[Source:] Loss of Latin -v- is regular in French in some situations (compare alleger from alleviare; neige from nivea; jeune from juvenis. A different sound evolution from the Latin word yielded Italian citta, Catalan ciutat, Spanish ciudad, Portuguese cidade.

Why might the Latin -v- have been lost in French? The following discusses the influence of the Gaulish language on French. So did Gaulish cause this loss?

[Source:] One example of a substrate language is Gaulish, from the ancient Celtic people The Gauls. The Gauls lived in the modern French-speaking territory before the arrival of the Romans namely, the invasion of Julius Caesar's militia. Given the cultural, economic and political advantages that came with being a Latin speaker, the Gauls eventually abandoned their language in favor of the language brought to them by the Romans, which evolved in this region until eventually it took the form of the French language that is known today. The Gaulish speech disappeared, but remnants of its vocabulary survive in some French words (approximately 150) as well as place-names of Gaulish origin. [...] In the case of French, for example, Latin is the superstrate and Gaulish the substrate.

  • 1
    Wasn't Latin V generally more of a semi vowel like [w]? That'd be a pretty easy sound to be lost intervocalicly, especially if one of them is a vowel near [u] Aug 17, 2015 at 20:08
  • 1
    Right. Latin did not have a phonemic (nor an orthographic) distinction between the high back rounded vowel [u] and the high back rounded semivowel [w]. They were both written V; Latin had no letter U. Similar remarks apply to the high front unrounded vowel and semivowel [i] and [j], respectively; they were both written I. So V was always [w], as in Caesar's famous quote Veni vidi vici, pronounced "Waynie, weedie, weekie" as we would hear it in English. And [w] sounds come and go with the vowels they orbit.
    – jlawler
    Aug 17, 2015 at 20:14
  • 1
    It isn't true that "V was always [w]": it started out as [w] but ended up as [v] (in Romance, where it survives), presumably going through [β]. The chronology isn't precisely known, but there is inscriptional evidence for a [β] pronunciation as early as the first century CE, and it's thought that by the third century a fricative pronunciation of some kind was standard.
    – TKR
    Aug 17, 2015 at 20:46
  • 1
    @jlawler: I think it was phonemic, though marginally. Wikipedia says that u and v could be exchanged in poetry in words like SILVA and GENVA, but I'm not sure if this actually shows phonemic identity -- Romans were capable of these distinct pronunciations. And then there are a lot of words with nuV like "insinuo," but any word starting with v prefixed with "in" has nvV instead like "involvo". One of the Claudian letters, Ⅎ, was for consonantal u. The distinction between "i" and "j" was definitely phonemic, because of the existence of compound words. Aug 17, 2015 at 20:49
  • 1
    I think it's pretty clear that Latin [w]/[u] and [j]/[i] were separate phonemes, not allophones. A minimal pair for the former is volui 'I wanted' vs. volvi 'I rolled'. And it's clear that e.g. SILVA is generally disyllabic while GENVA or BELVA are trisyllabic.
    – TKR
    Aug 17, 2015 at 21:22

2 Answers 2


It's probably not a substrate effect, but a phonetically caused deletion, seeing as the sound was not lost in all contexts but only in specific conditions.

A substrate-based loss would presumably be due to Gaulish not possessing a sound equivalent to that written with Latin v, which may or may not have been the case (depending on what exactly that sound was at that time and place, and on what exact sound the symbol V in Gaulish inscriptions represented). But if so, this would presumably have caused complete loss of the phoneme rather than conditioned deletion.

In fact v is regularly lost next to round vowels: pavo:rem > peur, avunculum > oncle, iuvenis > jeune. Between other vowels it generally remains: lavare > laver, levare > lever. Loss of a rounded consonant next to a round vowel is a very common type of assimilation.

Cases like alleviare > alléger are part of a different rule, namely that v is lost before most consonants: in this case it was followed by a [j] (which is how the i would have been pronounced in this word in Late/Vulgar Latin). Similarly for civ(i)ta:tem > cité (where the second i would have been lost by syncope).


Why might the Latin -v- have been lost in French?

Why were the Latin intervocalic -v- and -b- lost in Romanian as well?

alleger from alleviare; neige from nivea; jeune from juvenis.

Notice that in all these instances the -v- is intervocalic.

So did Gaulish cause this loss?

Technically, everything is possible, but, given the fact that Romanian has a Thracian/Dacian substrate, I'd say that it's unlikely at best. Of course, it is not theoretically impossible that, for some mysterious reason, these two distant Indo-European branches (Gaulish and Dacian) might have shared some phonetic similarity insofar -v- and/or -b- are concerned, but it seems much more likelier that Latin itself is the cause, since, as far as we know, this is the only common denominator between French and Romanian.