Re: English "another"/Conlang Question

Been meaning to get back to this for awhile.

On Fri, 10 Aug 2007 03:42:11 -0700, David J. Peterson <dedalvs@...> wrote:

>You know, if you just completely eliminate syntax--just pretend
>it doesn't exist--language (both nat and con) starts to make a lot
>more sense.  For example, imagine you have a toaster.  You plug
>the toaster in, put bread in the slots, press down, wait a bit, and
>you have toast.  Then let's say someone comes along and says,
>"No, you have to sacrifice a goat after you plug in the toaster,
>but before you put in the bread.  Otherwise nothing happens."
>That goat sacrifice is syntax, and that someone is a syntactician.
I'm a bit late to the discussion of this, I guess, but here goes: is it
really that much of an anathema to have a thing called syntax?  I was
intending to use the word without much theoretical baggage; there's probably
a word you could put there that would make you happier and still convey my
(poorly-expressed) sentiment, though I'm not sure which it is.

I could believe that drawing a boundary between 'syntax' and 'morphology' is
a Bad Idea, but still it seems like it's useful to have a name for the sort
of ru--- sorry, patterns that tend to deal with combining words as opposed
to those combining things within a word, for whatever definition of word is
useful, nebulous or bright-lined.  For that matter, I'm not sold on the
crucial difference between a 'rule' and a 'pattern'.  Is not this (from a
later message)

>{X [V] <-> should X [V (oblig. Z)] <-> should be Xing [V (oblig.
>progr. Z)]}
a rule, if one whose applicability is limited?  Is the problem just that
making rules absolute gets you into trouble later on with intransigent data?
or just that the sort of rules syntacticians like pile on abstractions until
taking care of them is more troublesome than the data itself? or what?

>Back to the matter at hand...
>
>Alex:
><<
>I don't know if you remember the time we were trying to coin a word for
>'command' in Kalusa, and among the proposals were things like _kagorota_
>|IMP.+speak|?  That felt wrong in largely the same way, which I did an
>equally poor job of describing then.
> >>
>
>This was wrong for reasons that had absolutely nothing to do
>with syntax.  The reason /kagorota/ didn't work was because
>it was totally and absolutely unprecedented.  There wasn't any
>pattern that was even remotely close to anything like this, so
>it just didn't make sense in the context of the language--even if
>there was a story for it (e.g., a particle used for imperatives is
>related semantically to commanding someone to do something).
Well, there were not a few things which were unprecedented when they first
appeared in Kalusa but came to be accepted.  My gut-felt objection ran
deeper than that -- I think I would have objected to /kagorota/ whatever
lang it was proposed in.  But I don't think I'm making any headway stating
my objection generally.

>The construction in Kamakawi might seem bizarre, but it's not
>unprecedented (and this is part of the background that's missing).
Ah, background, what I was after all along.  (Only a few small comments of
mie interspersed henceforth -- you were quite thorough.)

>First of all, /i/ is an honest to goodness preposition.  This is its
>canonical use:
>
>Mata ei i nawa.
>/see I OBJ. fish/
>"I see a fish."
>
>But it does have other uses:
>
>Ka hava ei i nawa i ia.
>/past eat I OBJ. fish for you/
>"I ate a fish for you."
>
>And like just about all the prepositions of Kamakawi, it started
>out life as a verb:
>
>A i nawa i aila.
>/pres. exist fish LOC. ocean/
>"There's a fish in the ocean."
>
>It's a fun little multi-functional preposition.  If it has a basic
>meaning anymore, I'm unaware of it.
Fun indeed; I like.

>Kamakawi has many prepositions that began life as verbs (and
>which can still act as full verbs, in fact, with little difference in
>meaning [e.g., /te/ (p.) on; /te/ (v.) to be on).  One pattern
>that's emerged in Kamakawi is the pattern of prepositions
>becoming prefixes to words, something like Latin (e.g., ex-, in-,
>ad-, ab-).  Some of the preposition > prefix pairs are not highly
>productive (e.g., a few are genitival constructions), but collectively,
>it's so common to see these guys, that the pattern of {X[P] <-> X-
>[Pref.]}
>is highly...what do I want to say, lit up?  It's in there and
>accessible.
"salient" ftw!

>So the move from having a preposition /i/ to a prefix /i-/ wasn't
>that difficult.
It'd be interesting to know -- no, I understate: it's probably the crux of
this whole matter to understand how the first instance of prefixing a
preposition (or verbs with positional senses) started.  If prepositions
whose meanings were more refined were doing this for a long time I agree
entirely that it's natural for /i/ to follow, and that the things you
described below would happen.  To answer your ultimate question about
whether /i i-/ should occur I'd want to know whether the first <Prep+X> was
ever a collapse of <Prep X> with the same semantics (literally /ikatava/ 'in
a palm tree', say), or whether the preposition was doing something different
from the beginning.  Only in the latter case would the preposition be
justified in going missing for "another"'s reasons.

Of course, it's conceivable that the double preposition might have gone
missing later, by something in the vein of what Language Log, after Fowler,
calls cannibalism
  (http://itre.cis.upenn.edu/~myl/languagelog/archives/004106.html).

Can anyone tell me how the Latin prep+verb thing got started?  I don't even
know whether or how productively IE had it, though I feel like I should.
Celtic did this extensively too, yeah?  Where's it attested in IE?  Is the
Germanic version (as in OE _for-giefan_, _be-cwe[th]an_, _and-swerian_, ...)
cognate to the Latin one?

>So, taking a step back, this preposition /i/ marks direct objects
>(or sometimes agents, in certain verbs), beneficiaries, sometimes
>recipients, general locatives, etc.  So then let's say you have a
>pair like this:
>
>katava = palm tree
>ikatava = ?
>
>What could be something that's for a palm tree, or on a palm
>tree, or at a palm tree?  Palms, for one.  Maybe coconuts, but
>palms seem pretty good.  Thus, /ikatava/ = palm.  Hopefully
>this doesn't seem like too much of a semantic leap.
No, not at all.  One small step for semantics.  I had somehow gotten it in
mind at first that it was (object marker) + 'palm tree' => 'something a palm
tree does things to', whereas /i katava/ in a sentence was 'palm tree' as
object, so that any derivation that went straight from one to the other
would have to get the argument roles backwards.  But if it's 'for/in/at' +
'palm tree' = 'for/in/at a palm tree' => 'something for/in/at a palm tree',
there's no changing the roles required, which makes much more sense to me.

What's a palm, btw?  Is it what I'd call a palm branch?  For me 'palm' =
'palm tree'.

>Now, if you look at a palm tree as a single thing composed of
>many parts, one part of a palm tree might be a palm.  So if you
>have this pattern {X[N (Z)] <-> iX[N (part of Z)]}, instantiated,
>in this case, by /katava/ <-> /ikatava/, then you have something
>upon which to base a series of derivations, e.g.:
>
>tikiki "scarlet honey creeper" > itikiki "feathers from the same"
>teli "flower" > iteli "flower petal"
>tava "banana" > itava "banana peel"
>polao "blowfish" > ipolao "blowfish quill/spine"
>eka'i "abalone" > ieka'i "abalone shell"
>
>Those are concrete and fairly easy to explain.  Next, think of
>an action as an object (a fairly simple metaphor found in probably
>every language).  If an action is an object, then it, too, can be
>composed of parts, as well:
>
>olomo "to walk" > iolomo "a step"
>
>This is one way of thinking of the composition of an action.
>Walking is a repetitive act composed of many steps.  Thus,
>part of walking is a step.
>
>Another way of imagining the composition of an action is to
>think of what's involved.  In the act of drinking, several things
>are involved: a drinker, a drink, perhaps a container, one's
>throat, etc.  For an action like this, possibly the most relevant
>part is the object:
>
>nivu "to drink" > inivu "a drink"
>
>A drink or a step can be seen as a single instance of a prolonged
>action.  Thus, we can get:
>
>tei "to dance" > itei "a dance"
Yeah, all of these extensions are very natural, I think.

>Of course, lexical pressure can help move the process along.
>With "dance", for example, /tei/ is the verb; its associated
>adjective is "dancing"; and its associated noun is "dancer" (most
>verbs can be divided into those whose associated noun is
>agentive; those whose is verbal; and those that don't have one).
>As the act of dancing is kind of a repetitive act, the reduplicated
>form seemed the most logical for "dancing", the verbal noun:
>/teitei/.  For a single dance?  You could use a convoluted
>expression, but why not /itei/?
So (zero-derived?) associated nouns came first.  Are there any patterns to
what sort of associated noun each verb selected that are relevant here?

>So, back to this:
>
>Alex:
><<
>Or was the _i_ just supposed to contribute the semantics of 'object of'
>giving rise to e.g. _ikavaka_ 'book = object of writing'?  That just
>feels
>*really* wrong to me in a way I can't quite describe convincingly.
> >>
>
>Of course, prefixes don't contribute semantics: they're just sounds.
>Whether /ikavaka/ is a good instantiation of the pattern, though,
>is a fair question.
It is, but now that you've filled in the gamut of nuances of the pattern
between /ikatava/ and /inivu/, with the part-of-whole and subsequent
semantics the /i/ picked up, it looks completely unobjectionable.

>Once you have these action verbs with derived nouns like "drink",
>a new generalization can be made.  There are a whole host of
>these zero-derived nouns in English that show that a good many
>nouns are actually transitive, if you have the right object:
>
>I cried a good cry.
>I danced a dance.
>I slept a long sleep.
>I swam a good swim.
>
>Even as a native English speaker, I can't tell which ones are good
>and which aren't ("I whistled a good whistle"?  "I googled a good
>google"?  "I hummed a good hum"?  Does "sing a song" count?),
>but the pattern is that a verb begets a zero-derived noun with
>essentially the same meaning, and it's treated like the direct object
>of the verb.  As you have pairs like this:
>
>nivu "to drink" > inivu "a drink"
>
>Where the derived noun is a canonical object of the old verb,
>the pattern can be expanded to include derived nouns that are
>canonical--or simply usual--objects of the associated verb.  As
>a book (or a pamphlet or a letter) is a typical object of writing,
>you get /ikavaka/ "book (or pamphlet, or written thing, but
>commonly 'book', as that's what it gets used most for)".
>
>Once you get all the way to here, then, a speaker may well imagine
>that it's because things are objects that they get the /i-/.
Gotcha.  Nicely done; much deeper than the "hey, it's 'object of'!  we have
a preposition that does something kinda like that!  let's glom!" that was
one of the things I thought (feared?) might be going on at first.

>But I
>now see that the analogy I used with respect to my initial objection
>was accurate.  Rather than "another", how about Latin with its
>prepositional prefixes?  Can you get something like "in incipio"?
>I feel like I don't know enough Latin to phrase this question
>correctly...  Seems like you do, though.
Well, I don't believe you can get _in incipio_, since _incipio_ is a verb
form, but  there's nothing wrong with _in initio_, for instance (_initium_
n. 'beginning').  So that's another argument in favor of /i i-/, if you like.

Alex

Reply