Self Segregation

["mcslason" <mcslason@adelphia.net>]

Hello group!

This is my first post. I have been studying Lojban, and I was 
wondering if someone could clarify something for me. As I understand 
it, Lojban self-segregates into cmavo, brivla and cmene based on the 
pattern of stresses, consonant clusters, pauses, consonant- or vowel-
finalness and syllable count (for brivla, n > 1). 

Thus in the sentence fragment "... BAcru le ..." we know 
that "bacru" is the gismu and "le" is the cmavo because stress falls 
on "ba" and the next syllable contains a consonant cluster and thus 
must be the final syllable of the word containing "ba." 

What I don't understand at this point is how the language 
disambiguates longer brivla which start with a cmavo-look-alike 
followed immediately by a legal initial consonant cluster. Pulled 
from the current lujvo list, consider:

[1] (brivla) backemselRERkru ?= (cmavo) ba + (brivla) ckemselRERkru
[2] (brivla) dicka'uDENmi "electrically negative" ?= (cmavo) di + 
(brivla) cka'uDENmi
[3] (brivla) guSMINra ?= (cmavo) gu + (brivla) SMINra

There are many more examples.

Considering [3], what is to prevent me from analyzing "gusminra" as 
cmavo "gu" followed by hypothetical brivla "sminru"? Note 
that "sminru":

- ends in a vowel;
- contain a consonant pair in the first five letters;
- is stressed on the next-to-the-last (penultimate) syllable;

thus meeting all the requirements for a brivla. 

I am sure it is I who is missing something here; I just can't figure 
out what. 

Thanks for your time :-)

--- Mike