On Thu, 17 Apr 2014 02:08:09 -0400, Anthony Miles <[log in to unmask]> wrote:

>/li/ and /yo/ are not the whole verb - the superlative directional /na/ has to be present to render the meaning 'eat'. 
>With the ablative directional /su/ it is 'vomit', with the allative directional /tu/, 'swallow'. 
>Or taking another tack, /eyoputuna/ 'he swallowed it' and /eyoputuma/ 'he will listen to it' might look the same, but a Siye-speaker would know that the perfective suffix /-na/ indicates the verb li/yo, and the imperfective suffix /-ma/. indicated the verb yo/sa. 
>It was very complicated to set up (thank God for dictionary software!), but believe me, I weeded out most of the homophony by cross-referencing aspect, transitivity, directionality, and use of the extensive case system (what's the use of designing an original 10-case system if I wasn't going to use it) 

Ah, beautifully assembled, then!  I love me a language with bipartite stems (cf. my own Sabasasaj).  

>There are plenty of two- and three-syllable verbs, but it is easier for me to keep the examples short. If it would be more helpful, I can use them in examples from now on.

Oh, that wasn't my intent in saying that.  Short examples are fine.  

I can, though, name a thing or two that would help me understand your examples: a big one is horizontally aligning the morphemes with their glosses in the interlinear.  Or if that's too much work, then the poor man's version would be to make the dashes separating morpheme glosses more salient somehow.  Otherwise it's hard to pick out the correspondence between morpheme and gloss, and if I want to know what a given morpheme is I have to just count dashes into it, which is errorprone.  

On Thu, 17 Apr 2014 12:39:54 +0200, BPJ <[log in to unmask]> wrote:

>You say there are no final consonants -- other than /m/ I take it?

IIRC, written coda /m/ is actually the marker of vowel nasalisation, not a coda consonant.   -- ah, yeah, .

>Also why would final syllables homophonous with a number suffix be a
>problem? Most languages tolerate that kind of small ambiguities, especially
>if the suffix in question has to be added anyway. Take the Swedish word
>_öken_ 'desert': it looks like the definite form of a common noun _*ök_,
>yet it is indefinite, its definite form being _öknen_.

This puzzles me too.  

I had been about to say that it could be that all verbs which historically ended in such syllables had them reanalysed away as number suffixes.  This sort of reanalysis in a verb system would have rough precedent in e.g. Chechen class markers.  In Chechen, a subclass of verbs have a root-initial consonant which alternates to show the noun class of the subject.  There are four class markers, /d- b- v- j-/.  Verbs not in this class have an invariant root initial consonant, but (IIRC) no native verbs start with any of /d- b- v- j-/ as an invariant consonant; apparently all the verbs that would have done have been reanalysed as class-marking verbs.

But!  But then I saw in the Frath documentation
that the number suffixes are obligatory, and none of them is zero!  This is the best possible situation for _avoiding_ confusion between stem-final syllables and number suffix: if there is just one syllable with the form of a number suffix, it must be a number suffix and so cannot be part of the stem; if there are two such syllables, the first one must be part of the stem and the second the number suffix (unless the latter one could be taken for some suffix of class even further right).  That is, this is the _worst_ possible situation for the posited reanalysis to take place.  So now I agree with Benct again, I don't see how there's a problem with those final syllables.  

Even if the fiction-internal basis for this rule is some kind of Strunk-and-Whitish false overgeneralisation on the part of whoever legislate correct Siye grammar, that still doesn't account for where the seed of the rule came from.  I've never heard even the kookiest pedants say there's something wrong with words like _öken_.