Talk:Tiki forum archive/Forum 5/8664
Source: 2013 Tiki forum 5, thread #8664. Original authors, UTC times, post IDs, and reply-parent IDs are preserved. This is an archive, not an active article discussion.
How many possible gismu?
How many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:05 UTC · Tiki post #8664
suggesting or considering making any new gismu; however, I am not saying that I would absolutely never approve of it during the baseline period, but it would take some highly unusual (and extremely unlikely) events for me to do so. What I am imagining is a future where Lojban has a large base of users, and on rare occasions a concept is found to be so commonly used and so important that it would unreasonable to deny it a gismu.
The question I really want answered is: What is the _smallest_ number of new gismu that could fill up gismu space? -- remember any gismu, current or proposed, blocks up other possible gismu: "If the proposed gismu was identical to an existing gismu except for a single consonant, and the consonant was 'too similar' based on the following table, then the proposed gismu was rejected." This is in addition to the rule that two gismu can not differ in only the final vowel; the brod-series is an exception. (In retrospect, it would have probably have been better to use rafsi cmavo for the assignable pro-bridi, but it's to late to change now.)
So as I've said I really want to know: What is the _smallest_ number of new gismu that could fill up gismu space? However, I don't see a way to answer that except by a brute-force algorithm, so I thought to answer the simpler question: What is the _smallest_ number of gismu that could fill gismu space, starting from an empty gismu space? However, I have not been able to think of a way to answer that without a brute-force algorithm, either; because it is possible for two gismu to block up some of the same "possible" gismu.
The reason that I ask this is that in the future that I imagine in the first paragraph, I want to know how "extensible" the set of gismu is. I ask for the _smallest_ number on the assumption that maximizing the number of possible gismu will not be a consideration early on; so finding the worse possible filling would, if anything, underestimate the number of gismu that would have to be created in this hypothetical future to fill gismu space.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
stevo · 2009-12-03 12:05 UTC · Tiki post #8665 · reply to Tiki post #8664
[email protected] writes:
> What I am imagining is a future
> where Lojban has a large base of users, and on rare occasions a
> concept is found to be so commonly used and so important that it
> would unreasonable to deny it a gismu.
>
This is precisely what lujvo and fu'ivla are for. New gismu would be useful for brand-new *fundamental* concepts. That's not likely to happen very often.
stevoHow many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:05 UTC · Tiki post #8666 · reply to Tiki post #8664
> In a message dated 8/24/2009 19:02:46 Eastern Daylight Time, > [email protected] writes: > > > What I am imagining is a future > where Lojban has a large base of users, and on rare occasions a > concept is found to be so commonly used and so important that it > would unreasonable to deny it a gismu. > > > This is precisely what lujvo and fu'ivla are for. New gismu would be useful > for brand-new *fundamental* concepts. That's not likely to happen very > often. > > stevo
I agree that even at the rate at which I am imagining new gismu to be created -- which may be ridiculously high -- there would be no actual problem in this regard for millennia. However, even though I can write a program to implementing the brute force algorithm that I have been able to come up with to answer my second question -- What is the _smallest_ number of gismu that could fill gismu space, starting from an empty gismu space? -- the program would based on my initial guess be O(n!); I don't expect that the program would finish in my lifetime. The first question -- What is the _smallest_ number of new gismu that could fill up gismu space? -- might not be answerable except by the brute-force approach; but there has got to be a better way to answer the second -- What is the _smallest_ number of gismu that could fill gismu space, starting from an empty gismu space? -- I've just not thought of it. Consider that question as a mathematical puzzle question with the rules for Lojban being the set-up for the question; one doesn't have to consider what inspired me to ask the question; I mentioned because I thought the notion of what inspired me to ask that question would be interesting, _not_ because I thought that it had any real _practical_ applications -- Mathematicians look for higher and higher pairs of amicable numbers (http://http://en.wikipedia.org/wiki/Amicable_numbers) but as far as I know it is not for practical applications of them.
Also, is a coincidence that your name "stevo" has the morphological form of a gismu, but does not show up in even the "experimental gismu" on the Lojban dictionary web search (http://www.lojban.org/cgi-bin/dict.pl?Form=dict.pl1&Query=
&Strategy=*&Database=jbo->en&submit=Submit+query), as "gumri" does.
I personally will not use "la gumri" or the "la marca" that I remember seeing discussed a long time ago on a different message forum. "la marca" could be considered taking liberties with the morphology, so if you are using "la stevo" in the same way, you can hardly criticze me for my imagination since I'm just personally curious about the mathematical answer to my question, and don't intend to actually propose new gismu.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
stevo · 2009-12-03 12:05 UTC · Tiki post #8667 · reply to Tiki post #8664
[email protected] writes:
> Also, is a coincidence that your name "stevo" has the morphological
> form of a gismu, but does not show up in even the "experimental gismu"
> on the Lojban dictionary web search
> (http://www.lojban.org/cgi-bin/dict.pl?Form=dict.pl1&Query=
> &Strategy=*&Database=jbo->en&submit=Submit+query), as "gumri" does.
> I personally will not use "la gumri" or the "la marca" that I remember
> seeing discussed a long time ago on a different message forum.
> "la marca" could be considered taking liberties with the morphology,
> so if you are using "la stevo" in the same way, you can hardly
> criticze me for my imagination since I'm just personally curious
> about the mathematical answer to my question, and don't intend to
> actually propose new gismu.
>
My name "stevo" deliberately has the form of a gismu, but when I use it in Lojban, I always use "stevon". I never try to use "stevo" as a gismu. What would be its meaning? "x1 is the person called "stevo" "?
mu'o mi'e stevonHow many possible gismu?
stevo · 2009-12-03 12:07 UTC · Tiki post #8668 · reply to Tiki post #8664
[email protected] writes:
> I agree that even at the rate at which I am imagining new gismu to be
> created -- which may be ridiculously high -- there would be no actual
> problem in this regard for millennia. However, even though I can write
> a program to implementing the brute force algorithm that I have been
> able to come up with to answer my second question -- What is the
> _smallest_ number of gismu that could fill gismu space, starting
> from an empty gismu space? -- the program would based on my initial
> guess be O(n!); I don't expect that the program would finish in my
> lifetime. The first question -- What is the _smallest_ number
> of new gismu that could fill up gismu space? -- might not be
> answerable except by the brute-force approach; but there has got to
> be a better way to answer the second -- What is the _smallest_
> number of gismu that could fill gismu space, starting from an empty
> gismu space? -- I've just not thought of it. Consider that question
> as a mathematical puzzle question with the rules for Lojban being the
> set-up for the question; one doesn't have to consider what inspired
> me to ask the question; I mentioned because I thought the notion of
> what inspired me to ask that question would be interesting, _not_
> because I thought that it had any real _practical_ applications --
> Mathematicians look for higher and higher pairs of amicable numbers
> (http://http://en.wikipedia.org/wiki/Amicable_numbers) but as far as
> I know it is not for practical applications of them.
>
This is an interesting problem. Doing it by hand might take a few days or a week or two. How do you get O(n!) for it?
stevoHow many possible gismu?
lukebergen · 2009-12-03 12:07 UTC · Tiki post #8669 · reply to Tiki post #8664
Could you put the requirements (input, expected output, rules, etc...) into a short, concise list?
On Mon, Aug 24, 2009 at 9:47 PM, <[email protected]> wrote:
> In a message dated 8/24/2009 21:08:35 Eastern Daylight Time, > [email protected] writes: > > > I agree that even at the rate at which I am imagining new gismu to be > created -- which may be ridiculously high -- there would be no actual > problem in this regard for millennia. However, even though I can write > a program to implementing the brute force algorithm that I have been > able to come up with to answer my second question -- What is the > _smallest_ number of gismu that could fill gismu space, starting > from an empty gismu space? -- the program would based on my initial > guess be O(n!); I don't expect that the program would finish in my > lifetime. The first question -- What is the _smallest_ number > of new gismu that could fill up gismu space? -- might not be > answerable except by the brute-force approach; but there has got to > be a better way to answer the second -- What is the _smallest_ > number of gismu that could fill gismu space, starting from an empty > gismu space? -- I've just not thought of it. Consider that question > as a mathematical puzzle question with the rules for Lojban being the > set-up for the question; one doesn't have to consider what inspired > me to ask the question; I mentioned because I thought the notion of > what inspired me to ask that question would be interesting, _not_ > because I thought that it had any real _practical_ applications -- > Mathematicians look for higher and higher pairs of amicable numbers > (http://http://en.wikipedia.org/wiki/Amicable_numbers) but as far as > I know it is not for practical applications of them. > > > > This is an interesting problem. Doing it by hand might take a few days or > a week or two. How do you get O(n!) for it? >
> stevoHow many possible gismu?
clsn · 2009-12-03 12:07 UTC · Tiki post #8670 · reply to Tiki post #8664
> In a message dated 8/24/2009 19:02:46 Eastern Daylight Time, > [email protected] writes: > > >> What I am imagining is a future >> where Lojban has a large base of users, and on rare occasions a >> concept is found to be so commonly used and so important that it >> would unreasonable to deny it a gismu. > > > This is precisely what lujvo and fu'ivla are for. New gismu would be > useful for brand-new *fundamental* concepts. That's not likely to > happen very often. Well, IIRC, gismu were never claimed to be truly "fundamental concepts"; that's just asking for trouble (are sheep more fundamental than gazelles? and don't even start on the cultural gismu). They're primitive only in the internal-to-Lojban sense; we would be foolish to claim that they are exhaustively "primitive" in any universal sense.
Conversely, I can't see why anyone would feel somehow compelled to add gismu; if you allow end-stage fu'ivla (unmarked) what really is the difference between a fu'ivla and a new gismu? The only functional difference is rafsi, and the original question doesn't even deal with rafsi. So far as I can tell, if new concepts are needed, however fundamental, fu'ivla ought to be able to handle anything.
Now, if we start considering rafsi, all kinds of extraneous details start mattering all of a sudden. For example, I noticed a while ago that there is really no need for a gismu that has a CCV rafsi to have any other. CCV rafsi are superior to the others: they can be initial, medial, or final (CVC can't be final) and never need -r- or -n- hyphens to form a consonant cluster if used at the beginning (as CVV or CV'V might), and are always one syllable. When I realized this, I wondered if we weren't "wasting" good rafsi on words that didn't need more than one, while other gismu went without. I was reassured that these rafsi weren't wasted; that the other rafsi for gismu which had CCV rafsi were unavailable for use by any other word anyway. If we start coining new gismu and giving them rafsi, does this have to be taken into account? Surely it would be too late to reassign the "wasted" rafsi, but would that not have a negative impact on the value of the new gismu?
Anyway, don't mess with adding gismu. Nothing a really really official fu'ivla can't handle.
~markHow many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:07 UTC · Tiki post #8671 · reply to Tiki post #8664
> My name "stevo" deliberately has the form of a gismu, but when I use it in > Lojban, I always use "stevon". I never try to use "stevo" as a gismu. What > would be its meaning? "x1 is the person called "stevo" "? > > mu'o mi'e stevon
As I remember the argument was that the person named "bunre" did not actually have to be brown, perhaps they are named "Emmett Brown". Since cmeme do not actually imply meaning, though they can suggest meaning. (Even in English, most of the people who have a last of "Smith" could trace it back to an actual ancestor who was an actual smith.) Since, cmeme are not related to meaning, the expression "la marca" can be unambiguously de-composed in to "la" followed by a brivla, even if that brivla is meaningless, so as long as the "la" is followed by a morphologically correct tanru, brivla, or cmevla, the use of meaningless brivla (inculding meaningless gismu) is not ambiguous, as long if it is morphologically correct; that what I mean about "taking liberties with the morphology"; it's technically legal, but pushing the boundaries of what was meant to be done.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:07 UTC · Tiki post #8673 · reply to Tiki post #8664
> I've read your previous emails and am also not sure where O(n!) came from. > Could you put the requirements (input, expected output, rules, etc...) into > a short, concise list?
You'll be happy to know that I found a lower bound, search for the text string "LOWER BOUND" to quickly reach it, but a brief summary is that there is no way to use less than 7429 additional gismu to completely block up the gismu list:
"So even given the rate at which I imagine that new gismu might be created; it would be extremely unusual for there to be more than 10 per century, usually less -- 10 per century would be 100 per millennium, leading to at least 7429/100 = 74+ millennia before there is even a possibility of running out of gismu."
You think that this is too small; it is; as I discuss under the title mentioned above, there is no chance that the "limit" of 7429 is the actual limit; I have merely proved it to be a lower bound; the actual value can not be less than 7429; there is no doubt in my mind that it is larger; I just can't determine how much larger.
There are two questions I am considering: 1) What is the _smallest_ number of new gismu that could fill up
gismu space?
2) What is the _smallest_ number of gismu that could fill gismu
space, starting from an empty gismu space?
The rules are that a gismu blocks the use of some possible gismu that differ by one letter. The blocked possible gismu (eg, "gismu") are: 1) gismu that differ only in the last letter (eg, "gisma", "gisme",
"gismi", "gismo"
2) The gismu that differ by one letter for the sets below:
b p, v c j, s d t f p, v g k, x j c, z k g, x l r m n n m p b, f r l s c, z t d v b, f x g, k z j, s (eg, "kismu", "xismu", "gicmu", "gizmu", "gisnu")
There are two gismu forms CVCCV and CCVCV. Since the 2nd and 3rd characters are VC in the CVCCV and CV in to CCVCV, they can not differ in just one letter. So a CVCCV can not block a CCVCC, nor can a CCVCV block a CVCCV, so at least these two sets can be considered separately.
Consider the CVCCV. The CC is a permissible consonant pair; there are 7 unvoiced consonants, 6 voiced consonants, and 4 syllabic consonants. A permissible consonant pair can be made with: 1) Two unvoiced consonants; given the restriction on double
consonants: this means 7*6 pairs or 42 of this type.
2) Two voiced consonants; given the restriction on double
consonants: this means 6*5 pairs or 30 of this type.
3) A unvoiced consonant followed by a syllabic consonant:
7*4 = 28
4) A syllabic consonant followed by a unvoiced consonant:
4*7 = 28
5) A voiced consonant followed by a syllabic consonant:
6*4 = 24
6) A syllabic consonant followed by a voiced consonant:
4*6 = 24
7) Two syllabic consonants; given the restriction on double
consonants: 4*3=12
The sum of these numbers is 188. However, the otherwise legal pairs; "cs" ,"sc" ,"jz" ,"zj", "cx", "xc", "kx", "xk", and "mz" are forbidden. Subtracting out these 9 leaves 179 possible values for the CC in CVCCV. For each V there are 5 possibilites, for the solitary C there are 17. The total possible CVCCV gismu is 17*5*179*5 = 76075. In order to make sure that I get the worst possible case; ie, the one that uses the fewest gismu, I have to test all possibilities.
Let's use the second question, I make a list of all the possibles: "babda" to "zuzvu". Then start by placing "babda" in place and marking off all gismu that prevents: "babde", "babdi", "babdo", "babdu", "pabdu", "tabdu". The gismu "bapda" and "babta" are already probitted because "pd" and "bt" are not permissible consonant pairs in the first place. So then, I place the next possible word on the list; it turns out to be "babga", mark the now unusable gismu, and so on. Finally I get to count the gismu when space runs out. Then I have remove the last added gismu remove the mark from the ones that excluded and try the next possible one of the now available gismu, and see if that fills gismu space, then when I finish counting all possibles that are made by taking one off the list. Then I need to take not only the last one added, but the next to the last one added. Then I place a new one in the next available space left in alphabetical order. Then try other possible next gismu for that. Because remember, it is not likely that I will get the _smallest_ number to fully pack the CVCCV gismu space on the first try; so eventually I have to test all possibles and see what the smallest count is. It has been a long time since I first considered this problem; I have forgotten why I concluded that the algorithm was O(n!), rather than O(e^n), which is worse.
The procedure for packing CCVCV gismu space is the same, except there are only 48 permissible initial pairs: 48*5*17*5 or only 20400 slots there.
It seems to me though that at least in the case that one starts with empty gismu space, one should be able to use the fact that there are 6 collision cycles: b-p-f-v-back to b, c-j-z-s-, g-k-x-, d-t-, l-r-, m-n-. Notice all the cycles but the first end with "-" because each cycle catches its own tail; the first catches its own tail, too, I just explicitly noted that in the cycle back at the end. The first problem is that while "nz" in a place should block "mz", "mz" is already forbidden as a permissible pair. One could try to use the fact that in a dual voiced pair; eg, "bg", "b" can not block "p" because the other consonant, "g", is voiced and "p" is unvoiced; but in "tablu", "b" does block "p", because "pl" is a permissible pair; so "taplu" would otherwise be permitted. The problem is made worse by the fact that some pairs that would be permitted under the unvoiced/voiced/syllabic rule are forbidden; I mentioned "mz"; there are 8 more. This problem is even worse for the CCVCV gismu space, since there are 128 otherwise legal spaces that are forbidden; and the rest don't follow any truly regular pattern.
There may be a way to figure out a "most efficient"; ie, using the fewest gismu; to block up gismu space without searching through all possibles; I'd think something could be done with ranking the consonant pairs by how efficiently they block other consonant pairs; ie, how many consonant pairs does "bl" block? When used as a permissible pair in the CVCCV space problem, and then when used as a permissible initial pair in the CCVCV space problem? And I haven't even considered the first question, in which the current gismu are already "fixed" in place, in the analysis.
LOWER BOUND
I can readily believe that the first question -- What is the _smallest_ number of new gismu that could fill up gismu space? (Assumes that the current gismu are "fixed") -- is too computationally intensive to be solved with current computers. The second question -- What is the _smallest_ number of gismu that could fill gismu space, starting from an empty gismu space? -- is not likely to be so; I strongly suspect that I am missing some trick that could vastly reduce the number of operations required. Remember, of course, these are really only interesting question as mathematical puzzles; even if someone does come up with the answer, or even comes up with an idea that helps me reach an answer; the answer won't be used for any practical purpose.
The one "minimum" calculation that I have been able to come up with is that each of the 3 consonant can block _at most_ 2 gismu each, the final vowel rule means that each gismu blocks another 4 gismu, so no gismu can block more than 10 other gismu; using this worst-worst case analysis each gismu can hold at most 11 slots in gismu space; 1 for the gismu itself, 10 for the other blocked slots. Given 76075 CVCCV slots, this means that there is no way to cover that space with less than 6916 gismu; 6915 could block up only 6915*11 = 76065; there are now 10 slots left, so, in theory, the 6915th gismu could cover them all. In CCVCV space, there are 20400 slots; this leads to an absolute minimum 1855 gismu to lock out all possible slots; 1854 could block up 1854*11 = 20394 slots, leaving the 1855th one to block up the remaining 6 open slots. That makes a total of 1855+6916 = 8771. Having thought of this analysis, there is room for the number of gismu to expand by another 5 times its current size; including the brod-series, there are 1342 current gismu; even if there were room for only 8771 or 8771-1342 = 7429. 7429/1342 = 5.5+. So even given the rate at which I imagine that new gismu might be created; it would be extremely unusual for there to be more than 10 per century, usually less -- 10 per century would be 100 per millennium, leading to at least 7429/100 = 74+ millennia before there is even a possibility of running out of gismu. Of course, there is no way in censored that each gismu is going to block up 10 slots of its very own in addition to itself, so this is a very conservative minimum. I'm more than comfortable with the room to grow.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:08 UTC · Tiki post #8674 · reply to Tiki post #8664
> gismu to expand by another 5 times its current size; including the > brod-series, there are 1342 current gismu; even if there were room > for only 8771 or 8771-1342 = 7429. 7429/1342 = 5.5+.
Change "by another 5 times its current size" to "_at least_ by another 5 times its current size".
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:08 UTC · Tiki post #8675 · reply to Tiki post #8664
> [email protected] wrote: > Anyway, don't mess with adding gismu. Nothing a really really official > fu'ivla can't handle. > > ~mark
I assure you that I have no intention of actually doing so; the idea of doing so was a theoretical question; it is just that the actual mathematical question I asked has a solution, and I wanted to know if anyone knew what it was.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Arran <[email protected]> · 2009-12-03 12:08 UTC · Tiki post #8676 · reply to Tiki post #8664
sounds will be added to the language maybe within 500 years, if it were to 'live' so to speak.
On Tue, Aug 25, 2009 at 4:59 PM, H. Felton <[email protected]> wrote:
> On Mon, Aug 24, 2009 at 10:28 PM, Mark E. Shoulson<[email protected]> wrote: > > [email protected] wrote: > > Anyway, don't mess with adding gismu. Nothing a really really official > > fu'ivla can't handle. > > > > ~mark > > I assure you that I have no intention of actually doing so; the idea of > doing so was a theoretical question; it is just that the actual > mathematical question I asked has a solution, and I wanted to know > if anyone knew what it was. > > > To unsubscribe from this list, send mail to [email protected] > with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if > you're really stuck, send mail to [email protected] for help. >
>How many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:08 UTC · Tiki post #8677 · reply to Tiki post #8664
> pe'i Given the time frame your thinking of it's more a mute issue. New > sounds will be added to the language maybe within 500 years, if it were to > 'live' so to speak.
As soon as I had the actual number of 7429, even though it's only a lower bound, I knew that my fancied future was possible. (Actually, I realized that before I finished calculating it.) I hadn't thought of the possibilty of adding new vowels or consonants to the language, but as you say, in even as short a time as 500 years that's not unreasonable; and that would expand rafsi space as well as gismu space.
I'm the kind of person who feels more comfortable when I can put numbers to things, rather than relying on a gut notion; eg, "of course, there's plenty of space for new gismu to grow in to" vs "there's absolutely at least enough gismu space to accomodate another 500% of growth of the gismu set".
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Jorge Llambías <[email protected]> · 2009-12-03 12:10 UTC · Tiki post #8681 · reply to Tiki post #8664
> > The one "minimum" calculation that I have been able to come up with > is that each of the 3 consonant can block _at most_ 2 gismu each, the > final vowel rule means that each gismu blocks another 4 gismu, so no > gismu can block more than 10 other gismu; using this worst-worst case > analysis each gismu can hold at most 11 slots in gismu space; 1 for > the gismu itself, 10 for the other blocked slots.
You can improve this bound a little by noting that a CC pair can block at most 3 other CC pairs (never 4), so no gismu form can block more than 9 other gismu forms.
mu'o mi'e xorxes
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
stevo · 2009-12-03 12:10 UTC · Tiki post #8682 · reply to Tiki post #8664
[email protected] writes:
> Since, cmeme are not related to meaning, the expression
> "la marca" can be unambiguously de-composed in to "la" followed by
> a brivla, even if that brivla is meaningless, so as long as the
> "la" is followed by a morphologically correct tanru, brivla, or
> cmevla, the use of meaningless brivla (inculding meaningless gismu)
> is not ambiguous, as long if it is morphologically correct; that what
> I mean about "taking liberties with the morphology"; it's technically
> legal, but pushing the boundaries of what was meant to be done.
>
Since I don't want to push that particular boundary at the moment, I change "stevo" to the cmevla "stevon" in Lojban.
stevoHow many possible gismu?
stevo · 2009-12-03 12:10 UTC · Tiki post #8683 · reply to Tiki post #8664
[email protected] writes:
> "So even given the rate at which I imagine that new gismu
> might be created; it would be extremely unusual for there
> to be more than 10 per century, usually less -- 10 per
> century would be 100 per millennium, leading to at least
> 7429/100 = 74+ millennia before there is even a possibility
> of running out of gismu."
>
You're assuming that change in the world is linear. It's not. It increases more than exponentially. So 10 new gismu per century is a very (hugely) conservative estimate. After the singularity occurs (in about 2045), we might use up all those new gismu in no time at all, since we can't predict what the world will be like after that point.
stevoHow many possible gismu?
stevo · 2009-12-03 12:10 UTC · Tiki post #8684 · reply to Tiki post #8664
[email protected] writes:
> I assure you that I have no intention of actually doing so; the idea of
> doing so was a theoretical question; it is just that the actual
> mathematical question I asked has a solution, and I wanted to know
> if anyone knew what it was.
>
Apparently no one has thought of the question before. Bravo to you!
stevoHow many possible gismu?
"H. Felton" <[email protected]> · 2009-12-03 12:11 UTC · Tiki post #8695 · reply to Tiki post #8664
> In a message dated 8/25/2009 02:53:53 Eastern Daylight Time, > [email protected] writes: > > >> "So even given the rate at which I imagine that new gismu >> might be created; it would be extremely unusual for there >> to be more than 10 per century, usually less -- 10 per >> century would be 100 per millennium, leading to at least >> 7429/100 = 74+ millennia before there is even a possibility >> of running out of gismu." >> > > You're assuming that change in the world is linear. It's not. It > increases more than exponentially. So 10 new gismu per century is a very > (hugely) > conservative estimate. After the singularity occurs (in about 2045), we > might use up all those new gismu in no time at all, since we can't predict > what > the world will be like after that point. > > stevo
If we hit that point we could well speculate about communicating by exchanging "core dumps" of the relevant brain states; I could not only communicate to you that I saw something blue, but also _exactly_ what shade of blue it was, by copying that part of my memory and and including it in the "core dump"; I could also communicate to you everything I know about a subject encoding as a "core dump" the brain states of the region(s) where that information is stored. (I don't say neuron states because I don't assume that we will be using a structure that involves neurons at that point.) IF the singularity occurs, there is, as you say, no way to predict anything beyond that point. If the world is to become less comprehensible to us than our culture and lives would be to a Neanderthal, we can't reasonably predict how that will affect things; so there is no way to plan for it.
I put those speculations in the same catagory as http://www.xibalba.demon.co.uk/jbr/futurese.html puts: If twenty-third century computer geeks have cybernetic implants to let them offload cognitive processes to specialised hardware, new slang possibilities arise such as inhumanly complex versions of "rot-13ed Pig Latin". or In a "world state" with an anglophone bureaucracy of artificial intelligences, ANSI-standard English could function as a sort of unnaturally preserved lingua franca even if the human native speakers died out. in his speculations about the future of English in the next millennium, and for pretty much the same reason: "because they don't make the language's future form any more predictable".
But yes a singularity-type of event could make Lojban as obsolete as pre-hominid's pointing and "grunting"; but I just don't think that I could in any real sense reasonably speculate about how a culture less comprehensible to us than our culture and lives would be to a Neanderthal, would affect Lojban _in particular_. I will grant that the "prediction" of "74+ millennia" is not worth a lot because of the likelyhood of these types of changes in 74 millennia; my speculations about the future of Lojban might be accurate out to a few millennia from now, but I'll not even take that for granted -- imagine the predictions that I would have made about what computers would be like in 2000, if I lived in 1950.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Alex-J · 2009-12-03 12:11 UTC · Tiki post #8696 · reply to Tiki post #8664
<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /> Sorry for not responding sooner, I learned a trick awhile ago that relates to figuring out how many possible gismu their can be (or at least getting a very close estimate).
Gismu can be C V C C V or C C V C V
In lojban there are 18 consonants and 6 vowels.
For now lets assume that C V C C V is the only way to have a gismu, and that within a gismu a letter cannot be repeated.
For my first letter I can have any consonant. So I have 18 letters to choose from For my second letter I can have any vowel. So I have 6 letters to choose from For my third letter I can have any of the remaining consonants. So I have 17 letters to choose from For my fourth letter I can have any of the remaining consonants. So I have 16 letters to choose from. For my fifth letter I can have any of the remaining vowels to choose from. So I have 5 vowels to choose from.
So C V C C V Becomes 18 * 6 * 17 * 16 * 5
Which equals 146,880 Now we subtract the 1300 current gismu
This equals 145,580 possible gismu
Now there are 146,880 more combinations using C C V C V But they would be so close to the 145,580 others that I won’t count them (even though some of them have to possibility to work)
Some of the 145,580 possible gismu I have counted will be somewhat hard to say. But because of the equation I used their will be no letters that repeat themselves, so the gismu that can’t be spoken won’t make up too large of a portion of the 145,580. Plus if I couldn't say the word sdaro (CCVCV) I could be able to say sadro (CVCCV) this uses the same letters, but swaps the "d" and "a", which isn't another combination I have already counted because its part of the CVCCV which I discarded.
Since some gismu have a repeated letter (like cmene) there are more possible gismu that have repeated letters (For Example: my equation could turn into 18*6*18*17*5). The added gismu I would have haven't been included in my 145,580.
So the gismu I have counted but don’t work will probably be made up for by the gismu I haven’t counted but do work with extra gismu to spare
Because of this I’m not very concerned about us running out of gismu, even if we make-up 5 new gismu per year, because we wouldn’t run out for 29,116 years.
Also if we run out of gismu that fit within the limits we have created we could just expand the limits of what letter combinations we consider gismu (so we could continue to make new gismu). It would be hard, we would have to make some serious changes in the language but it is possible. (For example 8 letter words could become gismu)
I know I haven’t proved how my equation works; it is very hard to explain unless you’re here with me. But if anyone is interested they can e-mail me at [email protected] and I will try to explain, and show them a few examples as to why this does work. ____________________________________________________________ Let great B to B marketing solutions propel your brand to new heights! Click now!
http://thirdpartyoffers.juno.com/TGL2131/fc/BLSrjnsKrsomdzkCOUTx77j4QA74JJq9mQlKRyFKQTIuJBqQgbJvEUnpzQs/How many possible gismu?
stevo · 2009-12-03 12:11 UTC · Tiki post #8697 · reply to Tiki post #8664
[email protected] writes:
> In lojban there are 18 consonants and 6 vowels.
>
I know of only 17 consonants: b c d f g j k l m n p r s t v x z. y'y bu doesn't count, especially for gismu.
What's the other consonant?
stevoHow many possible gismu?
stevo · 2009-12-03 12:12 UTC · Tiki post #8698 · reply to Tiki post #8664
[email protected] writes:
> For my second letter I can have any vowel. So I have 6 letters to choose
> from
>
gismu don't have y bu. Are you making a new rule?
stevoHow many possible gismu?
Alex-J · 2009-12-03 12:12 UTC · Tiki post #8699 · reply to Tiki post #8664
Which is 122,400 minus 1300 (current gismu) Which is 121,100 possible gismu. Everything else in the other e-mail should be true. Except now if we added 5 gismu every year we'd only have 24,220 years. ____________________________________________________________ Click for a wide selection of quality scales.
http://thirdpartyoffers.juno.com/TGL2131/fc/BLSrjnsEGFKYbhJ8KKOFgTf7vPgpOtpeTyZVtC3BSvu7z5PdfOS5jWmPq9q/How many possible gismu?
Alex-J · 2009-12-03 12:12 UTC · Tiki post #8700 · reply to Tiki post #8664
I'm really sorry, I haven't even been thinking, I just totaled up my consents and vowels, I wasn't even thinking about it when I added y!!!! Which should be now (if I haven't overlooked anything) 17 * 5 * 16 * 15 * 4 which is 81,600 .u'u ____________________________________________________________ Online Medical Insurance Get free online medical insurance quotes and save more money today.
http://thirdpartyoffers.juno.com/TGL2131/c?cp=VA-PaAFoNloApIg0lQmbaAAAJz3-M2YWXjKzYSbRbfae-XjmAAQAAAAFAAAAAJqZGT4AAAAAAAAAAAAAAAAAElA1AAAAAA==How many possible gismu?
Minimiscience · 2009-12-03 12:12 UTC · Tiki post #8701 · reply to Tiki post #8664
skamyxatra. > Sorry for not responding sooner, I learned a trick awhile ago that relates to > figuring out how many possible gismu their can be (or at least getting a very > close estimate). > [snipped] .skamyxatra
Your "trick" appears to consist of blatantly ignoring the rules of Lojban morphology, imposing an arbitrary new rule about "no repeating letters," considering only half of your new pseudo-{gismu} space, and claiming that it all evens outs magically.
If you want the real number of possible {gismu}, ignoring conflicts, consider that there are 17 consonants, five vowels, 48 initial consonant pairs, and 131 valid consonant pairs. A {gismu} is either "initial pair, vowel, consonant, vowel" or "consonant, vowel, consonant pair, vowel"; this comes out to 48 * 5 * 17 * 5 + 17 * 5 * 131 * 5 = 76075 possible {gismu} (again, ingoring conflicts).
mu'omi'e .kamymecraijun.
-- jicmu traji zifre fa loi remna lonu senpi
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
lukebergen · 2009-12-03 12:12 UTC · Tiki post #8702 · reply to Tiki post #8664
the trailing vowel?
Could that be taken care of by just taking off the last "* 5"?
On Tue, Aug 25, 2009 at 6:48 PM, Minimiscience <[email protected]>wrote:
> de'i li 25 pi'e 08 pi'e 2009 la'o fy. [email protected] .fy. cusku zoi > skamyxatra. > > Sorry for not responding sooner, I learned a trick awhile ago that > relates to > > figuring out how many possible gismu their can be (or at least getting a > very > > close estimate). > > [snipped] > .skamyxatra > > Your "trick" appears to consist of blatantly ignoring the rules of Lojban > morphology, imposing an arbitrary new rule about "no repeating letters," > considering only half of your new pseudo-{gismu} space, and claiming that > it > all evens outs magically. > > If you want the real number of possible {gismu}, ignoring conflicts, > consider > that there are 17 consonants, five vowels, 48 initial consonant pairs, and > 131 > valid consonant pairs. A {gismu} is either "initial pair, vowel, > consonant, > vowel" or "consonant, vowel, consonant pair, vowel"; this comes out to 48 * > 5 * > 17 * 5 + 17 * 5 * 131 * 5 = 76075 possible {gismu} (again, ingoring > conflicts). > > mu'omi'e .kamymecraijun. > > -- > jicmu traji zifre fa loi remna lonu senpi > > > To unsubscribe from this list, send mail to [email protected] > with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if > you're really stuck, send mail to [email protected] for help. >
>How many possible gismu?
Minimiscience · 2009-12-03 12:12 UTC · Tiki post #8703 · reply to Tiki post #8664
> where conflicts are when two gismu must not be exactly the same except for > the trailing vowel? > > Could that be taken care of by just taking off the last "* 5"? .skamyxatra
Two {gismu} may not differ by a single consonant when the different consonants are related according to the table in item 4 of section 4.14 of the CLL; up until about an hour ago, this entire thread was about how to take that restriction into account when calculating the number of possible {gismu}. Dividing 76075 by 5 to get 15215 would take care of the final-vowel conflicts but not the similar-consonants conflicts.
mu'omi'e .kamymecraijun.
-- na'e du'u mi lazni .i du'u mi na'e selra'u sa'u
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Arran <[email protected]> · 2009-12-03 12:12 UTC · Tiki post #8704 · reply to Tiki post #8664
> If we hit that point we could well speculate about communicating by > exchanging "core dumps" of the relevant brain states; I could not only > communicate to you that I saw something blue, but also _exactly_ what > shade of blue it was, by copying that part of my memory and and > including it in the "core dump"; I could also communicate to you > everything I know about a subject encoding as a "core dump" the brain > states of the region(s) where that information is stored. (I don't > say neuron states because I don't assume that we will be using a > structure that involves neurons at that point.) IF the singularity > occurs, there is, as you say, no way to predict anything beyond that > point. If the world is to become less comprehensible to us than our > culture and lives would be to a Neanderthal, we can't reasonably > predict how that will affect things; so there is no way to plan for > it. >
Speaking of 'blue' we actually do seem to refer to a specific shade of blue, maybe like some words (ie the gender) we need a word for a specific shade of blue and something that is close to the specific shade of blue.. A lujvo
would do, until you want to put it inside another lujvo. :PHow many possible gismu?
dbrock · 2009-12-05 11:20 UTC · Tiki post #8880 · reply to Tiki post #8664
"fundamental" about them (except for the tiny, tiny detail of having rafsi and thus being able to form additional words).
But then they always go on to say that never, never, never, never ever even think of creating additional gismu.
So here's a question to you all: given that we should NEVER even consider creating new gismu, well, then why don't we allow fu'ivla to have CVCCV or CCVCV form?
-- Daniel Brockman [email protected]
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Minimiscience · 2009-12-05 11:20 UTC · Tiki post #8881 · reply to Tiki post #8664
> So here's a question to you all: given that we should NEVER even > consider creating new gismu, well, then why don't we allow fu'ivla to > have CVCCV or CCVCV form? .skamyxatra
Because those {fu'ivla} would be new {gismu}. The CVCCV/CCVCV form is the defining feature of {gismu} that sets them apart morphologically from other {brivla}.
mu'omi'e .kamymecraijun.
-- lo panojauvaipa moi krinu be lonu cilre fi la lojban. zo'u noda ca'o se bangu la gliban.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
Daniel Brockman <[email protected]> · 2009-12-05 11:20 UTC · Tiki post #8882 · reply to Tiki post #8664
>> consider creating new gismu, well, then why don't we allow fu'ivla to >> have CVCCV or CCVCV form? > .skamyxatra > > Because those {fu'ivla} would be new {gismu}. The CVCCV/CCVCV form is the > defining feature of {gismu} that sets them apart morphologically from other > {brivla}.
So if gismu is just a morphological property, then why not create new ones?
What's the reason?
--
Daniel Brockman
[email protected]
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
stevo · 2009-12-05 11:31 UTC · Tiki post #8887 · reply to Tiki post #8664
[email protected] writes:
> People always say that gismu are not special and there is nothing
> "fundamental" about them (except for the tiny, tiny detail of having
> rafsi and thus being able to form additional words).
>
When people say that gismu are not special, they mean semantically. Morphologically, of course, they are special indeed, having rafsi and a special shape (CVCCV or CVCCV). It's often claimed that gismu are not intended to cover all semantic space, but it would be nice if they did indeed form some kind of fundamental semantic basis.
mu'o mie stevonHow many possible gismu?
Daniel Brockman <[email protected]> · 2009-12-05 11:31 UTC · Tiki post #8891 · reply to Tiki post #8664
-- Daniel Brockman [email protected]
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
arj · 2009-12-05 11:31 UTC · Tiki post #8892 · reply to Tiki post #8664
> I'm just wondering about the rationale for not creating new gismu.
Bob's position on this can be found here:
http://www.lojban.org/lists/lojban-list/msg05357.html
-- Arnt Richard Johansen http://arj.nvg.org/ Dwi'n gallu llefaru pob llinell heb atal, oherwydd does dim tafod gyda fi.
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
lagejyspa · 2009-12-05 11:32 UTC · Tiki post #8900 · reply to Tiki post #8664
gismu -- the way they are derived (from a weighted survey of the 6 source natlangs).
--gejyspaHow many possible gismu?
stevo · 2009-12-05 11:32 UTC · Tiki post #8902 · reply to Tiki post #8664
[email protected] writes:
> I noticed that no one else mentioned another distinguishing feature of
> gismu -- the way they are derived (from a weighted survey of the 6 source
> natlangs).
>
> --gejyspa
>
That's how most were derived. The remaining gismu fall into several
categories: uniquely Lojban concepts (e.g., "gismu"), metric prefixes ("decti",
"kilto"), basic metric and math units ("xampo", "tanjo"), the six Lojban
source languages ("jungo", "glico"), some other languages and geographical or
cultural gismu ("bengo", "merko"), and the religion gismu. (CLL, 4.15)
How many possible gismu?
Arran <[email protected]> · 2009-12-05 11:32 UTC · Tiki post #8907 · reply to Tiki post #8664
with rafsi. (If there can be such a thing.) What disappoints me is the lack of taxonomy for plants and animals is the gismu..
(Warning, offtopic, idiosyntratic, potentially not ver well thought out part) I have seriously wondered if there should be junk gismu/rafsi which only have meaning in combination with other words, ie for xruba, (not a good example, but I have some of this growing in my garden) you (can) have xunre stani xruba, but wouldn't it be interesting to have hrexub where hre is a some fictional rasfi which is almost predicate in meaning.
On Fri, Sep 25, 2009 at 2:51 AM, <[email protected]> wrote:
> In a message dated 9/24/2009 10:32:10 Eastern Daylight Time, > [email protected] writes: > > > I noticed that no one else mentioned another distinguishing feature of > gismu -- the way they are derived (from a weighted survey of the 6 source > natlangs). > > --gejyspa > > > > > That's how *most *were derived. The remaining gismu fall into several > categories: uniquely Lojban concepts (e.g., "gismu"), metric prefixes > ("decti", "kilto"), basic metric and math units ("xampo", "tanjo"), the six > Lojban source languages ("jungo", "glico"), some other languages and > geographical or cultural gismu ("bengo", "merko"), and the religion gismu. > (CLL, 4.15) > > mu'o mi'e stevon
>How many possible gismu?
Minimiscience · 2009-12-05 11:32 UTC · Tiki post #8908 · reply to Tiki post #8664
> (Warning, offtopic, idiosyntratic, potentially not ver well thought out > part) I have seriously wondered if there should be junk gismu/rafsi which > only have meaning in combination with other words, ie for xruba, (not a good > example, but I have some of this growing in my garden) you (can) have xunre > stani xruba, but wouldn't it be interesting to have hrexub where hre is a > some fictional rasfi which is almost predicate in meaning. .skamyxatra
If you mean {gismu} & {rafsi} which could theoretically be used alone but are virtually only ever used in {tanru} or {lujvo}, there are already some like that; there are the numeric/metric prefixes like "{megdo}" & "{zetro}," color {gismu} as used by speakers beyond beginner level, "{simxu}," and probably a bunch of others that I can't think of right now. If, however, you mean {gismu}/{rafsi} whose meanings are "context sensitive" and determined entirely by (and change in accordance with) the enclosing {tanru}/{lujvo} and cannot be used alone, that's just crazy talk.
mu'omi'e .kamymecraijun.
-- lonu ca rere'u finti le xislu cu ba sidju do lonu baza finti le vi'e xislu
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
PierreAbbat · 2009-12-05 11:32 UTC · Tiki post #8909 · reply to Tiki post #8664
> IMHO valsi referring to other cultures, geography, ect should be fu'ivla > with rafsi. (If there can be such a thing.) What disappoints me is the lack > of taxonomy for plants and animals is the gismu..
There is such a thing. The Book has an experimental proposal of a finite set of rafsi fu'ivla. Both xorxes and I have worked out how to make an infinite class of rafsi fu'ivla, using "y" instead of the last vowel (my proposal) or adding "'y" after the last vowel (xorxes' proposal) to join them. As far as the proposal in the Book goes, if a word like "mraigo" (Welsh) is preceded by a rafsi, the Book allows words like "sicybermraigo", but to allow an infinite number of rafsi fu'ivla, such words have to have "y" inserted regardless of phonotactics: "sicyberymraigo".
As to taxonomy: there are two gismu for birds in Anseriformes and two for birds in Galliformes, but the rest of the birds, including the by far largest order Passeriformes, are deserted. There are two gismu for rodents, but mice and rats are fairly close within the rodents, so it's not clear which gismu to use for other kinds of rodents. And there are no gismu at all for kinds of fish, just the general "finpe".
In plants, "xruba" and "guzme" denote families, "rozgu" denotes a genus, and "latna" doesn't denote a taxon. I'm not sure if "cunmi" is taxonomically sound.
Pierre
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
How many possible gismu?
lojbab · 2009-12-05 11:32 UTC · Tiki post #8910 · reply to Tiki post #8664
> As to taxonomy: there are two gismu for birds in Anseriformes and two for > birds in Galliformes, but the rest of the birds, including the by far largest > order Passeriformes, are deserted. There are two gismu for rodents, but mice > and rats are fairly close within the rodents, so it's not clear which gismu > to use for other kinds of rodents. And there are no gismu at all for kinds of > fish, just the general "finpe". > > In plants, "xruba" and "guzme" denote families, "rozgu" denotes a genus, > and "latna" doesn't denote a taxon. I'm not sure if "cunmi" is taxonomically > sound.
There was no attempt to make gismu taxonomical, nor taxonomically sound.
None of the gismu for plants or animals necessarily correlates to any
standard taxon ("usage will decide"), and thus finpe might include whales and dolphins (in appropriate lujvo of course). The goal was to incorporate concepts of cultural significance and/or words that might be used in lujvo.
lojbab
To unsubscribe from this list, send mail to [email protected]
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
