic. ** with no intervening Ole Veyored
Silluk	Little Zakef, Great Zakef or Tifkha	Atnakh or Zakef †	† Little or Great Zakef
Atnakh	Little Zakef, Great Zakef or Tifkha	Segolta or Zakef †	† Little or Great Zakef
Segolta	Zarka	Revia, Pashta, Yetiv or Zarka
Little Zakef	Revia, Pashta or Yetiv	Revia, Pashta or Yetiv ‡	‡ with no intervening Segolta, Little Zakef or Tifkha
Tifkha	Tevir	Revia, Pashta, Yetiv or Tevir ‡	‡ with no intervening Segolta, Little Zakef or Tifkha

No not preceded by restrictions apply to ole veyored, segolta, great shalshelet, and all earls. Results could be displayed thus:

Abbreviations

Cantillizer