Dynamic Cultural meltingpots.

[Arona]

I think Cultural metingpots should be more dynamic. There are som historical merged cultures what will give cultures a new name. But overally dynamic cultures should emerge where New culture has both cutures in name. Like Gaelic-Norse, Pictish-Norse. Or frank-breton. As different cultures merge they should also merge some of cultural features, some features will be lost and some will be avaible in new culture.

 

[Pied-Noir]

This would cause horrendous problems for name linking and scripting triggers.

 

[Zhetone]

A mechanic which sounds nice on paper but which would have many undesirable consequences

 

[DreadLindwyrm]

I think Cultural metingpots should be more dynamic. There are som historical merged cultures what will give cultures a new name. But overally dynamic cultures should emerge where New culture has both cutures in name. Like Gaelic-Norse, Pictish-Norse. Or frank-breton. As different cultures merge they should also merge some of cultural features, some features will be lost and some will be avaible in new culture.

There are problems with dynamic melting pots.

Names are one of the most problematic.

In a hypothetical Anglo-Lithuanian merged culture, which name list does it take?

If you get an Italo-Arab merged culture which name list does it take?

And then there's the question of if your gallo-norse culture merges with the franco-breton one, what do we call the culture?

 

[BrokenSky]

There are problems with dynamic melting pots.

Names are one of the most problematic.

In a hypothetical Anglo-Lithuanian merged culture, which name list does it take?

If you get an Italo-Arab merged culture which name list does it take?

And then there's the question of if your gallo-norse culture merges with the franco-breton one, what do we call the culture?

Presumably there would be some distinction between Norse-in-Ireland and Irish-in-Norway but for most of these they probably wouldn't be called "Norse-Gael", but rather, something based on the region they ended up in.

For example, let's say, something based on the name of the duchy they originated in in the language of the conquerors - e.g. Norse invade Anglo-Saxon York, create melting-pot culture "Jorvik", taking the title name-list from Norse and using a mix of Norse names and Anglosaxon names using Norse character replacements where possible. At the same time the french king francifies the Bretons, creating a culture in the french group which uses a mix of french and Breton names. This culture is named after the adjective for the french name version of the duchy of Brittany. Now our Jorvik culture goes out and conquers Brittany, and creates a New New culture, mixing the two melting pot cultures into a third, which would then use mostly Breton names with Norse character swaps and be named after the duchy of Brittany - but now there's a problem since the Brittany already has a melting-pot culture associated with it (and for the sake of argument isn't called something else in anglosaxon or norse), so instead the name is based on one of the holdings, which also becomes the new duchy capital and name, (prioritising the default capital), e.g. Nantsian.

Now I'm not saying this system is the best available, but it's something I came up with on the fly while writing this response over the course of like.. twenty minutes? CK2 already had a system for dynamically naming cultures, which is did for the randomised cultures in the random history system (so there's no need at all for something to ever be called Gaelo-Norse-Franco-Breton, thank goodness). Solutions exist, is my point.

Ideally though if they could make the system for creating the dynamic names lists as scriptable as possible, they'd probably only need to wait a year or two to look at how modders made it better for inspiration.

 

[Tschobo]

I made such a proposal once, but after looking a bit into scripting and everyting. Its not worth it. I think historical and some realistic alt history ones would be sufficient.

 

[DreadLindwyrm]

Presumably there would be some distinction between Norse-in-Ireland and Irish-in-Norway but for most of these they probably wouldn't be called "Norse-Gael", but rather, something based on the region they ended up in.

For example, let's say, something based on the name of the duchy they originated in in the language of the conquerors - e.g. Norse invade Anglo-Saxon York, create melting-pot culture "Jorvik", taking the title name-list from Norse and using a mix of Norse names and Anglosaxon names using Norse character replacements where possible. At the same time the french king francifies the Bretons, creating a culture in the french group which uses a mix of french and Breton names. This culture is named after the adjective for the french name version of the duchy of Brittany. Now our Jorvik culture goes out and conquers Brittany, and creates a New New culture, mixing the two melting pot cultures into a third, which would then use mostly Breton names with Norse character swaps and be named after the duchy of Brittany - but now there's a problem since the Brittany already has a melting-pot culture associated with it (and for the sake of argument isn't called something else in anglosaxon or norse), so instead the name is based on one of the holdings, which also becomes the new duchy capital and name, (prioritising the default capital), e.g. Nantsian.

Now I'm not saying this system is the best available, but it's something I came up with on the fly while writing this response over the course of like.. twenty minutes? CK2 already had a system for dynamically naming cultures, which is did for the randomised cultures in the random history system (so there's no need at all for something to ever be called Gaelo-Norse-Franco-Breton, thank goodness). Solutions exist, is my point.

Ideally though if they could make the system for creating the dynamic names lists as scriptable as possible, they'd probably only need to wait a year or two to look at how modders made it better for inspiration.

There is a problem with simply trying to use character replacements though.
A given character can be pronounced multiple ways in the same language - and it's usually clusters of characters you have to look for. Even then though, you have the problem that the same cluster of characters can represent multiple phonemes. "ough" for example has 8 11 different phonemes associated with it.
This isn't as much of an issue with related languages (such as Norse/Saxon), but when you start looking at putative combinations that cross language groups (Anglo-Andalusian, Italo-Berber, Franco-Egyptian), it can rapidly become a problem.

The "culture named for county" mechanic *sort of* worked, but you can end up with some horribly unpronouncable results.

Practically there isn't a compact way to be able to cleanly transliterate by machine across all the languages by applying simple rules *and* have the game be a reasonable size and speed.
You'd need a stronger transliteration program than is used professionally to be able to do it right.

 

[ArVass]

Anyone who believes that dynamic melting pots is a realistic idea has no idea how hard programming the code responsible for the merging of the namelists would be.

 

[BrokenSky]

There is a problem with simply trying to use character replacements though.
A given character can be pronounced multiple ways in the same language - and it's usually clusters of characters you have to look for. Even then though, you have the problem that the same cluster of characters can represent multiple phonemes. "ough" for example has 8 11 different phonemes associated with it.
This isn't as much of an issue with related languages (such as Norse/Saxon), but when you start looking at putative combinations that cross language groups (Anglo-Andalusian, Italo-Berber, Franco-Egyptian), it can rapidly become a problem.

The "culture named for county" mechanic *sort of* worked, but you can end up with some horribly unpronouncable results.

Practically there isn't a compact way to be able to cleanly transliterate by machine across all the languages by applying simple rules *and* have the game be a reasonable size and speed.
You'd need a stronger transliteration program than is used professionally to be able to do it right.

The same is true of every aspect of the game. Simulating an AI capable of acting like a real human person is beyond the capacity of current programmers, but nevertheless paradox has probably hundreds of thousands of characters in CK2, ostensibly acting as "people". The difference here is a question of "perfect" vs. "good enough", and the fact they you say "this can't be done" and my response is "here's a way to do it I came up with in twenty minutes" is clearly indicative of a similar problem; you're saying we can't do a perfect system which works 100% of the time? I agree. But I'm saying we can make do with something which is roughly good enough, especially if it run off a mod-able scripting system.

Because honestly I think if we applied the same standard to AI, War, Plots, Religion, Province development -basically anything- we'd have to give up and scrap the feature, but if we say "how can we do this simply?" then we get lots of options which aren't fantastic but kind of work, mostly. From there it's just a case of working out what we actually need to work well and what we can afford to leave up to abstraction, smoke and mirrors and the other myriad of tricks game makers use to make a handful of random numbers and deterministic algorithms look and feel like a real-ish world.

 

[DreadLindwyrm]

The same is true of every aspect of the game. Simulating an AI capable of acting like a real human person is beyond the capacity of current programmers, but nevertheless paradox has probably hundreds of thousands of characters in CK2, ostensibly acting as "people". The difference here is a question of "perfect" vs. "good enough", and the fact they you say "this can't be done" and my response is "here's a way to do it I came up with in twenty minutes" is clearly indicative of a similar problem; you're saying we can't do a perfect system which works 100% of the time? I agree. But I'm saying we can make do with something which is roughly good enough, especially if it run off a mod-able scripting system.

Because honestly I think if we applied the same standard to AI, War, Plots, Religion, Province development -basically anything- we'd have to give up and scrap the feature, but if we say "how can we do this simply?" then we get lots of options which aren't fantastic but kind of work, mostly. From there it's just a case of working out what we actually need to work well and what we can afford to leave up to abstraction, smoke and mirrors and the other myriad of tricks game makers use to make a handful of random numbers and deterministic algorithms look and feel like a real-ish world.

I'm saying that the method you suggested would be deeply inadequate, and thus not even remotely good enough.

The transliterations would be terrible, and often wrong, and badly so. Even trying to do this between closely related languages is going to result in utter rubbish a significant portion of the time.

Take for example the name of an English town. Loughborough. (broadly "Luff-buh-ruh")
If you take a simple machine transliteration, it *will* get the name wrong, and fail to render a name that will sound even remotely similar, due to the fact the "ough" parts of the name make two different sounds.
Then there's Slough. (rhymes with "cow"). Same ough combination, another sound entirely.
Woughton (Wuf-tun), Loughton (Lau (again, rhymes with cow )- ton), and Broughton (Braw-ton) all lie near each other, being parishes in the same town, and all have different uses of "ough".
To make a program that is capable of spotting those different sounds *and* getting it right is difficult, even as a professional piece of work.

The same applies to other combinations.

"Celt" has to pronunciations. "Kelt" is a member of the pre-saxon population of Britain. "Selt" is a stone knife.
"Celtic" can either be "Keltic" - appertaining to "Kelts", or "Seltic" - a football team.

For a simple transliteration program spotting when C is meant to be hard (K) or soft (C) is not easy.

"Ch" can be soft (church, Chester), or hard (school, choir), or almost a "j" sound (Greenwich, Norwich)

Functionally your 20 minute solution is another example of a solution to a problem that is "simple, neat, and wrong".
I'm not saying it isn't a good attempt at the problem, just that it is one that cannot work as simply as you appear to think it can.

You'd have to fill it with so many exceptions and special cases that you'd be essentially hand crafting the melting pots instead of making them happen dynamically.


*Or* everything would have to programmed in IPA behind the scenes and rendered into appropriate text, whilst not necessarily having the historical registers for a given culture or dialect, and with compromise registers being able to be created on the fly for where a given sound just doesn't exist in a language but they need to absorb it as part of the melting pot *and* assign it an orthography.

 

[Riamus]

I'm not certain that how it's pronounced really matters here. If the game were to make a melting pot name of something like Broughwich (just picking two parts of your examples here), does it matter if the player pronounces it in their head as Brow-which or Bra-wij or Broh-wik or something else entirely? I'm sure many (most?) players mispronounce many of the counties and duchies within the game as it is.

I do get that you are meaning that for a realistic-sounding mix, it is likely to pick something that makes sense in terms of how it sounds. The question then becomes, how bad is it if that's the case? If you happen to be from the area where the melting pot occurs, you might think it sounds bad, but is a bad-sounding name such a bad thing if it means you had the option for dynamic melting pots? Would that option be worth having weird sounding names considering the names wouldn't be spoken aloud anyhow unless the player says it aloud for some reason? I'm sure some would be on both sides of that. I am curious which side would have the most people.

That said, I do think that it's not anything even remotely easy to accomplish and in the end is probably not worth the effort involved to get it to even look remotely plausible. You'd probably end up with a lot of stuff like Jimatjisawty or some weird thing that would be so bad it would be jarring to see and extremely hard to even guess at a pronunciation. Trying to get it to work well enough to not be jarring, even if the sound of it doesn't fit very well, is probably something that would require so much work as to be pointless. It would probably be faster for them to just manually make all the possible melting pot names and then let the game dynamically create the melting pot and use the manually created names rather than dynamically create the names. And that's saying something as you'd need a lot of names in that list for it to work.

 

[Alejandro Cauchy]

Ahistorical cultural hybrid is pointless.

 

[ArVass]

*Or* everything would have to programmed in IPA behind the scenes and rendered into appropriate text, whilst not necessarily having the historical registers for a given culture or dialect, and with compromise registers being able to be created on the fly for where a given sound just doesn't exist in a language but they need to absorb it as part of the melting pot *and* assign it an orthography.

This is the only thing that would make it fine, and it's just too much work.

 

[Kainser]

What exactly would control when a dynamic culture would be formed? Would you end up having tiny blended cultures like Anglo-welsh, Czech-german or Franco-breton pop up every single game which sounds terrible? If not, how do you suggest them not being formed every time a ruler conquers a province with a different culture?

The same is true of every aspect of the game. Simulating an AI capable of acting like a real human person is beyond the capacity of current programmers, but nevertheless paradox has probably hundreds of thousands of characters in CK2, ostensibly acting as "people". The difference here is a question of "perfect" vs. "good enough", and the fact they you say "this can't be done" and my response is "here's a way to do it I came up with in twenty minutes" is clearly indicative of a similar problem; you're saying we can't do a perfect system which works 100% of the time? I agree. But I'm saying we can make do with something which is roughly good enough, especially if it run off a mod-able scripting system.

Because honestly I think if we applied the same standard to AI, War, Plots, Religion, Province development -basically anything- we'd have to give up and scrap the feature, but if we say "how can we do this simply?" then we get lots of options which aren't fantastic but kind of work, mostly. From there it's just a case of working out what we actually need to work well and what we can afford to leave up to abstraction, smoke and mirrors and the other myriad of tricks game makers use to make a handful of random numbers and deterministic algorithms look and feel like a real-ish world.

Yes, but it is enormously hard to simplify a concept like "dynamic cultures". I have yet to see an idea for them that doesn't end up involving namelists where couples will cheerfully name their three children Muhammad, Hrolfr and Aethelbald while belonging to the wonderfully named Anglo-norse-andalusian culture.

Names and ethnicity is the primary effect of cultures in CK. Ethnicity would be easy enough to handle with CK3 but the technology is just not there for dynamic names to be anything other than at best cringeworthy.

 

[Karlington]

Let's not forget that technology is tied to culture in CK3, and that cultures can have their own inventions as well, so dynamically generated cultures might cause a truckload of issues because of this.

 

[BrokenSky]

I'm saying that the method you suggested would be deeply inadequate, and thus not even remotely good enough.

The transliterations would be terrible, and often wrong, and badly so. Even trying to do this between closely related languages is going to result in utter rubbish a significant portion of the time.

Take for example the name of an English town. Loughborough. (broadly "Luff-buh-ruh")
If you take a simple machine transliteration, it *will* get the name wrong, and fail to render a name that will sound even remotely similar, due to the fact the "ough" parts of the name make two different sounds.
Then there's Slough. (rhymes with "cow"). Same ough combination, another sound entirely.
Woughton (Wuf-tun), Loughton (Lau (again, rhymes with cow )- ton), and Broughton (Braw-ton) all lie near each other, being parishes in the same town, and all have different uses of "ough".
To make a program that is capable of spotting those different sounds *and* getting it right is difficult, even as a professional piece of work.

The same applies to other combinations.

"Celt" has to pronunciations. "Kelt" is a member of the pre-saxon population of Britain. "Selt" is a stone knife.
"Celtic" can either be "Keltic" - appertaining to "Kelts", or "Seltic" - a football team.

For a simple transliteration program spotting when C is meant to be hard (K) or soft (C) is not easy.

"Ch" can be soft (church, Chester), or hard (school, choir), or almost a "j" sound (Greenwich, Norwich)

Functionally your 20 minute solution is another example of a solution to a problem that is "simple, neat, and wrong".
I'm not saying it isn't a good attempt at the problem, just that it is one that cannot work as simply as you appear to think it can.

You'd have to fill it with so many exceptions and special cases that you'd be essentially hand crafting the melting pots instead of making them happen dynamically.


*Or* everything would have to programmed in IPA behind the scenes and rendered into appropriate text, whilst not necessarily having the historical registers for a given culture or dialect, and with compromise registers being able to be created on the fly for where a given sound just doesn't exist in a language but they need to absorb it as part of the melting pot *and* assign it an orthography.

I feel like you're once again missing the point. I'm not suggesting we implement a solution which I came up with in twenty minutes - I'm suggesting we actually consider trying to solve the problem. In my experience there have been many things people have said are impossible which nevertheless achieved by people who are willing to put in the time and effort, and to compromise on the results to get something achievable.

It feels a lot like you're setting the highest possible bar - to transliterate Loughborough, for example, you need only consider the two parts; lough (lake presumably? since historically ough was more pronounced like the ch in loch in many cases) and borough (Borough, presumably). So any germanic hybrid could easily make this become Lochburg, say. For names, for example, you could have a 'root name' for all names (Julius / Julian), (Andrew/Andreas/Andrea), (Karl/Charles), (Peter/Petre/Petros) and for namelists which had different versions with the same route, keep either one, the other or both randomly, which would cut down the work a lot for most cultures in the same continent.

Again, I'm not saying this is the best way to go, but it is a possible route to a workable option. The point still stands - 'solutions exist' is what I'm saying.

 

[BrokenSky]

Yes, but it is enormously hard to simplify a concept like "dynamic cultures". I have yet to see an idea for them that doesn't end up involving namelists where couples will cheerfully name their three children Muhammad, Hrolfr and Aethelbald while belonging to the wonderfully named Anglo-norse-andalusian culture.

Or even Keith (Irish), Lukus (Latin) and Joshua (Hebrew), while belonging to the insanely named Norse-French-German-Welsh-Roman culture which modern people call English.

 

[DreadLindwyrm]

I feel like you're once again missing the point. I'm not suggesting we implement a solution which I came up with in twenty minutes - I'm suggesting we actually consider trying to solve the problem. In my experience there have been many things people have said are impossible which nevertheless achieved by people who are willing to put in the time and effort, and to compromise on the results to get something achievable.

It feels a lot like you're setting the highest possible bar - to transliterate Loughborough, for example, you need only consider the two parts; lough (lake presumably? since historically ough was more pronounced like the ch in loch in many cases) and borough (Borough, presumably). So any germanic hybrid could easily make this become Lochburg, say. For names, for example, you could have a 'root name' for all names (Julius / Julian), (Andrew/Andreas/Andrea), (Karl/Charles), (Peter/Petre/Petros) and for namelists which had different versions with the same route, keep either one, the other or both randomly, which would cut down the work a lot for most cultures in the same continent.

Again, I'm not saying this is the best way to go, but it is a possible route to a workable option. The point still stands - 'solutions exist' is what I'm saying.

And you are missing my point.

There isn't a simple way to extract these name elements cleanly and reliably. *and automatically*.

You'd have to find the initial etymology of each placename to do this sort of analysis, and mark the fragments of the names so they were correctly translated. But here you've added another standard. We're translating the etymology, not transliterating the sounds as originally suggested. You're assuming that a particular "ough" was pronounced in a particular way (when there are 11 surviving pronunciations just in standard English, many of which are old "historic" ones), and assuming an etymology for it in order to assume a sound.
The "dynamic" system almost certainly cannot have the intelligence to do this just from a list of names.

Loughborough, by the way, is considered to be derived from a personal name + borough.

And as for setting the highest bar?
Not even close. Loughborough is *difficult* certainly, but there are towns with the same spelling of their name which have different pronunciations. *If* these are significant enough to show us on the map, then you would need a way to distinguish between the two, and give them their proper pronunciation in the transliteration. "Southwell" (variously Suth-ul and Sow (as in the female pig)-th-well) would need to be correctly handled. You are not going to get a simple transliteration program to pick up that difference.

The root name system is in place for a lot of personal names in CK2, and hopefully carried across, but those names aren't really the issue. It's when you try to force names that don't have equivalents into a new hybrid list, and have to do it with a set of spelling rules (so that it's completely dynamic and not handled by pre-scripting).
It needs to be able to take a new name that is *completely* missing from the culture list of one parent culture and change the spelling from the other parent to fit the orthography of the merged culture. For example, Anglo-Saxon merging with Andalusian and acquiring classically Moorish/Islamic names which now need to be rendered into Anglo-Saxon phonemes - or Anglo-Saxon names needing to be rendered into Andlusian phonemes - and this being done just from the spelling.


Solutions *may* exist - but considering that professional programs have trouble with getting "correct" transliterations for modern, living languages that have access to the IPA system, it's not something that is a practical goal to add to a game like CK3 without a lot of bloat and hand crafting.

 

[BrokenSky]

And you are missing my point.

There isn't a simple way to extract these name elements cleanly and reliably. *and automatically*.

You'd have to find the initial etymology of each placename to do this sort of analysis, and mark the fragments of the names so they were correctly translated. But here you've added another standard. We're translating the etymology, not transliterating the sounds as originally suggested. You're assuming that a particular "ough" was pronounced in a particular way (when there are 11 surviving pronunciations just in standard English, many of which are old "historic" ones), and assuming an etymology for it in order to assume a sound.
The "dynamic" system almost certainly cannot have the intelligence to do this just from a list of names.

Loughborough, by the way, is considered to be derived from a personal name + borough.

And as for setting the highest bar?
Not even close. Loughborough is *difficult* certainly, but there are towns with the same spelling of their name which have different pronunciations. *If* these are significant enough to show us on the map, then you would need a way to distinguish between the two, and give them their proper pronunciation in the transliteration. "Southwell" (variously Suth-ul and Sow (as in the female pig)-th-well) would need to be correctly handled. You are not going to get a simple transliteration program to pick up that difference.

The root name system is in place for a lot of personal names in CK2, and hopefully carried across, but those names aren't really the issue. It's when you try to force names that don't have equivalents into a new hybrid list, and have to do it with a set of spelling rules (so that it's completely dynamic and not handled by pre-scripting).
It needs to be able to take a new name that is *completely* missing from the culture list of one parent culture and change the spelling from the other parent to fit the orthography of the merged culture. For example, Anglo-Saxon merging with Andalusian and acquiring classically Moorish/Islamic names which now need to be rendered into Anglo-Saxon phonemes - or Anglo-Saxon names needing to be rendered into Andlusian phonemes - and this being done just from the spelling.


Solutions *may* exist - but considering that professional programs have trouble with getting "correct" transliterations for modern, living languages that have access to the IPA system, it's not something that is a practical goal to add to a game like CK3 without a lot of bloat and hand crafting.

I don't believe I am missing your point - I'm saying it's not relevant to CK3 because you're trying to create a system which would work with any place name or any personal name in any language, but to get something which worked you don't need the system to be able to tell that Loughborough comes from a person name, not a lake - you can tell it that when you set the name based on culture, because there are a finite, small number of provinces in CK3 which the game knows exist from the outset, so you tell it "Loughborough = Lough + Borough" and a dynamic German based derivative maybe calls the places Loughburg instead, while a slavic-hybrid might use, say, "Loughgorod".

The script doesn't need to know how to take the word Loughborough appart - it only needs to know what the root setup for Loughborough is (set in the holding name) and how to substitute the prescripted key parts into certain things. Of course this could end up with weirdly named places with mixed names which seem kinda weird, but then IRL how many places are called "The River River" etc. when translated because of people doing just that?

That's what I mean by the highest possible standard - you're trying to find a solution which works in an absolute sense; which actually properly takes the words apart and makes them work properly. And yes, something like that is outside the scope of the game, but to make something with works in the highly limited case within the game is obviously far simpler - similar to how the AI gets away with being little more than a few decisions made at random, using traits and circumstance to skew the odds to make it look like there's a real AI in there. Most of the time.

 

[Viridianus]

OK, at this point I should intervene as a linguist.

First off, English spelling is somewhat outstanding in its difficulty and inconsistency, so Loughborough debate may not be the one you need to think of (although having that -burg/-gorod/-grad/-vár/… part to translate properly is, certainly, a subissue).

Secondly, despite the first point, it will, as @DreadLingwyrm noted, be enormously difficult to create a dynamic naming melting pot. First of all, languages never do a proper mix (even creoles are not one), one wins, taking loanwords and some minor patterns from the other languages involved and adapting them (see French loanwords in English, Church Slavonic loanwords in Russian, Swedish borrowings in Finnish, Celtic 20-based numbering system in French, Celtic progressive in English, and so on). So, you'd have to choose the "winner" each time. Then you have to decide with how it deals with borrowed names, which is near-impossible to do automatically if you only have orthography of the two cultures. Also, in order not to have your name lists blob enormously, there should be some cutting (unifying the names with the same base name like Pedro/Peter is both not enough and not actually accurate: currently popular Russian Oksana seems to be a Ukrainian loan from their version of Greek->Russian Xenia, and that's for closely related, I doubt Louis-Ludvig general rulers (not scholars) would even recognize each other as having the "same" name). And all subpoints here except for the last one, if disregarded, will lead not to a "working well enough" solutions but to a disastrous monsters. A simple example: you mash a culture with names having a sh-sound with Greek or Finnish culture which doesn't have it, and the Greek/Finnish language is the winner. Should it: 1)magically acquire the sound; 2)adapt it as s; 3)do different things for different names/spellings? "Obviously not 3" sounds optimal, but... I assure you that most automatic machineries will set to 3 or, in the best case, 1, whereas the most probable result linguistically is 2. And sh, at least, has a single adaptation. Were the sound th (which can be adapted as f/v, t/d, or s/z in different communities/words/...) or f, the problem would be much larger.

Thirdly, as some others noted, it is also not clear what should the firing conditions be if we don't want to spam our map with melting pots all over the place on day 2 (something like CKII de jure drift could work, but more gradual, and graduality is always difficult).

Fourthly, a minor issue which can be dealt with, but we know that newly generated characters come with preset ethnicities corresponding to their culture. Should we derive a preset ethnicity of a, say, Italo-Welsh mix by just averaging between Italian and Welsh preset ethnicities?