Home ] Up ]

Dagesh is a stop


Dagesh is not a morphological device, used arbitrary to create the word classes, but rather occurs for predictable phonetic reasons.

Absence of dagesh after silent schwa in constructus plural (citvei) and presence of dagesh in word-initial consonants with no leading schwa proves that schwa does not cause dagesh.

Pious Masoretes would not change sacred letters by restricting aspiration and their numeric value by gemination.

The Masoretes scrupulously invented marks for every minute detail. They could not use the same mark, dagesh, for two different phenomena, gemination (hazak) and loss of aspiration (kal). Both dageshes have a common meaning: stop.[1] Depending on the environment, stop causes loss of aspiration and gemination.

Post-tonic stop after the open syllable (dagesh hazak), common in many languages, naturally appears through gemination, hitlAb.besh. Interconsonantal stop (dagesh kal) naturally appears in singing.

The choice of dot for dagesh is modeled upon the use of dot as word separator and full stop in sentences. One dot, dagesh, is a silent pause; two dots, schwa, is a longer vocal pause. Dot in waw makes it shuruk, [.ww]. Dot as mappiq reduces aspiration; hey after stop becomes syllable-initial and sounds like consonant h, dvare-h. Dot in waw, shin and sin is not positioned as a stop, but still relates specific ways to limit the aspiration.

One stop, dagesh hazak, prevents jamming of post-tonic vowels (hitlAb’sh – hitlAb.besh).

Gemination appeared before the Masoretes, and the LXX reflects it. The speakers empirically discovered that gemination produces an internal stop, and used gemination to prevent post-tonic reduction (hitlAb’sh – hitlAb.besh). Gemination produces stop only between the plosives (b.b, but vv), and geminated consonants remained non-aspirated.

The Masoretes realized that the real need is for stop and the gemination is etymologically incorrect. They introduced explicit post-tonic stop instead of gemination (hitlAb.besh became hitlA.besh). Spoken conventions, however, prevailed, and gemination replaced stop in speech (hitlAbbesh).

Heavy syllable with forcefully pronounced geminated consonant pulled the accent (hitlabbEsh). The stress shift occurred before the Masoretes: they could only hear the long tzere (hitlabbEsh) if accented. Gemination, no longer post-tonic, disappeared in speech (hitlabEsh).

Gemination also occurs after secondary accents which are often syntactical (hA-bbOker, wA-yyOmer). Secondary accent prevents blurring of naked vowels (a-mOn – ‘-mOn, but A-mOn – Am.mOn).

Gutturals cannot be pronounced as two distinct sounds if geminated, making rather a single elongated sound. Gutturals, therefore, cannot be geminated to produce internal stop. A solution to prevent post-tonic reduction after gutturals is to extinguish the breath on accented syllable by elongating its vowel, and produce a soft stop effect. Ashkenazic similarly extinguishes open accented syllables, and closes them with iod for soft stop (davAr – dAvar – dOivor).

No language, not even Hebrew cognates, has, or ever had dagesh kal. The Hebrew could not be the only language that has it. Dagesh kal quickly dropped out in Modern Hebrew, and could not survive for millennium to be heard by the Masoretes.

Presuming the Masoretes did not invent dagesh kal—they were too honest, e.g., marked the words they did not understand instead of changing them—they could hear it in a single phonological environment we have no records of in other languages, in chanting. Dagesh kal is a cantillation mark.

The Masoretes introduced dagesh kal to break the words into syllables of comparable weight (hi-tlAb.besh – hit.lab.besh; ni-chnas – nich.nas; ca-tav.ta) which are easy to chant. Hence no dagesh in possessive nouns (d’va-rchA or d’va-r’cha is convenient to sing; unequally weighted *d’var.cA or Secunda’s form devarec are not).

Singers are familiar with the problem of clear pronunciation of consonantal clusters in accented syllables: the first consonant blurs (ni-zchAr – ni-‘chAr; hi-thcAtev – hi-‘cAtev). Singers’ commonly introduce vocal pause inside clusters to re-syllabify the words (ni-zchar – niz’car). Silent (dagesh kal) and vocal (schwa) stops prevent jamming of consonants in clusters in accented or uptone syllables.

Blurring of consonants (schwa, kal) and vowels (hazak) is similar. Only CV syllables are distinct.[2] Consonants not followed by vocal sounds blur (ni-zchar – ni-‘char, but ni-z’-char), and vowels not preceded by consonants blur (hi-thlAb-esh – hi-thlAb-‘sh, but hith.lAb.besh).

Singers open the closed syllables with vocal sound (niz.car – ni-z’.car). The Masoretes marked that vocal sound with schwa. Intraconsonantal schwa is quickly lost in speech in any language. The Masoretes could only hear it in chanting.

Artificial syllabification added secondary accent, and changed the accent pattern, ni-zchAr – nIz.cAr.

Dagesh kal and schwa before it are chanting devices. The Masoretes did not write grammar, but chanting guide. Even the long/ short/ ultra-short vowels are distinctive only in chanting, not in speech. The vowel length distinction does not exist in LXX, Secunda, or poetic transliterations.

The Masoretes were concerned with phonetics, not grammar. They introduced blank space where heard, not where grammatically proper. They did not space prepositions which are pronounced together with the relevant words.


The Masoretes introduced dagesh kal in word-initial consonants to avoid word concatenation (catav davar – cata-vdavar – cata-‘davar) and provide clear pronunciation. Preceded with stop, words became pronounced each on a new breath. Word-initial syllable became uptone. Word-initial consonant in uptone syllable after a stop loses aspiration.

Concatenation only works forward: the word-final consonant is attached to the word-initial (cata-vdavar). Concatenation happens because unvocalized (word-final, catav) consonants are hardly pronounceable. Even a trace of vocalization available to them in clusters (cata-vdavar) allows for easier pronunciation.

Stop naturally occurs in the beginning of phrases, and so the Masoretes put dagesh kal in phrase-initial words.

The words are joined by maqef when they form a single semantic unit. They are properly sung on a single breath, and the Masoretes put no stop between such words.

Clusters blur in heavy closed syllables, but not in the light open syllables. The Masoretes saw no need for dagesh kal in noun suffixes. In d’va-rchA, cluster in the final accented syllable is clearly pronounced.

The haial form shows post-tonic gemination (dAbbar).

Segholate nouns with word-initial accent did not geminate like haial, and their post-tonic vowel reduced (cAlav – cAl’v – cAlv – cal.b).

Gemination after the definite article ha- is post-tonic (hAd.davAr). That holds whether the preposition is ha- or a-. Preposition ha- is semi-accented for syntactical reasons, and preposition a- also to avoid the reduction (adavAr – ‘davAr).

Gemination after the preposition mi is similarly post-tonic.

Distinct vocal schwa after the prepositions c, b, l breaks clusters and makes word-initial stop (dagesh kal) unnecessary. Semi-accented schwa, unlike vowel, causes no gemination.

Dagesh kal in c, b, and natural stop in l prevent concatenation (nicna-sbeheder – nicnas.beheder).

Gutturals were silent by the time of the Masoretes, and functioned like soft stop, similarly to iod (korE’). Gutturals caused no concatenation, thus no need for word-initial stop after gutturals (korE’ bakar versus nata-nbakar).

Constructus plural (citvei) have no interconsonantal stop (dagesh kal) because the first word of constructus is virtually unaccented (ci-tvEi, but cit-vei-gdo-lIm).

No dagesh kal in dvash nouns because they are conveniently weighted for singing (ccvc), and re-syllabifying them (*d.bash) would make them harder to chant.

Gemination in gdullA form was originally post-tonic (gadOl - gadOllah – g’dullAh).

Gemination in piel relates to its originally command-like pronunciation (hence the sense of intense action). Bi-syllable commands are pronounced with strong emphasis on the second syllable, and so was piel (davAr – da-vA:r! – d’b.bEr! - dib.bEr!).

Dagesh kal in the future tense of piel is an interconsonantal stop: d’.ber! (di.ber) – td’.ber – t.d’.ber (a stop added to break the cluster) – t’.d’.bEr (non-emphatically pronounced t’.da.ber). Contrast paal: catav – tcatav – tactav (transposition) – t(e)chtov – tichtov. Piel, unlike paal, lacked a vowel for transposition; the first vocal sound in piel was merely vocal schwa which later elongated to [a].

Dagesh kal in the future tense of niphal is an interconsonantal stop left from the original form tncanes – tin.canes.

Note the absence of dagesh hazak in the second root letter of the future tense of niphal, tin.zacher, as opposed to tit.la.bbesh. Dagesh hazak was fixed as morphological device already in the past tense, and retained when the future tense was formed.

LXX sometimes has dagesh hazak where the Masoretes have kal (Cal.neh – Halanni; Riv.kah – Rebekka; Japh.neh – Japunneh). Greeks geminated post-tonic consonant after open syllable. The Greeks could not pronounce certain clusters (ln, bk, pn) and broke them with epenthesis which caused stress shift according to the Greek conventions, and the post-tonic gemination. When the Greek rules did not cause the stress shift (Heshbon – EsebOn, the first vowel is open and very short), there is no gemination.





[1] Accented vowels pull the adjacent sounds on both sides. Strong accent virtually makes the entire word a single syllable. Both vowels and consonants are jammed and blurred. Stop syllabifies the word, “walls” the accented syllable, and makes other syllables unaffected by the accent pull.

[2] Scholastic habit of reading some consonants as VC (en, em, el) is unwarranted, and is disregarded by colloquial speakers and mediocre students. CV (ne, me, le) is more distinct.