With 14 languages you could speak with half the people on earth in their native tongue.

[rdearman]

I'd be interested in someone doing the math of recursive subtraction.

I'd be interested in someone else doing it too. Of course the real difficulty is that you'd not just need the percentage of native speakers, but also the percentage of Native speakers of language A who speak language, B, C, D, E, F ... That is the information which would be difficult to get. The actually recursive function would be trivial when compared to the information gathering exercise.

[zenmonkey]

I'd be interested in someone doing the math of recursive subtraction.

I'd be interested in someone else doing it too. Of course the real difficulty is that you'd not just need the percentage of native speakers, but also the percentage of Native speakers of language A who speak language, B, C, D, E, F ... That is the information which would be difficult to get. The actually recursive function would be trivial when compared to the information gathering exercise.

I was thinking about that.

There is data on the total number of bilinguals (and maybe trilingual) in many countries - you can use that to guesstimate the overlap of A & n. Then you basically ignore the rest because the number of 4+ speakers (that can actually speak 4 well) is statistically small ...

It would be really cool to have a dynamic web site where you can enter your languages (or any language) and see the overlap and overall ability to reach a country population...

hmmm. hmmmm. hmmmmmmmm.

[zenmonkey]

So today, couldn’t get this out of my head and decided to write out the math using Vann diagrams. This led down the rabbit hole of combinatorial probability... anyway, if anyone is interested in the problem of calculating the probability of total speakers of language A given the percent of monolinguals, bilinguals in a country and the number of speakers of each language ... well, one way to look at that is using inclusion exclusion principle...

wikipedia Inclusion-Exclusion principle article

It’s still pretty data intensive....

[zenmonkey]

Which 14 Languages?

These nine languages are absolutely necessary to get over 50% with 14 languages (again, counting Hindi and Urdu as two distinct but inseperable languages):

Rank	Language	Family	Native Speakers (million)	Percentage of world population
1	Mandarin (entire branch)	Sino-Tibetan	935	14.10%
2	Spanish	Indo-European	390	5.85%
3	English	Indo-European	365	5.52%
4	Hindi	Indo-European	295	4.46%
5	Arabic	Afro-Asiatic	280	4.23%
6	Portuguese	Indo-European	205	3.08%
7	Bengali (Bangla)	Indo-European	200	3.05%
8	Russian	Indo-European	160	2.42%
21	Urdu	Indo-European	66	0.99%

Why is Urdu on here? Why not Japanese with 128 million users?
Then these 12 get you to 50%:

Mandarin Chinese (incl.Standard Chinese)
Spanish
English
Hindi
Arabic
Bengali
Portuguese
Russian
Japanese
Western Punjabi (Lahnda)
Marathi
Telugu

[Deinonysus]

Which 14 Languages?

These nine languages are absolutely necessary to get over 50% with 14 languages (again, counting Hindi and Urdu as two distinct but inseperable languages):

Rank	Language	Family	Native Speakers (million)	Percentage of world population
1	Mandarin (entire branch)	Sino-Tibetan	935	14.10%
2	Spanish	Indo-European	390	5.85%
3	English	Indo-European	365	5.52%
4	Hindi	Indo-European	295	4.46%
5	Arabic	Afro-Asiatic	280	4.23%
6	Portuguese	Indo-European	205	3.08%
7	Bengali (Bangla)	Indo-European	200	3.05%
8	Russian	Indo-European	160	2.42%
21	Urdu	Indo-European	66	0.99%

Why is Urdu on here? Why not Japanese with 128 million users?
Then these 12 get you to 50%:

Mandarin Chinese (incl.Standard Chinese)
Spanish
English
Hindi
Arabic
Bengali
Portuguese
Russian
Japanese
Western Punjabi (Lahnda)
Marathi
Telugu

I explain this in the very beginning of my post, but to expand on it a bit: the source I use counts Hindi and Urdu as separate languages but they are so similar, especially in informal speech, that if you learn one you basically get the other for free, and you'd might as well put in a bit of extra work to learn the writing and formal registers of both. So I am treating them as two separate but inseparable languages.

You need to learn Hindi to be the most efficient, so Urdu is there as well. There was a good amount of discussion about this in previous pages.

This has nothing to do with Urdu being more important Japanese. It is because my source splits the Hindustani language in two but I'm not sure that makes sense.

[zenmonkey]

Which 14 Languages?

These nine languages are absolutely necessary to get over 50% with 14 languages (again, counting Hindi and Urdu as two distinct but inseperable languages):

Rank	Language	Family	Native Speakers (million)	Percentage of world population
1	Mandarin (entire branch)	Sino-Tibetan	935	14.10%
2	Spanish	Indo-European	390	5.85%
3	English	Indo-European	365	5.52%
4	Hindi	Indo-European	295	4.46%
5	Arabic	Afro-Asiatic	280	4.23%
6	Portuguese	Indo-European	205	3.08%
7	Bengali (Bangla)	Indo-European	200	3.05%
8	Russian	Indo-European	160	2.42%
21	Urdu	Indo-European	66	0.99%

Why is Urdu on here? Why not Japanese with 128 million users?
Then these 12 get you to 50%:

Mandarin Chinese (incl.Standard Chinese)
Spanish
English
Hindi
Arabic
Bengali
Portuguese
Russian
Japanese
Western Punjabi (Lahnda)
Marathi
Telugu

I explain this in the very beginning of my post, but to expand on it a bit: the source I use counts Hindi and Urdu as separate languages but they are so similar, especially in informal speech, that if you learn one you basically get the other for free, and you'd might as well put in a bit of extra work to learn the writing and formal registers of both. So I am treating them as two separate but inseparable languages.

You need to learn Hindi to be the most efficient, so Urdu is there as well. There was a good amount of discussion about this in previous pages.

This has nothing to do with Urdu being more important Japanese. It is because my source splits the Hindustani language in two but I'm not sure that makes sense.

Ok, I was confused. I get it now. Thanks.