Native American verbs vs. nouns

[rlarson at unlnotes01.unl.edu]

> I didn't want you to feel that no one was paying attention, though
> it's a time of the year when people really aren't paying much
> attention, I'm afraid.  Social activities tend to overwhelm
> everything else.  Not to mention the end of the Semester.

No kidding!  I've been locked out of the library most of the week,
helpless to respond!


>> Definitions needed, preferably with examples:
>>
>>   Head ...
>
> Also, the noun described by a relative clause is the head of
> the relative clause, which is what I was referring to here.
> The concept of head is a central element of current syntactic
> theory though not exactly an innovation within it.  What is,
> I think characteristic of modern approaches is an insistence
> that a head be endocentric, i.e., a variety of the thing of
> which it is a head.  This impacts the assessment of what might
> be the head of a construction, and of what the boundaries of
> the construct might be.

I don't follow this.  What does it mean, e.g., to say that a
preposition is a variety of a prepositional phrase?

In any case, a listing of "head" relationships doesn't answer
my question: What is a "head" in essence?  What chain of
reasoning leads us to the concept of a "head"?

I don't have any problem with using the term "head" for the
noun in a noun phrase that all the determiners, adjectives,
prepositional phrases, relative clauses, genitives and
qualifying nouns attach to; it seems like a useful word here.
My issue is with extrapolating this to verb chains and
prepositions, etc.


>> > As far as I can see the Omaha verbs are still perfectly finite.
>> > In  fact, I'd argue that there are essentially no non-finite
>> > verbs in Siouan languages.
>>
>> You'd need to define exactly what your criterion is for "finite verb".
>
> Any verb which is inflected for person is finite.  As far as I know,
> this is not a practice unique to me.  I would consider an OP third
> person verb to be inflected, albeit there is no prefix (or suffix)
> to indicate it.  This is one of the crosses that students of
> languages with "zero" pronouns(or unmarked personal categories)
> have to bear.  But in particular, there is nothing in Omaha-Ponca
> that I have noticed that seems analogous to a participle or gerund
> or infinitive or verbal noun.  I could be wrong, because I don't
> feel that I control all of the complementizing verb structures
> (I hope that, I think that,  ..., etc.).

I agree.  As I said in the previous post, we are in
substantive agreement here, though communication is a little
rocky due to my lack of grasp of the jargon.  Now let me
coin the term I should have been using:

  finite verb - any verb which is inflected for person.
                All verbs in Siouan appear to be finite.

  demand verb - any verb which signals the demand of a
                sentence.  I am claiming that no verb in
                OP/(Siouan?) is a demand verb, while most
                sentences in English/(IE?) depend on one.
                When the grammatical demand is anything
                other than a command, the English/(IE?)
                demand verb is equivalent to the (single)
                finite verb of the main clause.

The basic modern English pattern for signalling demand:

  [subject] [finite/demand verb] ...  (+ falling tone)
     ==> Statement

  [question-word phrase] [finite/demand verb] ...
     ==> Question word question

  [finite/demand verb (esp. aux)] ... (+ rising tone)
     ==> Yes/no question

  [unmarked/demand verb (esp. lex, stressed)] ... (+ falling tone)
     ==> Command

The basic ideal MVS pattern for signalling demand:

  [concept] [statement demand particle]
     ==> Statement

  [concept including question-word] [question demand particle]
     ==> Question word question

  [concept without question-word] [question demand particle]
     ==> Yes/no question

  [concept] [command demand particle]
     ==> Command

Decoding the demand from the English pattern depends largely
on the demand verb.  Once we have heard this word, we usually
know the demand of the sentence.  The MVS pattern is simpler
and more logical, but you never know what the demand is until
you reach the end of the sentence, where it is (ideally) coded
explicitly.  As with 3rd person verbs, it may be unmarked for
a default situation.


>>> I know what you mean about the embedding thing.  It was and sometimes
>>> still is difficult for me, too.  It didn't lead me to any new theories,
>>> but it is a bit disconcerting the way embedding and heads work.
English
>>> extracts the heads into the context and adjoins the remainder of the
>>> embedded clause, which may acquire a trace like a relative pronoun,
while
>>> Omaha leaves the head in place and appends the context, roughly
speaking.
>>> The reference to the head in the context clause is precisely the
embedded
>>> clause, with the focussed element determined by the context, though, if
>>> there's a determiner it may give a hint as to the identity of the head.

Thanks for your lengthy explanation of terms.  I can
actually read the above paragraph and understand what
you're saying now!   :)

However, I don't necessarily accept this analysis.
But to argue it, I think we need to double-check
where we stand in terms of our basic assumptions
about linguistics.

Much of your discussion involved breaking down sentences
into a hierarchy of concentric bracket-pairs like Chinese
boxes, e.g.:

> [I hope [[[[[the [beautiful [white horse]] that I have]
> which is sick] and lying sound asleep], which I love],
> will recover]]

To me, this looks like an approach intended to fit a
human language into a computer compiler according to
a set of formal grammar rules, along the lines of:

  S -> NxO
  N -> (p|dA|NC)
  A -> (n|aA)
  O -> (S|Nxv)
  C -> t+S-N ??? @#%^&**!!

I understand that this is the approach taken by Noam
Chomsky and Steven Pinker.  I believe this is also the
point where they get stuck, which necessitates some
rather Ptolemaic adjustments for handling relative
clauses.  I suppose the issue of "heads" has something
to do with these adjustments.  Is this the sort of
approach you are taking here?

In my view, while the grammar rules of a human language
may sometimes approximate a formal grammar, the former
are much too heuristic and opportunistic for us to derive
much of value from trying to force them into terms of the
latter.  Instead, I think we need to give more consideration
to the creatures that actually use human languages, and to
what they use them for.

In studying languages, there are three "worlds" involved,
which we need to keep distinct:

  1) the "real" world, things that exist objectively;
     e.g., a flesh-and-blood horse.

  2) the conceptual world, the mental model of the world,
     constructed in our heads, which we use as a map by
     which to get around; e.g., your or my perception or
     conception of that horse.

  3) the verbal world, the linear stream of codewords that
     we use to pass demands and references to the real
     world from one person to another, together with our
     traditional standards for coding and decoding; e.g.,
     the word "horse", or "Flicka".

In this light, consider your statement below:

> [(the) [horse [that I have]] is sick]
>
      [...]
>
>>   Focussed element
>
> The thing that a sentence (or clause) is about, to which
> attention is directed, e.g., in this case the horse.
> The focus of a relative clause is its head.

The sentence, including the word "horse", lives in World 3,
the verbal world.  But what the sentence is actually about,
is the flesh-and-blood horse that lives in World 1, the
real world.  If I take what you say here literally, then
the "head" of the relative clause "that I have" is a horse
and not a word.

We need to understand what is actually going on in a verbal
transaction.  If I say to you:

  Flicka is sick.

this involves you and me and allegedly a real-world horse.
I have a mental conception of this horse (World 2).  I
believe that you also have a mental conception that
references the same (World 1) horse.  I believe that I
have a piece of information about her that you lack and
ought to know; that is, I think that your (World 2)
conception should be modified.  I believe that you accept
the protocol of referencing this particular (World 1)
horse via the (World 3) pointer word "Flicka".  So I fire
off the (World 3) sentence to you, placing an appropriate
finite verb immediately after the subject word "Flicka",
which signals to you that the utterance is a statement
and that my demand is to update your (World 2) mental map
with the incoming (World 3) coded information.  You may
choose to do so, and can call me a liar should you
discover that the revision I demanded to your (World 2)
mental map did not improve its correspondence to the
situation in World 1.

A pronoun, or a name like "Flicka", is a direct reference
pointer to some presumed World 1 entity.  These are
generally short and handy, and we use them wherever
possible.  Unfortunately, they cannot be used or are not
available for all situations.  Often we have to perform
a search through the other party's mental database to
pull up the (World 2) reference we want:

  The big pink pig out in the barn
  that loves potato peels is sick.

This knowledge transfer would be much easier if we just
had "Wilbur" to use in World 3.  If not, I have to build
a noun phrase to do that search to find your World 2
reference.  "The" tells you that I think the (World 2)
reference is either already in your head or otherwise
freely available to you.  The word "pig" is an entity
class name which is also used as a conditional reference
to the (World 1) entity.  The modifiers "big", "pink",
"out in the barn", and "loves potato peels" are all
addressing criteria that are meant to help you locate,
among all the pigs you know about in your World 2,
the reference to that particular World 1 pig that I'm
trying to tell you about.

Using the formal grammar bracketting scheme you used
for my long sample Omaha sentence, we would probably
get:

  [[[[The [big [pink pig]]] out in the barn]
  that loves potato peels] is sick].

But why should we bracket it exactly this way?  Why
not just as well:

  [[The [big [pink [[pig out in the barn]
  that loves potato peels]]]] is sick].

Or:

  [[The [[[big [pink pig]] out in the barn]
  that loves potato peels]] is sick].

The only agreement here not imposed by word order and
consistency in which way we go in grouping modifiers
is that "pig" sits in the innermost grouping and that
everything in the noun phrase ultimately resolves
down to one thing exclusive of the rest of the sentence.
In fact, other than a desire to satisfy the unnecessary
constraint of trying to fit this into a formal grammar,
I see no reason for sub-bracketting the noun phrase at
all.  From a computer perspective, the point of the
sentence is this:

  SELECT pig FROM TABLE Pig
    WHERE pig.is-big
      AND pig.is-pink
      AND pig.is-out-in-the-barn
      AND pig.loves-potato-peels;
  UPDATE pig WITH is-sick;
  COMMIT;

The noun phrase in this case is just a select clause.
It makes no fundamental difference what order the
selection criteria come in.  Each of these is an
independent attribution of information regarding the
entity conditionally referenced by the noun "pig".
There is no progressive "embedding" about it.  There
is only the noun "pig", and a list of attributions
pragmatically linked to that noun according to some
more-or-less loose standards intended to prevent
confusion of intent given that the whole complex
needs to be pulled apart into a linear sequence for
transmission by air waves.

A noun phrase truly showing progressive embedding
would be:

  The man that rode the horse that kicked the dog
  that chased the cat that lived in the house
  that Jack built.

which would resolve as:

  [The man that rode [the horse that kicked [the dog
  that chased [the cat that lived in [the house
  that Jack built]]]]].

where each relative clause modifies a noun that sits
inside another relative clause.  This is different
from our horse and pig examples, where the relative
clauses and other modifiers all modify the same noun;
i.e., they all have the same "head", if you will.
We would do better to conceive these phrases as a
non-linear linked structure of what, at this level,
are items of three logical classes:

  Determiner    Noun         Attribution
  ----------    ----         -----------

     The -----> pig
                 ^---------- big
                 ^---------- pink
                 ^---------- out in the barn
                 ^---------- loves potato peels

A noun phrase is always intended to resolve into a
World 2 reference to an entity.  As a pronoun or a
name, it directly indicates that reference.  Where
we cannot use these, we use a noun together with a
list of attributions and determiners that tell the
listener where to look to get the reference.

An attribution, in whatever form it takes, can be used
either as an addressing criterion or as a new item
of information.  In the example sentence, "big",
"pink", "out in the barn" and "loves potato peels"
are all used as addressing criteria, while "is sick"
is presented as an item of new information about the
entity referenced by the noun phrase.

Whatever we attribute to an entity I'm calling an
attribution.  A predicate is an attribution to its
subject, and any adjectives, genitives, qualifying
nouns, prepositional phrases or relative clauses
pertaining to a noun in a noun phrase are also
attributions to the entity referenced by that noun.
In English, the former is an explicit, or overt,
attribution, while the latter are implicit, or
covert, attributions.  This is a fairly rigid
dichotomy in English.  My question is whether this
distinction is equally rigid, or even made at all,
in Siouan.

Suppose I use "A" instead of "The":

  A big pink pig out in the barn
  that loves potato peels is sick.

Now I'm implying that I don't expect you to have a
(World 2) reference to my (World 1) pig.  To a computer,
this is an awkward insert + update statement:

  INSERT pig INTO TABLE Pig
    VALUES (pig.is-big,
            pig.is-pink,
            pig.is-out-in-the-barn,
            pig.loves-potato-peels);
  UPDATE pig WITH is-sick;
  COMMIT;

This is a tad inefficient, but it's the way our
English speaking minds work.  We have to do an
update here because every statement requires a
pre-existing subject and an updating predicate.
This requirement, in turn, is due to the fact
that we depend on a special verb to signal that
this is a statement, and the attribution controlled
by that verb is the update demand.

A more normal way of presenting this as a new record
of Pig would be:

  There's a big pink pig out in the barn
  that loves potato peels that is sick.

In this case, we use a dummy subject "there", and
slur the following finite verb together with it.
The rest is a noun phrase for inserting.  This is
introduced with the determiner "a", which implies:
"you-don't-have-a-reference-to-the-entity-I'm-
thinking-of-or-if-you-do-I-don't-know-how-to-find-it".
Then we have the noun, surrounded by five separate
attributions that each link to it independently of
the others.

>>From the perspective of basic English grammar, this
is a subject "there", followed by the finite/demand
verb "is", which signals that the utterance is a
statement, and which controls the predicate as an
attribution to the subject.  However, we might also
consider this a radical (if limited) departure from
standard English grammar, in which we have evolved
a demand particle "therza", having roughly the
salience of the "there exists" reversed capital 'E'
sign in mathematical logic, which sits at the front
of the sentence and signals to the listener that the
following noun with attributions represents an entity
for which they are to add a new record to their
mental database.  From this perspective, we have a
demand particle and a noun with several attributions
tied to it, but no subject, no predicate, and no
demand verb.  In pseudo-SQL, it is simply:

  INSERT pig INTO TABLE Pig
    VALUES (pig.is-big,
            pig.is-pink,
            pig.is-out-in-the-barn,
            pig.loves-potato-peels,
            pig.is-sick);
  COMMIT;

Or is it?  If someone actually spoke this sentence:

  Therza big pink pig out in the barn
  that loves potato peels that is sick.

would we not feel that the attribution "is sick" is
likely what the speaker is really driving at, as
important new information, while the preceding
attributions are simply addressing devices aimed at
helping the listener mentally locate the pig?  And
if we stripped off the "is sick" attribution:

  Therza big pink pig out in the barn
  that loves potato peels.

then it would be the fact that he loves potato peels
that would be important and new.  And if we took
that off:

  Therza big pink pig out in the barn.

the important and new information is that he's
out in the barn.

Nevertheless, there is no grammatical basis for
such distinctions.  These are simply made on the
assumption that the last attribution was the new
information and the earlier ones were addressing
criteria.  This is fuzzy, and in fact several of
the attributions could be new and important.

Now consider my original sentence about the horse,
and our attempts to translate it into English:

  ShoN'ge ska u'joN abdhiN' wakhe'ga zhoN't?e khe
    xta'adhe khe gini' koNbdhe'goN.

  I hope the beautiful white horse that I have which
    is sick and lying sound asleep, which I love,
    will recover.     (my translation)

  [I hope that [[[[[[[the horse which is white]
    which is beautiful] that I have] which is sick]
    which is asleep] which I love] will recover]]
                      (John's translation)

Perhaps a better translation would be:

  Therza horse that is white that is beautiful that I have,
    that is sick, that is lying sound asleep, that I love,
    that I hope will recover.

This version would be true to the fact that all the
attributions are grammatically equal and independent.
One is free to guess that the last attribution is the
point of the sentence, with the preceding ones being
for addressing purposes, but one can just as well take
the whole stream of attributions as new and important
information, added one step at a time.  How one takes
it depends on context, not grammar.

We are still left with a long sequence of relative
clauses in the English version.  These relative clauses
are clearly marked as such by a relative pronoun, which
does not appear in the OP version.  I believe that the
supposition that the attributions in the OP version are
subtly packaged relative clauses, except for the final
one which corresponds to the English demand verb around
which the sentence hinges, is in error, and that that
error derives from attributing our English grammatical
restrictions to the Omaha.  In English, the demand of
a sentence is signalled by a single demand verb that is
unique to that sentence.  The attribution controlled by
that verb is the explicit point of the sentence.  We
cannot freely make other attributions in the same way
without confusing the demand.  Hence, if we do need to
make other attributions, perhaps for addressing, we need
to mark them to make them distinct from the demand verb
and its predicate.  If we need to reference a noun with
an entire "sentence" that is not governed by the demand
verb, then we have to alter that "sentence" into a
relative clause that attaches to the noun in a way that
does not confuse its verb with the demand verb.  This
grammatical system is deeply embedded in our speech-thought,
and we have a very hard time imagining a language that does
not use demand verbs, with a consequent dichotomy between
explicit attribution (predication) and implicit attribution
(relative clauses).

In Siouan, the demand is coded by special particles that
come at the end, separate from the verbs.  This means that
the verbs are freed from the function of indicating demand,
and therefore that their functionality does not have to be
distinguished according to whether they are the demand verb
or not.  They function the same way all the time, and you
can pile up as many as you like on the same subject,
inflecting them any way you wish for person, because they
are not subject to the English/IE constraint of having to
also indicate demand.  This is handled separately, for the
entire sentence, by particles that specialize in it.
Relative clauses are simply not needed, because whatever
they do for us in English can be handled by perfectly
normal verb attribution in Siouan.

Broadly speaking, it seems to me that English/(IE?), and
probably other languages that use a demand verb system,
lean toward establishing a subject and updating it, while
OP/(Siouan?), and probably other languages that use a demand
particle system, lean toward building a record on the fly,
and updating or inserting according to whether the listener
has recognized the reference or not.

Rory