Theiling Online    Sitemap    Conlang Mailing List HQ   

Case, Innateness, Almost Allnoun, NGL.

From:Gerald Koenig <jlk@...>
Date:Friday, July 30, 1999, 2:08

       <H2>What an Innatist Argument Should Look =

[Main Point glk]:
xxxx  an argument for a localist hypothesis of case roles and
clause structure as part of the genetic endowment of the human

This post is a commentary on a web document with the above url, pointed
out on Conlang by Charles, which I find very interesting and I want to
share my reactions to it. It has a lot to say about case, and I have
added some conclusions of my own about a verb-minimalist approach to
language. I have made very large unmarked deletions in it; it is best
to read the original. VST stands for Vector Space Tense as I have
developed it in the Conlang posted series and Writekit. NGL stands for
Next Generation Language as published at

Throughout this web document my (Gerald Koenig's) comments appear
initiated by my initials, glk, and ended by a line, thus:

Scott DeLancy's words.
my comments
Scott DeLancy's words.

                         Scott DeLancey
                      University of Oregon

<H3>1.    The Issue of Innateness</H3>

The so-called "innateness" issue has been a focus of controversy in
linguistics for over a generation, although the precise issues at
stake are not clearly delineated in much of the argument.  The
basic innatist claim is that expressed by Chomsky:

     ... that certain aspects of our knowledge and
     understanding are innate, part of our biological
     endowment, genetically determined, on a par with the
     elements of our common nature that cause us to grow arms
     and legs rather than wings.  This version of the
     classical doctrine is, I think, essentially correct.
     (Chomsky 1988:4)

     Our understanding of language has suffered greatly from the
structuralist pretense, born with Saussure and nurtured into
virulent growth in the age of Logical Positivism, that language has
nothing in particular to do with human beings.  This pretense is
far from dead; we see it manifested, for example, in the bizarre
concept of objectivist semantics, and in the related conception
prevalent in much of "cognitive science" of human cognition as an
information-processing module mathematically equivalent to a Turing
machine.  Many of our errors in thinking about language stem from
the fundamental error of conceiving it as essentially a
mathematical problem--of imagining language as a disembodied
system, something that could in principle be implemented in any
computing machine.

  Quite obviously to anyone who has ever spoken--or played
various childish or adult games, or even chewed and swallowed food-
-the front end of the tongue is far more agile than anything else
in the oral tract, definitely including the lips.  This is, in
fact, an evolutionary feature of the primates, probably dating back
to some pre-lemurian ancestor that used its tongue to dig insects
out of tree bark or some such.  This being the case, it is
virtually inconceivable that human vocal language would <I>not</I>
show a tendency to take special advantage of this agility.

<H3>2.    Modes of Explanation</H3>

issue).  I do, however, want to suggest that it is a dangerously
bad idea to assume a priori that any universal or widely-attested
set of facts about linguistic structure must automatically be
attributed to some innate, specifically linguistic neural
structure.  (For specific criticisms of particular arguments for
innateness in generative grammar, see e.g. Giv=A2n 1979, Itkonen
1994, 1996).  There are obviously motivations for universal
patterns for which we do not need to seek innatist explanations,
autonomously linguistic or otherwise.  That every language has some
way of expressing concepts like 'water' or 'sleep' clearly does not
need to be written into linguistic, or to more general cognitive,
theory.  The salience of these notions is a matter of human
biology, the universality of their linguistic expression a matter
of simple pragmatics--all humans will devise ways to talk about
things that all humans need to talk about.  It's worth thinking
about notions like Agent and Recipient here.  Bickerton (1990)
suggests on the basis of its universality that the Agent case role
is innately defined (though he is wisely equivocal as to whether it
should be ascribed to the human linguistic capacity or to pre-
linguistic cognitive structure).  But it is hardly necessary to
appeal to innateness here--agency, on the part of both the self and
others, is a phenomenon of human behavior that a growing child can
hardly miss noticing and wanting to talk about.  Likewise, as
anyone who has had dealings with a child in its first two years
will understand viscerally, the concept of "Recipient" is built
into the pattern of human parenting.  And a child with siblings
will very quickly find a need for rather complex disquisitions on
the subject.
     Certainly few universal or widespread patterns of
morphosyntactic structure are going to be amenable to the sort of
simple-minded explanation that applies to a phenomenon like the
universality of words for water.  But it does not automatically
follow that they must necessarily be explained in terms of an
innate linguistic endowment.  Some, such as well-known universals
     Turning to still more general principles, it is
     reasonable to speculate that the possibility of forming
     complex constructions with an embedded clausal complement
     involves no learning at all.  Rather, this possibility is
     simply available as a principle of the language faculty.
     (Chomsky 1988:17)

I have already suggested problems with this in the case of
agentivity, and here is part of the reason why generative grammar

     Perhaps the best way to pursue this argument is to discuss an
example of an innatist claim which <I>does</I> have this kind of
biological plausibility.

As an example of how an innatist argument
properly must proceed, I will in the remainder of this paper

[Main Point glk]:
outline an argument for a localist hypothesis of case roles and
clause structure as part of the genetic endowment of the human

<H3>3.    Localist Case Theory</H3>

It has long been noted (see e.g. Marty  1910, Hjelmslev 1935) that
there is much in the formal and semantic behavior of case systems
to suggest that they are organized in terms of the conceptual and
grammatical categories used to express spatial relations.  I will
not take the space here to present the full evidence and argument
for localist case (see e.g. Anderson 1971, 1976, Gruber 1976, Diehl
1975, Jackendoff 1990), or for the particular, highly restricted
version of it which I will assert here (see Diehl 1975, DeLancey
1991), but the sorts of evidence which can be adduced should be
evident from the examples which I will discuss.

     The most direct argument is based on the use of forms which
primarily encode local cases for certain more "grammatical" functions.
Roughly speaking, we can say that languages will mark the arguments of
a particular construction either in terms of their underlying case
roles [external referents, denotation glk] or in or in terms of
syntactic grammatical relations.[self,sentence-referential glk]  When
marking is in terms of underlying case roles, it is extremely common
that the marking used will be that associated with a local [spatial
glk] case.  We can consider here several cross- linguistically robust

     Probably the clearest, and most widely attested, pattern is
the marking of the "Patient" and "Recipient" of a trivalent
predicate as THEME and LOCATION (these notions will be defined
below).  Note that this is literally true for many ditransitive

     1)   She handed me the book. 6

"me" is a fixed location. The book moves.
[out of order DeLancy quote:]
     This approach requires a rather more drastic innovation from
standard approaches to case theory, though one which is anticipated
in the work of Gruber (1976).  The definition of Theme and Loc in
terms of one another entails the claim that every proposition, and
thus every clause, contains both a Theme and a Loc argument.


If this claim is true, then it follows that every proposition can be
written as a vector type with appropriate adverbialization chosen from
the set of vector adverbs, or from nouns converted to adverbs, to
particularize the general-purpose vector verb. This is equivalent to
saying that every proposition can be expressed as either a fixed
position or as a motion of some object, the "Theme". The VST verb for
fixed position is <dis>, from displaced to or located at; the VST verb
for motion on a path is <pas>, from path. We reach the conclusion that
only two basic verb forms are necessary, one for states and one for
events. Actually there are more vector verbs than that, but they are

I don't propose to use Vector Space Tense in this way for ordinary
conversation, rather I am here so using it to test the hypothesis
asserted. VST is used below in combination with the NGL grammar and
vocabulary which neatly marks case, (-ac, accusative, -ad, dative),
however it could be any language.

ziux N    Za ad pas be ziux umi ku bokac. 10
         AGENT              LOC    THEME?
          She past move by hand to me the book.   OR
          Zad mi pas ziux-ig bok. 7
          She-past meDAT move hand-ly bookACC.

          Za pa 4' disju ziuxig umi ibok.
          She past 4'displace forward (+4j) hand-ly to me the book.
          Handly is an VST adverb of manner.

          Za pa <3i,4j>pas umi moyac.
          She moved the ball 5 meters, at a 45 degree angle to herself,
          to me-LOC-DAT-RECIPIENT.
          "Me" is at the tip of the resultant vector of 5 meters.
          The Dative object is Located of the end of the resultant vector.
          This is why the Dative is considered Locative-based.

          4  5
          4 /
          333 i>

          [not to scale] 3,4,5 right triangle. "Mi" is located
          coincident with the 5, the resultant. "Za" is located at the
          lower left corner facing j^.
          To say she kicked the ball, the adverb paig, footly, is used:
          "pas paig" says "moved footly".

          The mathematical convention is to write the ordered pair or
          triplet between < and >: <x,y,z> is a math vector. In VST the
          math vector preceeds the vector verb (dis, pas, ves, ses,..)
          and then the adverbs are attached or placed after the vector
          verb. A vector sentence in NGL then can look like this:

 Agent-NOM  <x,y,z> OBJ-dat Vector-verb-[adverb] |[adverb] Obj-ACC

"Perhaps those causes can be moved on"

-Ted Kennedy. Old quote heard tonite Aug 28 99 on TV while writing this.
A Synchronistic example of an object "cause" which is not physically
located but nonetheless is treated as though it were. An abstract

Ro ka ke inle'es  re pasju.
It is possible that an unknown agent is able to make it true that
these causes repeatedly move ahead.

De Lancy resumes:
For this sentence to be true the book must literally change its
location, coming to rest at a location defined by the recipient.=20
     When recipients have surface marking which reflects case role,
they are typically marked locative, as in modern Romance and Indic
languages, or Tibetan:

Tibetan   2)   blo=3Dbzang-la deb  de   sprad
               Lobsang-LOC  book that give
               'give that book to Lobsang'

NGL translations:
                Vu ace bok xa deur uLobsang.
                You that book must give to Lobsang.
                Vu xa deur ace bok Lobsangad.
                You shall give that book to Lobsang.
                Xav ace bok pas deurig uLobsang.
                Make true that: that book move gively  LobsangDAT.
                So the claim is that "uLobsang" is a locative case as
                well as a THEME?

(If the language distinguishes locative from allative, this is
likely to be allative, as in English; note however that this
distinction is very far from being universal).  Otherwise, if they
are marked in terms of grammatical relations, they are treated as
     "Dative subjects", i.e. the subjects of verbs or emotion,
cognition, or perception (the case role sometimes referred to as
"Experiencer"), when they receive semantic case marking, are
typically marked the same as datives, that is, as locative:

Tibetan   3)   khyed=3Drang-la sbrul rmi=3Dlam btang-yong
               you-LOC       snake dream   EMIT-PRED.FUT
               'You will dream about snakes.'

glk3:          uvu fu pas tibeig inhapxuom aslises.
               to-you future move come-ly dreams snake-adj.
               Fis vu fu pas inhapxuom aslis.
               destination you will move dreams snake-type.

make better word for dream.

In many languages, the lexical encoding of situations of this type
may be even more explicitly localist, as in Newari (a Tibeto-Burman
language of Nepal):

Newari    4)   j   sw=FE-ya-gu      bas      khaya
               I.ERG flower-GEN-CLS smell(N) took
               'I smelled the flower.' (Agentive) 5

               Mi ga fis 'ecxi be oyo ad pas umi. 12
glk4:     I-NOM-AG (=destination) perception-essence of flower moved to me.

          5)   ji-ta  sw=FE-ya-gu      bas      wala
               I-DAT  flower-GEN-CLS smell(N) came
               'I smelled the flower.' (non-agentive)
               (lit. 'The smell of the flower came to me.')

                Umi ku 'ecxi bo oyo am pas.
                to-me the perception-essence of flower past moved.

need smell word.

And parallel evidence can sometimes be found even in languages with
no distinct "dative subject" construction; cf. English sentences

          6)   A while ago a crazy dream came to me. 11

glk6:          Lehi de mu [dem] ol inhapxu *loxo pas [kin] umi. 12
             Short time-interval left of [before] now a crazy dream
             move [coincident with] me.

             [abstract object, dream]
Note: put lehi, leho in writekit from Vector Anatomy post.
*id-::- prefix or particle makes object specificly abstract.

     When they are marked for their semantic role, possessors in
the possessive clause construction are also treated as locatives:

Russian   7)   u  menja   kniga
               at me(DAT) book
               'I have a book.'

                Dis umi bok.
                located at me book.

Tibetan   8)   khyed=3Drang-la ngul-tsam  yod-pas
               you-LOC       money-some COP-INTERROGATIVE
               'Do you have any money?'
               (lit. 'Is there money at you?')
                *Tin Koin dis uvu?
                 False/true?: Money is located at you.
*tin::- natu/ti; false/true?

Again, peripheral evidence for this equation can be found even in
languages which treat the possessor in a possessive construction as
an ordinary subject; cf. English:

          9)   You got any money <I>on you</I>?

Otherwise, if they are marked in terms of grammatical relations,
they are treated as subjects.

<H3>4.    The Correct Theory</H3>

Several lines of argument point to a maximally constrained localist
theory of case, in which core clausal arguments, at least
("Instrument" and such like are something different; see DeLancey
1991), are accounted for with an inventory of only three underlying
cases:  Agent, Theme, and Loc(ation).  I have dealt with problems
of Agentivity elsewhere (DeLancey 1984, 1985a, b, 1990a, b); my
concern here is with the Theme/Loc system (see also DeLancey 1991).=20

These categories are mutually defining--essentially, a clause
represents a proposition, and a proposition must express a Theme-
Loc relation, i.e. a Theme located at or coming to be located at a

glk:  Writing the above last sentence in VXT form I have:

Ex1: "stative"

       x epsilon A. (x is a member of the set A.)[example proposition]
       [x is a Theme, A is a Location]
       x is displaced at/in/inside/with set A.
       [In case of an infinite set the x is "adjoined" with A.]

    *  x disnir A.

       x is in A.
       [VXT formulation, "dis" means located at or displaced to
       A={u,v,w,x}. A set.]
       Thus set membership is a locative relation and conforms to the
       hypothesis that all predications have a THEME and LOC.

(Located vs. coming to be located is, of course, the
basis of the stative/eventive distinction).

Ex: "eventive, becoming"
        x epsilon A.

        x is moving on a path to (the set) A. [ex. proposition]

        @--------->x{_, u,v,w }  [diagram of x moving to set A]

        <X,0,0>; is a path vector, an ordered triple <x,y,z>, about to
        enter the set A as the variable x changes. x is the immigrant,
        A is the country. The journey is the path.

     *  x paspir A.

        x is moving on a path now penetrating: the set A.
        [The path need not be linear.]
        VXT formulation demonstrating eventive proposition.
        "pas" means moving on a path in VXT, a path vector; pir means
        penetrating. It could be adjoining.

<Dis> is "stative", <pas>, <ves>, <ses> are "eventive" in this terminology.

This particular example is important because many mathematicians believe
that all mathematics can be formularized in, if not discovered by, set
theory with its basic membership function. The membership relation
in math destroys the distinction between pas and dis because it does not
consider time, hence motion. Set theory is untensed for better or worse.


Clearly this works
only if "location" can be taken in several rather abstract senses.=20
For example, possessors and "experiencers" must be considered
abstract Locations, with possessed and thoughts, sensations, etc.,
Themes.  This makes a certain intuitive sense, but the primary
arguments in its favor are empirical, in particular the patterns of
case "syncretism" that I have discussed above.  The analysis of
ordinary transitive clauses in terms of this scheme will require
the interpretation of states as abstract locations (see below).=20

VST does not have abstract locations, an abstractor particle <id> will
need to be introduced for this to work.
Arguments for this have been made for years (e.g. Anderson 1971,
Diehl 1975, Jackendoff 1990); for the present I will point out only
that, as we will see directly, it makes for a neat as well as
intuitively plausible system.

     This approach requires a rather more drastic innovation from
standard approaches to case theory, though one which is anticipated
in the work of Gruber (1976).  The definition of Theme and Loc in
terms of one another entails the claim that every proposition, and
thus every clause, contains both a Theme and a Loc argument.

glk, repeating in context:

Then it follows that every proposition can be written as a vector type
with appropriate adverbialization with the set of vector adverbs to
particularize the general-purpose vector verb. This is equivalent to
saying that every proposition can be expressed as a fixed position or
as a motion of some object, the "Theme". I don't propose to use Vector
Space Tense in this way for ordinary conversation, rather I am here so
using it to test the hypothesis asserted.

of course, there are clause types in every language which have only
one NP argument.  The theory can be reconciled with the data only
if we can allow one of the two to be lexicalized in the verb,
rather than represented as a NP.  E.g:

     10a) The door is open. 4
            THEME     LOC
      Ku  'an dis cewig. 5
cew V             LOC
      The door is displaced openly.
      Contrary to Delancy's formulation, the locative sense is carried
      by the adverb in VXT. If I read Delancy well.

      The door is green. Green is a pred. adj.
      No locative? Or is existence a locative?.

stative case; no delta position.
       b) The door opened.
             THEME  LOC

          Ku 'an ad pas cewig.
          The door past moved (path) openly.
          openly: a VST adverb of manner.
          Again, the locative sense is carried by the adverb.

eventive case; change of position.

       c) She opened the door
          AG  LOC    THEME
         Za ad pas cewig ku 'an.
         She past moved open-ly the door.
         AGENT         LOCATIVE    THEME

eventive case: change of position with agent.

It appears so far that the assertion by DeLancy that there is a
locative sense to each predication is borne out by the VST forms which
use a combination of a locative (dis) or motion (pas) vector verb with
an associated adverb to carry the same sense as the ordinary verb which
carries the combined meaaning.

But there is abundant evidence for this move.  Consider Fillmore's
(1970) analysis of <I>hit</I> and <I>break</I> verbs.  Fillmore
shows that the object of a change-of-state verb like <I>break</I>
has some sort of patient or undergoer type role (Fillmore
provisionally uses "object"; I'll call this role Theme, following

but the object of what he calls "surface contact" verbs--
generally, verbs of affectionate or hostile physical contact like
<I>hit</I>, <I>hug</I>, <I>kick</I>, <I>kiss</I>--is some sort of
locative.  Fillmore notes, for example, a peculiar use in English
of a locative prepositional phrase which is unique to this class of
verbs.  With any other kind of clause an oblique locative can only
denote the place where the overall event occurred:

break verb:  Change of state for object.

Hit verbs need or have a locative Prepositional Phrase. Non
hit verbs use their PP to uniquely locate "the place where the
overall event occurred"

Ex: normal verb.

Ex" I saw the glass in the sink.
ie: It was inside the sink that the glass was seen. No other location is

change of state  verb type:

1)   I broke the glass in the sink.

     The point of this is that the verb "break" has the special
     "locative PP": "in the sink". It normally modifies the verb.

     Mi am pum dis-nir-oad be tiwe ku reyla[a]c.
  I past break locate-in-ADV-PP(locate-in-ly-ed)[linkword]sink the glassACC
 OR:    Miam pum dIDisnir tiwe Ireyla
                LOC          ACC
     I past break located-in sink glass-ACC.
    [The sense is that the breakage took place in the sink.]

(The reading in which the PP belongs to the object NP is irrelevant
glk: PP= in the sink.
     NP= the glass

glk: here is the "irrelevant" reading:

        Mi am pum ku reylac gua gol am disnin ku tiwe.
        I past-break the glassACC [call it it1], it1
        past-locate within the sink.
        This is two separate assertions:
        1. I broke the glass.
        2. It was at some past time in the sink.
         Another sense:
        Mi am pum ku raylac ke am disnin ku tiwe.
        I broke the glass that was located in the sink.
        [Here it may or may not have been broken _in the sink.]
With <I>hit</I>-class verbs, however, an oblique locative
can be added which specifies more precisely the part of the object
toward which the action is directed:

"Oblique locative" is starting to look like a tighter photographic/
visual focus. Zoom-in/out to the place of action.???  "on the lips" is
the OL here.
2)   I kissed her on the lips.

     Mi am gape iza ton ku gules.  [Calque translation.]
     Mi am diston zas gulesad gapeac.
     I past located-on her lips-dat  kiss-acc.

     Mi am gape dIDiston zas gules.
     I kissed located-on-lyADVERB her lips.

Apparently VXT has the stuff to express the oblique, if I understand
it. It seems to be a spatial restriction on the verb's action, hence an
adverb. "On the lips" would seem to have an adverbial phrase function.

Another piece of evidence which Fillmore does not mention is that
this class of verbs in English is uniquely eligible for a
productive light verb construction with the verb stem used as a
noun and <I>give</I> used as the verb:

3)   I gave her a kiss.

As we have already seen, the recipient argument of a trivalent verb
is underlyingly a Locative.  Thus <I>her</I> in (3) is
transparently a Locative argument,
Transparently, heh, these arguments still has clothes for me.

I displaced a kiss, destination: (to) her:
Mi pa dis ol gape fis iza.
I vectored a kiss to her.
Mip vek ol gape iz.

Again verifying that LOC case underlies what is commonly called DAT,
if we allow that "I gave [to] her a kiss" is the same thing as "I kissed
her", where "her" looks ACC.

lending further credence to
Fillmore's suggestion that it is likewise in (2).
     Again, of course, we have the problem of the missing argument-
-if <I>the glass</I> in (1) is a Theme argument, where is the Loc?=20
And if <I>her</I> in (2) is Loc, where is the Theme?  <I>Break</I>,
of course, is the same kind of verb as <I>open</I>, and like
<I>open</I> it names the change of state which the Theme undergoes,
and can thus be interpreted as providing the Loc argument.  For
"surface-contact" verbs like <I>kiss</I>, the <I>give</I>
construction furnishes direct evidence that the verb incorporates
a Theme argument, with the object as its Loc.
     Additional evidence for this interpretation is found in
Tibetan.  Tibetan overtly reflects the change-of-state vs. surface
contact distinction which is covert in English:  both classes of
verbs take an Agent argument marked with ergative case; the other
argument of a change-of-state verb is zero-marked, like other
Themes, while the second argument of a surface-contact verb is
marked as locative:

4)   shing-la sta=3Dre gzhus-pa
     tree-LOC axe    hit
     hit the tree with an axe

5)   sta=3Dre-s  shing 'chad-pa
     axe-INSTR tree  cut
     cut down the tree with an axe

Both of these still can be expressed as vector verbs with adverbs
though. Change of state can be expressed as position change or

In modern Tibetan, the vast majority of surface-contact verbs are
light verb constructions:

6)   thub=3Dbstan-gyis blo=3Dbzang-la <I>kha   bskyal</I>-song
     Thubten-ERG     Lobsang-LOC  mouth delivered-PERF
     'Thubten kissed Lobsang.'

7)   thub=3Dbstan-gyis blo=3Dbzang-la <I>mur=3Drdzog gzhus</I>-song
     Thubten-ERG     Lobsang-LOC  fist      hit-PERF
     'Thubten punched Lobsang.'

These thus transparently treat the Theme which is part of the
meaning of the English verb stem as a distinct argument (zero-
marked, as befits a Theme).

<H3>5.    How to Argue for Innateness</H3>

I have not, of course, provided sufficient syntactic and
typological evidence here to establish the superiority of this
     It turns out that this theory looks very much like the
fundamental structural construct of perception--Figure and Ground:

          One of the simplest and most basic of the
          perceptual processes involves what the Gestalt
          psychologists call <I>figure-ground
          segregation</I>.  Every meaningful perceptual
          experience seems to require in its description
          the property of "figuredness."  That is,
          phenomenally, perception is more than a
          collection of unrelated, unintegrated, sensory
          elements.  The units of perception are,
          rather, figures, or things, segregated from
          their backgrounds.  (Dember 1963:145-6)

In their concrete spatial use, Theme and Loc correspond directly to
Figure and Ground.  Nothing is intrinsically Theme or Loc; these
are relational notions.  A speaker presents one referent in
relation to another; the first we call Theme, and the second Loc.=20
Thus, despite some argument to the contrary in early literature on
Case Grammar (see Huddleston 1970), (8) and (9) are by no means

8)   The bank is next to the Post Office.

9)   The Post Office is next to the bank.

(8) describes the location of the bank, using the Post Office as a
reference point; (9) describes the location of the Post Office,
using the bank as a reference point.  Thus the subject of each
sentence denotes the referent to which the speaker wishes to draw
the addressee's attention, and the oblique NP denotes a referent
used as a background against which the subject can be identified.
     Now, figure-ground organization is, self-evidently, not a
feature of the physical universe; rather, it is a pattern imposed
on a stimulus by the process of perception.  Much work in
perception has been concerned with what we might think of as
prewired determinants of figure-ground identification.  All other
things being equal (e.g. in a properly designed experimental
context), humans will make a moving stimulus a figure, and the
stable environment against which it moves the ground.  Other
factors which increase the eligibility of some part of the visual
percept for figure status include defined boundaries, brightness,
color, centrality in the visual field, and, of course, lack of
competition from other areas of the perceptual field sharing these
     But in ordinary life other things are not often equal; any
perceiver in any real-life circumstance is predisposed by her
existing cognitive structures, and long-term and transient
"interests", to focus on certain types of structure as opposed to
others.  A universal pattern, which is probably innate, is that a
percept interpretable as a human figure has a higher eligibility
for figurehood than anything else, and a human face the highest of
all.  There is abundant evidence for what is sometimes called a
motivation effect in perception, i.e. the fact that a perceiver,
being more interested in some types of information than others,
will tend to organize the perceptual field so that relevant
information counts as figure.
     As any introduction to perceptual psychology will point out,
beyond the simple neurophysiology of edge detection, color
perception, etc., perception is a cognitive process.  In fact, it

basic level involves mental manipulation of representations of
objects (or, at the next higher level, categories of objects), and
the discrimination of objects is the basic task of perception.=20
Indeed, the figure-ground opposition is fundamental to--we could
even say, is--object discrimination.  The process of discerning an
object is the process of perceiving it as figure.
     It thus makes eminent sense that the evolution of cognition
should work from preadapted perceptual structure, and that the
opposition of figure and ground should be carried over from its
origins in the perceptual system to higher-order cognitive
structures which evolved to process, store, and manipulate
information obtained from the perceptual system.

     Indeed, if we think of language functionally, in the most
basic sense, it is almost inevitable that fundamental aspects of
its structure should mirror the structure of perception.  The same
philosophical tradition which gives us the peculiar conception of
intelligence as information-processing, inclines us to imagine that
what is passed from one mind to another in the course of
communication is some sort of pure information.  It is, of course,
no such thing.  In its communicative function, language is a set of
tools with which we attempt to guide another mind to create within
itself a mental representation which approximates one which we
have.  In the simplest case, where we are attempting to communicate
some perceived reality, the goal is to help the addressee to
construct a representation of the same sort that s/he would have if
s/he had directly perceived what we are trying to describe (cf.
DeLancey 1987).  Clearly all of the necessary circuits and
connections will be much simpler if that input, which is thus in a
very real sense an artificial percept, is organized in the same way
as an actual percept.  This involves many other aspects which are
also conspicuous in linguistic structure--deixis, to take one
striking example--but must, fundamentally, involve figure-ground
organization, since that is fundamental to perception.
     Thus the hypothesis that Figure-Ground structure might inform
the basic structure of syntax has exactly the sort of
<I>biological</I> plausibility that any innatist hypothesis must