The constants e and π in linguistics

What can I say? It doesn't fit above, put it here. Also the location of board rules/info.
Post Reply
User avatar
eldin raigmore
korean
korean
Posts: 6387
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

The constants e and π in linguistics

Post by eldin raigmore » 11 Sep 2018 04:47

I have recently noticed patterns in linguistics related to the transcendental consonants e (the base of the natural logarithms) and π (the ratio of a circle’s circumference to its diameter).
I have not yet been able to figure out how these constants should be part of universal grammar; but the tendencies are too widespread to be mere coincidences.
Maybe someone better-versed than I in universal grammar can figure it out.

In very many languages, the number of values of some grammatical feature, is some whole number between e and π.
For instance:
The number of persons;
The number of tenses;
The number of genders;
The number of voices.

Why do you suppose that would be?
Edit: this is a joke.
Last edited by eldin raigmore on 17 Oct 2018 16:56, edited 1 time in total.

User avatar
Thrice Xandvii
mongolian
mongolian
Posts: 3722
Joined: 25 Nov 2012 10:13
Location: Carnassus

Re: The constants e and π in linguistics

Post by Thrice Xandvii » 11 Sep 2018 06:23

People really seem to gravitate toward the number 3, both in the way in which we interloquate, as well as just in terms of structural organization, etc. As such, any significant constants that are in the range of 3 will seem more significant than they actually are, at least, IMO.
Image

User avatar
gach
MVP
MVP
Posts: 702
Joined: 07 Aug 2013 01:26
Location: displaced from Helsinki

Re: The constants e and π in linguistics

Post by gach » 11 Sep 2018 11:44

To make this more rigorous, you certainly need a comparison with the amount of languages where the number of values that these grammatical features get is a whole number below e. For instance, how common are three gender systems when compared to two gender systems or the lack of gender (i.e. "one gender systems")? That's a nice idea for someone to play with WALS for an evening, looking through different grammatical features.

You'll have to deal with a fair bit of small number statistics when doing this, but I wouldn't be surprised to still find a good number of power laws out there. Some years ago I played a bit with the Princeton University WordNet database to figure out the distribution of polysemy in the English lexicon. What you find there, is that the distribution of definitions per word follows a fairly neat decreasing power law for all the major word classes, with an exponent -3.8 for nouns, -3.4 for adjectives, -3.5 for adverbs, and -2.7 for verbs. In other words, verbs are more polysemous than the other lexical word classes. The amount of data I had to work with here is significantly larger than what's possible when numbering the values of grammatical features, but I would suspect the underlying processes to be fairly similar there as well.
ImageKištaLkal sikSeic

User avatar
sangi39
moderator
moderator
Posts: 3299
Joined: 12 Aug 2010 01:53
Location: North Yorkshire, UK

Re: The constants e and π in linguistics

Post by sangi39 » 11 Sep 2018 23:57

eldin raigmore wrote:
11 Sep 2018 04:47
@shimobaatar: this is a joke.
Eldin, try not to put words in other people's mouths, or single someone out and assume their reaction to your thread unnecessarily. Thank you.
You can tell the same lie a thousand times,
But it never gets any more true,
So close your eyes once more and once more believe
That they all still believe in you.
Just one time.

User avatar
Lambuzhao
korean
korean
Posts: 7782
Joined: 13 May 2012 02:57

Re: The constants e and π in linguistics

Post by Lambuzhao » 15 Sep 2018 12:35

Why not just ask "Why do you suppose that there are there between 2-4 / mostly 2 or 3 of the following…?"

I mean, statistically, it's prolly much closer to use e and π as parameters, but IMHO √-1 natlangs
would have π number of tenses, or e number of tenses.

:wat:

Then again, I wasn't all that good with higher maths anyway.
[:$]

User avatar
Lambuzhao
korean
korean
Posts: 7782
Joined: 13 May 2012 02:57

Re: The constants e and π in linguistics

Post by Lambuzhao » 15 Sep 2018 12:41

sangi39 wrote:
11 Sep 2018 23:57
eldin raigmore wrote:
11 Sep 2018 04:47
@shimobaatar: this is a joke.
Eldin, try not to put words in other people's mouths, or single someone out and assume their reaction to your thread unnecessarily. Thank you.
I think he's trying to tell Shimo that the underlying proposition of e < # of x natlang verb tenses, etc < π is more quip or witticism than ad hominem.

But might have misread/misunderstood the spoiler.

User avatar
eldin raigmore
korean
korean
Posts: 6387
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: The constants e and π in linguistics

Post by eldin raigmore » 17 Jan 2019 01:28

Lambuzhao wrote:
15 Sep 2018 12:35
....
√-1 natlangs
....
Aren’t √-1 ‘langs conlangs?

User avatar
eldin raigmore
korean
korean
Posts: 6387
Joined: 14 Aug 2010 19:38
Location: SouthEast Michigan

Re: The constants e and π in linguistics

Post by eldin raigmore » 17 Jan 2019 01:29

eldin raigmore wrote:
17 Jan 2019 01:28
Lambuzhao wrote:
15 Sep 2018 12:35
....
√-1 natlangs
....
Aren’t √-1 ‘langs conlangs?
gach wrote:
11 Sep 2018 11:44
To make this more rigorous, you certainly need a comparison with the amount of languages where the number of values that these grammatical features get is a whole number below e. For instance, how common are three gender systems when compared to two gender systems or the lack of gender (i.e. "one gender systems")? That's a nice idea for someone to play with WALS for an evening, looking through different grammatical features.

You'll have to deal with a fair bit of small number statistics when doing this, but I wouldn't be surprised to still find a good number of power laws out there. Some years ago I played a bit with the Princeton University WordNet database to figure out the distribution of polysemy in the English lexicon. What you find there, is that the distribution of definitions per word follows a fairly neat decreasing power law for all the major word classes, with an exponent -3.8 for nouns, -3.4 for adjectives, -3.5 for adverbs, and -2.7 for verbs. In other words, verbs are more polysemous than the other lexical word classes. The amount of data I had to work with here is significantly larger than what's possible when numbering the values of grammatical features, but I would suspect the underlying processes to be fairly similar there as well.
That’s interesting, @gach!

Wouldn’t you also have to compare them with those languages that have more than pi of whatever?

Post Reply