The Place of Representation in the Minimalist
Program
The impact of generative linguistics on
philosophy has been of some substance beyond the resuscitation of nativism and
the general cognitive revolution.
In particular, there has been an influence in
the philosophy of language, going from the early work of Fodor and Katz up to
the current day in the work of thinkers such as Higginbotham, Larson, Segal,
Stanley, Ludlow, Stainton, etc.
The basis for this: linguistic
representation/knowledge of language.
This is an error, which comes into focus with
MP. Here, I shall just give the groundwork for diagnosing the error.
Intro:
MP as less
representational than its predessors.
Q1: What does this mean?
Q2: What sense of representation is involved in
MP?
GB Architecture
D-Structure
[X-bar schemata, theta assignment]
S-Structure [BT,
Case, Bounding theory]
A-P Ü PF LF Þ I-C
S-Structure,
D-Structure, PF and LF are levels of
representation.
What
is a representation?
Philosophical
sense: Representation as Intentional
R[S, T(L)],
where
‘S’ is a subject and ‘T’ is a theory of language L.
The
Reflex thought: Representation implies represented.
…the
grammar is not itself a characterization of a system of mental representation;
it is the object of a speaker’s knowledge, not a description of how that object
is represented by the speaker (if it is)… The grammar is what is represented,
not what is doing the representing.
¾
A. George, 1989: How not to become confused about linguistics, p.91.
…
the conflation of grammar with… the mental representation of grammatical
objects is widespread and can lead to misrepresentation of the contents of
linguistic theory.
¾
J. Higginbotham, 2001: On referential semantics and cognitive science,
pp.152-3.
The
Real Problem: Representations have no represented. There is no grammar / language
distinction.
Second
Sense of Representation
Chomsky
(1955-6) on Levels:
L = [L , f
, R1 ,…, Rm
, l
, F
, w1
,…, wn]
(1)
L = set of primes (primitive
elements)
(2)
f
= concatenation
(3)
R1 ,…, Rm = set of
classes/relations defined over (1)-(2)
(4)
l
= set of constructed objects by (1)-(3)
(5)
F
= map to descriptive conditions
(6)
w1
,…, wn
= set of classes/relations defined over L and L¢,
L¢,…
On
the philosophical sense, it would be meaningless to think of the system as
being less representational, but…
[LEXICON]
[NUMERATION]
Merge/Move
Merge/Move
A-P Ü
PF *
[SPELL-OUT]
Move
LF Þ
I-C
PF and LF are the two remaining levels of representation.
Level-Free
Architecture
[LEXICON]
[NUMERATION]
Merge/Move
A-PÜ
·
Þ
I-C
Merge/Move
A-PÜ
·
Þ
I-C
Merge/Move
A-PÜ
·
Þ
I-C
…
Consequence?:
All representational conditions naturally follow from the derivational
procedure from the lexicon to A-P and I-C.
E.g., Epstein on C-Command
Representational
c-command
a
c-commands b
iff
(i) the first branching node dominating a
dominates b,
and
(ii)
a
does not dominate b,
and
(iii)
a
¹
b
vP
DP v ¢
Bill’s friend
likes+v VP
<likes> himself
Interpretive
rule: the antecedent of himself is a matching
c-commander.
Bill
c-commands friend alone.
The
DP Bill’s friend c-commands himself.
Derivational
C-Command:
a
c-commands all and only those elements of b
with which a
was paired by Merge/Move in the course of the derivation.
If a
is a merged constituent of an object constructed independently of b,
a
is not in a c-command relation with b.
Is
this a representation-free model?
We
can raise a question - at least, an apparent question - about the interpretation
of L [i.e., a state of the language faculty].
One might construe L as a step-by-step
procedure for constructing Exps [i.e., <PF,
LF>], suggesting that this is how things work as a real property of the
brain, not temporally but as part of its structural design. Assumptions of this
nature constitute a derivational approach to L. The strong derivational approach dispenses with the expression
altogether, assuming that information is provided to interface systems “dynamically”
[i.e., there is a cyclic, level-free transfer to A-P and I-C]…
With richer set-theoretic assumptions, a
recursive definition can be restated as a direct definition, in this case, of
the following form: E is an expression of L iff …E…, where …¾…
is some condition on E. One might, then, take, L to be a direct definition of
the set {Exp}, adopting a representational
approach….
The apparent alternatives seem to be
mostly intertranslatable, and it is not easy to tease
out empirical differences, if there are any.
Surprisingly, there is reason to believe
that the questions may be real.
¾
N. Chomsky 2000: Minimalist inquiries: the framework, pp. 98-9
Computational
Efficiency
(i) All relations are local (understood via Merge), none are
global.
(ii)
Step-by-step property: structure is built cyclically with (i)
loss of information and (ii) structure always extending from the root.
(iii)
Counter-cyclicity contravenes LCA.
The
principled elements of S0 are the conditions imposed on FL by the
systems with which it interacts. If language is to be useable at all, its
design must satisfy on “interface condition” IC… The goal is to determine just
what aspects of the structure and use of language are specific to the language
faculty, hence lacking principled explanation at this level.
[I]nitial
conditions…
(i) unexplained
elements of S0
(ii)
IC (the principled part of S0)
(iii)
general properties
Principled
explanation, going beyond explanatory adequacy, keeps to (ii) and (iii). An
extremely strong minimalist thesis: (i) is empty.
¾
N. Chomsky, 2001: Beyond explanatory adequacy, pp.2-3.
Consequence?
The
representational approach falls short of principled explanation, for the
conditions on a given representation can’t fall under (iii), which are general
properties of an organic system. Nor can they fall under (ii), which are external/independent
of FL proper. They must fall under (i). A principled
explanation would derive the
conditions from (ii)-(iii).