Skip to main content
Loading Explore guides…
Romance Language Sound Evolution · Acedra
Home
Explore
Languages
Romance Language Sound Evolution
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
60 segments
Copy
Hide video
0:00
Hey guys, what’s up. If you remember the introduction to the video where I talked about the evolution of sounds in the Germanic languages, you knew this was coming eventually. That’s right, since I’ve now learned Romanian, that means I’ve fulfilled my requirements to make a video about the sounds of the Romance languages! Well, technically I didn’t need to learn Catalan for this, but I have learned it, so nothing’s stopping me from using it. Just like last time,
0:17
I’m gonna start with the history of the family and its branches, then I’ll explain the phonology of the starting language, and then I’ll go through the sounds and how they changed. Oh ya, and about the starting language, I’m gonna do something that’s possibly a bit controversial here- I’m gonna use Classical Latin as the starting point for the sound evolution. There are two main reasons for this- first, there are some sound changes
0:33
that had already been taking place in various different ways by the time of Vulgar Latin, which makes it easier to just start earlier, in the more unified Classical Latin. But the bigger reason is that I’m not gonna squander the opportunity to use a well-documented, attested language as the starting point in a language family evolution video. So with that out of the way, let’s go! So first off, we have the history section. The Romance languages are a branch of the
0:52
Indo-European language family. Specifically, they’re the languages descended from Latin, but what exactly was the timeline of that? We should probably start in 509 BC, with the beginning of the Roman Republic in Rome, because that’s when the Romans started expanding. Through a timeline of getting sacked and sacking other places, by 218 BC they’d captured most of what’s now Italy, and then just kept going on to eventually the rest of the Mediterranean and then
1:11
some. During the Roman Republic, they spoke Old Latin, and kept speaking it until near the end, about 100 BC. At this point, the era of Classical Latin is considered to have started, more of the expansion happened, and later in the century in 27 BC, the Roman Republic became the Roman Empire, followed by even more expansion, as the peak territorial extent of the Roman Empire was in 117 AD. Anyway, Classical Latin was spoken until sometime in the 200s AD, at which point the
1:34
general populace started speaking Vulgar Latin, a more evolved form of Latin, but here’s where it gets confusing. The elites of this time preferred instead to write in Late Latin, somewhat of an extension of Classical Latin, even though they very much spoke Vulgar Latin natively. After the fall of the Roman Empire in 476 AD, the language was much freer to evolve every which way. But before we talk about that evolution, I need to point something out about Vulgar Latin.
1:54
It was a dialect continuum across southern Europe. Therefore, the “branches” I’m about to talk about are fluid and somewhat arbitrary. So starting from the west, there’s Ibero-Romance, spoken on the Iberian Peninsula. It got a lot of Arabic influence due to Muslims having control over the region in the early middle ages, but then the Catholics took it back in the 1100s, and the Ibero-Romance languages today are Portuguese, Spanish, and several other
2:14
regional languages. Then there’s Gallo-Romance, named after the region of Gaul. Especially in the north of the region, they got more Germanic influence due to the Franks, and, you guessed it, the modern Gallo-Romance languages are French, Catalan, and some other regional languages. By the way, Ibero-Romance and Gallo-Romance are often combined into Western Romance by linguists. So then for the east, we have Italo-Dalmatian (which I may call “Italo-Romance” sometimes
2:35
to clarify that it’s a Romance sub-branch), and that branch has, of course, Italian and most of the other Italian dialects, although not Sardinian, which is considered to make up its own branch. And finally, there’s Eastern Romance, which has a lot more Slavic influence, and the main language from that branch is Romanian, of course, with other minority languages. Alright so now we’ll just look at the phonology of Classical Latin, to see what we’re starting from.
2:54
In terms of consonant phonemes, Classical Latin had 17 of them, with eight plosives, [p] [t] [k] [kʷ] [b] [d] [g] and [gʷ], three fricatives, [f] [s] and [h], two nasals, [m] and [n], three approximants, [ɫ] [w] [j], and a trill, [r]. In addition, [z] was sometimes present in Greek loanwords. As far as allophones, /ɫ/ was dark by default, but became light [l] when either before [i] or geminated. Speaking of which, yes, Latin had consonant
3:17
gemination, where the consonant was held twice as long. This will be a very important point at various times in this video. There was also just, the standard allophone of /n/ becoming [ŋ] before velar consonants. Now for the vowels, there was also, you guessed it, vowel length distinction, between short and long. There were five basic vowels, ten total with the lengths, although, depending on who you ask, there may have been quality differences between the lengths,
3:36
giving us [a] [aː] [ɛ] [eː] [ɪ] [iː] [ɔ] [oː] [ʊ] and [uː]. Latin also had a few diphthongs, being these, and there were two places where vowels would become nasal: Before M at the end of a word, and before a nasal consonant that was itself before a fricative. As for stress rules, they’re a bit complicated. But basically, the stress was normally on the penultimate syllable, unless the word was three syllables or longer, and the penultimate syllable was short, in
3:58
which case the antepenultimate, or third-to-last syllable, gets the stress. To close this section, normally I wouldn’t have to do this, but I do here since the origin language was written- I need to go over parts of Latin orthography that may be confusing to modern readers. Basically, here’s the original Latin alphabet, and you’ll notice it only has 20 letters. Ok well 23 if you count the
4:14
letters specifically used for Greek loanwords, K, Y, and Z. First up is the reason why some letters you may know from modern times aren’t there- I and “U”, which looks like modern V, could both be either the vowels /i/ and /u/, or the consonants /j/ and /w/, and you’d know based on where it was in the word. The letters J and U were added later when sound changes made them necessary in various
4:33
languages, while W isn’t even used in native words by any modern Romance language. As for other notes, the QU digraph was still used to make the [kʷ] sound, X did, in fact, represent /ks/, and H was pronounced, as /h/. Finally, C and G were always pronounced /k/ and /g/- there was no such thing as their “soft” variants, and spoiler alert, that will get addressed today. So now it’s time to see what happened to the consonants, starting with the nasals,
4:54
liquids, and trills- or, as they’re collectively known, sonorants. I’ll start with the trill, /r/, because it introduces a lot of necessary concepts. So first off, and probably the one y’all are the most familiar with, in French it became uvular /ʁ/. Easy! But the rest of the changes to Latin /r/ are a bit less direct, and in order to set the scene, I’ll talk about Spanish. Technically
5:11
nothing changed with the /r/ in Spanish, BUT other stuff changed around it. Basically, even back in Latin, when between vowels and ungeminated, /r/ was pronounced more as a tap, /ɾ/. But then, Spanish lost consonant gemination as a feature, but this tap vs. trill distinction remained, which made them phonemically different in Spanish. In fact, the only notable modern Romance language that still has consonant gemination is Italian. So for the other languages,
5:32
Catalan did basically the same thing as Spanish, but they also dropped the /r/ at the ends of words (outside of Valencia), giving us for example, the word aterrar, meaning “land”. Meanwhile, Portuguese actually went the extra mile and changed the trilled form into /x/ (the exact pronunciation of which varies by dialect). The /x/ form also took over at the beginnings of words, since they were naturally trilled, and also in a couple other positions only in Brazilian
5:52
Portuguese. So now for the comparative examples of this sound. First, there’s Latin terra, which became Portuguese terra, Spanish tierra, Catalan terra, Italian terra, French terre, and Romanian țară. For an example that was ungeminated, there’s Latin hōra, which became Portuguese hora, Spanish hora, Italian ora, and French heure. And since we were just talking about gemination, I’ll continue on this theme with /ɫ/ and /n/. First off,
6:14
the whole dark L thing didn’t really get passed on to any modern Romance languages. Portuguese does have a dark L, but that’s its own system. But as for the gemination, the Western Romance languages all did palatalization of geminated /l/ and /n/ to some extent, to /ʎ/ and /ɲ/ respectively. It’s most regular in Spanish and Catalan, but Portuguese, French, and even Italian all had their
6:31
part. (It should also be mentioned that in most dialects of Spanish, [ʎ] has now merged into [ʝ], in a sound shift known as yeísmo, but I happen to speak a dialect that didn’t do that.) For one example, there’s Latin stēlla, which became Spanish estrella and Catalan estrella, but also Portuguese estrela, French étoile, Italian stella, and even Romanian stea. Things do get more interesting when the Latin geminated /l/ had /i/ after it tho. In that situation,
6:53
most Romance languages are more likely to have palatalized it- for example, there was Latin allium, which became Romanian ai, Italian aglio, French ail, Catalan all, Portuguese alho, and Spanish… ajo. Wow so uhh, not that reliable of a sound shift for Spanish anymore. And for /n/, there’s also Latin annus, which became Spanish año and Catalan any, but Portuguese ano, French an, Italian anno, and Romanian an. However, Catalan also palatalized word-initial /l/, like
7:20
with Latin lupus, which became Catalan llop, but Spanish lobo, Portuguese lobo, French loup, Italian lupo, and Romanian lup. So now onto the nasals. The main thing to say about them, other than the palatalization on /n/, is that when any nasal consonant was in coda position, in both French and Portuguese, they disappeared, but made the vowel before them nasal. For example, there was Classical Latin quīnque, which became Vulgar Latin cīnque, and then Portuguese cinco and
7:43
French cinq, but Spanish cinco, Italian cinque, and Romanian cinci. For an example at the end of a word, there was the Latin accusative form leōnem, which became French lion and Portuguese leão, but Spanish león and Italian leone. So to end the sonorants, we have /w/ and /j/, neither of which are really gonna stick around. For /w/, it first shifted over to a voiced bilabial fricative, [β],
8:03
and then went one of two directions. In Spanish and Catalan, it stayed like that between vowels, but elsewhere, especially word-initially, it became [b]. And that just so happens to be the phonemic profile of /b/ in those languages, so… that’s a merger! In the other languages, [β] simply shifted over to [v]. So for example, there was Latin uentus, which became Romanian vânt, Italian vento, French vent, and Portuguese vento, but Spanish viento and Catalan vent. As for /j/,
8:28
at least word-initially, it became [x] in Spanish, [ʤ] in Italian, and [ʒ] in the other four languages. For example, there was Latin iocārī, which became Spanish jugar and Italian giocare, but then Portuguese jogar, Catalan jugar, French jouer, and Romanian juca. Meanwhile in the middles of words, it still became [ʤ] in Italian and [ʧ] in Catalan, but otherwise it pretty much stayed [j] (except in Spanish, where it slightly intensified to [ʝ], and French, where it somehow
8:51
merged into the surrounding vowels). So in Latin there was Maius, and that gives us modern Italian maggio, Catalan maig, Spanish mayo, French mai, Romanian mai, and Portuguese maio. Alright, we’ve gotten to the fricatives, and there are only three of them, so let’s do this. First off, we have [h], and uhh… it’s gone. However, it should be mentioned that Romanian does still have a [h] sound, but it’s due to an abundance of Slavic loanwords. This is much the
9:13
same situation that French had from roughly 800 to 1700 AD with the H aspiré thing, but in that case, it was due to Germanic loanwords. So for example, we have Latin hominem, which became Romanian om, Italian uomo, French homme, Catalan home, Spanish hombre, and Portuguese homem. Also, notice that the western languages actually still write an H there, but it’s silent. In Spanish tho,
9:32
the silent H could have a different origin- F. Ya, in Spanish, original Latin [f] first became [h], and then disappeared altogether, largely considered to be due to Basque influence- sometimes. The pattern isn’t 100% consistent due to various competing influences, but generally, [f] was dropped unless it was before [e], [i], or another consonant. To give an example where [f] did disappear in Spanish, there’s Latin farīna, which became Romanian făină, Italian farina,
9:55
French farine, Catalan farina, and Portuguese farinha, but Spanish harina. And for an example where Spanish kept the [f], there’s Latin flōrem, which became Romanian floare, Italian fiore, French fleur, Catalan flor, Portuguese flor, and Spanish flor. And for the final fricative, there’s [s]. Remember back when I was talking about how [r] split up phonemically in Spanish due to loss of consonant gemination? Well a similar thing happened to [s], but not in Spanish (or
10:19
Romanian, which was just generally not affected by degemination shifts). However, in Portuguese, Catalan, French, and Italian from northern Italy, [s] became voiced [z] when intervocalic and ungeminated, while staying [s] if geminated. This effectively made [z] a new phoneme in Portuguese, Catalan, and French, while it’s still an allophone in Italian due to still having gemination. So for an ungeminated example, there was Latin casa, which became Romanian casă and Spanish casa,
10:43
but Portuguese casa, Catalan casa, Italian casa, and French… chez. Ya, interesting. And for a geminated example, there was Latin tussīre, which became Italian tossire, but Portuguese tossir, Spanish toser, Catalan tossir, French tousser, and Romanian tuși. So ya, about the Romanian word, there was a general pattern of palatalization before front high vowels, like in this example with [s] becoming [ʃ]. But this will become a bit more relevant with the plosives.
11:09
So now we’re at the plosives, and we’ll pick right up where we left off with Romanian palatalization, an obviously Slavic-influenced feature. So we saw [s] becoming [ʃ], but there were also two plosives affected by this: [t] became [t͡s], and [d] became [z]. So for an example with [t], there was Latin tenēre, which became Portuguese ter, Spanish tener, Catalan tenir, French tenir, and Italian tenere, but Romanian ține. And for an example with [d], there was Latin dīcere, which
11:33
became Portuguese dizer (or dizer in Portugal), Spanish decir, Catalan dir, French dire, Italian dire, and Romanian zice. But aside from the [d] becoming [z] in Romanian, among those examples you saw many different ways that Latin [k] behaved, so this is a perfect segway into the broader fronting patterns that happened when [k] or [g] was before [e] or [i]. For [g] in those situations, it actually did the same thing that [j] did, where it became [x] in Spanish,
11:56
[ʤ] in Italian, and [ʒ] in French, Portuguese, and Catalan (or at least, outside of Valencia). The one exception is Romanian, because instead of becoming [ʒ] like [j] did, [g] became [ʤ], like in Italian. For example, there was Latin gentem, which became Spanish gente, Portuguese gente, Catalan gent, French gens, Italian gente, and Romanian gint, which is now obsolete. But it’s a bit more interesting what happened to [k] before [e] and [i]. In Italian and Romanian, it became
12:19
[ʧ], while in the western languages, it initially became [t͡s], and then split up further. In Portuguese, Catalan, and French, this sound simply merged into [s]. But in Spanish, it initially shifted over to a laminal S, made with the blade of the tongue instead of the tip, that being [s̻]. From here, in Latin American Spanish, this sound merged into regular “apical” [s] just like in the rest of the western Romance languages, but in most of Spain, it instead became a voiceless
12:40
dental fricative, [θ]. So for example, there was Latin caelum (the ae diphthong merged into [ɛ] in Vulgar Latin), and it became Romanian cer, Italian cielo, French ciel, Catalan cel, Portuguese céu, and for Spanish, in Latin America it’s pronounced cielo, and in Spain it’s pronounced cielo. But French also did something else with [k]. When [k] was before [a] in Old French, it initially became
12:59
[ʧ], and then later, [ʃ]. So there was Latin capra, which became Romanian capră, Italian capra, Portuguese cabra, Spanish cabra, and Catalan cabra, but French chèvre. But also on that word, notice the variation on the center consonant. That’s our next topic. The voiceless consonants, when between vowels (or sometimes liquids) and ungeminated, became voiced in the western Romance languages, and then in French, the [b] became [v], and the rest of them… disappeared. You just saw
13:22
the example with capra becoming chèvre, so for an example with [t], we have Latin vīta, which became Italian vita, Portuguese vida, Spanish vida, Catalan vida, and French… vie. It also led to Romanian viață through a different form in Vulgar Latin. And for an example with [kʷ], there was Latin aqua, which became Italian acqua, Portuguese água, Spanish agua, Catalan aigua, and French… eau, of course. Well, it was [ˈja.wə] in Old French, if that helps explain it. And also very
13:49
naturally, it became Romanian apă. Wait, what? Ya, as it turns out, Romanian had a sound change where the labial aspect of [kʷ] and [gʷ] completely took over, and they became [p] and [b]- at least around [a]. For an example with [gʷ], there was Latin lingua, which became Italian lingua, Portuguese língua, Spanish lengua, Catalan llengua, French langue, and Romanian… limbă. However, there is one more thing to talk about with [kʷ] and [gʷ], and that’s when they merged into [k] and
14:14
[g]. In Spanish, Portuguese, and Catalan, the merger basically just happened before [e] and [i], as this is commonly believed to have been part of a chain shift also involving the [k] to [s] thing from earlier. Meanwhile, French basically did it everywhere, and Italian did the merger almost nowhere. As for Romanian, when before [e] or [i], they merged [kʷ] and [gʷ] into [k] and [g] before
14:31
they became [ʧ] and [ʤ], because only in Romanian were Latin [kʷ] and [gʷ] affected by the fronting. So for an example of all this, there’s the Latin word quīndecim, which became Portuguese quinze, Spanish quince, Catalan quinze, French quinze, Italian quindici, and Romanian cincisprezece. So now we’ve seen all the consonants by themselves, but there are several clusters that defy expectations, so we gotta address them. First off, there’s [tj], or realistically,
14:54
any time there was the Latin cluster TI before a vowel, and it’s especially important due to one suffix in particular. Basically, this cluster behaved how you’d expect in Romanian, becoming [t͡s], but it also became [t͡s] in Italian. Now here’s a super interesting connection that I somehow didn’t realize until recently, and I feel kinda stupid about it. You know the Italian administrative region where Rome is? Ya, it’s called Lazio. You know the ancient region
15:14
that gave the Latin language its name? That’s Latium, and ya, it’s literally the same word, just evolved into Italian, and I spent most of my life thinking the namesake region of Latin didn’t survive anywhere. Anyway, this cluster also became [t͡s] in the western Romance languages, where it then followed the same trajectory as [k], with the laminal [s̻] and whatnot. So for example, there
15:31
was Latin statiōnem, which became Romanian stație, Italian stazione, French station, Catalan estació, Spanish estación, and Portuguese estação. Ok well most of those were technically loanwords but the sound shift still applies. Next up is [sk]. First off, in Spanish and French, this cluster became [s] (or [θ] in Spain Spanish- it basically just joined the other sounds we’ve talked about). But
15:50
in Portuguese, Catalan, Italian, and Romanian, it became [ʃ] instead. So there was Latin piscem, and that became Spanish pez (or pez in Spain), French poisson, Portuguese peixe, Catalan peix, Italian pesce, and Romanian pește. So the next cluster is [kt], and it kinda went all over the place. The most common change happened in Portuguese, Catalan, and French, where it became [it], while in Spanish this cluster changed even further, to [ʧ]. Meanwhile,
16:12
in Italian it became geminated [tː], and in Romanian, the [k] changed to [p]. So for example, we have Latin noctem, which became Portuguese noite (which I said in European Portuguese to avoid that confusing Brazilian allophone), Catalan nit, French nuit, Spanish noche, Italian notte, and Romanian noapte. And finally, we have [kl], [gl], [pl], and [bl]. To start simply, these clusters survived unaffected into Catalan and French. As for the rest of the shifts,
16:35
they’re highly inconsistent, mostly due to the upper classes trying to undo sound changes that happened in their languages, resulting in the sound changes applying to some words, but not others. One of the easier sound changes is that in Italian, all these clusters switched the [l] to [j], while only [kl] and [gl] did that in Romanian. Similarly in Portuguese, [pl] and [bl] became [pɾ] and [bɾ], while [kl] actually became [ʃ]. I don’t… think [gl] changed in Portuguese,
16:56
but, ya know, I can’t be sure of that. Meanwhile in Spanish, at least [kl] and [pl] became [ʎ], but [kl] did it more consistently. For an example that actually behaves well, there was Latin clāvis, which became French clé, Catalan clau, Italian chiave, Romanian cheie, Portuguese chave, and Spanish llave. For an example that behaves less well, there was Late Latin blancus, which became Spanish blanco, Catalan blanc, French blanc, Portuguese branco, Italian bianco, and… nothing
17:22
in Romanian, because this is a Germanic loanword from well after the Roman Empire. The Romanian word for “white” is alb, which does in fact come from the Classical Latin word for “white”, albus. So now we’re on to vowels, but of course, they’re way too confusing to actually go into detail on, so I’ll just briefly discuss two main themes of them. First off, I’ll specifically talk about how the length distinction evolved into Italian, as that’s the only one of the languages we’ve been
17:42
discussing that still has it. The thing is, the long vowels in Italian aren’t actually descended from the long vowels in Latin. Modern Italian vowel length is effectively an allophonic system based on whether the consonant after it is geminated or not, and modern Italian consonant gemination is descended from the Latin feature. But basically in Italian, short vowels come before geminated consonants, and long vowels come before ungeminated consonants,
18:03
so they work in tandem with each other. If you’ve watched the Swedish language overview, this may sound familiar to you, and it turns out, I’m not the only person who’s noticed this before. For example, let’s revisit the Latin word stēlla from earlier. It has both a long vowel and a geminated consonant. But it evolved into Italian stella, which still has a geminated consonant, but the vowel is now short, since it comes before the geminated consonant. The other thing about
18:24
vowels is the diphthongization that happened in Spanish and Romanian. Ya, interesting combo, right? So in both of those languages, mid vowels became opening diphthongs when they were stressed, with Spanish having the extra complication that it only happened to formerly open-mid vowels. In Spanish this largely resulted in ie and ue, while in Romanian the results were ea and oa. So for example, there was Latin porta, which became Italian porta,
18:44
Catalan porta, French porte, and Portuguese porta, but Spanish puerta and Romanian poartă. So of course at the end here, I gotta talk about new things that have arisen in the Romance languages that weren’t covered in any of the previous sections. First off, I talked about all the changes that happened to Latin [j], but the modern Romance languages largely still have [j]s of their own, so how did those develop? To demonstrate one of the most common ways this would
19:05
happen, there was the Latin pronoun egō̆, with a G in the middle. But later on, the G got dropped, leaving just those two vowels, and some languages chose to make the first one a consonant, like Spanish with yo, and Italian with io. French and Catalan interestingly also did this, but then that sound merged in with [ʒ] instead, making this point completely moot for them- they have je and jo respectively. Other than that, another thing that happened was that,
19:25
when a word began with S followed by another consonant in Latin, the western Romance languages added an epenthetic E beforehand. And then in French, the S got deleted. So for example, there was Latin scūtum, which became Romanian scut, Italian scudo, Portuguese escudo, Spanish escudo, Catalan escut, and French écu. Now here’s something interesting: While Italian didn’t do the epenthesis in the same way its western relatives did, the general idea that
19:47
S-consonant clusters can’t start a word is still felt there- namely, in terms of the articles. For masculine singular nouns, the definite article is normally il- but before the aforementioned clusters (along with some others that are considered “impure” in Italian morphophonology), the article is lo, which noticeably ends with a vowel. I can’t be 100% sure this is the same process as the E-epenthesis, but it definitely seems like more than just a coincidence. And for
20:05
the final topic of this video, I’ve got something that has probably been on a lot of y’all’s minds to this point- sounds disappearing off the ends of words. Of course, French is the language most famous for doing it, but Catalan and Romanian also did their fair share. As for how it worked, in the beginning, all three of those languages dropped [o] and [u] off the ends of words. But then, for French only, they dropped many consonants off the ends of words. And then after that,
20:27
French also reduced and dropped [a] sounds off the ends of words. This whole process will become relevant when we talk about Romance grammatical evolution in a future video. But it should be mentioned that Catalan and Romanian also did vowel reduction to their [a]s- in Catalan it’s still fully allophonic today, while in Romanian the reduced vowel has now become phonemic due to… their article system, of all things.
20:45
So that’s it, that’s a general summary of how sounds evolved in the Romance languages. Obviously I was only able to talk about the languages I’ve studied, but if you speak another Romance language, feel free to talk about these sound shifts in your language in the comments. Hope y’all liked it, thanks for watching, and I’ll see y’all next time!
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat