Thanks. I'll definitely use this when editing Ṫirdonic.


-Yash Tulsyan


On Tue, Sep 10, 2013 at 9:59 AM, Alex Fink <[log in to unmask]> wrote:

> On Sun, 8 Sep 2013 16:12:32 -0700, Padraic Brown <[log in to unmask]>
> wrote:
>
> > Is Tirdonic supposed to be an IE language, or some kind
> > of sister to it? Or just taking inspiration from?
>
> Yeah, this inspiration certainly shows.
>
> > At first glance, the nominal morphology strikes me as extremely regular.
> That might
> > be fine and dandy for a 19th century ophthalmologist of some note, but
> for the
> > naturalistic language of a bunch of wandering barbarians, this might be
> a tad too regular!
> >
> > Haven't looked at the verbs and so forth yet.
>
> In the verbs it's even extremer.  I hope not to come off too harsh, but
> the current incarnation of Thirdonic strikes me as what David was warning
> against when he inveighed against Lego-block morphology (see his LCC1 talk
> "Down with morphemes!", and its slides, both online).
>
> For one, they're way too long!  The octosyllable _bidirekameolidodh_ for
> "saw" is outside the pale of plausibility for me.  Surely speakers would
> have already worn them down, one way or t'other, to a more manaģable size.
>  Languages which can be reasonably analysed in morphemes tend to make heavy
> use of zero morphemes for the most common value of any category, like what
> you currently have for the active voice and imperative mood.
>
> My high-level recommendation would be to leave the morphemic mould.  Have
> you *seen* the PIE verbal system?  Four unrelated ways to form the aorist,
> with which one (or more) each stem uses basically lexical and not
> predictable by rule; elèven ways to form the present, similarly (and 27 if
> you count subclasses!).   <https://en.wikipedia.org/wiki/PIE_verbs>
>
> But concretely, in the near term, I'd suggest throwing out the exponence
> of your third-person marker: if the presence of a gender marker implies
> third person, then a separate marking for third person is overkill.  In
> subject number, every morpheme having a _-dh_ means that every finite verb
> ends in a _-dh_, and it's just as reasonable to analyse that _-dh_ as a
> marker of nothing more than "this is a finite verb" unrelated to the
> subject number; under that analysis it's not really pulling its weight and
> may as well be dropped.
> Then there's aspect, tense, and mood, which are rarely so cleanly
> segregated into three separate categories as you have them here: this could
> not only benefit from some zero morphs, but from some rules which say
> "certain aspect (tense, mood) distinctions are only made in certain tenses
> (moods, aspects), and otherwise collapsed", as well as perhaps from some
> rules like "which aspect of a given verb is zero-marked depends on its
> lexical aspect <https://en.wikipedia.org/wiki/Lexical_aspect>" to shave
> off a fraction more of a morpheme on average.
>
> As for the forms, I see you have a predilection for -CV and -VC
> suffixes... but not either particular one!  Having ònly CV forms, or ònly
> VC forms, would have ensured that assembled verbs consisted of alternating
> ...CVCVC... which might be phonotactically desirable.  But having both
> doesn't give you a similarly nice phonotactic outcome, and there seems to
> be little reason not to make some of the exponences just one segment (with
> maybe some phonotactic repair rules if necessary).
> Incidentally, this formal pattern also makes -sase- a bit of a sore thumb
> as a steal from Japanese.
>
> As for case, here you dò actually have number morphemes which are single
> V, and case morphemes which are single C; you just haven't analysed them as
> such in your tables!  Here my concern is, kinda, the opposite: it's weird
> that, given so many case markers are just C, no two of them were
> accidentally the same.  (You could do a birthday paradox computation: with
> 10 nonzero-marked cases and 23 consonants, there's only a (23 choose
> 10)/23^10 = 10% chance of this.)  You might make some of them longer --
> after all, it's common for case markers to still synchronically be a bit
> like cliticised adpositions, and adpositions are rarely just one consonant
> long.  Or you could build some of them off a common oblique stem.
>
> Alex
>