When did the consonant U (i.e., V) begin to be pronounced as the fricative [v] instead of [w]?

Latin Language Asked by Nathaniel is protesting on August 25, 2021

It’s well established that the consonantal u (or v) was pronounced as [w] in Classical Latin (i.e., w as in wine). Of course, Romance languages developed voiced fricatives out of this u-consonant, like the bilabial [β] (Spanish) or the labiodental [v] (Portuguese and French; v as in English vine).

Did this shift begin in Latin? If so, when? Allen and Greenough seem to suggest that it may have existed even in the era of Classical Latin:

The ordinary English sounds of j and v did not exist in classical Latin, but consonant u perhaps approached English v in the pronunciation of some persons. [§5]

If this is correct, who were these “some persons” who began pronouncing the u-consonant as a fricative?

3 Answers

There is indeed evidence for the u-consonant being pronounced as a voiced fricative during the Classical period, even as early as the middle of the 1st century. A wax tablet dated to AD 39 records a transaction by merchant Gaius Nouius Eunus, about which Clackson and Horrocks write:

Eunus’s text provides us with one of the earliest examples of the confusion of b and the consonantal u attested in Latin, in his writing Iobe for Iouem and dibi for diui, where the written b probably represents a bilabial fricative [β]. [The Blackwell History of the Latin Language, page 242]

So why, then, is there such a strong consensus that the u-consonant was pronounced as [w] in Classical Latin? It is because this shift toward a voiced fricative was largely limited to the Vulgar Latin of the lower classes:

The confusion between b and consonantal u is a feature of many sub-elite documents after this date

Spelling mistakes of this kind were common in the first several centuries AD:

spelling confusion between b and u (usually writing b instead of u, rather than the other way round) was common in word-initial position and after a consonant as well. [Herman, Vulgar Latin, 45–46]

W. Sydney Allen cites a 2nd century description of this shift, in the writings of Velius Longus, in which "the sound is specifically referred to in terms of friction" (K. vii, 58: 'cum aliqua adspiratione'; Vox Latina, 41).

The transition continued over subsequent centuries, and though the [w] pronunciation was apparently not completely wiped out by the 5th century (cf. Consentius, K. v, 395):

the fricative pronunciation was so general that Priscian has to give rules about when to write u and when b (K. iii, 465) [Allen, 41]


Thus, the earliest evidence we have for this shift in the pronunciation of the consonantal u toward a voiced fricative more similar to the English [v] appears in the 1st century, making inroads first in the lower classes and becoming widespread by the 5th century.

Correct answer by Nathaniel is protesting on August 25, 2021

As to digamma it would seem easier for a "w"to disappear than a "v". Based on my extensive Hebrew studies, it is virtually universally acknowledged that the Hebrew "vav" or"waw" was originally "w".A few Hebrew ethnic traditions preserve it. The majority of European Jews, who used Hebrew only for prayers and for textual study- not speech- pronounced it "v", presumably because their vernacular (first and ordinarily spoken language) did not contain /w/. Witness the attempt of someone learning English as a second language. They typically have great difficulty in pronouncing /w/ and it often is realized as/v/. Even when European Jews migrated to English speaking countries, they stuck with the tradition of /v/. Modern (Israeli) Hebrew, established by Europeans who had no /w/, retained /v/ for "vav" notwithstanding their attempt to establish a "sephardic" (non-European) Hebrew pronunciation which they did superficially. Thus the /v/ realization for /vav/ has remained standard in popular descriptions and transcriptions of Hebrew.

Answered by marc berlove on August 25, 2021

§ 1. Latin, Greek, and Hebrew

I would argue that the consonant V by itself was never pronounced as a W, and that something near to the W sound only occurred depending on the position of the letter within a word, of which V [ U ] would then act as a semi-vowel.

Speaking of consonantal V, Walter Blair says:

...the evidence which we have to adduce points to a normal sound for this consonant, which is a medium between English v and w. It is soft like w when, after s, g, and q, in the same syllable, it is followed by a vowel; hard like v in other situations. Thus, suetus, sanguis, quisquam, sounded swetus, sangwis, qwisqwam, and seruus, uulgus, servus, vulgus.

Blair, Walter. Latin Pronunciation: An Inquriy into the Proper Sounds of the Latin Language During the Classical Period. New York and Chicago: A. S. Barnes & Company, 1873. Page 40.

This is comparable to the English words queen and suite, which are pronounced “qween” and “sweet”, respectively.

Blair also points out how the Roman grammarians state that V as a consonant is quite different from V as a vowel:

...Nigidius (ap. Gell. XIX, 14, 6): “V in Valerius, Volusius, etc., is not a vowel at all.” Quintilian (I, 7, 26) says that the writing ceruum with the sign which belongs to the vowel u, does not represent the sound that is heard.


Blair goes on for several pages, pointing out how the Roman grammarians compare V directly to the Greek digamma, and then concludes:

In view of all that has been said, therefore, we infer, that in Latin the consonant V sounded like English V when it began a word or a syllable (unless in combination with preceding s, g, or q). When, after s, g, and q, it began a syllable, the intonation of the vowel u was partially supplied to it, yielding a sound like English w.

Ibid. Page 48.

The Latin consonantal V corresponded to the Greek digamma. The Greek alphabet descends from the Hebrew / Phoenician, with the ancient Greek digamma Ϝ corresponding directly with the Hebrew Vav:


Though the proponents of the W sound will argue otherwise, evidence for the pronunciation of the digamma points more to the sound of the English V:

In addition to the smooth and aspirated breathings, the ancient [Greek] language had another, which remained longest among the Æolians. This is most commonly called, from the appearance of the character Ϝ, used to denote it, Digamma, that is a double Γ [Gamma]. It was a true consonant, and appears to have had the force of f or v. [...] The whole doctrine, however, of the Digamma, for want of literary monuments remaining from the period when it was most in use, is exceedingly obscure.

Buttman, Philip[p]. Greek Grammar For The Use Of Schools. Second Edition. Translated by Edward Everett. Boston: Cummings, Hillard, and Company, 1826. Page 328.

The sounds that the letters F and V make are very close to each other. Say the following syllables and notice how your lips and mouth assume almost the same positions for each:

  • fa, va
  • fe, ve
  • fi, vi
  • fo, vo
  • fu, vu

Philip Buttman's son, Alexander Buttman, who revised and enlarged his father's work, eschews the “f or v” option and sides solely with a V sound for the digamma:

Note 3. Along with these two breathings the earliest language had still another aspirate, which was longest retained by the Æolians. This is commonly called Digamma, from its shape Ϝ, i. e. a double Γ; [...] It was strictly a real consonant with the sound of v....

Buttman, Philip, and Alexander Buttman. Greek Grammar For The Use Of Schools And Universities. Translated by Edward Robinson. New York: Harper & Brothers, 1872. Page 11.

See also Appendix B of the aforementioned work starting on page 463, where he talks about the history of the Greek alphabet, how it originally descended from the Hebrew / Phoenician, and how the digamma's ancient name was Ϝαυ (which is the same as the Hebrew ו Vau—Vau being an alternate spelling of Vav), and was later known as Βαυ. That the digamma was later called Βαυ shows the close relation of the V and B sounds. But more on this in a moment.

Frances Ellen Lord addresses the Cicero anecdote commonly used to promote the W sound of the Latin V:

[Dic. de Div. XL. 84.] Dum M. Crassus exercitum Brundisii imponeret, quidam in portu caricas Cauno advectas vendens “Cauneas!” clamitabat. Dicamus, si placet, monitum ab eo Crassum caveret ne iret, non fuisse periturum si omini paruisset.

[When Marcus Crassus was embarking his army at Brundisium a man who was selling Caunian figs at the harbour, repeatedly cried out 'Cauneas, Cauneas.'231 Let us say, if you will, that this was a warning to Crassus to bid him 'Beware of going,' and that if he had obeyed the omen he would not have perished.

231 i.e. "Caunian figs," but might be heard as cave ne eas. This illustration of the identity of sound between cavneas, i.e. cave ne eas, and cauneas has been the subject of some interesting discussion in Latin phonetics. Cf. Moser, Div., ad loc.

(English translation by Bill Thayer)]

Now when we remember that Caunos, whence these particular figs came, was a Greek town; that the fig-seller was very likely a Greek himself (Brundisium being a Greek port so to speak), but at any rate probably pronounced the name as it was doubtless always heard; and that u in such a connection is at present pronounced like our f or v,[*] and we know of no time when it was pronounced like our u, it is difficult to avoid the conclusion that the fig-seller was crying “Cafneas!”—a sound far more suggestive of Cave-ne-eas! than “Cauneas!” of Cawe ne eas!

But beyond the testimony, direct or indirect, of grammarians and classic writers, an argument against the w sound appears in the fact that this sound is not found in Greek (from which the vau [digamma] is borrowed), nor in Italian or kindred Romance languages.

Lord, Frances Ellen. The Roman Pronunciation of Latin: Why We Use It and How to Use It. Boston: Ginn & Company, 1894. Page 35.

[* The modern Greek diphthong alpha upsilon αυ has an “af” or “av” sound, depending on which letter follows the diphthong. See any modern Greek grammar about this, such as Essential Modern Greek Grammar by Douglas Q. Adams, page 7 (if you get a message saying the page is unavailable for viewing, reload the page and it should display).]

Additionally the Hebrew words גַּו gav and גַּב gab (transliterated as “gab”, but pronounced somewhat softer as to sound more like ‘gav’) are both used to mean “back” in several places (“gav” 1 Kings 14:9 / Nehemiah 9:26 / Ezekiel 23:35; “gab” Psalm 129:3 / Ezekiel 10:12). The letter ב without a dagesh, or dot (such as בּ), having a somewhat softer pronunciation that tends sounds more like V instead of its normal B sound (this is a simplified explanation).

And as Gesenius shows, they are “of the same sense” as each other:


Gesenius, Wilheim. Gesenius's Hebrew and Chaldee Lexicon to the Old Testament Scriptures. Translated by Samuel Prideaux Tregelles. New York: John Wiley & Sons; London: Samuel Bagster & Sons, 1893.

This is very strong evidence for the Hebrew ו vav, the Greek Ϝ digamma, and the Latin consonantal V having the English V sound.

§ 2. Sanskrit

The interchange of B and V occurs in Latin as well, such as ferveo, ferbui and bovile, bubile. This close relation of B and V can also seen in Sanskrit, with even the shape of the letters closely resembling each other:

ब Ba
व Va

While the orthography of Sanskrit bears no resemblance to Latin, Sanskrit and Latin are actually very similar in many ways, so much so that Sir William Jones said:

The Sanscrit language, whatever be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either; yet bearing to both of them a stronger affinity, both in the roots of verbs, and in the forms of grammar, than could possibly have been produced by accident; so strong, indeed, that no philologer could examine them all three without believing them to have sprung from some common source, which, perhaps, no longer exists.

Jones, William, Sir. Discourses Delivered Before the Asiatic Society: and Miscelleneous Papers, on the Religion, Poetry, Literature, etc. of the Nations of India. Edited by James Elmes. London, 1824. Page 28.

Here are a few examples of similar words (note that the “a” in the words below have the sound of ‘a’ as in “father”):

  • Sanskrit: देव “devá”, दैव “daíva or daivá1 ; Latin: “divus”, “divinus” ; English: “divine” (Greek: δῖος “dios” from the more ancient δῖϝος “divos”2)

  • Sanskrit: पितृ “pitṛí3 ; Latin: “pater” ; English: “father” (Greek: πατήρ “pater”)

  • Sanskrit: मातृ “mātṛí3 ; Latin: “mater” ; English: “mother” (Greek: μήτηρ “meter”)

  • Sanskrit: सामि “sāmí” ; Latin: “semi”, “semis” ; English: “semi” or “half”, e.g. semicircle

  • Sanskrit: सर्प “sarpá” ; Latin: “serpens” ; English: “serpent”

  • Sanskrit: नव “náva” ; Latin: “novus” ; English: “new” (Greek: νέος “neos” from the more ancient νέϝος “nevos”4)

  • Sanskrit: नस् “nás or nās” (for nās see “नास् 2. nā́s5) ; Latin: “nasus” ; English: “nose”

  • Sanskrit: दम् “Dám”, also “Dáma” ( दम ) ; Latin: “domus6 ; English: “a house” (Greek: δόμος “domos”)

  • Sanskrit: नौ “naú”, also नाव “nāva” ; Latin: “navis7 ; English: “a ship (or boat)” (Greek: ναῦς “naus”)

  • Sanskrit: जानु “jā́nu” ; Latin: “genu” ; English: “knee” (Greek: γόνυ “gonu”)

  • Sanskrit: वाच् “vā́c” ; Latin: “vox” ; English: “voice, sound, speech”

1 “dai-” sounds like dye, and rhymes with sky and why.

2 Note that δῖος “dios” originally had the digamma as δῖϝος, which with the digamma sounding like V would be “divos”.

3 “pitṛí” and “mātṛí” will also be seen as “pitṛ” and “mātṛ”, depending on who is transliterating and how strictly you are abiding by the International Alphabet of Sanskrit Transliteration (IAST) scheme. See the second paragraph on page xxix for his reasons for transliterating this way.

4 Note that νέος “neos” originally had the digamma as νέϝος, which with the digamma sounding like V would be “nevos”.

5 नास् “nās” will sometimes be seen more loosely transliterated as “naas” to signify the slightly longer ‘a’ sound (as in “father”), as compared to the very short ‘a’ in नस् “nas” (which sounds almost like the vowel sound in “Tom” if said quickly).

6 Other words such as domestic and domicile come from this.

7 Other words such as naval and navy come from this. Compare also the Latin words “navigo” and “navigatio”; “nauticus” and “nauta”, and also the English word “nautical”.

And there are many more.

The Sanskrit letter व Va is very similar to the Latin V, being pronounced as the English V when it begins a syllable unless it comes directly after another consonant in the same syllable with no vowel between them—called a conjunct consonant—where it is then pronounced more like the English W. Thus for ऋग्वेद “Rigveda”, the व् v of वेद veda is pronounced as V since it begins the second syllable. Note that Sanskrit letters inherently have a short ‘a’ sound after a consonant (‘a’ as in father), so that if you only want to specify the consonant letter itself, a diagonal line, called Virama, is written underneath the letter. Thus व Va and व् V :

When व् v is the last member of a conjunct consonant it is pronounced like w, as द्वार[*] is pronounced dwára; but not after r, as सर्व sarva.

Williams, Monier. A Practical Grammar of the Sanskrit Language, Arranged with Reference to the Classical Languages of Europe, for the use of English Students. Fourth Edition. Oxford: at the Clarendon Press, 1877. Page 13.

[* This word contains a conjunct consonant and is written in a special way, and will not display as printed in the book unless your fonts include proper ligature support, such as Nirmala UI or Noto Sans Devanagari. See section 5 beginning on page 4 to learn more about conjunct consonants.]

Another example of this is स्वद् “svad” or स्वाद् “svād”, pronounced as “swad” and meaning

  • to taste well, be sweet or pleasant to;
  • to taste with pleasure, enjoy, like, delight in;
  • to make palatable, season;
  • to make sweet or pleasant or agreeable;
  • to be pleasant or wholesome;
  • to propitiate, conciliate

Compare the Latin “suavis”, meaning ‘sweet, pleasant, agreeable, grateful, delightful’ or otherwise pleasant to the senses in one or more ways, pronounced as “swavis” and being originally written as “svavis”. See also “suadeo”, ‘to persuade’, pronounced as “swadeo” and originally written as “svadeo”; and “Suada” / “Suadela”, the goddess of persuasion (persuading you with her sweet words and pleasing appearance), pronounced “Swada / Swadela” and originally written as “Svada / Svadela”.

The Latin root “vid”, which helps form Latin words such as “video”, meaning ‘to see, discern, perceive’, is linked with the Sanskrit root विद् “vid”, which also means ‘to know, understand, perceive, learn, (etc.)’. Both of these are linked with the Greek root ἰδ “id”, which originally had the digamma and was ϝιδ, which with the digamma sounding like V would be “vid”. From these come the English “wit”, which is the later sound. You can see this to a certain extent in Hindi as well, where the “व Va” is often pronounced as “Wa”, though this is regional and there are still Hindi speakers that pronounce it as “Va”. Hindi is a descendant of Sanskrit, and the later sound of the Sanskrit “व Va” has in many areas become “Wa”.

In his Vedic Grammar, Arthur Macdonell brings out how B is interchanged with V in some very early Vedic writings:

45. [...] a. [...] 3. In a few examples it takes the place of or interchanges with v; thus páḍbīśa- (RV.), beside páḍvīśa- (VS.); bāṇá- beside vāṇá- ‘arrow’; -blīna- (AV.) ‘crushed’, beside -vlīna_- (B.). [...]

Macdonell, A. A. Vedic Grammar. Strassburg, 1910. Page 36.

The Vedic period was before the classical Sanskrit period, and in this period the occasional interchange of B and V again shows the close relation between the two sounds in ancient times, similar to the Hebrew גַּו gav and גַּב gab examples above.

§ 3. Vox Graeca and Vox Latina

William Sidney Allen, who promotes the W sound for the digamma in his Vox Graeca, says:

[w] (Ϝ, ‘digamma’). In early Greek this sound existed as an independent phoneme; in the Cyprian and Mycenaean (Linear B) syllabaries there are signs for wa, we, wi, wo, and most of the dialects show epigraphic evidence in the form of a special letter, of which the most common shape is of the type Ϝ. This was a differentiated form of the Semitic ‘waw’, [...]

Allen, W. Sidney. Vox Graeca. Third Edition. Cambridge University Press, 1987. Page 47.

This is what all the proponents of the W sound inherently rely on: the “Semitic ‘waw’”. But as shown above in § 1, there is strong evidence it is a vav and not a waw. William Sidney Allen appeals to Sanskrit on page 48 of his Vox Graeca for additional support of the W sound, but only does so with the Sanskrit sva- combination, which as shown above in § 2, only has a W sound because the व Va is part of a conjunct consonant. The normal V sound of the व Va is weakened into a W sound by an adjacent strong consonant.

In his Vox Latina, William Sidney Allen references Nigidius Figulus in support of the W sound for the Latin V:

In the first century B.C. Nigidius Figulus (†Gellius, ⅹ, 4, 4) evidently referred to the consonant sound, like that of the vowel, in terms of lip-protrusion, which can only indicate a bilabial, semivocalic articulation (in a discussion of the origins of language, he points out that in the words tu and uos the lips are protruded in the direction of the person addressed, whereas this is not the case in ego and nos).

Allen, W. Sidney. Vox Latina. Second Edition. Cambridge University Press, 1978. Page 41.

This does not prove a W sound for “uos” (as he spells it, instead of the more proper “vos”), it only proves the long ‘o’ sound for the vowel, as in snow, no, and so. Here is the actual quote:

“When we say vos, or ‘you,’” says Nigidius, “we make a movement of the mouth suitable to the meaning of the word; for we gradually protrude the tips of our lips and direct the impulse of the breath towards those with whom we are speaking. But on the other hand, when we say nos, or ‘us,’ we do not pronounce the word with a powerful forward impulse of the voice, nor with the lips protruded, but we restrain our breath and our lips, so to speak, within ourselves. The same thing happens in the words tu or ‘thou,’ ego or ‘I,’ tibi ‘to thee,’ and mihi ‘to me.’ For just as when we assent or dissent, a movement of the head or eyes corresponds with the nature of the expression, so too in the pronunciation of these words there is a kind of natural gesture made with the mouth and breath. The same principle that we have noted in our own speech applies also to Greek words.”

Aurlus Gellius. Attic Nights. Book X, 4, 4.

English Translation by John C. Rolfe in his: The Attic Nights of Aurlus Gellius. Volume 2. 1927. Reprint, Harvard University Press, 1984. Page 229.

Pronounce “vos” with a long ‘o’ sound and notice how your lips still protrude, whether you pronounce it as ‘vos’ or ‘wos’. So this particular point can't be used to prove a V or W sound, as ‘vos’ and ‘wos’ are too close to each other in their pronunciation. Also, William Sidney Allen leaves out the part where Nigidius refers to “a powerful forward impulse of the voice,” which is a description that would much more likely be used to illustrate a long ‘o’ sound. Say the following words and notice how the long ‘o’ has a much more powerful forward impulse of the voice than does the short ‘o’:

  • vōs & wōs (long ‘o’) , rhymes with boast and coast
  • vŏs & wŏs (short ‘o’), rhymes with floss and toss

William Sidney Allen appeals to Sanskrit for many other Latin letters in his Vox Latina, but when it comes to the Latin consonant V, he is completely silent on the matter. He even tries to compare the Indo-European root “wid-” (as he spells it, instead of “vid”) of the Latin “uideo” (as he spells it, instead of “video”) to the English “wit” in the first paragraph of page 41. But this ignores the Sanskrit root विद् “vid” of which I have already talked about above in § 2, and there is no evidence that the Sanskrit व Va was ever pronounced as anything other than with a V sound.

Speaking of confusion between V and B in Latin, William Sidney Allen makes the following statement on page 42:

However, there is no evidence for any such development before the first century A.D., and the [w] value of consonantal u must be assumed for the classical period.

Allen, W. Sidney. Vox Latina. Second Edition. Cambridge University Press, 1978. Page 42.

This statement is actually incorrect as can be seen in the writings of Varro and Cicero, who both lived in the first century B.C.:


Varro, Marcus Terentius. On the Latin Language. English Translation by Ronald G. Kent. Volume 1. London: William Heinemann LTD; Cambridge, Massachusetts: Harvard University Press, 1938. Page 94.


Cicero, Marcus Tullius. De Legibus Libri Tres. Edidit C. F. W. Mueller. Lipsiae In Aedibus B. G. Teubneri, 1915. Page 426.

Varro spells it as “verbex” (meaning “a castrated ram”) while Cicero spells it as “vervex” (“vervecibus” being a declined plural form of “vervex”). This again demonstrates the close relation between V and B.

And even Polybius, who lived in the second century B.C., spells the Latin name “Livius” with a β multiple times in his Histories:

Page 512-513: Λίβιον

Page 514-515: Λιβίῳ

Page 516-517: Λίβιον 3 times

Polybius. The Histories, Book VIII. English Translation by W. R. Patton. Volume 3. 1923. Reprint, London: William Heinemann LTD; Cambridge, Massachusetts: Harvard University Press, 1966.

Page 228-229: Λίβιος

Polybius. The Histories, Book XI. English Translation by W. R. Patton. Volume 4. 1925. Reprint, London: William Heinemann LTD; Cambridge, Massachusetts: Harvard University Press, 1968.

There is also an inscription from Delos, dated at 179 B.C., that names two Romans. Line 86 mentions Γαίου Λιβίου Ῥωμαίου ἀνάθεμα “an offering of the Roman Gaius Livius”, spelling Livius with a β:

Gaius Livius

ID 442, face B, line 86.

(Note that I have activated the word wrap feature available on the webpage for this picture.)

And line 130 mentions ἀνάθεμα Βιβίου Ῥωμαίου “an offering of the Roman Vivius”, spelling Vivius with two β's.


ID 442, face B, line 130.

(Note that I have activated the word wrap feature available on the webpage for this picture.)

(Due to character limits, this answer is continued here.)

Answered by Bʀɪᴀɴ on August 25, 2021

Add your own answers!

Related Questions

Ask a Question

Get help from others!

© 2022 All rights reserved. Sites we Love: PCI Database, MenuIva, UKBizDB, Menu Kuliner, Sharing RPP, SolveDir