Robin wanted those two new gismu and so I started preparing their sounding.
At first I asked myself "Are those frequencies taken from Atlas of the world true or correct? Are there any flaws in the method of generating gismu from 6 languages?"
So here is my new method.
1.At first I opened this wikipedia
article. Unfortunately Wikipedia just removed the data that we need in it's newer revisions of the page.
2.Write out the number of native (I'll call them L1) speakers of first most common languages. Determine the number of L2 speakers as L2=(total number of speakers - L1 speakers).
3. In one special case we have L3 speakers.
4. Determine the corrected frequency as =L1+L2 / 2 + L3 / 3
5. Leave first 12 languages (I'll tell you later why 12)
5. Convert them to fractions so that the sum=1.
6.Therefore we get (3-letter ISO-codes of languages in the 1st column)
| |
l1 |
l2 |
l3 |
total |
cor.freq. |
fraction |
| cmn |
845 |
180 |
|
1025 |
935 |
0.246117 |
| eng |
375 |
375 |
750 |
|
812.5 |
0.213872 |
| spa |
329 |
61 |
|
390 |
359.5 |
0.09463 |
| ara |
232 |
220 |
|
452 |
342 |
0.090024 |
| hin |
240 |
165 |
|
405 |
322.5 |
0.084891 |
| ben |
181 |
69 |
|
250 |
215.5 |
0.056725 |
| rus |
144 |
106 |
|
250 |
197 |
0.051856 |
| por |
178 |
15 |
|
193 |
185.5 |
0.048829 |
| jap |
122 |
1 |
|
123 |
122.5 |
0.032245 |
| pun |
109 |
0 |
|
109 |
109 |
0.028692 |
| deu |
90 |
28 |
|
118 |
104 |
0.027376 |
| fra |
68 |
52 |
|
120 |
94 |
0.024743 |
| jav |
85 |
0 |
|
85 |
85 |
|
| wu |
77 |
0 |
|
77 |
77 |
|
| mar |
75 |
3 |
|
|
76.5 |
|
| tel |
70 |
5 |
|
75 |
72.5 |
|
| vie |
69 |
0 |
|
69 |
69 |
|
| |
|
|
|
|
|
|
| |
|
|
|
|
|
1 |
7.Now some general stuff. I stopped at French cuz it's a Romance language similar to Poruguese and Spanish so it might shift the frequencies of phonemes a bit.
8.Punjabi and Bengali are no longer considered as variations of Hindi.
9.Now let's find etymologies for our gismu.
I can recommend the following links (warning! unsorted)
10.Some dictionaries contain audio samples.
11.Notice that Punjabi, Bengali and Hindi (together with Urdu) are quite similar. You usually should get the same sounding.
12. So prepare the sounding using standard methods of it's adapting to lojban phonology as described in the CLL.
13.Now let's use scoreGismu perl app.
For "stomach" we get
| cmn |
vei |
0.253388 |
| eng |
beli |
0.22019 |
| spa |
bientre |
0.097425 |
| ara |
muada |
0.092683 |
| hin |
pet |
0.087398 |
| ben |
pet |
0.058401 |
| rus |
jeludyk |
0.053388 |
| por |
bariga |
0.050271 |
| jap |
xara |
0.033198 |
| pun |
pet |
0.028692 |
| deu |
magen |
0.028184 |
| fra |
vontr |
0.025474 |
With first 4 languages it outputs
| vreli
velti venli velbi |
0.691286 (rating) |
| With all 12 languages it outputs |
|
| vetli |
0.535
(rating) |
I'd choose {vetli} as it's similar to Chinese vei, English belly, Indian "pet" and Romance "ventr-"
For "back of human body" we get
| cmn |
beiji |
0.253388 |
| eng |
bak |
0.22019 |
| spa |
espalda |
0.097425 |
| ara |
vyxra |
0.092683 |
| hin |
pic |
0.087398 |
| ben |
pic |
0.058401 |
| rus |
spin |
0.053388 |
| por |
kostas |
0.050271 |
| jap |
se |
0.033198 |
| pun |
pic |
0.028692 |
| deu |
riuk |
0.028184 |
| fra |
dos |
0.025474 |
| Top score is |
|
| bekpi bajdi |
0.50684 |
4 langs |
|
|
|
| bekpi |
0.4697 |
12 langs |
I'd choose {bekpi} which reminds me of Chinese bei, English back, Indian pic and Russian sPIna.
Robin approved them several hours ago in chat but I'm free to suggestions.
Notes.
Still 6 languages is enough. 4 languages are not enough. More than 12 languages is a useless waste of time (but If you can collect translations please do).