<< . .

( 19)

. . >>

ral language syntax, a core of computational processes and relations is
fundamentally local, i.e. it can only take place within a limited amount
of structure. Locality can reasonably be construed as an economy prin-
ciple, in that it limits the amount of structure to be computed in a
single application of a local computational process, thus contributing
to reducing the complexity of linguistic computations. For instance,
the locality principle known as Subjacency, mentioned in connection
with the introduction of the concept of parameters (see above), limits
the search for the target of movement to the portion of structure con-
tained within two adjacent bounding nodes (Chomsky 1973, 1986b).
Subjacency unifies under a single formal statement much classical
work on Island Constraints (Ross 1967, 1986); its effects may now be
subsumed, in ways that remain to be fully developed and implemented,
under the Phase Impenetrability Condition. This principle, assuming
derivations to take place in distinct “phases,” corresponding to the
computation of major clausal categories (VP and CP), states that only
the edge of a phase (its specifier and head) is accessible to operations
taking place in higher phases (Chomsky 2000a, 2001a). So, in a higher
phase a computational process cannot look too deeply inside a lower

On nature and language

Relativized Minimality is another locality principle which limits
the search for the target of a local relation to the closest potential
bearer of that relation (Rizzi 1990); according to this principle, in the
following configuration:

(45) . . . X . . . Z . . . Y . . .

a local relation cannot hold between X and Y if there is an intervening
element Z which is of the same structural type as X, so that Z somehow
has the potential of entering into the local relation with Y (there is a
clear family resemblance here with the “minimal distance” principles
for control, Rosenbaum (1967), and other analogous ideas for anaphor
binding). Therefore, local relations must be satisfied in the smallest
environment in which they can be satisfied; the amount of structure to
be scanned in the computation of a local relation is correspondingly
restricted. Consider again the impossibility of extracting an adjunct
from an indirect question:

(46) How do you wonder [who solved the problem t]

Under Relativized Minimality, how cannot be locally connected to its
trace t because of the intervention of another Wh element in the em-
bedded complementizer, an element of the same structural type as
how (both in the rudimentary typology of positions distinguishing be-
tween A and A™ specifiers, and in the more sophisticated feature-based
typology of Rizzi (2001a)); so that the antecedent“trace relation fails
in this environment, and the structure cannot be properly interpreted.
This mode of explanation has been extended to the analysis of all Weak
Islands, environments selectively barring extractability to certain types
of elements, basically along the argument/adjunct divide (see Szabolcsi
(1999) for an overview); consider the sharp contrast between the
following examples in Italian:

Editors™ introduction

(47) a. Quale problema non sai come risolvere t t™?
Which problem don™t you know how to solve t t™?

b. Come non sai quale problema risolvere t t™?
How don™t you know which problem to solve t t™?

On the factors determining the selective extractability from Weak
Islands, in apparent violation of Relativized Minimality, see the ap-
proaches in Rizzi (1990, 2001a, 2001b), Cinque (1990), Manzini (1992),
Starke (2001). The original formulation of Relativized Minimality is
representational: a local relation fails at LF in a configuration like (45);
Chomsky (1995a, 2000a) offers derivational formulations in terms of
the Minimal Link Condition on Attract, and locality on the Agree opera-
tion; see also Rizzi (2001b) on the derivational/representational issue.

6.5 The copy theory of traces
All the research directions mentioned in the previous sections sug-
gest that language design is sensitive to economy principles, and well
adapted to make linguistic computations simple and smooth. How
far can these observations lead? The Minimalist Program pursues this
question by exploring the strongest thesis that can be envisaged: could
it be that language is an optimally designed system, given certain
criteria? The minimal need that linguistic computations must satisfy
is to connect interface representations, the representations through
which the language faculty “talks” to other components of the mind:
Phonetic Form, which connects language with the sensorimotor sys-
tems of perception and articulation, and Logical Form, which connects
language with the thought systems of concepts and intentions. So,
could it be that language is an optimally designed system to connect
representations legible to sensorimotor and thought systems?

On nature and language

The difficult task that the Minimalist Program has put on the
research agenda is to review all the results achieved in the study of
Universal Grammar to see if they can be meaningfully reconstructed as
meeting minimalist requirements. In some cases, it has been possible
to show that the adoption of a more “minimal” set of assumptions
can even improve the empirical adequacy of the analysis. A case in
point is the copy theory of traces and the explanation it provides for
reconstruction effects. Consider the following sentences:

(48) a. Which picture of himself does John prefer t?

b. Which picture of John does he prefer t?

(48a) is fine with the anaphor himself bound by John, and (48b) does
not allow coreference between John and he (the sentence is of course
possible if he refers to a different individual mentioned in previ-
ous discourse). Both properties are somewhat unexpected though:
anaphoric elements like the reflexive himself must be in the domain of
(c-commanded by) their antecedents; if this does not happen, as in
(49a), the structure is excluded. Reciprocally, a name and a pronoun
are free to corefer if the name is not in the domain of the pronoun, as
in (49b):

(49) a. This picture of himself demonstrates that John is really
b. This picture of John demonstrates that he is really sick

Why is it that we get reversed judgments in (48)? It appears to be the
case that, in configurations of this sort, with a complex phrase moved
to the front, the mental computation of the binding principle takes
place as if the phrase was in the position of its trace, and had not
moved at all: in fact the judgments on (48) are the same that we get
with the unmoved phrases:

Editors™ introduction

(50) a. John prefers [this picture of himself ]

b. He prefers [this picture of John]

This is the phenomenon called “reconstruction”: a moved phrase be-
haves, in certain respects, as if it was in the position of its trace. Previ-
ous proposals involved an operation “putting back” the moved phrase
into the position of its trace in the computation of LF, or a more com-
plex computation of c-command relations in the relevant environment
(Barss 1986). In fact, Chomsky points out in the first minimalist paper
(Chomsky 1993), the solution emerges at once if we go back to the
basic ingredients of the movement operation. Moving a phrase in-
volves copying the phrase into a higher position, and then deleting the
original occurrence. Suppose that, instead of being deleted, the origi-
nal occurrence is simply left unpronounced, without phonetic content,
but visible to abstract computational operations. So the representation
of (49) is the following, with the original unpronounced occurrences
within angled brackets:

(51) a. Which picture of himself does John prefer <which picture
of himself>
b. Which picture of John does he prefer <which picture
of John>

The binding principles apply on these richer representations
giving the right result: the anaphor is bound by the name in (51a), the
name cannot enter into a coreference relation with the c-commanding
pronoun in (51b). No complex theory of reconstruction is needed, and
the empirically correct result is achieved by simply tracing “movement”
back to its elementary computational components (on the adjust-
ments needed to get appropriate operator-variable structures at LF
see Chomsky (1993), Fox (2000), Rizzi (2001b); on the fact that it is

On nature and language

apparently sufficient to bind only one occurrence of the anaphor in
(51a) see the references just quoted, and also the discussion in Belletti
and Rizzi (1988); on the different behavior of arguments and adjuncts
under reconstruction, Lebeaux (1988)).
Other cases of complex empirical patterns are not so easily re-
ducible to elementary computational principles and their interactions.
Nevertheless, the successful reduction of the theory of reconstruction
is indicative of a mode of explanation that may be generalizable to
other domains of the language faculty.
To the extent to which the fundamental minimalist question
can be positively answered, large portions of UG, as they have been
determined in decades of empirical studies, may be amenable to a
further level of explanation, which may in turn guide further inquiry on
neighboring cognitive systems, and set sharper conditions for future
attempts at unification with the brain sciences.

Chapter 2

Perspectives on language and mind

It would only be appropriate to begin with some of the thoughts of the
master, who does not disappoint us, even though the topics I want to
discuss are remote from his primary concerns. Galileo may have been
the first to recognize clearly the significance of the core property of
human language, and one of its most distinctive properties: the use of
finite means to express an unlimited array of thoughts. In his Dialogo,
he describes with wonder the discovery of a means to communicate
one™s “most secret thoughts to any other person . . . with no greater
difficulty than the various collocations of twenty-four little characters
upon a paper.” This is the greatest of all human inventions, he writes,
comparable to the creations of a Michelangelo “ of whom Galileo
himself was a virtual reincarnation according to the mythology con-
structed by his student and biographer Viviani, memorialized in Kant™s
image of the reincarnation of Michelangelo in Newton through the
intermediary of Galileo.
Galileo was referring to alphabetic writing, but the invention
succeeds because it reflects the nature of the language that the little

Galileo Lecture, Scuola Normal Superiore, Pisa, October 1999

On nature and language

characters are used to represent. Shortly after his death, the philos-
opher-grammarians of Port Royal took that further step, referring
to the “marvelous invention” of a means to construct “from 25
or 30 sounds that infinity of expressions, which bear no resemblance
to what takes place in our minds, yet enable us to reveal [to others]
everything that we think, and all the various movements of our soul.”
The “infinity of expressions” is a form of discrete infinity, similar to
that of the natural numbers. The Port Royal theorists recognized that
“the marvelous invention” should be the central topic of the study
of language, and pursued the insight in original ways, developing
and applying ideas that became leading topics of inquiry only much
later. Some were revived and reshaped in Frege™s concept of Sinn and
Bedeutung, others in the phrase structure and transformational gram-
mars of the latter part of the twentieth century. From a contemporary
point of view, the term “invention” is of course out of place, but the
core property of language that Galileo and his successors identified is
no less “marvelous” as a product of biological evolution, proceeding
in ways that lie well beyond current understanding.
The same property of human language, and its apparent bio-
logical isolation, also intrigued Charles Darwin when he turned his
attention to human evolution. In his Descent of Man, Darwin wrote
that with regard to the understanding of language, dogs appear to
be “at the same stage of development” as one-year-old infants, “who
understand many words and short sentences but cannot yet utter a
word.” There is only one difference between humans and other an-
imals in this regard, Darwin held: “man differs solely in his almost
infinitely larger power of associating together the most diversified
sounds and ideas.” This “association of sounds and ideas” is the
“marvelous invention” of seventeenth-century commentators, which

Perspectives on language and mind

Darwin hoped would somehow be incorporated within the theory of
The theory of evolution, not necessarily the workings of natu-
ral selection; and surely not these alone, since, trivially, they operate
within a physical “channel,” the effects of which are to be discovered,
not stipulated. It is also worth recalling that Darwin firmly rejected
the hyperselectionism of his close associate Alfred Russell Wallace,
which has been revived in some contemporary popular versions of
so-called “neo-Darwinism.” Darwin repeatedly emphasized his con-
viction “that natural selection has been the main but not the exclusive
means of modification,” taking explicit note of a range of possibili-
ties, including non-adaptive modifications and unselected functions
determined from structure, all topics that are alive in contemporary
theory of evolution.
An interest in the nature and origins of the “marvelous inven-
tion” leads to investigation of the component of the human brain that
is responsible for these unique and indeed wondrous achievements.
This language organ, or “faculty of language” as we may call it, is a
common human possession, varying little across the species as far
as we know, apart from very serious pathology. Through maturation
and interaction with the environment, the common language faculty
assumes one or another state, apparently stabilizing in several stages,
finally at about puberty. A state attained by this faculty resembles what
is called “a language” in ordinary usage, but only partially: we are
no longer surprised when notions of common sense find no place in
the effort to understand and explain the phenomena they deal with in
their own ways, another achievement of the Galilean revolution, now
taken for granted in the hard sciences but still considered controversial
beyond “ inappropriately, I think.

On nature and language

The internal language, in the technical sense, is a state of the
faculty of language. Each internal language has the means to construct
the mental objects that we use to express our thoughts and to interpret
the limitless array of overt expressions that we encounter. Each of these
mental objects relates sound and meaning in a particular structured
form. A clear understanding of how a finite mechanism can construct
an infinity of objects of this kind was reached only in the twentieth cen-
tury, in work in the formal sciences. These discoveries made it possible
to address in explicit ways the task that was identified by Galileo, the
Port Royal theorists, Darwin, and some others “ a scattering of others,
as far as I have been able to discover. For the past half century, a good
part of the study of language has been devoted to the investigation
of such mechanisms “ called “generative grammars” in the study of
language “ an important innovation in the long and rich history of
linguistics, though as always, there are precedents, in this case tracing
back to ancient India.
Darwin™s formulation is misleading in several respects. It is now
understood that the linguistic achievements of infants go far beyond
what Darwin attributed to them, and that non-human organisms have
nothing like the linguistic capacities he assumed. Furthermore, asso-
ciation is not the appropriate concept. And his phrase “differs solely”
is surely inappropriate, though “primarily” might be defensible: the
property of discrete infinity is only one of many essential differences
between human language and animal systems of communication or
expression, for that matter other biological systems rather generally.
And of course, the phrase “almost infinite” must be understood to
mean “unbounded,” that is, “infinite” in the relevant sense.
Nonetheless, Darwin™s point is basically correct. Essential char-
acteristics of human language, such as the discrete-infinite use of
finite means that intrigued him and his distinguished predecessors,

Perspectives on language and mind

appear to be biologically isolated, and a very recent development in
human evolution, millions of years after the separation from the near-
est surviving relatives. Furthermore, the “marvelous invention” must
be present in Darwin™s one-year-old, indeed in the embryo, even if not
yet manifested, just as the capacity for binocular vision, or undergoing
puberty, is part of the genetic endowment, even if manifested only at a
particular stage of maturation and under appropriate environmental
conditions. Similar conclusions seem highly plausible in the case of
other aspects of our mental nature as well.
The concept of mental nature underwent an important revision
in the Galilean era. It was formulated in a novel way, in fairly clear
terms “ and I think it can be argued, for the last time: the concept soon
collapsed, and nothing has replaced it since. The concept of mind was
framed in terms of what was called “the mechanical philosophy,” the
idea that the natural world is a complex machine that could in principle
be constructed by a skilled artisan. “The world was merely a set of
Archimedian simple machines hooked together,” Galileo scholar Peter
Machamer observes, “or a set of colliding corpuscles that obeyed the
laws of mechanical collision.” The world is something like the intricate
clocks and other automata that excited the scientific imagination of
that era, much as computers do today “ and the shift is, in an important
sense, not fundamental, as Alan Turing showed sixty years ago.
Within the framework of the mechanical philosophy, Descartes
developed his theory of mind and mind“body dualism, still the locus
classicus of much discussion of our mental nature, a serious misunder-
standing, I believe. Descartes himself pursued a reasonable course. He
sought to demonstrate that the inorganic and organic world could be
explained in terms of the mechanical philosophy. But he argued that
fundamental aspects of human nature escape these bounds and can-
not be accommodated in these terms. His primary example was human

On nature and language

language: in particular, that “marvelous invention” of a means to ex-
press our thoughts in novel and limitless ways that are constrained
by our bodily state but not determined by it; that are appropriate to
situations but not caused by them, a crucial distinction; and that evoke
in others thoughts that they could have expressed in similar ways “ a
collection of properties that we may call “the creative use of language.”
More generally, Descartes held, “free will is in itself the noblest
thing we can have” and all that “truly belongs” to us. As his followers
expressed the thesis, humans are only “incited and inclined” to act
in certain ways, not “compelled” (or random). In this respect they are
unlike machines, a category that includes the entire non-human world,
they held.
For the Cartesians generally, the “creative aspect” of ordinary
use of language was the most striking illustration of our noblest gift.
It relies crucially on the “marvelous invention,” the mechanisms re-
sponsible for providing the “infinity of expressions” for expressing
our thoughts and for understanding other people, though it relies on
far more than that.
That we ourselves have these noble qualities of mind we know by
reflection; we attribute them to others, in the Cartesian model, by “best
theory” arguments, as they are now called: only in this way can we deal
with the problem of “other minds.” Body and mind are two substances,
one an extended substance, the other a thinking substance, res cogitans.
The former falls within the mechanical philosophy, the latter not.
Adopting the mechanical philosophy, “Galileo forged a new
model of intelligibility for human understanding,” Machamer argues
plausibly, with “new criteria for coherent explanations of natural phe-
nomena” based on the picture of the world as an elaborate machine. For
Galileo, and leading figures in the early modern scientific revolution
generally, true understanding requires a mechanical model, a device

<< . .

( 19)

. . >>

Copyright Design by: Sunlight webdesign