Tech question: Is there anything I can do on my end so that the
reply-to address for the mails I send to the list isn't my address?

On 7/19/05, David J. Peterson <[log in to unmask]> wrote:
> Here's a question that hasn't been asked yet: What are the phonemes
> you've got?  That might suggest something.

I tried to come up with a system that my friend could pronounce, but
that's also somewhat exotic from an English-speaking point of view.
(I'm still learning X-SAMPA, so if any of the following seem really
bizarre, it's probably a mistake.)

a e i o u @







It's open to tweaking, but I don't want to change it too much.
Basically, plosives can be palatalized/non-palatalized, nasals can be
voiced/voicless, and there are two retroflex consonants: s` and l`. I
might get rid of the retroflex consonants because I'm not sure that I
like having them and the palatalized consonants at the same time.

In the first draft of the phonology, there was some sort of
velar/retroflex harmony: all velars or retroflexes would turn
following nasals, /s/ and /l/ into velar or retroflex.

*kamas -> kangash 
*salhatan -> salhatang

(I've been using "sh" and "lh" for the retroflexes; I want to avoid
too many diacritics since this is for a book.)

I'm not sure how much I like this idea or even how realistic it is, so
I haven't developed it any further yet. I really like the idea of
"blocking" consonants, since that would allow me to have more variety
and avoid some "ugly" words that would otherwise be unavoidable.

Another thing I thought of was voicing harmony, and I might still go
that way, but I'm not sure yet. Palatalization harmony, I decided not
to do, because I already have a conlang with a lot of palatalization
assimilation (though not harmony) and I wanted to do something new.

> An idea that I got from the example already stated has to do with
> different types of sounds and blocking.  So, remembering that
> language with the [s]/[S] deal (what was it called?  It had a name
> that I swore I'd never forget.  But, then, here we are...), the way
> it worked was that the sibilant that was furthest right in the word
> determined the quality of all previous sibiliants.  So if you had:
> /sanaSil/ > [SanaSil]
> /se-sanaSil/ > [SeSanaSil]
> /se-sanaSil-s/ > [sesanasils]
> But, this wasn't the end of it.  The glottal fricative [h] blocked any
> kind of spreading:
> /sahaSil/ > [sahaSil]
> So you could have a word with both types of sibilants.
> This gave me an idea for a language of mine called Tan Tyls that
> really has more to do with vowels.  Specifically, vowels are nasalized
> after nasal onset and before nasal codas.  This nasalization, then,
> spreads to all other vowels in the word:
> /nakup/ > [na~ku~p]
> However, nasalization can't spread beyond what I call a "back"
> consonant.  For Tan Tyls, this is any consonant further back
> than the velum: [q], [?\] and [h].  Thus:
> /nahup/ > [na~hup]
> And you can get a bizarre situation like this:
> /naqamu/ > [na~qamu~]
> However, if you have, say, [+back] and [-back] consonants, then
> you could conceivably come up with a consonant harmony
> system that closely resembles a vowel harmony system.  Thus:
> k [-back] vs. q [+back]
> x [-back] vs. X [+back]
> r [-back] vs. R\ [+back] (uvular trill)
> N [-back] vs. N\ [+back] (uvular nasal)
> And with the obstruents, you can have voiced and voiceless
> series.  Anyway, in this system, anything unpaired would be
> neutral.  So, here are some sample words:
> karta "bear"
> qaR\ta "replacement air filter"
> xisloN "fish"
> XisloN\ "to forget one's cousin's birthday"
> And, with neutral elements, you can have wildcards:
> paxi "cat"
> paXi "tiger-striped cloak"
> Anyway, these are just lexemes.  The real fun would come with
> morphology.  So, let's say the first person possessive morpheme
> is /-K/ (which stands for [-k/q]).  Then you get:
> karta-k "my bear"
> qaR\ta-q "my replacement air filter"
> And this type of split can be done at any level.  An easy one to
> do would be voicing harmony.  In this type of system, nasals,
> vowels and approximants could be neutral (unless you wanted
> to get wild and allow for voiceless nasals, voiceless vowels, and
> voiceless approximants [the latter being the most likely]).  So
> let's say it works like the Salish system, where the last eligible
> consonant determines the voicing of all previous.  Some examples:
> kanta "song"
> ganda "dentist's drill"
> However, things can get very interesting when you add morphology.
> So let's say the plural is /-t/, and the dual is /-d/.  Then you'd get
> the following:
> kanta-t "songs"
> kanta-t "denist's drills"
> ganda-d "two songs"
> ganda-d "two dentist's drills"
> And if you just wanted to get downright ridiculous, this language
> could also have a word-final devoicing rule which applies *after*
> the harmony, giving you:
> ganda-t "two songs"
> ganda-t "two dentist's drills"

I enjoy the downright ridiculous, which means I really like this. =)

> And, of course, you can have something that blocks the spread
> (nasals might do it, actually), other things that trigger voicing
> all by themselves (e.g., obstruents must be voiced if they follow
> a nasal), and bizarre things which happen with neutral consonants
> (e.g., a word that starts with a stop and ends with a nasal with
> no other stops intervening would give you a voiced stop, but
> the same wouldn't be true of an approximant).
> Finally, others have mentioned things like ejective spreading and
> aspiration spreading (or dissimilation).  This can work with any
> secondary articulation: labialization, palatalization,  
> pharyngealization,
> velarization, creaky voicing, aspiration...  That's almost too easy.

Indeed. One of the reasons I've avoided simply doing voicing harmony
is that it's not very fun. But adding additional rules is a good way
to *make* it fun.

> Oh, and another natlang example that comes to mind is Latin.
> In Latin (and there are lots of Latin buffs on the list, so jump in
> here if you've got some more info), you could never have two
> /l/'s or two /r/'s in close proximity.  This is how we get "moral"
> and "molar".  Now, correct me if I'm wrong, but I think this went
> all the way through the word, didn't it?  Thus, "floral", but *froral
> would be impossible?  Not sure about that, but it's beyond
> plausible.