Re: Korahamla lives!
| From: | Amanda Babcock <langs@...> |
|---|
| Date: | Wednesday, November 13, 2002, 14:59 |
|---|
Yay, questions!
On Wed, Nov 13, 2002 at 08:27:44AM -0500, Muke Tever wrote:
> From: "Amanda Babcock" <langs@...>
> >
> > ków "self"
> > plus -ki- "my" gives kókiw "myself, I"
> > plus topic marker i- gives ikókiw
>
> Ah, you have infixes.
Yes. Was supposed to have mutation as well, but I need computer assistance
to derive words as it is, and didn't want to mess with programming
mutation :)
> infixes and I would interlinearize your word thusly:
>
> i-kó[ki]w
> TOP-self[1GEN]
Ok, thanks.
> The next word would be more like
>
> jo-m[e[la]l]é[j][ki]ro
> COMM-cat[DIM][love][AGT][1GEN]
Heheh :) Thanks, I like the way that looks :) There's something about
nested brackets that warms a mad infixer's heart...
> > Next, the comment:
> >
> > méro "cat"
> > plus diminutive -l- gives meléro "kitty"
> > plus -ki- "my" gives melékiro "my kitty"
> > plus -a- "to love" gives melalékiro "to love my kitty"
> > plus -j- "agent of" gives melaléjkiro "one who loves my kitty"
> > plus comment marker jo- gives jomelaléjkiro
>
> Interesting. What determines where in the word the affixes go?
Information I left out. Each affix consists of its phonemes, plus its
place and manner of insertion.
In an effort to make it as non-decomposable as possible, I came up with
the following possible infix behaviors:
# p prefix
# s suffix
# < insert before
# x insert within
# d duplicate first item in cluster and insert between that
# t insert between initial liquid/glide and other cons
# q insert between cons, if dup. shift first to stop
# f insert between cons, if dup. shift first to fricative
# > insert after
# x< insert in cluster before
# (i.e. x<c-1 insert in vowel before last consonant)
# t< insert between initial liq/glide and other cons in cluster before
# d< duplicate first item in previous cluster
# q< insert between cons before, if dup. shift first to stop
# f< insert between cons before, if dup. shift first to fricative
# << insert before cluster before
# x> insert in cluster after
# d> duplicate first item in next cluster
# t> insert between initial liq/glide and other cons in cluster after
# q> insert between cons after, if dup. shift first to stop
# f> insert between cons after, if dup. shift first to fricative
# >> insert after cluster after
and locations (used in conjunction with all of the above except prefix
and suffix):
# v1 first vowel
# va accented vowel
# v-1 last vowel
# c1 first cons clust
# c-1 last cons clust
Infixes can also carry a preceding or following stress, or take a stress
themselves, or take a stress themselves but only if followed by the same
vowel in the next syllable.
There's a lot of redundancy here. 1-syllable words have the same position
for first, last, and stressed vowel, which could lead to multiple derivations
being realized the same way on those words. More problematically, the
order in which infixes are applied is not always preserved, if they go into
different sites in the word and don't change each other's position.
> -j- and -ki-
> seem to go after the stressed vowel
-j- does; -ki- ends up there coincidentally, as it actually goes before
the final consonant cluster.
> and -l- and -a- before--
Sort of... -l- technically goes *within* the stressed vowel, but since
vowel clusters don't exist and a word can only have one stress, it actually
ends up reduplicating the vowel in an unstressed form to its left.
-a- goes into the consonant cluster before the stressed vowel, in a
consonant-doubling manner. It doubles the first consonant in that
consonant cluster and inserts itself between the two. (The other manner
of insertion would have the infix going between two existing consonants
if it falls on a two-consonant cluster, and causing reduplication if it
falls on a single consonant.)
> Is there a
> grammatical or phonological rule for this, or is it lexicalized?
It goes with the morpheme, so I guess that would be lexicalized. However,
there are phonological constraints (no initial consonant clusters, no
vowel clusters, internal consonant clusters cannot exceed an initial
liquid or glide plus two more consonants, final consonant clusters can only
be a liquid or glide plus a non-liquid/glide consonant) which are applied
after infixation.
> What positions
> are possible? (I see before the word, after the word, before the stress, and
> after it... any others?)
As many as I could think of :) If there gets to be too much ambiguity as
the lexicon grows, I may simplify.
> When is reduplication invoked?
When the infix requires it. An infix can call for simple insertion, which
only reduplicates if the cluster has an odd number of consonants; post-
liquid/glide insertion, which does not reduplicate but may cause a nearby
vowel to be drafted for double duty if the resulting consonant cluster is
too big; insertion which reduplicates a consonant as the appropriate stop
if performed on an odd-numbered cluster; insertion which reduplicates a
consonant as the appropriate fricative if performed on an odd-numbered
cluster; and specific reduplication, which reduplicates the first consonant
regardless of the size of the cluster.
If I weren't depending on perl scripts to sort this out for me, I'd have
some more natural and complex rules, like "reduplicate the first stop, if
any, from the cluster", but I didn't want to code that. (And why *am* I
depending on perl scripts? Because my only defense against the potential
for ambiguity is mass derivation of words, followed by sorting for
duplicates...)
> involved... what're the phonological constraints?] In the case of
> vowel-insertion what determines which vowel to insert?
Well, not counting the cases where the vowel *is* the infix:
If a consonant is inserted "within" a vowel, that vowel is duplicated.
If a consonant is added to a consonant cluster and the result is too big,
the following vowel (if any) is inserted, or the previous vowel if there
is no following vowel, or 'e' if there are no vowels to choose from.
But the word is joined in pieces, so this rule may get applied strangely.
> Does anything in
> particular prevent the reduplication from happening, i.e. from having
> *<jomaléjkiro>? Do all non-CV affixes invoke reduplication?
This may have been answered above, but specifically: in the case of these
two affixes, they will always involve reduplication, because -l- is applied
"within" the stressed vowel and since vowel clusters are illegal, this
always leads to reduplication of the vowel; and the -a- infix for "love"
specifically calls for duplication. Affixes also exist which apply a vowel
within a consonant cluster, and these do not reduplicate if the consonant
cluster contains two consonants.
> Would it be possible to rearrange these affixes? i.e., apply the "my" to
> "agent of" and come up with "my kitty-lover" ? If so, how would that work,
> systematically speaking?
Well, let's see whether that is ambiguous or not. If it is, then I should
probably change some rules, as that could be important :)
méro: cat
meléro: kitty
melaléro: kitty-love
melaléjro: kitty-lover
melalékijro: my kitty-lover
Ok, it is not ambiguous, but only because my rules consider the j to be part
of the consonant cluster :) Whew :)
Amanda