Re: A Survey

<<1. Does your language(s) distinguish between active ("X breaks Y"), middle
("X breaks (apart)"), and passive ("X is broken (by Y)")?>>

Yeah!   ~:D

<<2. If the answer to #1 was "yes," what method(s) does your language(s) use
to make some/all of the above distinctions?>>

Unfortunately, for the passive, my languages (all of them) are rather
unimaginative: They all use an affix.   (Well, except for Kele, but I don't think
I've come up with a way to do passives in that...   It's a creole; it may not end
up having a passive.)   Some examples:

Zhyler:
sexa pettir mat.   /man-NOM. king-ACC. see-PRES./ "The man sees the king."
sexa matnes.   /man-NOM. see-PASS.-PRES./   "The man is seen."

Kamakawi:
a mata ei i nawa.   /PRES. see 1sg. PRED. fish/   "I see a fish."
a mata?u nawa.   /PRES. see-PASS. fish/   "A/the fish is seen."

Njaama:
sangi li heZi sa kaane sa.   /man D.O. sun 3sg. PRES.-see 3sg./   "The man
sees the sun."
heZi sa wakaane.   /sun 3sg. PASS.-PRES.-see/   "The sun is seen."

For the so-called "middle voice", I don't treat it as a voice, like active or
passive.   (For the example you listed, I'd just call that an intransitive,
and the base form of the verb, semantically.   Then "to break x" would be a
semantic causative, zero-derived in English.)   Rather, it's called an
anti-causative, and is treated just like a causative--but in reverse.   There is no
special affix for this in the languages I've developed thus far, but I've seen
natural language data with anticausative affixes in a paper I was reading on the
typology of anticausatives.   Here are a couple examples to compare:

Zhyler:
sexa say.   /man-NOM. die-PRES./ "The man dies."
sexa pettir sajas.   /man-NOM. king-ACC. die-CAUS.-PRES./   "The man kills
the king."
pettir sajas.   /king-ACC. die-CAUS.-PRES./   "The king is getting killed."

Hmmm...   Bad example.   "Kill" doesn't really work with an anticausative in
the way the "middle voice" is understood to work in English...   How about:

2lg2r man.   /egg-ACC. cook-PRES./   "The egg is cooking."

Now, for Sathir...

Sathir (Note: Sathir utilizes infixing prefixes):
Active: Nale pane t_heltoko.   /FUT-cook ABS.-fish DEF.-ERG.-man/   "The man
will cook a fish."
Causative: Netwale pane jistana t_heltoko.   /FUT.-CAUS.-cook ABS.-fish
DEF.-DAT.-woman DEF.-ERG.-man/   "The man will cause the woman to cook a fish."
Antipassive: Nolale t_hotoko.   /FUT.-ANTIP.-cook DEF.-ABS.-man/   "The man
will cook (something)."
Anticausative: Nale pane.   /FUT.-ANTIC.-cook ABS.-fish/   "A fish will be
cooking."

In the last, there actually is a process for the anticausativization.   Verbs
exhibit person morphology--it's just unmarked in the third person in the
future (which is why I chose that).   However, if we were to, say, do first person
non-future...

Normal: Nantale pane.   /N.F.-1sg.-cook ABS.-fish/   "I am/was cooking a
fish."
Anticausative: Natale pane.   /N.F.-cook ABS.-fish/   "A fish is/was
cooking."

<<3. What method(s) does your language(s) use to distinguish between basic
nouns and verbs of the same root (i.e. "a hit" vs. "he hits")?>>

As with any language, there should probably be a bunch.   In Kamakawi...

Kamakawi:
1.) Adding the prefix /i-/ (historically linked to the predicate marker) is
sometimes used to create what I once called "instance of action" nouns (what is
the proper name for them?): olo = "to sleep" (v.); iolo = "(a/the) sleep".

2.) Sometimes it's just zero-derivation, though: oloko = "to dream" (v.);
oloko = "dream" (n.).   (Kamakawi has no case marking, but strict VSO word order,
so there's no ambiguity.)

3.) There's a productive way to turn mass nouns into count nouns, where you
prepose the mass noun with "ape o" (lit., "one of"), and this is sometimes
used: lola = "to hope" (v.); ape o lola = "(a/the) hope".

4.) Sometimes they're completely unrelated: nemile = "to die" (v.); "death
(the abstract idea)" (n.); "dead" (adj.).   inemile = "corpse" (n.);
"corpse-like" (adj.); "to be a corpse (slang for "to be dead")" (v.).   But: ili?u =
"(a/the) death" (n.) (translates to "a taken thing", though coming from a word
that can be used to mean "to take" or "to give".   More like "a transferred
thing").

5.) There are some very rare instances where a common reduplication serves to
do the job: mata = "to look at" (v.); matata = "(a/the) look" (n.).

Now, the untold story here is that strategies one and two are the most
common.   The other three come into play when all the common strategies have been
used up for other lexical items.   (For example, the word "mata" means "eye" as
a noun, and "imata" means "to tell the truth".   *ape o mata just wouldn't
make sense, and there's really no need to have a completely different lexical
entry for "(a/the) look".)   They're there as irregularities, because all
languages need them, and they fit in with the language well, but if you look at my
lexicon, most of those types of nouns will either be zero-derived or prefixed by
/i-/.

Good survey.   Got me thinking about anticausatives in languages where I
never thought about them before.   :)

-David