With the possible exceptions of constructed languages, languages seem to evolve. As a real-world example, we note that Latin has evolved into Italian, French, Spanish, Portuguese, Romanian, etc.

What defines the transition between an ancestor language and a descendent language? At what point do we say "These people in the Italian peninsula are no longer speaking Latin; they are speaking Italian"? Similarly for the other Romance languages, and for the split between Spanish and Portuguese.

Carrying that further, at what point does Puertoriqueño cease to be a dialect of Spanish, and take on its own identity as a language? Similarly for Brazilian vs Portuguese, or the other various dialects of Spanish in Latin America, or even English/England vs English/Australia vs English/North America?

(Note: I'm interested in, for lack of a better term, the linguistic constraints that define the difference, rather than the politico-geographical factors that influenced the development of those constraints. For the historical developments of $ModernRomanceLanguage from Latin, or Portuguese from Spanish, I'm also curious about the when.)

  • 16
    In philosophy, this is known as "the demarcation problem." It occurs when we feel the need to draw a line in the sand between two things when there is actually a continuum between them. We run into the same issues in science with the idea of species. As far as we can tell, nature has no concept of "species." Everything is on a continuum. But it proves so terribly valuable to be able to differentiate between a dog and a horse that we draw the lines anyways.
    – Cort Ammon
    Commented Aug 7, 2018 at 19:05
  • 3
    See also the Ship of Theseus paradox Commented Aug 8, 2018 at 0:21
  • 7
    As soon as it can no longer reproduce with other ... wait, no that's a different thing Commented Aug 8, 2018 at 1:23
  • 1
    In the case of Romanian, the demarcation point is its encounter with Slavic tongues, and the subsequent linguistic influence.
    – Lucian
    Commented Aug 10, 2018 at 4:40
  • 1
    At what point do we say "These people in the Italian peninsula are no longer speaking Latin; they are speaking Italian"? - Such point doesn't exist; from Latin to Romance languages there are at least three rupture points (Latin/Vulgar Latin, Vulgar Latin/Romance, and Romance/Romance languages). Commented Aug 11, 2018 at 0:31

5 Answers 5


This is a difficult question. Greek is perceived as one language despite the fact that Classical Greek is no longer intelligible for a native speaker of Modern Greek without exposure to the classical language. For the Romance language, the split into several different descendants (Italian, Provençal, French, Spanish ...) surely helped to form different identities and not seeing Italian as kind of "popular Latin".

Chinese is seen as one language mainly for political reasons, although different "dialects" of Chinese are not mutually intelligible as spoken languages. On the other side, Croatian, Bosnian, Serbian, and Montenegrin are considered different languages for political reasons, and there are ongoing language-planning activities to tear the languages further apart.

All in all, the perception of one language or two language is not governed by linguistic criteria, but by political and cultural criteria.

  • 6
    So the answer is, "after the Romans start calling their language 'Italian'?".
    – user6726
    Commented Aug 7, 2018 at 16:51
  • 4
    In principle, yes. Commented Aug 7, 2018 at 17:04
  • 1
    An even more difficult case is Macedonian, which some claim is a empty name used for a variety of Bulgarian. This is very much a political dispute.
    – Colin Fine
    Commented Aug 9, 2018 at 20:40
  • 2
    "after the Romans start calling their language 'Italian'?" - To notice, "Romans" (as in, "people from Rome") never started calling their language "Italian". The language they call "Italian" is Tuscan, not the language they spoke before unification. You may call this a Dantesque situation, which is literally true, but Modern Italian is not in direct continuation of what used to be spoken in Rome... Commented Aug 12, 2018 at 12:50
  • 2
    I presume you understood my actual point, and can look past the limitations of blurb-length comments.
    – user6726
    Commented Aug 12, 2018 at 17:41

At what point do we say "These people in the Italian peninsula are no longer speaking Latin; they are speaking Italian"?

There used to be an official answer to this: Italians were taught that Dante's standardization of the literary Florentine dialect is "when the language became Italian". But that's a completely arbitrary answer.

In Dante's time, people in much of Italy outside of Tuscany were speaking languages that weren't mutually intelligible with his, or with modern Italian. Those languages were only gradually replaced with "Italian" over the course of centuries (finally completed over a few generations from Napoleon to the final unification).

And of course his Florentine (and likewise all those other dialects/languages that aren't direct ancestors of modern Italian), was already unintelligible with Latin much earlier. Bishops were complaining that the laypeople couldn't understand the Church Latin Mass, advanced students had to be taught international medieval Latin as a foreign language, etc. And meanwhile, there are pre-Dante poems that are reasonably easily intelligible to modern Italians. In fact, there are documents that go back to the 10th century that Italian university students might struggle to read, but they're still clearly much much more like Italian than it is like Latin.1

So, clearly, they picked the wrong year to divide Latin and Italian, right?

But there is no right year. Latin became Italian over the course of several centuries, so any year you pick is going to be arbitrary.

And this continuity problem isn't something specific to the Romance languages, and it's not there only because we have so little documentation of the intermediate steps.

Take a look at Ottoman Turkish changing to (modern, aka Istanbul) Turkish. This took place over a single generation in the last century, but it's just as impossible to pick any non-arbitrary point "when the language became Turkish".2

Ataturk ordered the reform process started in 1927. Every step in the process was planned out by TDK or other committees, and documented in detail. By 1942, TDK reported that many children3 of Ottoman-speaking families4 couldn't understand Ottoman Turkish.5 By 1963, Ataturk's 1927 declaration was so incomprehensible that it had to be translated for anyone to follow it.

Unintelligibility is our best characterization of "different languages" vs. "just dialects".6 So, we can say that Ottoman Turkish became Turkish between 1927 and 1942. But exactly at what point during that 15 years did it happen?

It's not like the children born before 1 January 1938 00:00:00 grew up natively speaking Ottoman and the children born a nanosecond later, modern Turkish. All of the children during that period learned a language that was somewhere between the two. And at any given point during that 14 years, some of the children were speaking a more Ottoman-like language than others. Even looking at a single child at a time, most of them were speaking more Ottoman-like languages early in their acquisition and less-Ottoman-like later.

There just is no answer. It happened over the course of a generation, in the same way Italian happened over the course of a few centuries. If you want an answer, you can arbitrarily pick the first language reform law in 1928, or the founding of TDK in 1932, or TDK's progress report in 1942, or anything else you want, so long as you realize that it's arbitrary.

1. See the Wikipedia article on Placiti Cassinesi, which conveniently shows the original text with Latin and Italian translations.

2. Of course there's a legal answer; I think a bill declaring a new Turkish language was passed in 1932. And most Turkish laypeople would probably say 1928, because that's when the alphabet was switched. But you're asking about actual linguistic criteria.

3. Of course many children were bilingual, even for generations after 1942. But the fact that there were also monolingual speakers is the point.

4. As an aside, outside of the urban elites, most of the country didn't speak Ottoman Turkish in the first place. And outside the cities, it took longer for everyone to switch from their regional languages to modern Turkish. But that isn't really relevant here.

5. And it's not just the different vocabulary (and orthography). For example, monolingual Turkish speakers have to learn a whole new set of morphological rules to be fluent in Ottoman Turkish.

6. It should be obvious that there's a continuum issue there as well…


(edited to reflect sumelic's comment below)

At what point a language changes so much that it becomes another language?

That is a difficult question, that probably can only be answered in hindsight. At no moment in the convoluted history of Romance languages, anyone woke up and realized, "gee, this thing I am speaking is not Latin, it is, wow, French". People just spoke the language of their time and place, or one or more of the languages of their time and place (in France, likely, Vulgar Latin and Frankish).

At that time in history, people certainly distinguished Latin - the language in which the priest said the mass, which was mostly incomprehensible to them - from the language they actually spoke, which was, for lack of a better word, generally called "Vulgar" or "Vernacular", and quite certainly didn't constitute a unified language - it was different in Lisbon or Catania, Nice or Santander. When in contact with people of different places, people would recognise that those from other places spoke a different language - sometimes so far apart as to be unintelligible, sometimes close enough to be described like, "those damn Genoese and their ridiculous and wrong way of speaking". But contact with foreigners was rare, and probably most accessible to people who could manage some Latin - Medieval Latin, already quite different from Classic Latin, but that remained the language of high culture throughout Europe during the Middle Ages.

When the Romans conquered Western Europe, there were already several differences between the language spoken by the Roman elite and the Roman populace, to which most of the conquering soldiers belonged. It is this variant of Latin - Vulgar Latin - that originated the several Romance languages (with the exception, I think, of Sardinian, which is more related to Classic Latin). To the extent that Romans were conscious of this dichotomy between their standard language and the language effectively spoken by common people, the divide is very old. Ciceron probably knew his language was different from the language of his servants, though he probably framed this as "proper Latin" versus "wrong", or "debased", Latin.

Vulgar Latin then evolved differently at the different parts of the Roman Empire/influence area. At some point in history, the differences were big enough that we can no longer speak of "Vulgar Latin" as one language. Unfortunately, this was mostly an undocumented process: people actually changing the language were illiterate, so they didn't write their language; and literate people wrote in Medieval Latin, not in Vulgar, and weren't really interested in writing about the commoners' language(s). But at some point the language spoken in the Iberic Peninsula was Iberic Romance rather than Vulgar Latin.

Apart from the scant documentation, we can use comparative method to understand such changes.

For instance, here we have the history of a single word, from Classic Latin to Modern Portuguese:

Classic Latin: genu- (/genu/)

Classic Latin (diminutive of the former): genuculu- (/genukulu/)

Vulgar Latin: genuclu (/dʒenuklu/)

Western Romance: genuclu (/ʒenuklu/) (while Southern Romance kept the /dʒ/, as in Italian "ginocchio")

Iberic Romance: *genollo (/ʒenoʎo/) (while Western Romance in the area that is now France saw a different evolution, which gave us French "genoux", I suppose with the "x" originally pronounced, and then gone mute in Old or Modern French)

Old Portuguese: jeolho (/ʒeoʎo/) (while the Spanish cognate, which would be something like "jenojo", with the "j" pronounced as /x/, unfortunately (or fortunately, if you think in euphonic terms) being dropped in favour of the unrelated "rodilla")

Modern Portuguese: joelho (/ʒoeʎu/ or /ʒweʎu/))

How do we know that the order of the changes is that, first /g/ to /dʒ/, then /kl/ to /ʎ/?

Apart from documentation (we have Old Portuguese texts with the word "jeolho", for instance), we know that Medieval Latin has /dʒ/, and not /g/, so this sound must have changed very early and was probably done in Vulgar Latin; we know that all Romance languages have changed /kl/ into something different, so that it must have happened later, when Vulgar Latin was morphing into several different Romances; we know that Portuguese alone dropped intervocallic "n" (compare Portuguese "geral" to Spanish "general", so that must have been even later, when Iberic Romance was dividing into Galician-Portuguese and Castillian.

This process then intertwined with the political process of the formation and unification of modern nation-States. At this point, the local elites of Portugal, Spain, France, decided that they would be better served by making their local languages literate than by trying to repopularise Latin, as it would help in constructing the borders, and in creating a national ideology that could justify the existence of the nation-State. At that point we have the earliest literature in those languages. This process was unfortunate for several Romance languages that didn't correspond to the political boundaries - Mirandese, Occitan, Catalan, Leonese, etc. And it originates the saying, "a language is a dialect with an army", from which it is often deduced that the differences between languages is more political than linguistic (but notice, the army, and the State in general, are, besides obviously political entities, also glossopoietic entities: military terms are coined by the army (and navy), and courts, police, universities, public schools, also coin a huge vocabulary).

In the specific case of Italy, the political process of unification of the nation-State happened much later, only in the 19th century, which favoured the permanence of several different local languages - Venetian, Sardinian, Sicilian, Tuscan, Neapolitan, Emilian, Lombard, Piedmontese, Calabrian, etc. - not all of which even belong to the same Romance branch (Sardinian is Insular Romance, Venetian is Western Romance, while most of the others are Southern Romance). So things in Italy are even more complicated than in Portugal, France, or Spain: while those have had national languages from the end of the Middle Ages, Italy chose Tuscan its national language, or based its national language on Tuscan (mainly due to the literary and cultural importance of Dante), only in the 19th century, much after everyone in the peninsula was very sure that what they spoke was not Latin.

So, at what point we say "These people in the Italian peninsula are no longer speaking Latin; they are speaking Italian"? At no point. They stopped speaking Classic Latin at the end of the Antiquity, when cultured Latin evolved into Late Latin (which in turn evolved into Medieval Latin later), while popular Latin evolved into Vulgar Latin. At the Middle Ages, they were speaking several different languages, which can be linguistically grouped into three different "families", Western Romance in the North, Insular Romance in Sardinia, and Southern Romance in most of the Peninsula and in Sicily. "Italian" as we know it was one of these Southern Romance languages, namely Tuscan, and spoken by a very small minority of the population up to the 1860, when Italy became a unified nation-State, and its State deliberately choose it as the national language, and enforced via public education (and active repression of the local languages, which became collectively know as "dialects").

(But there certainly was some point in history, much before this, when people in Italy would listen to the mass and realise it was not the same language they used to speak. Probably as early as the 7th or 8th century, I would venture.)

As for "the split between Spanish and Portuguese" (which is really the split between Castillian and Galician-Portuguese), it happened in the High Middle Ages, certainly before the split the Condado Portucalense and the Kingdom of Spain. But then the political split reinforced the linguistic divide, making the languages further evolve apart from each other (and sparing Portuguese the fate of the minoritary languages of Spain).

And "at what point does Puertoriqueño cease to be a dialect of Spanish, and take on its own identity as a language? Similarly for Brazilian vs Portuguese, or the other various dialects of Spanish in Latin America, or even English/England vs English/Australia vs English/North America?"

At what point in time it is impossible to know. Modern communication doesn't favour the division of languages - and, so, the process is bound to be slower, possibly much slower, than the split of Latin into several languages. Unless, of course, there is a collapse of modern civilisation, that breaks the communication system apart, so that people in Puerto Rico, Brazil, or Australia, are no longer in touch with Spain, Portugal, or England (which, as a second thought, is what happened to the Roman Empire too, so the odds are, if there is a civilisation collapse, in half a millenium, when finally Venezolans rediscover Spain, people in Latin America will be unable to understand Spaniards, and conversely). At what point in linguistic terms, when the spoken varieties of those languages become mutually incomprehensible (text is more conservative, and may allow for the fiction of linguistic unity, perhaps useful for political purposes).

  • This is a very interesting answer; would it be safe to say that an "accurate oversimplification" from a linguistic, ignoring political, point of view would be that you have separate languages instead of dialects when mutual intelligibility either ceases to exist, or is so low that each speech must be specifically studied to achieve intelligibility? Commented Aug 12, 2018 at 17:36
  • Gets tricky when, say, given a historical language X for cultural/political reasons they get termed, say, Y and Z, despite being incredibly similar (and remain often virtually identical, accent and orthography notwithstanding). But then Y breaks into Y1 and Y2, and while Y1 stays virtually identical to Z, Y2 becomes markedly different. Despite Y1 and Y2 being so different, for political/cultural reasons, they still are considered the same. I'm of course speaking here of pt-PT/BR and gl-ES. Commented Aug 12, 2018 at 21:02

The philosophical answer that it is a continuüm is correct, but natural borders like seas, rivers, mountain ranges and even city limits have their influence on languages. It is not only political and cultural, but also natural that changes in languages and mingling with other languages or not create enough differences that you can at a certain point get away with calling it a different language. In the most extreme case it could be said about two groups speaking an identical language with one of them having a word the other group does not use. That said a language is different when it has its own word base that differs from the language it mostly came forth from. Put the line wherever you wish, as long as it is founded. Even within city limits...


One keyword here is holotype. E.g. Carl von Linne used his brother as type specimen for the Homo Sapiens, and the first Archaeopterix finds were all named differently because the relation among them wasn't clear. Definitions for familiar relations are based on a set of similarities. You could even have a phylogenetic tree of clade systems.

Classical Latin had Cicero, Ancient Greek had Homer, Standard High German has the guy in the evening news, the Queen's English is well received and so on. So far this is easy, but language depends on communication and dialogue, not a single person. Hence groups of speakers are compared. The biggest, accepted grouping is the language family, for example Indo-European or Afro-Asiatic. Some go even further and assume all extent languages developed from a common source. Hence I assume proscription to some extend means what was said for the first time, description is copying, and prescription is what was said before. So a language changes by faulty description and other proscription.

The real question is, what is a language. My take is, language is uncountable, dialect is individual, and the plural, languages, is a multitude of instances of using the language. Which is a fuzzy set (see references e.g. in mathworld/fuzzy_sets

At the small scale, even a single individual will make use of code switching for example, make up new sentences everyday and learn new words. The pressing question is how even a single word is learned, defined and produced, reinterpreted, and reproduced. Impressions left by a forbearer are as important, in the simple case, as much as imagination is in the big picture, from baby speech to high poetry. Hence I don't completely subscribe to the single-mother hypothesis.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.