Vreleksá

eldin raigmore · Posted: Sat Jun 28, 2008 6:58 pm Post subject: A completely isolating conlang.

Suppose I wanted to make a conlang in which every morpheme was a whole word and every word was just one morpheme.

Suppose I also wanted every word to be just one syllable (and every morpheme to be an entire syllable).

Suppose I wanted it to have as many words in it as English does; about 225,000 to 625,000.

Suppose I wanted it to have relatively-simple syllable structures; or, rather, not-too-complicated syllable structures.

That is, suppose I allowed all and only the following patterns of syllable-types:
(C)(C)V(C)
(C)V(C)(C)
(C)V(V)(C)
(C)(V)V(C)

(1) How many phonemes would I need?

If I have C consonants and V vowels, I can make
(C+1)*(V+VV+CV+VC)*(C+1) different syllables, assuming there are no phonotactic restrictions (which isn't that likely).

For instance I have the following table:
C V (C+1)*(V+VV+CV+VC)*(C+1)
46 3 636,192
42 4 658,244
39 5 672,000
36 6 648,906
34 7 651,700
32 8 635,976
31 9 663,552
30 10 682,310
28 11 629,068
27 12 630,336
26 13 625,482
25 15 669,240
24 16 650,000
23 17 626,688
22 19 643,264

All of those require a total of between 41 and 49 phonemes, which seems like a lot to me.

Even for fewer syllables, I get the following table:
C V (C+1)*(V+VV+CV+VC)*(C+1)
33 3 242,760
29 4 226,800
27 5 235,200
25 6 231,192
24 7 245,000
23 8 253,440
21 9 226,512
20 11 252,252
19 12 244,800
18 13 234,650
17 15 243,000
16 16 226,576

requiring 32 to 36 phonemes altogether; which still looks like it might be more than I'd want, though in fact I think it may be close to English's phoneme-inventory-size, at least by some counts.

(2) Can get a good lexicon-size with a reasonable number of phonemes?

(3) Would any completely isolating language (no polymorphemic words) without polysyllabic words and with the syllable structure I've specified, require tones to get a large enough lexicon?

If I added tones, then with about seven tones I could get 625000 or more syllables with the following combinations:
C V (C+1)*(V+VV+CV+VC)*(C+1)
14 11 99,000
15 10 104,960
16 8 94,792
17 7 95,256
19 6 108,000
20 5 101,430
22 4 103,684
24 3 97,500

all with about 25 to 27 phonemes, which is reasonably small. But I don't really like tones at all, to tell the truth; and even if I give in and use them, I don't think I'd want to use as many as seven. For one thing, no natlang has more than five level tones, so the language would require some glide tones or contour tones. Suppose there were three levels, H and M and L; then there could be six glides, HM, HL and ML for falls and MH, LH, and LM for rises. So i'd need to either have four or more levels, or use level tones and glide tones both, or use glide tones and contour tones (peaks (rise-falls) and dips (fall-rises)). (With three levels there'd be five possible peaks and five possible dips.)

---------------------------------------------

Chinese comes closest to being this kind of language; almost completely isolating, almost completely monosyllabic, and not-too-complicated syllable-structure. The languages related to Chinese spoken geographically near Chinese also seem to fit. A sizeable majority of those languages are tonal. Is there a good reason?
_________________
"We're the healthiest horse in the glue factory" - Erskine Bowles, Co-Chairman of the deficit reduction commission

langover94 · Joined: 21 Aug 2007 Posts: 509 Location: USA

I would say, if you do not want to have tones, the phoneme inventory would have to be about the size of English, maybe even with more vowels. This is just an estimation however, I could be exaddgerating. I believe that it would be easier to have 4-6 tones would make it easier to have a larger lexicon size. I don't really like tones either, but in this situation I think it would help a lot.
_________________
Join us at: The Renewed Spirits Forum!
Please join for good discussion. (We need members!)

Hemicomputer · Joined: 04 Feb 2008 Posts: 610 Location: Calgary, Alberta

I don't really know if this would work BUT: perhaps you could play around with length. Have long, medium, and clipped vowels, and possibly apply it to some consonants. It's not quite as inventory-expanding as tone, but it's much easier to use in speech.
_________________
Bakram uso, mi abila, / del us bakrat, dahud bakrita!

yssida · Joined: 16 Sep 2007 Posts: 253 Location: sa jaan lang

Is this more of a question about tonality and the existence of one syllable morphemes? I would be inclined to agree with your idea eldin. That is not to say that I do not have any reservations. Chinese (Mandarin?), though with its relatively small phoneme inventory and only 4 contours and one level tone (i'm not sure about this part), and having a lot of monosyllabic (though I'd say there are probably a lot more polysyllabic) morphemes, still is a nightmare for homophones. I say you should go with the most comfortable number of phonemes for you. I don't think you really need to differentiate every single morpheme. Surely context will enable the speakers to understand each other. Still, against my own idea, I have to say that a totally isolating language using only monosyllables is less tolerant of noise, and has less built in redundancy.
_________________
kasabot ka ani? aw di tingali ka bisaya mao na

my freewebs site

Tolkien_Freak · Posted: Sun Jun 29, 2008 8:05 pm Post subject:

Summary of the above-
To make an isolating lang with mainly monosyllables, you need one or more of:
Tones
Length distinctions
Lots of phonemes
Complex syllable structure

Otherwise, you end up needing more polysyllabic morphemes.

(English actually does a pretty good job with this - you can say a whole lot with a very limited number of polysyllables, and there's very little inflection. It's got its syllable structure going for it.)

eldin raigmore · Posted: Sun Jun 29, 2008 9:42 pm Post subject:

Thanks to you all.
Since this is going to be a multi-response and the things I'm responding to are lengthy (not individually, but in total), I'm not going to quote you.
I believe you will be able to tell who said whatever I'm responding to at the moment.

A phoneme inventory in the forties is not really unreasonable in natlangs. Some natlangs have around 20 vowels; some have almost 100 consonants. English probably has around 12 vowels and around 32 consonants, depending on who's counting and what dialect they're looking at and what they count as "different phonemes". It's just that I personally prefer the mid-to-high-twenties for number of phonemes, and "around five" for number of vowels. But that might give me only around 16,875 different syllables of the CCVC structure, unless tones or some such distinguished. The middle option, with phoneme-inventory-size in the low-to-mid-thirties and around three tones, shoud be an acceptable compromise IMO.

The structure I talked about already has phonemic length for vowels, but only two lengths, not three. If I went two three lengths for vowels I might get anywhere from 1.5 to 4.5 times as many syllables, depending on what I allowed (and depending on how correct my math is!). So that might be a good suggestion; even if it didn't stop me from needing tones, it might cut down a lot on the number of different tones I'd need. (But to tell the truth I "dislike" phonemic length more than I "dislike" tones -- or, at least I have up 'til now; perhaps I'll quit disliking either one of them.)

Including two different phonemic lengths for consonants, however, could be a lot more productive than you seem to have thought. It would double the number of syllables which contained one consonant; quadruple the number of syllables which contained two consonants; and give me eight times as many syllables that contain three consonants. That could be a really big boost.

Yssida, since I asked four questions, I believe my answer to your "Is this more of a question about ..." question is probably "Yes". Wink

I asked about both of the topics you mentioned.

Yssida and Tolkien_Freak; your summaries are correct. Thanks. So are your remarks about Mandarin and English; and they are helpful, as well.
_________________
"We're the healthiest horse in the glue factory" - Erskine Bowles, Co-Chairman of the deficit reduction commission

Hemicomputer · Joined: 04 Feb 2008 Posts: 610 Location: Calgary, Alberta

Yes, THREE lengths is a bit too much. That would be hard to speak and it would mess with the orthography. Many lengths means diacritics, which you already need for tone.

Off the topic of questions, this is quite an interesting idea for a language. It'll be interesting to see it (or really any of the languages that elden has constructed).
_________________
Bakram uso, mi abila, / del us bakrat, dahud bakrita!

eldin raigmore · Posted: Tue Jul 01, 2008 12:17 am Post subject:

Hemicomputer · Joined: 04 Feb 2008 Posts: 610 Location: Calgary, Alberta

eldin raigmore · Posted: Tue Jul 01, 2008 6:01 pm Post subject:

About the "totally isolating conlang";

The reason for sticking to the slightly-complex syllable structures
(C)(C)V(C)
(C)V(C)(C)
(C)V(V)(C)
(C)(V)V(C)
is so they'll be writable as a syllable-block after the manner of Hangul.

For a C1C2V(C3) syllable I want the C2V part to be representable as a nucleus; I want the C2 to be a semivowel or glide or liquid or approximant, or some other highly sonorant consonant.

For a (C1)VC2C3 syllable, likewise, I want the VC2 part to be representable as a nucleus; so I want the C2 to be highly sonorant, for instance a nasal, approximant, liquid, glide, or semivowel.

For a (C1)V1V2(C2) syllable, I want the V1V2 part to be representable as a nucleus; either a diphthong or a long vowel.

- - - - - - - - - - - - - - - - - - - - - - - -

A just-slightly-more-complicated syllable structure would get me >=625,000 syllables without tones and with a smallish number of phonemes.
For instance, either
C(C)(C)V(V)
or
C(C)V(V)(C)
or
CV(V)(C)(C)

would get me >=625,000 syllables with just 29 phonemes; 15 to 20 consonants and 9 to 14 vowels. Or if 9 is too many vowels,
with 22 consonants and 8 vowels, or
with 24 consonants and 7 vowels, or
with 26 consonants and 6 vowels, or
with 30 consonants and 5 vowels, or
with 34 consonants and 4 vowels, or
with 42 consonants and 3 vowels. (Though IMO 42 consonants is kind of a lot, there are natlangs with that many.)

If I used the pattern
(C)(C)V(V)(C)(C)
(which is the simplest pattern that includes all four of the patterns I did use),
I could get >=625000 syllables with seventeen or eighteen phonemes;
ten to twelve consonants and five to seven vowels for seventeen phonemes, or nine to fifteen consonants and three to nine vowels for eighteen phonemes.

But neither of those patterns would be writable as a syllable-block, IMO.

-------------------------------------------------------------------------------------

eldin raigmore · Posted: Tue Aug 12, 2008 5:55 pm Post subject:

Correction;
With at least 39 phonemes, including at least 26 consonant phonemes and at least 11 vowel phonemes, you could get as many syllables as English has words using the (C) { V(C|V) | (C|V)V }(C) pattern; provided there aren't too many or too strict phonotactic constraints.

(That is, the union of the
(C)V(C)(C)
(C)V(V)(C)
(C)(C)V(C)
(C)(V)V(C)
patterns.
Nuclei can be V or VV;
codas can be absent, or C, or CC;
rimes can be V or VC or VCC or VV or VVC;
onsets can be absent, or C, or CC;
so syllables can be
V
CV
VC
VV
CCV
CVC
VCC
CVV
VVC
CCVC
CVCC
CVVC . )

C V Syllables
26 13 625,482
27 12 630,336
28 11 629,068
23 17 626,688
24 16 650,000
25 15 669,240
30 10 682,310
31 9 663,552
32 8 635,976
22 19 643,264
34 7 651,700
36 6 648,906
39 5 672,000
42 4 658,244
46 3 636,192
_________________
"We're the healthiest horse in the glue factory" - Erskine Bowles, Co-Chairman of the deficit reduction commission