Re: [lojban] Re: camxes's reaction to some fu'ivla

[Jorge Llambías <jjllambias@gmail.com>]

On Mon, Feb 8, 2010 at 1:29 AM, Pierre Abbat <phma@phma.optus.nu> wrote:
>
> Is there any indication in the Book that the initial "i" or "u" in a diphthong
> is considered part of the onset.

Not directly, the Book doesn't talk about syllable structure at all.
But the fact that there are no CVV cmavo with rising diphthongs
indicates that rising diphthongs are treated differently from falling
diphthongs. It's as if CiV and CuV start with a consonant cluster (or
a semi-cluster), so they can't be cmavo.

>> Maybe the implementation of camxes you are using is not based on the
>> current morphology as it appears in
>> http://www.lojban.org/tiki/BPFK+Section:+PEG+Morphology+Algorithm
>> (or maybe I'm missing something there.)
>
> I downloaded it from a link someone gave me on IRC. How do I find out what
> version it is?

I don't know. The formal morphology was written independently of any
implementation, and there was a time when it was being modified a lot.
If the file you have has a date, you could compare it with the date of
last modification of the wiki page.

> Here's another question to consider: Should a coda be allowed after a
> consonantal nucleus? My example is "bangrgbe" (some African languages, named
> for their word for "language"). Both vlatai and valfendi accept it; camxes
> doesn't. Valfendi, when asked to syllabify it, says "BA,ngrgbe", but I'd
> say "BAN,grg,be". (In Gbe, the "gb" is one sound.)

I would limit the consonantal syllables to the 64 syllables formed by
a consonant plus l,m,n,r. No codas and no heavier onsets. That's all
that is needed for type-3 fu'ivla, and type-3 fu'ivla are the only
ones where I would use consonantal syllables. And exclusively for the
hyphen syllable, so I would never actually use vocalic "m" in fu'ivla.

>> If weird codas are allowed for the last syllable, shouldn't they be
>> allowed for any syllable, rather than just the last one?
>
> Chapter 4 contains the name "tcarlz", which ends in three consonants,
> and "ralj" as an example of a cmevla formed from the long rafsi of a gismu.
> If any long or CVC rafsi is allowed to be used as a cmevla (except ones that
> begin with "la" - so how do we name lakes?), then "rafs" and "datk" should be
> valid cmene. Some real-world names that end in clusters that can occur
> medially are "Minsk" and "Yakutsk".
>
> What about cmevla beginning with clusters that aren't allowed initially in a
> brivla, such as "mkyveix"?

That's why I ended up dividing cmevla into two groups, the fully
lojbanized ones, consisting of ordinary Lojban syllables, and the
anything goes ones, for the rest. I had to resign myself that trying
to restrict cmevla to things that sounded lojbanic was a losing
battle. What I don't understand is, if we don't care whether the name
sounds lojbanic or not, why insist on imposing any cluster
restrictions on them at all, why not allow any string of Lojban
phonemes as long as it ends in a consonant?

>> camxes does not allow the combination of ts/tc/dz/dj with other
>> consonants in an onset. The only consonant-cluster onsets that it
>> accepts (besides those four) must fall within the pattern:
>>
>>    (s,c,z,j) (p,t,k,f,x,b,d,g,v,m,n) (r,l)
>>
>> so the maximum length for an onset is three consonants.
>
> Where did you get this set of clusters from?

Just from an examination of the permissible initial clusters. All of
them (except for the four affricates) fall within that pattern. "x"
and "n" are special in that "x" could have been put in the first
group, and "n" could have been put in the last group, but this way the
additional restrictions are fewer.

mu'o mi'e xorxes