Re: New Project

At 04:02 PM 1/4/99 +0100, Kristian Jensen wrote:
>Tim Smith wrote:
>>I'll tell you about this in more detail sometime soon (I hope).
>>
>Overall, an interesting project. I'm looking forward to more
>details. I'm especially looking forward to how you're applying
>applicatives to the relative-clause formation.
>
Thanks for your interest, and I'm sorry to take so long in responding.  What
follows isn't an attempt to give a general overview of the language (I'm
still working on that), but rather an attempt to specifically answer your
implied question about applicatives and relative clauses, while putting it
in enough context to make some kind of sense.  Therefore it risks being,
paradoxically, both overly verbose and hopelessly vague.

(Warning: this is a very long post!)

In order to be topicalized, a third-person noun phrase must be
marked as proximate, and in order to be marked as proximate, it
must be either a subject or a primary object.  This follows from
the requirements that: (1) either the subject or the primary
object _must_ be proximate (otherwise there would be no way to
distinguish between them); and (2) there can only be one
proximate NP per clause.  Thus, the underlying function of
applicatives, which promote non-primary objects to primary-object
status, is to allow non-primary objects to be topicalized.  As I
see it, there is a strong relationship, at some level of
abstraction, between topicalization and relativization, because
the relativized NP (the head of the relative clause) is in some
sense the "topic" of the relative clause.  It therefore makes
intuitive sense to me that the same restrictions that apply to
topics should also apply to heads of relative clauses, and that
the same means should be used to circumvent these restrictions
(namely applicatives).  One of the things that's esthetically
pleasing to me about this grammar is that this underlying,
abstract similarity between topicalization and relativization is
overtly manifested in the surface structure.

[Note:  This grammar distinguishes between primary and secondary
objects rather than between direct and indirect objects.  A
primary object can loosely be defined as the direct object of a
monotransitive verb or the indirect object of a ditransitive
verb, and a secondary object as the direct object of a
ditransitive verb.  In the English sentence "the man gave the
woman the book", "the woman" is the primary object and "the book"
is the secondary object, whereas in "the man read the book", "the
book" is the primary object and there is no secondary object.]

To illustrate this, let's take a simple sentence from my new and
still nameless conlang:

en t'ai kharash et sei zat tuval ("the man met the woman at the
house")

In this and the subsequent examples, I'm using the following
lexical and grammatical morphemes:

Determiners:
        en (definite 3rd-person singular proximate)
        et (definite 3rd-person singular obviative)

Nouns:
        t'ai (man)
        sei (woman)

Verb stem:
        kharash (meet)

Verbal prefixes:
        Subordination:
                a- (relative clause)
        Person/number agreement:
                0- (both arguments are 3rd person singular)
        Modality:
                0- (realis)
        Tense:
                0- (non-anterior)
        Aspect:
                0- (perfective)
        Orientation:
                0- (direct)
                sa- (inverse)
        Applicative:
                ze- (location -> primary object)

Preposition:
        za (at)

The phonology and orthography, and hence the lexicon, are very
tentative, so I'm not going to say anything about them here.

So the sample sentence above can be glossed as follows (ignoring
categories that aren't relevant to the issue at hand):

en   t'ai kharash et  sei   za-et  tuval
PROX man  met     OBV woman at-OBV house

Here the topic is _en t'ai_ ("the man") and there is no focus.
It's the proximate determiner _en_, not the sentence-initial
position, that marks this noun phrase as the topic.  We could
instead put a non-topic before the verb:

et sei kharash en t'ai zat tuval

et  sei   kharash en   t'ai za-et  tuval
OBV woman met     PROX man  at-OBV house

This sentence has the same truth-value "meaning" as the previous
one, and "the man" is still the topic, but now _et sei_ ("the
woman") is the focus, so a better translation would be "the man
met the WOMAN at the house" or "it was the WOMAN that the man met
at the house".  Putting an obviative NP before the verb marks it
as the focus, whereas a proximate NP is the topic regardless of
its position.  We could also focus an oblique (prepositional)
object in the same way:

zat tuval kharash en t'ai et sei

za-et  tuval kharash en   t'ai et  sei
at-OBV house met     PROX mai  OBV woman

("The man met the woman at the HOUSE" or "It was at the HOUSE
that the man met the woman.")

It's also possible for both the topic and the focus to be
fronted, in which case the topic goes first:

en t'ai et sei kharash zat tuval

en   t'ai et  sei   kharash za-et  tuval
PROX man  OBV woman met     at-OBV house

But this doesn't happen very often.  The normal constituent order
in main clauses is verb-second, and there are constraints on when
it's possible to deviate from this order, although I haven't
figured out exactly what those constraints are.  My intuitive
sense is that there are "strong topics" and "weak topics", and
that a strong topic is always fronted, but a weak topic is
fronted only if there's no focus.  But I haven't determined
exactly what makes a topic strong (that is, what makes its
topicality salient enough to override the tendency to put only
one constituent before the verb).  A contrastive topic is
definitely strong, but there may be other kinds of strong topics
as well.  At any rate, this sentence means something like, "As
for the man, it was the WOMAN that he met at the house."

In all the examples so far, the proximate NP is the subject.  We
know this because the verb is marked as direct rather than
inverse.  Direct orientation means that the subject is higher
than the primary object on the "topic-worthiness" hierarchy (2nd
person -> 1st person -> 3rd person proximate -> 3rd person
obviative); inverse orientation means the opposite.  If we want
the primary object to be the topic, we mark it as proximate and
the subject as obviative, and we mark the verb as inverse:

en sei sakharash et t'ai zat tuval

en   sei   sa- kharash et  t'ai za-et  tuval
PROX woman INV-met     OBV man  at-OBV house

This is probably best translated by the English passive voice,
one of whose functions is to topicalize the patient:  "The woman
was met by the man at the house."

But (here's where it gets complicated) what if we want "the
house" to be the topic?  As an oblique object, it can't be made
proximate.  There can only be one proximate NP per clause, and it
has to be either the subject or the primary object; otherwise
there would be no way to distinguish the subject from the primary
object.  The answer is to use an applicative: in this case, one
that promotes the location from oblique object to primary object
and (since there can only be one primary object) demotes the
original primary object to secondary object.  Thus, as an
intermediate step, we get:

en t'ai zekharash et tuval et sei

en   t'ai ze-  kharash et  tuval et  sei
PROX man  APPL-met     OBV house OBV woman

There isn't really a good way to translate this into English, but
it's something like "The man met-at the house the woman."  From
here, we can mark the primary object as proximate, the subject as
obviative, and the verb as inverse, giving:

en tuval sazekharash et t'ai et sei

en   tuval sa- ze-  kharash et  t'ai et  sei
PROX house INV-APPL-met     OBV man  OBV woman

("At the house the man met the woman.")

In these last two sentences, unlike any of the previous examples,
we have to rely partly on word order to disambiguate syntactic
roles.  The rule is that when two obviative NPs are immediate
constituents of a clause (that is, arguments of the verb rather
than objects of prepositions, possessors of other NPs, or
whatever), they go in the order: subject -> primary object ->
secondary object.

(One point that I perhaps haven't made clear is that there are
many applicative affixes, one for promoting each of a wide range
of non-primary objects to primary-object status: one for
locations, one for beneficiaries, one for instruments, etc.
They're sort of like the trigger-role-marking affixes in
languages like Tagalog, in that they specify the semantic role of
the affected NP.)

Now, finally, we come to relative clauses, which is what this was
originally supposed to be about.  A relative clause is basically
a sentence with the relativizing prefix _a-_ attached to its verb
and with its topic (its proximate NP) deleted.  Thus, "the man
who met the woman at the house" is:

en/et t'ai akharash et sei zat tuval

en/et    t'ai a-  kharash et  sei   za-et  tuval
PROX/OBV man  REL-met     OBV woman at-OBV house

Here, "the man" is the NP that would have been the topic of the
relative clause if the relative clause had been a full sentence.
We don't know whether it's proximate or obviative because that
depends entirely on its role in the matrix clause, which has
nothing to do with its role in the relative clause.

A primary object can be relativized in the same way:

en/et sei asakharash et t'ai zat tuval
("the woman that the man met at the house")

en/et    sei   a-  sa- kharash et  t'ai za-et  tuval
PROX/OBV woman REL-INV-met     OBV man  at-OBV house

But you can't relativize an NP that's neither a subject nor a
direct object.  In order to say "the house where the man met the
woman", we have to use an applicative form of the verb to make
"the house" the primary object, so that it can take the proximate
(topical) role within the relative clause:

en/et tuval asazekharash et t'ai et sei

en/et    tuval a-  sa- ze-  kharash et  t'ai et  sei
PROX/OBV house REL-INV-APPL-met     OBV man  OBV woman

-------------------------------------------------
Tim Smith
timsmith@global2000.net

The human mind is inherently fallible.  It sees patterns where there is only
random clustering, overestimates and underestimates odds depending on
emotional need, ignores obvious facts that contradict already established
conclusions.  Hopes and fears become detailed memories.  And absolutely
correct conclusions are drawn from completely inadequate evidence.
        - Alexander Jablokov, _Deepdrive_ (Avon Books, 1998, p. 269)