Re: Small Derivational Idea

On Feb 25, 2009, at 3∞20 AM, Andreas Johansson wrote:��> On Wed, Feb 25,
2009 at 12:46 AM, David J. Peterson�> <dedalvs@...> wrote:�> [snip]�>> On
Feb 24, 2009, at 1∞28 PM, Andreas Johansson wrote:�>>>�>>> What's absurd
about supposing -os to be an ending? Its etymology,�>>> the�>>> repeat
occurence in Marcos, and its disappearance in derivations�>>> like�>>>
_carlista_ are certainly suggestive of endingness.�>>�>> The question is if
it's an ending to *modern* speakers the way /-um/,�>> /-us/ and /-a/ were
endings to Latin speakers. I suspect it's�>> not, but�>> that it's used as
an analogical model for adding things like /-it/.�>�> I'm not quite clear
what what distinction you are making here - when a�> Latin speaker leapt to
the conclusion that an unfamiliar word ending�> in -a had a plural in -ae
they was presumably indulging in analogical�> modeling - but "I suspect it's
not" seems a far cry from "absurd".��This is what I mean. In Spanish, there
are two word sets, let's say,�each�of which contain two members. The first
set is those that end in /-�a/ or�/-o/, and those that don't. The second
set is those that are masculine�and those that are feminine. To the extent
that the /-o/ and /-a/�endings�can be reliably mapped to either masculine
or feminine (and aside from�some very narrow exceptions, they can be), these
endings /-o/ and�/-a/ are thought of as masculine and feminine endings. As
for the�other set, though, there's not much that can be said. There are
several�recognizable endings that are pretty reliably one or the
other�(e.g. /-ion/�is feminine, /-ad/ is feminine, /-al/ is masculine,
etc.), but it�would be�a mistake from the point of view of the language to
call, for example,�/-ion/ a version of "the" feminine ending. That idea is
simply not real�to the mind of a Spanish speaker.��Back to "Carlos", I
would say that "Carlos" goes into the class of�all other�words that don't
end in /-o/ or /-a/, and that to think of /-os/ as a�masculine "ending" is
about equivalent to referring to /-al/ as a�masculine�ending. In the
context of the Spanish language, that doesn't make�sense.��Despite that,
the fact that the name happens to end in /-os/ is probably�why you can add
/-it/, because, for example, you have plenty of examples�of the
following:��perro "dog" > perrito "little dog"�perros "dogs" > perritos
"little dogs"��Even thought that /-s/ on Carlos isn't plural (or anything
at all), the�phonological form is similar enough that it seems speakers used
the�plural pattern as a model for diminutive Carlos.��Now, did it
necessarily have to happen this way? Of course not. The�possibility existed,
though, and Spanish speakers took advantage of it.�Numerous other
possibilities exist. For example, instead of sticking�with the actual
phonemes, one might see a pattern like this:��...CV > ...CitV��That
pattern exists in Spanish. It exists in a very specific�context, but
it�exists, nonetheless. Spanish speakers might, then, have used *that*
as�the model, and come up with something like this:��Peru > Peritu�arte
> artite�Gaudi > Gauditi��The /-o/ and /-a/ endings, though, seem to be
tied quite strongly to�the diminutive, though. Consider, for example, that
all nouns ending�in a consonant that take the suffix also get an ending that
matches in�gender:��tamal > tamilito�flor > florita��>> More
importantly, if you posit, let's say, two masculine�>> endings, /-o/�>> and
/-os/, then there's no reason why you couldn't have /-men/ or�>> /-tifuli/,
or anything as a masculine ending. By positing these as�>> allomorphs of the
same underlying morpheme, you deny the similarity�>> in shape, or their
evolution--and the same goes for positing /-it/�>> and�>> /-it-/. Formally,
it looks like an accident that the forms are�>> similar.�>�> Clearly
there are some theoretical presuppositions at work here that�> I'm not aware
of. Why should the claim that /-os/ and /-o/ are�> allomorphs of a masculine
suffix be construed as saying anything�> whatsoever about their evolution?
What's the problem with saying�> Spanish *could* have had /-men/ as a
masculine ending? I would have�> thought the fact it does not is a historical
accident.��This is what I meant.��By positing that /-os/ from "Carlos"
and /-o/ from everywhere else�are variants of the same underlying ending,
that admits the possibility�that the two could be different in form (as,
indeed, many allomorphs�are--some quite, quite different). Then, for example,
/-os/ could very�well have been /-men/, or anything else. Let's say it was
/-ilma/, and�instead you had "Carlilma". If you go that route, then the
prediction�would be that the diminutive would just as easily be "Carlitilma",
as�"Carlitos" is the diminutive of "Carlos". That seems to be missing
a�*big*�generalization, and doesn't seem to me to accurately reflect
Spanish�as it's spoken.��> In my experience, it encompasses linguists,
conlangers, and language�> teachers. But it seems to me the association of
the term with those�> claims is far from complete even among professional
linguistics - I've�> seen ones speak of empty morphemes that carry no meaning
(eg. -t- in�> dramatic), of words whose meanings are not reducible to those
of their�> constituent morphemes (eg. "thriller"), and of words containing
what's�> clearly a morpheme combined with something that appears to have
no�> independent meaning (no English examples come to mind). What I can't�>
recall seeing is anyone explicitly accepting those claims.�>�> (Is this a
descriptive v. theoretical linguistics thing? I pretty much�> only read the
former sort.)��I think it is. I've certainly seen this before, but I've
also seen�theorists�tear their hair out when they hear stuff like this. For
example, a�morpheme�that has no meaning is pretty much the exact opposite
of what a morpheme�is or can be (a unique pairing of form and meaning). To
stretch the�definition like this is really to destroy the theoretical
concept.��Of course, in this case, that's a good thing. Take the analysis
of�"dramatic"�there. Let's say it was analyzed
thus:��drama-t-ic�/"drama"-?-adj./��Why /t/? Presumably, it could be
anything, yet /t/ keeps popping up:��ego > egotist/egotistic�fantasy >
fantastic�spasm > spastic��If you posit an analysis like the above,
there's no hope of an�explanation.�You can spot it, and give it a name, and
maybe predict where it will�occur�(though note: egoist/egotist), but it
won't answer the question: Why /�t/?��(Note: I think the standard
analysis of this, though, is that /-tic/�and /-ic/�are versions of the
same.)��> (Meghean has a plural marker that variously manifests as /-an/,
/-n/�> or /-n-/+fortification of the following consonant. I suppose you'd
say�> describing this in terms of morphemes would be absurd?)��Well, you
tell me. Here's how I would do it (the best way I can�think of for a
morphemic analysis):��Plural Forms: /-n/, /-n-/��Plural Rules:�Class
A: N + /-n-/ = plural N�Class B: N + /-n/ = plural N��Phonological
Rewrite Rule 1: ø > a / CC#�Phonological Rewrite Rule 2: C > [+hard] /
N_�Phonological Rewrite Rule 3: N > [alpha place] / _C[alpha
place]��Class A: words that end in weak consonants�Class B: all
others��What's lost here is the similarity between /-n/ and /-n-/
(the�first could very well be /-k/, and the second /-l-/).
Further�complications would arise if the epenthetic vowels found
elsewhere�were not /a/. That would necessitate a third allomorph,
/-an/,�which would lose the similarity between /-n/ and /-an/.���On Feb
25, 2009, at 7∞22 AM, Paul Kershaw wrote:��> I think you have missed *my*
point. If one sees morphemes as�> indivisible building blocks, than infixing
as a concept is a�> challenge, I agree. How can something that is indivisible
absorb�> something else? But if you're not using a building block notion
*at�> all*, then you wouldn't be using the terms "infix," "suffix," and�>
"prefix" *at all,* because those terms refer to affixing someting�> in,
after, or before. Regardless, though, by my reading, you *are*�> using a
building block concept, and that concept is tied to pieces�> with meaning,
you're just not taking out buckets and insisting that�> each sort of affix go
in a specific bucket, or that morphemes are�> innately inseparable except
when they're not. To use your example,�> "carlos" and "it" are still
identifiable chunks, even if you don't�> want to call them
"morphemes."��Sure. "Car" is an identifiable chunk, too. Anything a speaker
wants�to identify as such can be a chunk, whether certain linguists
would�analyze it as a morpheme or not. This is how we get things
like�/-aholic/ (chocoholic, shopaholic, silkoholic, etc.). They may call
it�a morpheme now, but what was it before? It's just the same way�someone
can say, "I need another '-urt' word", and an English speaker�can respond
with, "Let's see, there's hurt, curt, dirt, alert, pert..."��> By the way,
your example misses a possibility: -it- is an infix�> across the board, and
is infixed into "Paco" after -o as added.�> Regardless, though, I don't have
a theoretical qualm with getting�> rid of the notion of "infix," "prefix,"
and "suffix" and simply�> going with "basic units are combined into words
according to�> certain pholological and word-level rules" or some
such.��I didn't mean to suggest that affixes weren't useful. They seem
to�exist, after all. It's just that they are not necessarily
associated�with�a particular meaning. After all, if "sang" is just as past
tense as�"punted", why should /-ed/ have some special significance
attached�to it? It's just a strategy for filling out a cell.��> There's a
further possibility: Modern diminuatives aren't currently�> productive, but
rather are the result of centuries of�> fossilization, at least for common
names (like "Carlos"). This�> would explain the sporadic implementation. For
instance, English�> "children" must be learned as an exceptional plural, but
it's the�> result of two historic plurals (-er and -en); at the point that
-en�> was affixed, speakers had apparently forgotten that "-er" was an�>
earlier plural affix. These days, it's not unheard of to hear�> "childrens,"
which diachronically contains three plural affixes,�> but which to most ModE
speakers contains the standard plural affix�> combined with the fossilized
exceptional plural form. With regards�> to names, the common English
diminuatives largely came from some�> sort of productive rule a long time
ago, but have since become�> fixed forms (John -> Jack, Richard -> Dick,
William -> Bill, etc.);�> I see no reason to not suspect something similar in
other�> languages.��Certainly. If there were a new name dropped into
Spanish like�"Tofos", though, I bet it wouldn't be long before Spanish
speakers�were producing "Tofitos" simply by comparing it to "Carlos".��>>
I can�>> understand why a linguist might want to be constrained thus,
but�>> why a conlanger?�>�> This conlanger wants languages that are
constrained by linguistic�> principles. If you don't, that's your prerogative
as well. :)���Bear in mind, though, that while the goal of linguistics is
to�explain natural language phenomena, they're not there yet!�So if the
ultimate goal is to create a naturalistic language, one�is probably better
off studying other natural languages and�drawing one's own conclusions. By
adhering to a linguistic�framework, one will create a language that that
framework�can explain--or at least the corners of language that
the�framework can adequately explain. My guess is the result�will be rather
artificial.��-David�*******************************************************************�"A
male love inevivi i'ala'i oku i ue pokulu'ume o heki a."�"No eternal reward
will forgive us now for wasting the dawn."��-Jim
Morrison��http://dedalvs.conlang.org/�