Gleanings from yellowing LoCCan3 files.
# acceptable consonant frames (CC/C, C/CC) = 2975
# if only one of C1C2, C2C1 is allowed = 1836
# if only one of C1C2, C1'C2' is allowed (' being opposite voicing) = 1819
# if only one of C1C2/C3, C3/C1C2 is allowed = 2159
There are also notes about various combinations of these (cluster-singleton exchange with voicing and order reversal, say) but not numbers.
There are numbers for cases where only r or l can occur in a given place in a frame and similarly for the sibilants, but since these do nothing to solve the same problem in the cmavo, they are not mentioned here.
Notice, all these assume the standard initial cluster restrictions, which seem
unnecessarily strict (would tf or dv be a problem, or tl or dl? And there are probably others. Oh, yes, medial mz). Even so, all these further restrictions leave enough frames to allow every gismu to have its own frame and leave vowels to serve other redundancy increasing tasks (taking care of sibilants and r/l, for example), even if we have only a 3-way stressed vowel distinction and no distinguishing unstressed vowels.
Just fodder for the flames of the next generation (when that charismatic nutcase comes along).
From: Ian Johnson <blindbravado@gmail.com>
To: lojban@googlegroups.com
Sent: Thursday, August 2, 2012 4:27 AM
Subject: Re: [lojban] racli/ralci
Erm, woops, no, it doesn't force the letters to be distinct, as you can see from the very beginning :)
On Thu, Aug 2, 2012 at 5:10 AM, Ian Johnson
<blindbravado@gmail.com> wrote:
On Wed, Aug 1, 2012 at 9:38 PM, Jorge Llambías
<jjllambias@gmail.com> wrote:
I don't think so. "blanu" for example will block "planu", but it won't
block "plana", so by playing with the final vowel you can probably
accomodate all 20315 four-letter initial forms.
I see what you mean. It would be interesting to attempt this computationally; the "transitivity" if you will of the blocking rules makes this seem like it could be subtler than either of us are seeing. You're closer to right than I was, though, if not actually right.
> An interesting computation: assuming every combination of 2 vowels and 3
> consonants can serve as exactly one gismu (which lets words differ only in
> their final vowel; not doing so makes the current number of gismu just
> barely fit), you get 6800 possible words.
Three _different_ consonants and two _different_ vowels, right? But
you can have gismu with two repeated consonants or vowels (e.g.
"nanba"). So I think the number for different combinations comes out
as 14280.
Good point, I didn't think of that possibility. I actually just wrote some code that puts down a gismu list with all letters distinct. It's pretty readily extensible; in particular I can easily change the order of the nesting, which significantly alters the words that actually get generated. A variant in which initial consonants vary first, then the cluster, then the initial vowel, then the final vowel, with no initial clusters whatsoever, results in http://pastebin.com/SYfHQf8p, which I found pretty interesting.
> Almost all such combinations work,
> too: ignoring the 9 special exceptions, any 2 vowels obviously work, and any
> group of 3 consonants will have at least 2 with have the same voicing.
I would guess around 5-10% are blocked by the special exceptions.
Well, in the slightly naive computation earlier, it's 9/(191+48) = about 3.8%.
mu'o mi'e latros
--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.