![,
2 WILLIAM D. DAVIES AND STANLEY DUBINSKY
functions and determine whether they are interesting viable notions or should simply
be abandoned as epiphenomenal mnemonic terms.
The collection of papers in this volume addresses the role of subject and object
in the theory largely by making specific proposals about phenomena in a range of
languages. Most conclude that the properties of objects and subjects can be
attributed to functional or phrase structural properties and include novel proposals
for the representation of these properties. .
In this chapter, we outline some of the foundational developments in the role of
granunatical functions in syntactic description and analysis, beginning with
Chomsky 1965 (section 1). In section 2, we treat, in somewhat more detail, recent
developments in the treatment of subjects and objects. Section three elaborates on
points of convergence achieved by the chapters in this volume, and points the way to
a typology of language architecture and to further research in the area.
1. A POINT OF DEPARTURE
The assumption that GFs such as subject and direct object are definable in terms of
phrase structure configurations, rather than being primitives of the theory, is a
position which was first articulated in Chomsky 1965. Therein, Chomsky clearly
distinguished GFs from granunatical categories such as NP, asserting that "the
notions in question have an entirely different status" (p. 68) and that the functional
information is redundant and "can be exacted directly from the rewriting rules ofthe
base, without any necessity for ad hoc extensions and elaborations...to provide
specific mention of grammatical function" (p. 73). Thus, it is assumed in Chomsky
1965 that GFs are defined notions, defined in terms of phrase structure
configurations and perhaps only relevant to deep structure.I The general definitions
that Chomsky assumes for grammatical functions are as follows: (i) Subject-of =
[NP,S], (ii) Predicate-of = [VP.S], (iii) Direct-Object-of= [NP,VP]. and (iv) Main
Verb-of = [V,VP] (p. 71). Thus, by these definitions in a structure such as (1),
sincerity is the subject and the boy is the direct object.
(1) S
�
NP Aux VP
I I �
N M V NP
I I I /"-....
sincerity may frighten Det N
I I
the boy
This position is reasserted in Chomsky 1981, in reaction to certain advances in
the theories of Relational Grammar and the nascent Lexical Functional Grammar:](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-21-320.jpg)
![.
4 WILLIAM D. DAVIES AND STANLEY DUBINSKY
Aissen & Perlmutter 1976 and Gibson & Raposo 1986. Burzio, for his part,
formalized an account of unaccusativity first presented in Perlmutter 1978.
At the same time, the publication of the GS theory in Chomsky 1981 and 1982
involved the further intrusion of GFs as undefined (or inconsistently defined)
theoretical objects. In Chomsky 1981 (209-222), for example. the term SUBJECT is
introduced into the calculation of "governing category". SUBJECT is used there to
refer both to NPs occurring in the traditional. configurationally defined "subject
position" (i.e. the immediate NP daughter of a clause or NP node), as well as to
AGR (which is asserted to be the "nominal" content of INFL). On this formulation.
the notion SUBJECT is no longer a purely configurationally defined relation, but one
which identifies its referents variously on the basis of either configuration or
labelling properties. With the articulation of the Extended Projection Principle
(EPP). or more specifically the extended part of the EPP (Chomsky )982:10), GFs
came to play an even more critical role in the central assertions of the theory. Here
it is stated that ''the requirement that a clause have a subject is independent of the
Projection Principle" and that this requirement is "a general principle governing D
structures, hence also governing structures derived from them". The "subject" of the
EPP, however, is not the same as SUBJECT (since clauses of Romance languages are
held to have "subject positions" in addition to the SUBJECT [i.e. AGR] which they
would possess anyway). Putting this inconsistency aside, the EPP elevated the
notion "subject" (albeit a configurationally defined notion) to the level of a UG
determinant of core grammar (Chomsky 1982:13).
Since the mid-1980's, various developments in GS theory and what has evolved
into Minimalism have further unsettled the traditional view of GFs in
transformational syntax. One of these, the VP-lNTERNAL SUBJECT HYPOTIIESIS (see
Kitagawa 1986 and Sportiche )988 among others), resulted in the abandonment of a
singular. configurationally defined "subject position", inasmuch as this hypothesis
provides separately for deep (i.e. thematic) and surface (i.e. inflectional) subject
positions. This notion that all subjects are derived led subsequently to a
reexamination of the category INFL itself, and to proposals in which its various
functional components are separated out into independently headed projections. The
view (expressed first in Pollock 1989) that AGR and Tense reside in separate
functional categories has opened the door for further discussion of the positions and
properties of subjects, such as found in Guilfoyle, Hung, & Travis 1992, Sobaljik &
Jonas 1996, McCloskey 1997.
At the same time that the notion "subject" has come under renewed scrutiny,
there has arisen a corresponding interest in the traditional object functions of direct
object (DO) and indirect object (10). Drawing on the VP-internal subject hypothesis
discussed above, Larson (1988) proposed a view of multiple object constructions in
which the underlying positions of direct and indirect objects are comparable to that
of subjects and direct objects. That is, just as the clausal subject occupies a D
structure spec,VP position superior to the complement, the direct object of a
ditransitive verb is claimed to occupy a spec.VP (of a lower VP) and be superior to
the indirect object complement in D-structure. This analogy is illustrated in (2),
where VP1 in (2a) should be compared to VP2 in (2b).](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-23-320.jpg)
![,
6 WILLIAM D. DAVIES AND STANLEY DUBINSKY
2. RECENT VIEWS ON GRAMMATICAL FUNCTIONS
2.1 On the sub
ject o
fsub
jects
As noted above, the notion "subject" and its role in the grammar evolved
considerably over the past three decades. While a configurational definition of
subject remained intact and relatively unchanged from Aspects to LGB and into
Barriers (the only change coinciding with the introduction of the CP/IP
nomenclature), the role of subject became increasingly important. This is
epitomized by the Extended Projection Principle of Chomsky 1982, which explicitly
requires all clauses to have subjects, defined therein as [NP,S]. Since then,
speculation has arisen regarding whether all languages obey the EPP and whether
some languages satisfy the EPP in alternative ways. Additionally, developments in
clausal architecture have radically reshaped the configurational definition of subject
and, in fact, have cast doubt on whether a unitary notion of subject can be identified,
as taken up in part III of this volume.
An early step in what McCloskey (1997) tenns the "deconstruction" of a unitary
subject was the introduction of the VP·internal subject hypothesis. This notion, that
the subject originates in the VP and then in most (but not all) languages moves to
spec,IP, entails the positing of two subject positions. Some of the most striking
evidence for this hypothesis comes from Sportiche's (1988) analysis of quantifier
float.
(3)a. Tous les enfants ont vu ce film.
all the children have seen this movie
b. Les enfants (*tous) ont tous vu (*tous) ce film.
the children have all seen this movie
Sportiche adduces evidence for a VP·internal position for subjects in order to
explain the fact that the quantifier tous 'aU' in (3b) can appear between the auxiliary
and main verbs, but not in other plausible positions (such as before the auxiliary
verb or after the main verb). Sportiche suggests that for French (and the English
analog), the NP complement of the quantifier may raise independently to spec,IP,
leaving the quantifier stranded as shown in (4).
(4) [IP [NP les enfants]1 ont [vp [op tous tl ] vu ce film 11
Of course, (3a) is derived by having the entire QP move from spec,VP to spec,IP.
These facts, together with other considerations (e.g. adverb placement), provide
evidence of a VP·internaJ position for the subject. FUrther evidence for a VP·
internal subject position came from languages for which it was argued that the
subject of the sentence never raises to spec,IP but remains in the VP at S-structure.
Such proposals include Kuroda 1988 for Japanese and McCloskey 1 991 for Irish.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-25-320.jpg)
![GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 7
The notion of two subject positions was developed further by Guilfoyle, Hung,
and Travis (1992), who present an analysis of Austronesian languages that makes
crucial use of the two subject positions, spec,VP and spec,IP. In their analysis, they
extend the potential of a non-unitary notion of subject, and argue that the properties
most often associated with a single argument and identified as "subject properties"
can actually be split between two arguments in certain constructions in these
languages.· So, for example, in the Tagalog examples below (Guilfoyle, Hung, and
Travis 1992:(25» babae 'woman' possesses all subject properties in (Sa), while the
subject properties are split in (5b) between babae and bigas 'rice'.2
(5)a. Mag-aalis ng-bigas sa-sako para sa-bata' ang-babae.
AT-take.out Acc-rice OBL-sack for OBL-child TOP-woman
'The woman will take rice out ofthe sack for the child.'
b. AaHsin ng-babae sa-sako para sa-bata' ang-bigas.
TT-take.out GEN-woman oBL-sack for OBL-child TOP-rice
'The woman will take the rice out of the sack for the child.'
Under their analysis of (Sa), theVP-intemal subject babae moves from spec,VP to
spec,IP ( at the end of the clause), thus filling both subject positions. In (Sb), babae
remains in the spec,VP subject position, while the theme bigas moves to spec,IP.
This is illustrated in (6)-n.b. t2 is the trace orv movement.
(6)a.
b.
[IP [I' mag-aalis2 [yp t
ang-babae IP]
t2 ng-bigas sa-sako para-sa-bata' ]]
lIP [I' aalisin2 [yp ng-babae t2 tt
ang-bigast IP]
sa-sako para-sa-bata' ]]
Under Guilfoyle, Hung, and Travis' analysis, typical Tagalog subject properties
such as reflexive antecedence and being controlled are characteristic of the spec,VP
position, while the properties of floating quantifiers and extraction accrue to spec,IP.
When a single argument fills both positions (via movement), as is the case for babae
in (Sa), that argument exhibits all subject characteristics. In (5b), however, the deep
subject babae originates in spec,VP and resides there at S-strucrure, while the object
bigas moves out of the VP complement position coming to occupy the surface
subject position spec.IP at S-structure. Accordingly, the subject properties in (5b)
are split between these two arguments.
More recent accounts, particularly spurred on by the disassembling of INFL in
Pollock 1989, have proposed yet other subject positions. In work on the Icelandic
transitive expletive construction, Bobaljik and Jonas {I996) have argued that there is
a subject position outside ofVP but lower than spec,AgrSP (which Teplaced spec,IP
as the identified S-structure subject position following Chomsky 1991). They
identify this position as spec,TP. In the Icelandic construction, both spec,AgrSP and
spec,TP are occupied by phonetic material. As McCloskey (1997) points out, with](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-26-320.jpg)
![, i
8 WnLlAM D. DAVIES AND STANLEY DUBll'lSKY
two subject positions in the inflectional layer, this either brings to three the number
of identified subject positions or casts doubt on the VP-internal subject position.
Additionally, Branigan (1996), Cardinaletti (1997), Kiss (1996), and Haegeman
(1996) have proposed additional subject positions outside of the thematic and
inflectional layers. For example. Branigan (1996) suggests that EPP features drive
movement into spec,CP in verb-second languages. In such a proposal spec,CP thus
becomes a GF position in those languages which have EPP-motivated movement to
positions outside the inflectional layer.
Current theoretical proposals, calling for mUltiple "subject" positions and for the
parcelling out of "subject" properties among them, are reminiscent of RG proposals
of ten to twenty years earlier. TheVP-internal subject position, assumed in most
current work, has its analogue in the RG concept "initial subject". Analogous to
current theoretical proposals for derived "subject positions" (in the inflectional layer
of the clause), RG theory recognized the distinct status of "final subjects" in
triggering agreement and f�eding word order rules (among other things). For
example, Bell (1976) proposed distinct grammatical properties for initial and final
subjects in Cebuano, a proposal quite similar to that offered (in the current model)
by Guilfoyle, Hung. and Travis for other Austronesian languages.
2.2 Objects and other objects
Unlike subject, the definition of the GF notion direct object was less troublesome to
Chomsky's transformational theory at the outset. It was, by and large, assumed that
direct objects of simple clauses (such as her in (7a» are not derived. Subsequently,
Chomsky'S (1981) adoption of the Projection Principle entailed that no surface
direct object can be derived from any other position. This helped provide
justification for the exceptional case marking (ECM) analysis of accusative
infinitival subjects, by which him in (7b) is a deep and surface subject despite its
accusative case.l
. (7)a.
b.
James saw her.
Garth wants [5 him to leave]
Thus, the Projection Principle divided the class of apparent surface direct objects
into true direct object complements (7a) and accusative infinitival subjects (7b).
Under these assumptions, direct object is trivially defined as the NP sister ofV.
This situation changed with the publication of I.-arson 1988, in which the position
of a deep direct object was claimed to be dependent upon the valency of the verb.
As shown in section 1 (example (2», the direct object of a transitive verb is analyzed
as the NP sister of V, while the direct object of a ditransitive verb is claimed to
occupy spec,
VP (of a lowerVP). Thus. under a Larsonian approach to D-structure,
object properties can no longer be attributed to a singular, configurational definition
of the notion direct object.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-27-320.jpg)
![GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 9
While the Larson (and later. Pesetsky 1995) approach to lexical insertion did
muddy the definition of deep-subject, it left intact the notion that direct objects are
not derived (at least in contrast to the VP-intemal subject hypothesis which held that
subjects occupy distinct thematic and inflectional positions). However, following
Pollock's (1989) division of INFL into distinct AGR and Tense components,
Chomsky (1991) proposed an inflectional position for objects parallel to that of
subjects. Based on work by Kayne (1989), Chomsky takes Pollock's AGR to be a
"subject-agreement element" (AgrS), and suggests that there is an "object-agreement
element (AgrO) ... [that has] VP as its complement". On this view, (7a) would have
the following representation at LF.
As (8) illustrates, under the multiple Agr-projection hypothesis (as per Chomsky
1991) coupled with the VP-intemal subject hypothesis. all subjects and all direct
objects are derived (at le�st by LF).4 Rounding out the picture, Bowers (1993)
proposes a functional projection PredP which is instantiated inside the inflectional
layer (IP) and outside the thematic layer (VP) and which mediates the relation
between the subject and the VP. Unlike Larson's 1988 VP-shell analysis the
instantiation of PredP is independent of the thematic requirements of the verb
(which makes it distinct from the agentive vP proposed in Chomsky 1995, whose
presence is triggered by the thematic structure of the verb). Thus, by 1993 we find
proposals that collectively introduce functional projections corresponding to the
three GFs (subject, direct object, predicate) that Chomsky rejected as being part of
phrase structure in Aspects. Recent proposals (Franco 1993, Koizurni 1995. and
Sportiche 1996) have gone further, and provided for a separate agreement element
for indirect objects (AgrIO) which also projects a phrase.
Taken all together, these proposals lead to important questions about the place of
functional projections in Universal Grammar. One can ask whether there is a
universal stock of functional projections instantiated in every language or whether a
universal repertoire of such projections forms an inventory out of which individual
languages select a subset. Alongside this question is the issue of whether functional
projections are instantiated in some fixed and universal hierarchy or whether the
relative dominance of individual functional projections might vary parametrically
from one language to another. Indeed these questions have been the subject ofsome
discussion in the literature and the positions taken on these issues vary across a
spectrum (see Iatridou 1990, Nash & Rouveret 1997, and Tlmiinsson 1996 for
discussion). On one end of this spectrum is the more restrictive view that the set of
functional categories and their relative dominance is fixed by UG. At the other is
the suggestion that the inventory of functional projections and their relative
dominance varies from language to language. As an example of the kind of debate
that surrounds the instantiation of particular functional projections, one might
compare Koizurni 1995 (which maintains that English has an AgrIO projection, in
addition to AgrO) with Montrul 1998 (which assumes that AgrIO is specific to
languages that have dative clitics, such as French and Spanish). In this volume](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-28-320.jpg)
![GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX II
achieve greater uniformity of representation, Baker suggests reducing all
grammatical prominence (inter- and intra- clausal) to embedding prominence. He
implements this through the adoption of a lexical decomposition strategy that
requires at least one syntactic predicate for each subcategorized argument. By
assuming, for instance. that a subject is in the specifier ofa vP and that an object is
in spec,VP, the intraclausal superiority of subject over object can be treated as
arising out ofan embedding asymmetry.
While lexical decomposition can clearly be used as a vehicle for mapping out
relational prominence, Newmeyer (in chapter 3) points out that lexical semantic
representations (LSR) cannot take the place ofsyntactic grammatical functions. He
notes that there are important reasons for representing relational prominence
syntactically, and reminds us that syntactically unaccusative structures can arise out
of a varied set of lexical semantic representations (a point made early on in Rosen
1984). Since the set of possible argument structures is much smaller than the set of
possible semantic representations, and since the mapping from semantic
representations to argument structures is not fixed (as documented in Grimshaw
1979), there is still a need for an autonomous level of deep syntactic structure in
which "grammatical functions, thematic roles, and constituent structure are not
[necessarily] in alignment".
3. 1 Sub
ject attributes through the clause
There are a number of syntactic properties that are characteristic of a single
nominal in English. These properties include being the controllee in control
structures, bearing nominative case and triggering agreement on the verb. and
obligatorily undergoing raising in appropriate contexts.
(9)a.
h.
(10)a.
b.
Vanessa tried to phone Andrew.
*Andrew tried Vanessa to phone.
Amy appears to be out ofthe office.
*Appears Amy to be out ofthe office.
Such properties have come to be known as "subject" properties inasmuch as they
accrue to a single constituent referred to as the Subject in English. However,
looking over a range of languages one finds that these properties can be distributed
among distinct constituents in different languages. In order to understand the
mechanism by which these properties are concentrated in a single constituent or
divided among several. one must understand them to be reflexes of formally distinct
syntactic attributes. Reviewing the list ofproperties just presented, we can ascribe
controlleehood as arising from configurational or thematic salience, nominative case
and agreement as arising from the interaction of an argument and a functional head](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-30-320.jpg)
![GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 1 3
Thus it is the many-to-one mapping from syntactic attributes to a single argument,
which creates the illusion ofa single set of"subject" properties in English (and other
languages like it). The cross-linguistic evidence presented in this volume shows that
the notion ofsubject is epiphenomenal rather than primitive.
It is also the case that the syntactic attributes which contribute to the
epiphenomenon of subject are themselves parameterized rather than universal. For
instance, we. find that agreement heads (or features) can be obligatory or optional.
Irish clauses do not require the presence of a nominal argument, thereby allowing
f
or unaccusative predicates whose sole argument is prepositional.
(14) Bhreathaigh at: ab aimsir.
became-fine on the weather
'The weather became nice.•
(McCloskey, ch. 7)
In Niuean, on the other hand, the absolutive argument is obligatory.
Another parameterized syntactic attribute that plays a central role in the
determination of clausal . behavior is the EPP. We find, from cross-linguistic
evidence, that the presence of an EPP f
eature appears to be universal. That is, the
head ofTP (or its equivalent) is always found to bear a category feature that induces
movement into the inflectional layer ofthe clause. In this regard, though, languages
vary according to whether this f
eature is D or V. In D-prominent languages, such as
English, French, and Spanish. the EPP f
eature in T f
orces the appearance of a DP in
spec,TP. In V-prominent languages, such as Bulgarian, Irish, and Niuean, the EPP
feature induces the movement of V (or some projection of V) to T and yields (in
most instances) V-initial word order. In V-prominent languages. EPP properties
normally associated with the thematic subject arc absent, since the EPP does not
aff
ect the distribution of any nominal element. Accordingly, in these languages,
raising is optional as in (12), subject expletives are absent, and subject island effects
are frequently missing. For instance, Bulgarian allows extraction from sentential
sub
jects, as in (15) (Davies and Dubinsky, Chapter 1 0).
(15) na kakvol mislis [�e fda otide tl ] bese vaZno za nego]
to what you.think that to go was important for him
'To what do you think that to go was important f
or him?'
The choice of V- or D- prominence also interacts with another parameter that
detennines the relative superiority of T (or the EPP f
eature) and Agr/Case. This
parameter, as f
ar as we can determine, is only active in D-prominent languages.
Irish and Niuean, in addition to being V-prominent, also f
eature a T(-like) projection
higher than Agr/Case. In contrast, English is D-prominent and has Agr above T.
Spanish, while D-prominent like English, is like Irish in having T above Agr. This
difference between Spanish and English results in certain key differences between
these two languages. Unlike English, the thematic subject in Spanish need not
occupy the preverbal position, since it only need check Agr/Case features lower in
the clause. At the same time, unlike Irish, some DP element must check the 0-](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-32-320.jpg)
![1 4 WILLIAM D. DAVIES AND STANLEY DUBINSKY
feature in T higher up. This results in a obligatory "subject'.'-like position, which is
sometimes filled by a nominal locative or by the trace of a moved wh-operator. The
choice ofEPP feature coupled with the relative height of T and Agr leads to a range
oflanguage types, as expressed in the foHowing table.
D-prominent EPP V-prominent EPP
T/Agr Spanish Irish
Agr/T English --
3.2 Ob
jectproperties
Lasnik (chapter 5) proposes that, while overt movement to check features is in
general obligatory, AgrO in English is optionally present. Because of this, he
argues, direct objects in English can be underived. As an illustration of this, he
shows first that extraction from object NPs is ruled out when the object does raise to
spec,AgrOP in order to bind a reflexive pronoun, as in (16).
( 1 6) ??Which senatordid the special prosecutor question
[ two friends of_ ] during each other's trials.
When derivation is not motivated by binding requirements, the object may remain
underived, thereby f
acilitating extraction, as in (1 7).
( 1 7) Which senator did the special prosecutor question
[ two friends of_ ]
during the president's trial.
Lasnik concludes from this and other evidence that, at least for English, the AgrO
can be distinguished from AgrS by its optionality. Similarly, Travis (chapter 6)
proposes for Malagasy that the analog of the AgrO projection (AspP) is inert. She
provides evidence that a number of constructions, previously assumed to involve
derived objects, in fact involve derived subjects instead. She shows that Malagasy
fails to have the equivalent of AgrO in its clause structure. Thus, both Lasnik and
Travis agree that objects may be distinguished functionally from subjects, in at least
some languages, in that the former may or must remain in their base-generated
position within a basic transitive clause. This state ofaffairs is comparable to that of
Niuean, for which Massam (chapter 9) claims that the Ergative argument is
underived.
In their account of objects, both Lasnik and Travis take the lower Agr/Asp
projection to be intermediate between the two verbal projections (vP and VP). in
keeping with recent proposals by Koizumi (1995). According to this view of clause
structure the inflectional and thematic space cannot be divided into distinct layers,
since an inflectional projection intervenes between two lexical ones, as in (18).](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-33-320.jpg)
![, I
1 6 WILLIAM D. DAVmS AND STANLEY DUBINSKY
English is obligatory. What this means is that ALL English clauses must have the
higher Agr feature. that which has been associated with nominative Case and
agreement. Thus, regardless of transitivity, there must always be a nominative
argument in the clause. When the higher Agr feature is obligatory, the same Agr
feature associated with the thematic subject ofa transitive clause will need to also be
checked by the single argument of an intransitive clause (whether unergative or
unaccusative). This is schematized in (I9).
(19)a. Transitive clause:
[ Thematically-high-arg AgrX [ Thematically-Iow-arg (AgrY) ... JJ
b. Intransitive clause:
[ Single-argument AgrX ... ]]
This state of affairs results in a case/agreement system traditionally characterized as
nominative/accusative, in which AgrX is AgrS (or Nominative) and AgrY is AgrO
(or Accusative). The need, in English, for a nominative argument in each clause
might be seen as one side of the "coin" that motivated the Final I Law and other
similar proposals (with the D-prominence of the EPP being the other side of that
"coin").
In some ways, what Massam proposes for Niuean is the inverse of Lasnik's
proposal for English. Citing the optionality of the Ergative argument and the
obligatory nature of the Absolutive, Massam proposes an obligatory Absolutive
feature in vP. In Massam's analysis then, the obligatory Agr feature is the
configurationally lower one (in that it is checked by the lower oftwo arguments). In
this circumstance, the same Agr feature associated with the thematic object of a
transitive clause will need to also be checked by the single argument of an
intransitive clause (whether unergative or unaccusative). It is this same Agr feature
that is checked by the thematic subject of a transitive clause when the thematic
object undergoes noun incorporation with the verb. This case system is schematized
in (20).
(20)a. Transitive clause:
[ Thematically-high-ar8x [Thematically-Iow-argy AgrX AgrY ... ]]
b. Transitive NI clause:
[ Thematically-high-arg AgrY [ Thematically-low-arg . .. ]]
c. Intransitive clause:
[ Single-argument AgrY ... ]]
The result of this is, of course, a case/agreement system characterized as
ergative/absolutive, in which AgrX is Ergative and AgrY is Absolutive. The view
adopted here unifies derivation of Accusative objects in English and Absolutives in
Niuean, in that both reference the same (i.e. lower) Agr feature. The difference](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-35-320.jpg)
![,
18 WILLIAM D. DAVIES AND STANLEY DUBINSKY
issues, such that the resulting convergences dearly provide a path for future research
in this area.
For example, the EPP, however it may ultimately be characterized, is clearly
associated with subjects and not objects. It is also clear from some of the chapters in
this volume that the derivation ofsubjects and objects out of the thematic layer into
the inflectional layer ofa clause is detennined independently for each relation in the
grammar of a given language. Thus, to the extent that objects are optionally derived
in English, for example (Lasnik, chapter 5), their grammatical status is distinct from
subjects in ways that have yet to be determined. The resulting situation appears to
be as follows: (i) specific properties generally related to grammatical functions (such
as agreement) have autonomous reflexes in the grammar and are often implemented
here as distinct functional heads (e.g., Agr); (ii) at the same time, the relation
between a particular property type (such as agreement) and its associated
grammatical functions (such as subject and object) is not unifonn across all
grammatical functions. So, while (i) suggests that there is no discrete grammatical
reflex for any particular grammatical function, (ii) suggests that there is more to the
distinction between particular grammatical functions than "all and only" the
functional properties that have been taken account of thus far. This then surely
argues for further exploration into the underlying role of GFs as determinants of
grammatical structure.
The sort of conclusion that the work in this volume points to was in many
respects anticipated in Keenan's 1976b 'Towards a universal definition of
"subject"'. In that paper, Keenan attempts to arrive at a definition of the notion
subject by testing 30 some properties (some of which had been widely used as
traditional diagnostics for identifying subjects) against data from a variety of
languages. With respect to these properties he notes that no one of them is "both
necessary and sufficient for an NP in any sentence in any L [language] to be the
subject of that sentence." Keenan concludes that '''subject' does not represent a
single dimension of linguistic reality. It is rather a cluster concept, or as we shall
say, a multi-factor concept." Appealing to the metaphor of 'intelligence', which is
recognized as "a combination of abilities", he draws an analogy between it and the
properties that he attributes to the subject relation. He goes on to say, "Being a
subject is...more like being intelligent than...like being a prime number." At the
same time he suggests (hopefully) that "the concept of subject might coincide
with...groupings of [these] properties, though in the worst of cases each
[property]...would be an independent factor."
ENDNOTES
I
Despite the f
act that Chomsky uses thc tcons "logical sub
jecf' and "grammatical sub
ject" in his
discussion of grammatical functions, the structural definitions that he supplies for GFs are most directly
applicablc to deep structure configurations. As he notes (in note 32, pp. 220-221). the existence of
sentences containing PP sub
jects or two initial NPs may necessitate a distinct set of OF definitions
relevant to swface structure.
1
In the glosses, AT stands for AGENT TOPIC and 1T stands for THEME TOPIC. The other abbreviations
are: TOPOC), ACC(USATIVE), OBL(lQUE), and GEN{lTIVE).](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-37-320.jpg)
![,
22 MARK C. BAKER
1. THE ORIGINS OF THE CHOMSKIAN VIEW
To put these issues into context, let us begin by reviewing how the claim that
grammatical relations are expressed in phrase structure arose. In f
act, it emerged
rather naturally out of the sequence of topics that were investigated and came to be
partially understood by generative grammar. However, my f
ocus will be on the
idealized conceptual order of the ideas, rather than trying to trace their actual
historical order in detail. This conceptual order also reflects the way that
Chomskian syntax is often presented to students in textbooks and syntax courses,
and this may be of more current significance than the history itself.
The first step is motivating that there is such a thing as phrase structure f
or
languages like English, showing how this phrase structure can be established using
traditional constituency tests. For example, one of the less obvious f
eatures of
English syntax is that there is a verb phrase that contains the verb and the direct
object, together with some particles and PPs, but not the subject. (It is my inf
onnal
experience that when introductory linguistics students are asked out of the blue
which is more closely related to the verb, the subject or the object, they have no
clear intuitions: 50% chose one and 50% the other prior to seeing any arguments.)
However, one can show that the verb and the NP that f
ollows it f
onn a unit with a
familiar battery oftests, including movement (la), replacement with a pronoun (l b),
deletion (lc), and coordination ( I d).
(1)a. Sue said she would eat the octopus, and [vp eat the octopus] she
did - .
b. Sue will [vp eat the octopus], and John will [vp do so] too.
c. Sue will [vp eat the octopus], and John will [vp - ] too.
d. Chris will [vp chop the vegetables] and [vp eat the octopus].
Once this notion of phrase structure is firmly in place, it is a rather trivial
observation that sub
jects and ob
jects always come in a fixed position in the phrase
structure. In particular, the direct object is the only noun phrase that is contained in
the VP but not in any smaller phrase. The subject is consistently contained in the
clause but not inside the VP. Indirect objects in a sentence like Chris gave the book
to Pat are uniquely contained in a prepositional phrase headed by to. And so on.2
Since there is such a close relationship between the traditional grammatical
functions and phrase structure positions, it seems ulUlccessary to ref
er to both in
one's syntactic theory. Now, some representation of constituency is needed in the
grammar anyway, to characterize which sets ofwords can be moved as a unit, or can
be deleted, or can be replaced with a pronoun. Theref
ore, Chomsky (1965)
eliminated subject and object with the f
ollowing definitions in terms of phrase
structure:](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-41-320.jpg)
![PHRASE STRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 23
(2)a. Subject is "NP immediately dominated by S" ([NP, S]).
b. Object is ''NP immediately dominated by VP" ([NP, VPD.
c. Indirect ob
ject is "NP immediately dominated by PP headed by to"
([NP, PPD·
Logically speaking, the alternative reduction would have been to eliminate
phrase structure in f
avor ofgrammatical relations. However, that makes little sense.
For example, one wants to say that asflat as a pancake is a phrase in (3), even
though it has little to do with the traditional grammatical relations.
(3) Chris pounded the clay [as flat as a pancake].
Thus, phrase structure is more general and easier to motivate independently than
grammatical relations. So if one can only keep one, it should be phrase structure.
Basically, Chomsky's suggestion was an Occam's razor-style argument par
exceIIance.
Next, investigation into other f
ancier phenomena reveals that there are many
ways in which SUbjects, objects, and other phrases behave diff
erently from one
another. At the level of basic description, the operative generalizations can be stated
in terms of grammatical relations. The f
ollowing are f
amiliar examples from the
history ofthe field:
(4)a Only subjects can be controlled in nonfinite clauses.
b. The antecedent ofa reflexive must be a subject.
c. The direct object becomes the subject ofa passive verb.
d. Only the direct object can incorporate into the verb, etc., etc.
But using the equivalencies in (2), these statements can be rephrased in terms of
phrase structure, as follows.
.
(5)a. Only the highest NP in a nonfinite clause can be controlled.
b. The antecedent of a reflexive must be the NP immediately dominated
by S.
c. The NP in VP moves to the Specifier of IP ifthe verb is passive.
d. Only the nominal that is the sister ofthe verb can incorporate into it,
etc., etc.
,
: . = ",](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-42-320.jpg)
![" 1
28 MARK. C. BAKER
crosslinguistically without conceding that phrase structure itself is not present
crosslinguistically. Now suppose that English-like phrase structure exists in a
language like Malayalam even though it cannot be detected by English-like
movements and deletions. Then the grammatical relations ofsubject and object can
still be defined in terms ofthat phrase structure, and passive, control, and reflexive
binding in Malayalam can be analyzed in terms ofnotions like c-command.
It is instructive to see how these concerns about constituency tests arose internal
to Chomskian syntax itself, before considering further how they affect the
comparison with RG and its allies. For various reasons, Chomskian" theory in the
1980s and 1990s was led to enrich its stock of phrases beyond what was accepted
previously. For example. it was argued that bare adjectival predicates form a "small
clause" constituent with their NP subjects, at least in some contexts (Chomsky 1981.
Stowell 1983). Later, Abney (1987) argued the determiner in English heads its own
phrase, DP, distinct from the NP that is headed by the noun itself. As a third
example, Larson (1988) argued that the two complements of a triadic verb in
English constitute a phrase oftheir own, specifically a VP headed by a trace ofthe
overt verb.
(8)a. Mary considers [John overly proud ofhimself7*herseIf].
b. Mary bought lop that fNP picture ofJohn n.
c. Mary gave [each worker tv his paycheck].
Small Clauses
Functional heads
Larsonian shells
But if one accepts these kinds of phrase structures, then one is clearly forced to
acknowledge that not all phrases can be revealed by traditional phrase structure tests
in any simple way. There is sporadic constituency evidence in favor of these
innovations, but many ofthe traditional phrase structure tests do not apply to these
new phrases. For example. many phrase types in English can be clefted, including
DPs, PPs, APs, and CPs. However, one cannot cleft a small clause, the NP
complement ofa determiner, or a Larsonian shell:
(9)a. *It's [John overly proud ofhimselt] that Mary considers -- .
b. "'It's [picture ofJohn] that Mary Dought that --.
c. "'It's [each worker tv his paycheck] that Mary gave -.
Small Clauses
Functional heads
Larsonian shells](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-47-320.jpg)
![PHRASE SlRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 29
If these phrase structure proposals are at all on the right track, then there are
many phrases that are not readily detected by movement, even in highly
configurational English. The same holds true for pronominalization and ellipsis:
these phrases are not readily omitted or replaced by a pronoun.
( lO)a.
b.
*Mary considers [him smart]. and Bill considers (it) too.
*Mary thinks that [John is nice]. and Bill wonders if(it).
*Mary answered the third question correctly,
and Bill answered the (it) too.
Small Clauses
Functional heads
c. *Mary threw [John tv his jacket] and Bill handed (it).
Larsonian shells
Why does one get these patterns of facts, with some phrases being cleftable (or
elidable, or pronominalizable) and others not? In many cases, this is still not well
understood, but it is assumed that there should be an explanation internal to the
workings of(say) clefiing. In some cases reasonable proposals have been put on the
table. For example. (9a) and (9c) might be out because there is no source of Case
checking for the NP at the beginning of the clefied constituent within that
constituent; this could lead to a Case theory violation. On the other hand, (9b) might
be out because the functional head D or C is not strong enough to license a trace as
its complement, according to the Empty Category Principle or its descendents in
more recent theory. Similarly, Loebeck (1995) among others proposes substantive
syntactic conditions on which phrases can be elided and which cannot. In general.
these gaps in the distribution ofclefting and ellipsis are interesting syntactic puzzles,
but they do not look especially mysterious or problematic f
or the overall framework.
Rather. they should be amenable to the usual kind ofsyntactic inquiry.
These kinds of issues would need to be f
aced by Chomskian theory anyway in
the course of pursuing its own agenda, even if there were no language other than
configurational English. From this perspective, when the RG critique points out that
second position clitics in Warlpiri (Simpson 1991) and clefiing in Malayalam
(Mohanan 1982) do not reveal a VP constituent, there is no particular surprise. One
does not automatically assume that Warlpiri and Malayalam do not have a VP, but
only that the structure of phrases in those languages somehow interacts with the
nature ofcliticization or clefting to have this eff
ect. This could have as much to do
with the nature ofclefting in these languages as it does with phrase structure proper.
Now, if we no longer trust traditional constituency tests to reveal to us the basic
phrase structure ofa construction in a language in any straightf
orward way, how can
that phrase structure be determined? There could be many ways, some easier to
apply in one language and others easier to apply in another language. But if one is
committed to the idea that phenomena like passive. control, and reflexives are
structurally determined along the lines summarized in (5), then it is legitimate to](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-48-320.jpg)
![30 MARK C. BAKER
infer from these phenomenathe phrase structures that are needed to make them work
properly. If c1efting and cliticization and word order do not work smoothly from
these structures, those are problems to be solved, but not necessarily reasons to hold
back. In short, Chomskian syntax plunges ahead and assumes that the structures it
needs to work properly are present.
Many particular examples of this characteristic approach could be cited. An
early example is Chomsky's (1981) discussion ofJapanese, in which he uses [NP, S]
terminology that implies a certain kind ofphrase structure for Japanese grammatical
functions, even though these do not correspond to a directly observable "D
structure" or "S-structure." Hale's (1 983) approach to Warlpiri is similar: he argues
that a configurational "L-structure" exists for Warlpiri alongside the
nonconfigurational "C-structure" that determines word order and clitic placement.
This is similar to the "two independent levels" view of LFG, but with the significant
diff
erence that f
or Hale the representation of grammatical functions (his L-structure)
has the f
ormal properties of a phrase marker, over which predicates like c-command
are defined.
Similarly. when I studied noun incorporation in languages like Mohawk in Baker
( l988), I simply assumed that direct objects were the only NPs that were structural
sisters of the verb, and used that as the basis f
or a structural explanation of why only
direct objects can be incorporated. I then took the overall success ofthis account to
be evidence that languages like Mohawk do in f
act have phrase structures
comparable to English, not worrying about how this fit with other issues about word
order and constituency in the language (until later: see Baker (1991, 1996c» .
Another landmark in this changing attitude toward phrase structure is Larson's
(1988) study of double object constructions in English. In these constructions,
phrase structure does not seem to reveal any difference between the positions ofthe
Prime Minister and a letter in a sentence like I sent the Prime Minister a letter.
Both seem to be objects in the sense of being NPs immediately dominated by VP.
However, Barss and Lasnik (1986) showed that there are differences between the
two for most binding phenomena: the first object can bind the second but not vice
versa. Barss and Lasnik look at this apparent conflict between phrase structure and
binding theory and ask whether pure c-command is really the right condition on
binding. However, Larson takes a markedly different approach: he takes it as
established that c-command is the right condition on binding, and uses the binding
facts to infer a novel phrase structure for English double object constructions. Thus,
for him, GR-related phenomena are (almost) enough to establish phrase structure. In
fact, Larson does make an eff
ort to show that there is converging evidence for the
structures he proposes from traditional phrase structure tests, such as conjunction.
But his independent evidence is slender and debatable (see Jackendoff (1 990) vs.
Larson (I 990», and it is clear that this is not what is driving his account. Nor did
anyone ever arrive at these phrase structures purely by traditional constituency tests.
Conceptually rather similar is the work on scrambling and word order in
languages like Japanese and ' Hindi inspired by Saito (1 985) and Hoji (I 985), in
which configurational structure is inferred indirectly and freedom of word order is
derived from movement processes that fit within the internal logic ofthe system.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-49-320.jpg)
![ces services. Ils en ont profité dans cette circonstance. Ils ont rendu
l'or et reçu du blé; et la souffrance, au lieu de se manifester sous
forme d'inanition, s'est manifestée sous forme d'appauvrissement, ce
qui est moins dur. Mais cet appauvrissement, ce n'est pas
l'exportation du numéraire qui en est cause, c'est la perte des
récoltes.
C'est absolument comme le chapelier dont nous parlions tout à
l'heure. Il vendait beaucoup de chapeaux, et, se soumettant à des
privations, il réussit à accumuler de l'or. Sa maison brûla. Il fut bien
obligé de se défaire de son or pour la reconstruire. Il en resta plus
pauvre. Fut-ce parce qu'il s'était défait de son or? Non, mais parce
que sa maison avait brûlé.—Un fléau est un fléau. Il ne le serait pas
si l'on était aussi riche après qu'avant.
«Le blé consommé, l'argent exporté, que reste-t-il à l'acheteur?»
demande le National.—Il lui reste de n'être pas mort de faim, ce qui
est quelque chose.
Nous demanderons à notre tour: Si l'Angleterre n'eût consommé
ce blé et exporté cet argent, que lui resterait-il? des cadavres[26].
22.—DU COMMUNISME.
27 Juin 1847.
Les préjugés économiques ne sont peut-être pas le plus grand
obstacle que rencontrera la liberté commerciale. Entre hommes qui
diffèrent d'opinion sur un point, à la vérité fort important,
d'économie politique, la discussion est possible, et la vérité finit
toujours par jaillir de la discussion.
Mais il est des systèmes si complétement étrangers à toutes les
notions reçues, qu'entre eux et la science il ne se trouve pas un
terrain commun qui puisse servir de point de départ au débat.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-62-320.jpg)
![dans les classes que le travail manuel détourne de la méditation;
nous voulons parler du communisme[27].
On voit des hommes qui ont du superflu, d'autres qui n'ont pas le
nécessaire, et l'on dit: «Si l'on mettait toutes ces richesses en
commun, tout le monde serait heureux.» Quoi de plus simple et de
plus séduisant, surtout pour ceux qu'affligent des privations réelles;
et c'est le grand nombre?
Ce n'est pas notre intention de réfuter ici ce système, de montrer
qu'il paralyserait complétement dans l'homme le mobile qui le
détermine au travail, et tarirait ainsi pour tous la source du bien-être
et du progrès; mais nous croyons devoir prendre acte de la
réfutation décisive qui en a été faite, dans le dernier numéro de
l'Atelier, par des hommes qui appartiennent aux classes ouvrières.
C'est certainement un symptôme consolant de voir des systèmes
subversifs repoussés et anéantis, avec une grande force de logique,
par des hommes que le sort a placés dans une position telle qu'ils
seraient plus excusables que d'autres s'ils s'en laissaient séduire.
Cela prouve non-seulement leur sincérité, mais encore que
l'intelligence, quand on l'exerce, ne perd jamais le noble privilége de
tendre vers la vérité. Pour beaucoup de gens, le communisme n'est
pas seulement une doctrine, c'est encore et surtout un moyen
d'irriter et de remuer les classes souffrantes. En lisant l'article auquel
nous faisons allusion, nous ne pouvions nous empêcher de nous
rappeler avoir entendu un fougueux démocrate, appartenant à ce
qu'on nomme la classe élevée, dire: «Je ne crois pas au
communisme, mais je le prêche parce que c'est le levier qui
soulèvera les masses.» Quel contraste!
Une chose nous surprend de la part des rédacteurs de l'Atelier,
c'est de les voir s'éloigner de plus en plus de la doctrine de la liberté
en matière d'échanges.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-64-320.jpg)
![cette police que la société exerce sur elle-même. Là où les citoyens
comptent trop sur les autorités, ils finissent par ne pas assez
compter sur eux-mêmes, et la cause la plus efficace du progrès en
est certainement neutralisée[28].
Si ces idées se rapprochent de celles que l'Atelier a développées
dans l'article que nous avons en vue, nous ne devons pas être peu
surpris du ton d'irritation avec lequel il persiste à s'exprimer sur la
liberté du commerce et ce qu'il nomme l'école économique anglaise.
L'Atelier est plein de douceur pour les communistes, qu'il vient de
combattre et même de terrasser, mais il conserve envers nous les
allures les plus hostiles. C'est une inconséquence que nous ne nous
chargeons pas d'expliquer, car il est évidemment beaucoup plus loin
du communisme que de la liberté du travail et de l'échange. L'Atelier
croit la protection plus nécessaire que la liberté à la prospérité
nationale. Nous croyons le contraire, et il conviendra du moins que
les doctrines sur la propriété et la liberté, qu'il a opposées aux
communistes, mettent la présomption de notre côté. Si la propriété
est un droit, si la liberté d'en disposer en est la conséquence, la
tâche de prouver la supériorité des restrictions, l'onus probandi,
incombe exclusivement à celui qui les réclame.
Nous n'abandonnerons pas le sujet du communisme sans adresser
quelques réflexions aux classes qui tiennent de notre constitution le
pouvoir législatif, c'est-à-dire aux classes riches.
Le communisme, il ne faut pas se le dissimuler, c'est la guerre de
ceux qui ne possèdent pas, ou le grand nombre, contre ceux qui
possèdent ou le petit nombre. Partant, les idées communistes sont
toujours un danger social pour tout le monde, et surtout pour les
classes aisées.
Or ces classes ne jettent-elles pas de nouveaux aliments à la
flamme communiste quand elles font en leur propre faveur des lois
partiales? Quoi de plus propre que de telles lois à semer l'irritation](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-68-320.jpg)
![maintenues? Ne l'a-t-on pas vue prendre une telle résolution, sans
admettre même le débat, comme si elle avait eu peur de la lumière,
là où elle ne pouvait éclairer qu'un acte d'injuste égoïsme?
La bourgeoisie persévère dans cette voie, parce qu'elle voit le
peuple, impatient de beaucoup d'injustices chimériques, méconnaître
la véritable injustice qui lui est faite. Pour le moment, les journaux
démocratiques, abandonnant la cause sacrée de la liberté, sont
parvenus à égarer ses sympathies et à les concilier à des restrictions
dont il n'est victime qu'à son insu. Mais la vérité ne perd pas ses
droits; l'erreur est de nature essentiellement éphémère; et le jour où
le peuple ouvrira les yeux n'est peut-être pas éloigné. Pour le repos
de notre pays, puisse-t-il n'apercevoir alors qu'une législation
équitable[29]!
23.—RÉPONSE AU JOURNAL L'ATELIER.
12 Septembre 1847.
(Écrite en voyage et adressée à l'éditeur du Journal des
Économistes.)
Si j'ai eu quelquefois la prétention de faire de la bonne économie
politique pour les autres, je dois au moins renoncer à faire de la
bonne économie privée pour moi-même. Comment est-il arrivé que,
voulant aller de Paris à Lyon, je me trouve dans un cabaret par delà
les Vosges? Cela pourra vous surprendre, mais ne me surprend pas,
moi qui ne vais jamais de la rue Choiseul au Palais-Royal sans me
tromper.
Enfin me voici arrêté pour quelques heures, et je vais en profiter
pour répondre au violent article que l'Atelier a dirigé contre le Libre-
Échange dans son dernier numéro. Si j'y réponds, ce n'est pas parce
qu'il est violent, mais parce que cette polémique peut donner lieu à
quelques remarques utiles et surtout opportunes.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-70-320.jpg)
![Nous le disons hautement: ce que nous réclamons pour toutes les
classes, dans l'intérêt de toutes les classes, c'est la justice,
l'impartialité de la loi; en un mot, la propriété et la liberté. À cette
condition, nous ne voyons pas des classes, mais une nation. Malgré
la mode du jour, notre esprit se refuse à admettre que toutes les
vertus, toutes les perfections, toutes les pensées généreuses, tous
les nobles dévouements résident parmi les pauvres, et qu'il n'y ait
parmi les riches que vices, intentions perverses et instincts égoïstes.
S'il en était ainsi, si le bien-être, le loisir, la culture de l'esprit
pervertissaient nécessairement notre nature, il en faudrait conclure
que l'éternel effort de l'humanité, pour vaincre la misère par le
travail, est la manifestation d'un mobile à la fois dépravé et
indestructible. Il faudrait condamner à jamais le dessein de Dieu sur
sa créature de prédilection[30].
Il ne me reste pas d'espace pour réfuter la troisième accusation
formulée contre l'économie politique, celle fondée sur cette
assertion, qu'elle est l'expression du côté inférieur de l'homme. C'est,
du reste, un vaste sujet sur lequel j'aurai l'occasion de revenir.
Parce que l'économie politique circonscrit le champ de ses
investigations, on suppose qu'elle dédaigne tout ce qu'elle ne fait
pas rentrer dans sa sphère. Mais, sur ce fondement, quelle science
ne devrait-on pas condamner? L'économie politique, il est vrai,
n'embrasse pas l'homme tout entier; elle laisse leur part de cet
inépuisable sujet à l'anatomie, à la physiologie, à la métaphysique, à
la politique, à la morale, à la religion. Elle considère surtout l'action
des hommes sur les choses, des choses sur les hommes, et des
hommes entre eux, en tant qu'elle concerne leurs moyens d'exister
et de se développer. Exister, se développer, cela peut paraître aux
rédacteurs de l'Atelier chose secondaire et inférieure, même en y
comprenant, comme on doit le faire, le développement intellectuel et
moral aussi bien que le développement matériel. Pour nous, après ce
qui se rapporte aux intérêts d'une autre vie, nous ne savons rien de
plus important; et ce qui prouve que nous n'avons pas tout à fait
tort, c'est que tous les hommes, sans exception, ne s'occupent guère](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-76-320.jpg)
![d'autre chose. Après tout, il ne peut jamais y avoir contradiction
entre ce que les sciences diverses renferment de vérité. Si
l'économiste et le moraliste ne sont pas toujours d'accord, c'est que
l'un ou l'autre se trompe indubitablement. On peut réfuter tel
économiste, comme tel moraliste, comme tel anatomiste; mais la
guerre déclarée à l'économie politique me paraît aussi insensée que
celle que l'on ferait à l'anatomie ou à la morale[31].
24.—RÉPONSE À DIVERS.
1er
Janvier 1848.
Un journal émané de la classe laborieuse, la Ruche populaire, fait
remonter au travail l'origine de la propriété. On est propriétaire de
son œuvre. Nous pensons absolument comme ce journal.
En même temps, il attaque la liberté d'échanger. Nous l'adjurons
de dire, la main sur la conscience, s'il ne se sent pas en contradiction
avec lui-même. Est-ce être propriétaire de son œuvre que de ne la
pouvoir échanger sans blesser l'honnêteté et en payant l'impôt à
l'État? Suis-je propriétaire de mon vin, si je ne le puis céder à un
Belge contre du drap, parce qu'il déplaît à M. Grandin que j'use du
drap belge?
Il est vrai que la Ruche populaire ne donne pas d'autre raison de
son opposition au libre-échange, si ce n'est qu'il se produit dans
notre pays à l'encontre des journaux indépendants. En cela, fait-elle
preuve elle-même d'indépendance? L'indépendance, selon nous,
consiste à penser pour soi-même, et à oser défendre la liberté,
même à l'encontre des journaux dits indépendants.
La même considération paraît avoir décidé une feuille de Lyon et
une autre de Bayonne à se mettre du côté du privilége. «Comment
ne serions-nous pas pour le privilége, disent-elles, quand nous le
voyons attaquer par les journaux ministériels?» Donc, si le ministère](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-77-320.jpg)
![uns de ses membres, il faut que tous y contribuent; plus
économique, parce qu'il épargnerait beaucoup de frais de perception
et ferait disparaître beaucoup d'entraves; plus loyal, enfin, parce que
le public verrait clair dans l'opération et saurait ce qu'on lui fait
faire.»
Vous voyez, monsieur, que je n'ai pas attendu la lettre dont vous
avez bien voulu m'honorer pour reconnaître tous les mérites de la
protection directe.
Oui, comme vous, et par d'autres motifs, il me tarde qu'on nous
prenne notre argent sous une forme qui nous permette de voir où il
passe. Il me tarde que chacun de nous puisse lire sur son bulletin de
contribution à combien se monte la redevance que nous imposent
MM. tels ou tels[32].
Veuillez recevoir, monsieur, l'expression de mes sentiments de
considération et d'estime.
Frédéric Bastiat.](https://image.slidesharecdn.com/920029-250617113817-fe751e32/85/Objects-And-Other-Subjects-Grammatical-Functions-Functional-Categories-And-Configurationality-William-D-Davies-86-320.jpg)
Objects And Other Subjects Grammatical Functions Functional Categories And Configurationality William D Davies
- 1. Objects And Other Subjects Grammatical Functions Functional Categories And Configurationality William D Davies download https://ebookbell.com/product/objects-and-other-subjects- grammatical-functions-functional-categories-and- configurationality-william-d-davies-1840058 Explore and download more ebooks at ebookbell.com
- 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Objects And Other Subjects Grammatical Functions Functional Categories And Configurationality 1st Edition William D Davies https://ebookbell.com/product/objects-and-other-subjects-grammatical- functions-functional-categories-and-configurationality-1st-edition- william-d-davies-11778458 Realisms Interlinked Objects Subjects And Other Subjects Arindam Chakrabarti https://ebookbell.com/product/realisms-interlinked-objects-subjects- and-other-subjects-arindam-chakrabarti-50218432 A Collection Of Endangered Objects And Other Stuff Stack Steve https://ebookbell.com/product/a-collection-of-endangered-objects-and- other-stuff-stack-steve-8694656 21st Century Dodos A Collection Of Endangered Objects And Other Stuff Steve Stack https://ebookbell.com/product/21st-century-dodos-a-collection-of- endangered-objects-and-other-stuff-steve-stack-50468032
- 3. Data Structures And Other Objects Using C 4th Edition Michael G Main Walter J Savitch https://ebookbell.com/product/data-structures-and-other-objects- using-c-4th-edition-michael-g-main-walter-j-savitch-33556400 Data Structures And Other Objects Using Java 4th Edition Main https://ebookbell.com/product/data-structures-and-other-objects-using- java-4th-edition-main-55557912 Tibetan Silver Gold And Bronze Objects And The Aesthetics Of Animals In The Era Before Empire Crosscultural Reverberations On The Tibetan Plateau And Soundings From Other Parts Of Eurasia John Vincent Bellezza https://ebookbell.com/product/tibetan-silver-gold-and-bronze-objects- and-the-aesthetics-of-animals-in-the-era-before-empire-crosscultural- reverberations-on-the-tibetan-plateau-and-soundings-from-other-parts- of-eurasia-john-vincent-bellezza-49992276 Exploring Comets Asteroids And Other Objects In Space 1st Edition Nancy Dickmann https://ebookbell.com/product/exploring-comets-asteroids-and-other- objects-in-space-1st-edition-nancy-dickmann-51640130 Sons And Other Flammable Objects Porochista Khakpour https://ebookbell.com/product/sons-and-other-flammable-objects- porochista-khakpour-57633056
- 5. Studies in Natural.Language and Linguistic Theory VOLUME 52 Managing Editors Liliane Haegeman, University ofGeneva Joan Maling, Brandeis University James McCloskey, University of California, Santa Cruz Editorial Board Carol Georgopoulos, University of Utah Guglielmo Cinque, University ofVenice Jane Grimshaw, Rutgers University Michael Kenstowicz, Massachusetts Institute ofTechnology Hilda Koopman, University ofCalifornia, Los Angeles Howard Lasnik, University ofConnecticut at Storrs Alec Marantz, Massachusetts Institute ofTechnology John J. McCarthy, University ofMassachusetts, Amherst Ian Roberts, University ofStuttgart The titlespublished in this series are listedat the end ofthis volume.
- 6. OBJECTS AND OTHER SUBJECT Grammatical Functions, Functional Categories and Configurationality edited by WILLIAM D. DAVIES University ofIowa, Iowa City and STANLEY DUBINSKY University ofSouth Carolina. Columbia KLUWER ACADEMIC PUBLISfJERS DORDRECHT I BOSTON I LONDON
- 7. I �ol . 0 � , I A C.I.P. Catalogue record for this book is available from the Library ofCongress. ISBN 1-4020-0064-2 Published by Kluwer Academic Publishers, P.O. Box17, 3300AA Dordrecht,The Netherlands. Sold and distributed in North, Central and South America by KluwerAcademic Publishers. 101Philip Drive, Norwell. MA 02061, U.S.A. In all other countries. sold and distributed by Kluwer Academic Publishers, P.O. Box 32 2, 3300AH Dordrecht, The Netherlands. Printed on acid-free paper All Rights Reserved © 2001 Kluwer Academic Publishers No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical; including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner. Printed in the Netherlands.
- 8. DEDICATION For David Perlmutter and Paul Postal While they might not agree with many ofthe answers in this volume, were it not fQr their groundbreaking work, many ofthe questions might never have been asked. .
- 9. · I
- 10. TABLE OF CONTENTS LIST OF CONTRIBUTORS .. ......... .. .. . .. .... ... . . .ix FORWaRD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi ACKNOWLEDGEMENTS ............................ xv PART I: OVERVIEW 1. WILLIAM D. DAVIES AND STANLEY DUBINSKY / Remarks on Grammatical Functions in Transfonnational Syntax . . . . . . . . . I 2. MARK C. BAKER / Phrase Structure as a Representation of"Primitive" Grammatical Relations . . . . .. . . . . . . . . . . . . . . . . . 21 3. FREDERICK J. NEWMEYER / Grammatical Functions, Thematic Roles. and Phrase Structure: Their Underlying Disunity . . . . . .53 4.ALEX ALSINA / Is Case Another Name for Grammatical Function? Evidence from Obj�crAsymmetries . . . . . . . . . . . . . . . . 77 PART II: OBJECTS 5. HOWARD LASNIK / Subjects. Objects, and the EPP . . . . . . " 103 6. LISA deMENA TRAVIS / Derived Objects in Malagasy . . . .. . . 123 PART III; SUBJECTS 7. JAMES McCLOSKEY / The Distribution of Subject Properties in Irish . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. 157 8. GRANT GOODALL / The EPP in Spanish . . . . . . . . . . .. . . 193 9. DIANE MASSAM / On Predication and the Status of Subjects in Niuean . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . .225 10. WILLIAM D. DAVIES AND STANLEY DUBINSKY / Functional Architecture and the Distribution of Subject Properties . . . . . . 247 REFERENCES . . . . . . . . . . . . . . . . .. . . . . . . . . . .. . . . ..281 AUTHOR INDEX . . . . . . . .. . . . . . . . . . .. . . . . . . . . .. ...297 SUBJECT INDEX . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . 30 I
- 12. LIST OF CONTRIBUTORS Alex Alsina Department de Traducci6 i Filologia Universitat Pompeu Fabra La Rambla, 30-32 08002 Barcelona. SPAIN alex.alsina@trad.upf.es Mark Baker Linguistics Department Rutgers University 18 Seminary Place New Brunswick NJ 08901. USA mabaker@ruccs.rutgers.edu William D. Davies Linguistics Department University of Iowa Iowa City. IA 52242, USA william-davies@uiowa.edu Stanley Dubinsky Linguistics Program University ofSouth Carolina Columbia, SC 29208, USA dubinsky@sc.edu Grant Goodall Department ofLanguages and Linguistics The University ofTexas at EI Paso 500 W. University Avenue EI Paso. TX 79968, USA ggoodal1@utep.edu Howard Lasnik Department of Linguistics University of Connecticut 341 Mansfield Road Storrs. CT 06269, USA lasnik@sp.uconn.edu Diane Massam Department ofLinguistics University ofToronto 130 S1. George Street Toronto, ON, M5S 3Hl, CANADA dmassam@chass.utoronto.ca Jim �cCloskey Department ofLinguistics University of California, Santa Cruz Santa Cruz, CA 95064, USA mcclosk@ling.ucsc.edu Frederick J. Newmeyer Department ofLinguistics University ofWashington Seattle, WA 98195, USA fjn@u.washington.edu Lisa deMena Travis Department ofLinguistics McGill University 1085 Dr. Penfield Avenue Montreal. QC H3A lA7, CANADA ltravi@po-box.mcgill.ca
- 14. FORWORD The initial idea for this volume began with discussions with Brian Joseph and Geoffrey Huck about a book that would explore the place of grammatical relations/functions in Principles and Parameters Theory and Minimalism. It was clear to both of them that (i) grammatical functions played a much more essential role in these theories than was generally acknowledged, and (ii) those new to the discipline very often had no appreciation for the origins and significance of such notions. We briefly considered the possibility of writing a monograph surveying the recent developments in this area of syntactic theory, but we realized that there was a rich body of relevant work being done on this topic. It seemed that many working in P&P and Minimalism were unaware of the very significant work being pursued in this area, and of its relationship to earlier work and work in other paradigms. Thus was born the idea for a workshop that would bring together leading scholars in syntax to explore the place of grammatical functions (GFs) in . theories of syntax. This idea evolved into a three day workshop (supported by NSF grant and by the USC College of Liberal Arts), held in July 1999 at the Linguistic Institute in Urbana-Champaign (UI-UC). The workshop presentations and commentary from: Alex Alsina (Universitat Pompeu Fabra, Barcelona), Mark Baker (Rutgers University), Richard Campbell (Oakland U / Microsoft Research), William Davies (University of Iowa), Stanley Dubinsky (University of South Carolhia), Samuel Epstein (University of Michigan), Patrick Farrell (UC, Davis), Grant Goodall (University of Texas, EI Paso), Howard Lasnik (Connecticut University), James McCloskey (University of California, Santa Cruz), Frederick Newmeyer (University of Washington), Norvin Richards (MIT), Daniel Seely (Eastern Michigan University), and Lisa Travis (McGill University). The workshop took place 25 years after Perlmutter and Postal gave their lectures on Relational Grammar at the LSA Summer Institute. Their course on RG remains one of the most widely attended Institute courses on record and, perhaps, given the 15-20 years of vigorous research activity that it spawned, one of the more influential Institute courses ever given. Taking as its starting point the view that grammatical functions are primitive notions of the theory, work done in RG was a startling generative departure from the Standard Theory view, and attempted to explain a wide variety of clause-level phenomena in a wide variety of languages. In the time since that course, the role of grammatical functions has undergone some shifts in transformational syntax. These shifts have been in part in reaction to some of the discoveries in relationally-based theories and in part due to theory-internal developments. The 25th anniversary of the Amherst lectures seemed like an appropriate time to examine the current status of grammatical functions in transformational syntax. Some 100 institute participants attended the public presentations and discussion of the 1999 workshop, and these included scholars representing an exceptionally broad range of theoretical perspectives. One of the NSF reviewers suggested that Davies and Dubinsky might learn a lot about current theoretical perspectives on grammatical functions, and indeed we did (as most likely did many others in attendance). Some of the questions that were addressed in the workshop
- 15. , xii presentations included: • What are the number and status of subject positions? • How many of these positions might be active in any one language? • Can any properties of these positions be considered universal? • What is the status and content of the EPP? Is the EPP active in all languages? Is the EPP manifested in the same way in all languages in which it is active? Should the EPP be abandoned altogether? • What is the status of objects? Are subjects and objects to be distinguished or are they more similar than has been suspected? Do all languages make the use of the full range of objects ano object positions? • In what ways are the abstract configurational representations of the transformational theory representations of grammatical functions? • All in all what is the status of grammatical functions in current transformational theory and what are the acceptable levels of variation from language to language that allow these notions to be viable and interesting before they should be abandoned simply as epiphenomenal mnemonic terms? During and following the workshop, it became clear that a number of the presentations had converged in very interesting ways, and merited being brought together in print. After receiving encouragement for this venture from several quarters and enlisting the enthusiastic participation of several key contributors, we set about the task of pulling this volume together. We are proud of the product of the efforts of the workshop participants and contributors. a volume that we hope will be of value both in assessing recent progress in the field. and in framing important questions for future research. As this book neared completion. we found that it likely has a wider audience than we had originally planned. While we had originally conceived of this volume as a collection of papers on a special topic. it turns out that the book has potential as a reader for an advanced course in syntax (or a syntax seminar). In fact. both reviewers made just this point, suggesting that its value as a class text would be enhanced with the inclusion of additional background infonnation about RG. LFG, and HPSG, setting the stage for the papers included here and providing some theory comparison. However, rather than changing the focus of the book and delaying production significantly, we have opted to describe how the book might be used in a course, based on a successful course in which we used the papers in this volume. The course, as we taught it, involved a comparison of RG and P&P theory, however it would have been quite plausible to frame the course using LFG or HPSG as a theoretical counterpoint to the P&P model. The syllabus of the course was structured as follows: (1) A review of VP-shells and to Agr projections. This unit focused on chapters 9 and 10 of Radford's (1997) text. a. In covering the material on VP-shells (chapter 9), reference was made to Sportiche's (1988) article on the VP-intemal subject hypothesis, to Larson's (1988) article on VP-shells, and to the ensuing debate (in lackendoff 1990 and Larson 1990).
- 16. II xiii b. In covering the material on Agr Projections (chapter 10), we also examined Pollack's (1989) article in Linguistic Inquiry and Lasnik and Saito's (1991) CLSpaper. (2) A survey ofRelational Grammar theory. This part of the course used Blake's (1990) handbook as a guide, and included chapters from the Studies in Relational Grammar volumes and other key articles. a. In introducing the theory, we examined Perlmutter and Postal's (1983b) chapter on proposed laws of basic clause structure, as well as Keenan and Comrie's (1977) article on the noun phrase accessibility hierarchy. b. We next looked at "clause internal revaluations" and read Permutter (1984a) on the inadequacy of monostratal theories of passive, along with Bell (1983) on advancements and ascensions in Cebuano, Chung (1976) on an object creating rule in Bahasa Indonesia, Dubinsky (1990) on Japanese direct to indirect object demotion, and Perlmutter (1984b) on inversion in Italian, Japanese, and Quechua. c. In covering the topic of reflexives and impersonals, we read Perlmutter Postal's (1984) chapter on impersonal passives and Rosen's (1984) on the interface between semantic roles and initial grammatical relations. d. The unit on multnode networks and clause union covered Aissen Perlmutter's (1976) article on clause reduction in Spanish, Perlmutter Postal's (1983a) chapter on the Relational Succession Law, Gibson and Raposo's (1986) article on clause union, Rosen's (1983) paper on universals of union, Davies and Rosen's (1988) article on multi-predicate union, and Dubinsky's (1997) application ofmulti-predicate union to Japanese passives. e. The phenomenon of antipassive was covered by reading Davies' (1984) chapter in Studies in Relational Grammar 2 and Davies and Sam-Colop's (1990) article in Language. (3) Current research in the syntax ofgrammaticaljUnctions. The second half of the course used this volume as a text, and also made reference to a number ofother key chapters and articles dealing with grammatical relations from P&P perspectives. This other material included parts of Mark Baker's (1988a) book on incorporation, Bobaljik and Jonas' (1996) article on subject positions and TP, Cole and Hermon's (1981) article on subjecthood and islandhood. Guilfoyle, Hung, and Travis' (1992) article on SPEC of IP and SPEC of VP, Eric Haeberli's (2000) paper on the EPP and Case, sections from Teun Hoekstra's (1984) book on transitivity, Kyle Johnson's (1991) article on object positions, and James McCloskey's (1997) article on subject positions.
- 18. xv ACKNOWLEDGMENTS For their help and counsel in the early conceptualization ofthe workshop on "The Role of Grammatical Functions in Transfonnational Syntax" and the inception of the process that led to this book, we would like to thank Mark Baker, Paul Chapin, and Alec Marantz. We also owe thanks to the speakers and attendees of this workshop, for their presentations and their discussion of these, which ultimately enriched this volume. We thank Liliane Haegeman and one anonymous reviewer of this volume for their helpful comments, both on the individual chapters and on the overall organization of the book. We also wish to acknowledge Jacqueline Bergsma, Anne van der Wagt, and Susan Zwartbol-Jones ofK1uwer Academic Publishers for all their help in the editing process, as well as Theresa McGarry of the University of South Carolina Linguistics Program for her diligence and hard work editing the final typescript. We are indebted to the NSF (Grant #SBR 9817364), the College of Liberal Arts at the University of South Carolina, and the University of Iowa Arts and Humanities Initiative for support of the workshop and this volume. We want to thank our families (Melissa and Elijah Dubinsky; Pat Weir, Davies, and Kate Davies) for putting up with us, and for all their good vibes support, during our absences and unexpected presences. Finally, we would like to thank each other. We certainly had an awfully time putting this all together.
- 19. , 1
- 20. WILLIAM D. DAVIES AND STANLEY DUBINSKY REMARKS ON GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX The broad topic for this volwne is the role of grammatical functions in transformational syntax, that is, what role the notions subject and object play in the theory ofgrammar. There has been a long-standing debate in the field regarding the status of grammatical functions as theoretical primitives or defined notions. The Chomskyan school has consistently stated the position set out in Chomsky's 1965 Aspects ofthe Theory ofSyntax that grammatical functions are secondary notions that are definable on the basis of phrase structure representations of sentences. Work in other syntactic frameworks (first in Relational Grammar starting in the 1970's and later in Lexical-Functional Grammar starting in the 1980's) challenged this position, demonstrating the efficacy ofcapturing cross-linguistic generalizations in terms of grammatical functions and questioning whether or not these generalizations were available in theories with defined grammatical functions. Developments in Chomskyan theory involved more and more explicit reference to grammatical functions (particularly sub ject), and after the publication of Chomsky 1981 (Lectures on Government and Binding), the place of grammatical functions in work based on his theories evolved in distinct ways. Some authors challenged the basic claims of Relational Grammar, while others sought to incorporate into Government and Binding theory some of the insights that had crystalized in the RG and LFG literature. Further developments in Government and Binding theory in the 1980's and precursors to Chomsky's Minimalist Program in the late 1980's and early 1990's incorporated yet more references to grammatical functions, including the notion ob ject. It has been 25 years since the introduction of Relational Grammar theory by Perlmutter and Postal at the 1974 Linguistic Society of America Linguistics Institute, and many of the insights of linguists working in that theoretical framework have been incorporated in one way or another into the Chomskyan framework. And, with the advent of a more highly articulated phrase structure in the late 1980's and the 1990's and highly abstract configurational representations, there is the appearance that grammatical function-based notions permeate the formalism. However, explicit examination of the role of grammatical functions in the theory has been lacking. It is thus an appropriate time to re-examine the status of grammatical William D. Davies and Stanley Dubinsky (eds.), Objects and other subjects: Grammaticalfunctions,functional categories, and configurationality, 1-19. � 2001 Kluwer Academic Publishers. Printed in the Netherlands.
- 21. , 2 WILLIAM D. DAVIES AND STANLEY DUBINSKY functions and determine whether they are interesting viable notions or should simply be abandoned as epiphenomenal mnemonic terms. The collection of papers in this volume addresses the role of subject and object in the theory largely by making specific proposals about phenomena in a range of languages. Most conclude that the properties of objects and subjects can be attributed to functional or phrase structural properties and include novel proposals for the representation of these properties. . In this chapter, we outline some of the foundational developments in the role of granunatical functions in syntactic description and analysis, beginning with Chomsky 1965 (section 1). In section 2, we treat, in somewhat more detail, recent developments in the treatment of subjects and objects. Section three elaborates on points of convergence achieved by the chapters in this volume, and points the way to a typology of language architecture and to further research in the area. 1. A POINT OF DEPARTURE The assumption that GFs such as subject and direct object are definable in terms of phrase structure configurations, rather than being primitives of the theory, is a position which was first articulated in Chomsky 1965. Therein, Chomsky clearly distinguished GFs from granunatical categories such as NP, asserting that "the notions in question have an entirely different status" (p. 68) and that the functional information is redundant and "can be exacted directly from the rewriting rules ofthe base, without any necessity for ad hoc extensions and elaborations...to provide specific mention of grammatical function" (p. 73). Thus, it is assumed in Chomsky 1965 that GFs are defined notions, defined in terms of phrase structure configurations and perhaps only relevant to deep structure.I The general definitions that Chomsky assumes for grammatical functions are as follows: (i) Subject-of = [NP,S], (ii) Predicate-of = [VP.S], (iii) Direct-Object-of= [NP,VP]. and (iv) Main Verb-of = [V,VP] (p. 71). Thus, by these definitions in a structure such as (1), sincerity is the subject and the boy is the direct object. (1) S � NP Aux VP I I � N M V NP I I I /"-.... sincerity may frighten Det N I I the boy This position is reasserted in Chomsky 1981, in reaction to certain advances in the theories of Relational Grammar and the nascent Lexical Functional Grammar:
- 22. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX the less fully such notions as.•.grammatical relations... are reducible to primitives.., the greater the indeterminacy of grammars selected on the basis of primary linguistic data. There is. however, little reason to suppose that such indetenninacy exists beyond narrow bounds. Insofar as this is true, we should be skeptical about theories with a primitive basis containing concepts that cannot plausibly be assumed to enter into the detennination of the primary linguistic data. (p. 16) 3 While it appears that little has changed within transfonnational syntax with respect to the starus of GFs as theoretical primitives, in fact, the role of GFs has changed dramatically, albeit incremental1y, in part in reaction to developments in nontransfonnational theories and in part due to theory-internal developments. GFs played no appreciable role in early Standard Theory treatments; however, the notion of subject was accorded a new and significant status in Chomsky 1973b. where reference is made in the Specified Subject Condition, "where by 'specified subject' we mean a subject NP that contains either lexical items or a pronoun that is not anaphoric" (p.239). In this paper the GF definitions (in particular subject, p.239) that Chomsky (1965) provided for the base component were extended to derived structures. Thus, in "Conditions on Transformations", the GF subject took on a significantly different role than it had previously had: it was a central concept in a key grammatical principle and it also had a wider application, relevant to derived phrase strucrures rather than simply base strucrures. Concurrently and subsequently, new nontransformational paradigms arose in which GFs played a central role and were accorded the starus of primitives. Beginning with Perlmutter and Postal 1972 and gaining significant status at the 1974 LSA Linguistic Instirute (Perlmutter and Postal 1974), Relational Grammar (RG) sought to account for a wide range of cross-linguistic regularities in terms of GFs. In a spate of influential papers (Chung 1976, Pennutter and Postal 1977, Perlmutter 1978, Postal 1977), basing its empirical results on a much wider range of languages than ever before, RG presented a serious challenge to the notion that GFs are derived notions by providing insightful universal and language-particular analyses utilizing GFs as core primitives of the grammar (and, correspondingly, that the configurational positions of phrases in a sentence are merely artifacts of how a particular language presents these primitives). Work in RG not only gave a central role to the OF subject but also to the GFs direct object, indirect object, and others, examining a number of clause-level and sentence-level strucrures which crucially referred to these GFs. Additionally, the late 1970s saw the rise of another relationally-based theory, Lexical Functional Grammar, as embodied in Bresnan 1978, 1980 and the papers in Bresnan 1982b. Like RG, LFG accorded GFs a central role in the analysis of universal and language-particular phenomena, and treated them as undefined primitives of the theory. After the publication of Chomsky 1981, the place of grammatical relations in work based on his theories evolved in distinct ways. First, a few authors, such as Hoekstra (1984) and Marantz (1984), explicitly challenged the basic claims of RG theory. Others, such as Baker (1988) and Burzio (1986). attempted to build into GB some of the insights that had crystallized in RG and LFG literature. Baker's incorporation fonnalism was designed in large part to account for revaluations to object, which include the clause union/reduction phenomena treated extensively in
- 23. . 4 WILLIAM D. DAVIES AND STANLEY DUBINSKY Aissen & Perlmutter 1976 and Gibson & Raposo 1986. Burzio, for his part, formalized an account of unaccusativity first presented in Perlmutter 1978. At the same time, the publication of the GS theory in Chomsky 1981 and 1982 involved the further intrusion of GFs as undefined (or inconsistently defined) theoretical objects. In Chomsky 1981 (209-222), for example. the term SUBJECT is introduced into the calculation of "governing category". SUBJECT is used there to refer both to NPs occurring in the traditional. configurationally defined "subject position" (i.e. the immediate NP daughter of a clause or NP node), as well as to AGR (which is asserted to be the "nominal" content of INFL). On this formulation. the notion SUBJECT is no longer a purely configurationally defined relation, but one which identifies its referents variously on the basis of either configuration or labelling properties. With the articulation of the Extended Projection Principle (EPP). or more specifically the extended part of the EPP (Chomsky )982:10), GFs came to play an even more critical role in the central assertions of the theory. Here it is stated that ''the requirement that a clause have a subject is independent of the Projection Principle" and that this requirement is "a general principle governing D structures, hence also governing structures derived from them". The "subject" of the EPP, however, is not the same as SUBJECT (since clauses of Romance languages are held to have "subject positions" in addition to the SUBJECT [i.e. AGR] which they would possess anyway). Putting this inconsistency aside, the EPP elevated the notion "subject" (albeit a configurationally defined notion) to the level of a UG determinant of core grammar (Chomsky 1982:13). Since the mid-1980's, various developments in GS theory and what has evolved into Minimalism have further unsettled the traditional view of GFs in transformational syntax. One of these, the VP-lNTERNAL SUBJECT HYPOTIIESIS (see Kitagawa 1986 and Sportiche )988 among others), resulted in the abandonment of a singular. configurationally defined "subject position", inasmuch as this hypothesis provides separately for deep (i.e. thematic) and surface (i.e. inflectional) subject positions. This notion that all subjects are derived led subsequently to a reexamination of the category INFL itself, and to proposals in which its various functional components are separated out into independently headed projections. The view (expressed first in Pollock 1989) that AGR and Tense reside in separate functional categories has opened the door for further discussion of the positions and properties of subjects, such as found in Guilfoyle, Hung, & Travis 1992, Sobaljik & Jonas 1996, McCloskey 1997. At the same time that the notion "subject" has come under renewed scrutiny, there has arisen a corresponding interest in the traditional object functions of direct object (DO) and indirect object (10). Drawing on the VP-internal subject hypothesis discussed above, Larson (1988) proposed a view of multiple object constructions in which the underlying positions of direct and indirect objects are comparable to that of subjects and direct objects. That is, just as the clausal subject occupies a D structure spec,VP position superior to the complement, the direct object of a ditransitive verb is claimed to occupy a spec.VP (of a lower VP) and be superior to the indirect object complement in D-structure. This analogy is illustrated in (2), where VP1 in (2a) should be compared to VP2 in (2b).
- 24. GRAMMATICAL FUNCTIONS INTRANSFORMATIONAL SYNTAX (2)a. Greer likes Speer. VP, b. � NP V' 6. � Greer V NP I .6- like Speer Greer sent a letter to Speer. VP, � NP V' 6. � Greer V VP2 � NP V' �� a letter V I PP � 5 send to Speer Larson's approach is developed further in Pesetsky 1995, where VPs having non lexical heads (such as VPl in (2b» are done away with and replaced by lexical projections (sometimes having phonologically null heads). The essence of these proposals was to divorce phrase structure from traditional constituency tests, in favor of representations in which thematic hierarchies "map directly to a syntactic analog (much like the initial layer of structure in RG), a topic taken up by Baker in chapter 2. While the D-structure representation ofobject GFs evolved in the manner shown above, the inflectional status of objects also underwent a metamorphosis. Following Pollock's (I989) proposal to separate the nominal (AGR) properties of inflection from Tense, Chomsky (1991) proposed separate AGR projections for subject and object (AgrS and AgrO). Under this proposal, both subjects and objects are derived categories, each coming to occupy the Spec of its respective functional category. This proposal had the consequence of doing away with the traditional configurational difference between subjects and objects (inasmuch as both are now claimed to occupy a Spec position), and replacing this with a difference that resides either in the labeling function (i.e. AgrS vs. AgrO) or in the configurational position of the Agr phrases relative to lexical projections (whether an AgrP selects or is selected by vP).
- 25. , 6 WILLIAM D. DAVIES AND STANLEY DUBINSKY 2. RECENT VIEWS ON GRAMMATICAL FUNCTIONS 2.1 On the sub ject o fsub jects As noted above, the notion "subject" and its role in the grammar evolved considerably over the past three decades. While a configurational definition of subject remained intact and relatively unchanged from Aspects to LGB and into Barriers (the only change coinciding with the introduction of the CP/IP nomenclature), the role of subject became increasingly important. This is epitomized by the Extended Projection Principle of Chomsky 1982, which explicitly requires all clauses to have subjects, defined therein as [NP,S]. Since then, speculation has arisen regarding whether all languages obey the EPP and whether some languages satisfy the EPP in alternative ways. Additionally, developments in clausal architecture have radically reshaped the configurational definition of subject and, in fact, have cast doubt on whether a unitary notion of subject can be identified, as taken up in part III of this volume. An early step in what McCloskey (1997) tenns the "deconstruction" of a unitary subject was the introduction of the VP·internal subject hypothesis. This notion, that the subject originates in the VP and then in most (but not all) languages moves to spec,IP, entails the positing of two subject positions. Some of the most striking evidence for this hypothesis comes from Sportiche's (1988) analysis of quantifier float. (3)a. Tous les enfants ont vu ce film. all the children have seen this movie b. Les enfants (*tous) ont tous vu (*tous) ce film. the children have all seen this movie Sportiche adduces evidence for a VP·internal position for subjects in order to explain the fact that the quantifier tous 'aU' in (3b) can appear between the auxiliary and main verbs, but not in other plausible positions (such as before the auxiliary verb or after the main verb). Sportiche suggests that for French (and the English analog), the NP complement of the quantifier may raise independently to spec,IP, leaving the quantifier stranded as shown in (4). (4) [IP [NP les enfants]1 ont [vp [op tous tl ] vu ce film 11 Of course, (3a) is derived by having the entire QP move from spec,VP to spec,IP. These facts, together with other considerations (e.g. adverb placement), provide evidence of a VP·internaJ position for the subject. FUrther evidence for a VP· internal subject position came from languages for which it was argued that the subject of the sentence never raises to spec,IP but remains in the VP at S-structure. Such proposals include Kuroda 1988 for Japanese and McCloskey 1 991 for Irish.
- 26. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 7 The notion of two subject positions was developed further by Guilfoyle, Hung, and Travis (1992), who present an analysis of Austronesian languages that makes crucial use of the two subject positions, spec,VP and spec,IP. In their analysis, they extend the potential of a non-unitary notion of subject, and argue that the properties most often associated with a single argument and identified as "subject properties" can actually be split between two arguments in certain constructions in these languages.· So, for example, in the Tagalog examples below (Guilfoyle, Hung, and Travis 1992:(25» babae 'woman' possesses all subject properties in (Sa), while the subject properties are split in (5b) between babae and bigas 'rice'.2 (5)a. Mag-aalis ng-bigas sa-sako para sa-bata' ang-babae. AT-take.out Acc-rice OBL-sack for OBL-child TOP-woman 'The woman will take rice out ofthe sack for the child.' b. AaHsin ng-babae sa-sako para sa-bata' ang-bigas. TT-take.out GEN-woman oBL-sack for OBL-child TOP-rice 'The woman will take the rice out of the sack for the child.' Under their analysis of (Sa), theVP-intemal subject babae moves from spec,VP to spec,IP ( at the end of the clause), thus filling both subject positions. In (Sb), babae remains in the spec,VP subject position, while the theme bigas moves to spec,IP. This is illustrated in (6)-n.b. t2 is the trace orv movement. (6)a. b. [IP [I' mag-aalis2 [yp t ang-babae IP] t2 ng-bigas sa-sako para-sa-bata' ]] lIP [I' aalisin2 [yp ng-babae t2 tt ang-bigast IP] sa-sako para-sa-bata' ]] Under Guilfoyle, Hung, and Travis' analysis, typical Tagalog subject properties such as reflexive antecedence and being controlled are characteristic of the spec,VP position, while the properties of floating quantifiers and extraction accrue to spec,IP. When a single argument fills both positions (via movement), as is the case for babae in (Sa), that argument exhibits all subject characteristics. In (5b), however, the deep subject babae originates in spec,VP and resides there at S-strucrure, while the object bigas moves out of the VP complement position coming to occupy the surface subject position spec.IP at S-structure. Accordingly, the subject properties in (5b) are split between these two arguments. More recent accounts, particularly spurred on by the disassembling of INFL in Pollock 1989, have proposed yet other subject positions. In work on the Icelandic transitive expletive construction, Bobaljik and Jonas {I996) have argued that there is a subject position outside ofVP but lower than spec,AgrSP (which Teplaced spec,IP as the identified S-structure subject position following Chomsky 1991). They identify this position as spec,TP. In the Icelandic construction, both spec,AgrSP and spec,TP are occupied by phonetic material. As McCloskey (1997) points out, with
- 27. , i 8 WnLlAM D. DAVIES AND STANLEY DUBll'lSKY two subject positions in the inflectional layer, this either brings to three the number of identified subject positions or casts doubt on the VP-internal subject position. Additionally, Branigan (1996), Cardinaletti (1997), Kiss (1996), and Haegeman (1996) have proposed additional subject positions outside of the thematic and inflectional layers. For example. Branigan (1996) suggests that EPP features drive movement into spec,CP in verb-second languages. In such a proposal spec,CP thus becomes a GF position in those languages which have EPP-motivated movement to positions outside the inflectional layer. Current theoretical proposals, calling for mUltiple "subject" positions and for the parcelling out of "subject" properties among them, are reminiscent of RG proposals of ten to twenty years earlier. TheVP-internal subject position, assumed in most current work, has its analogue in the RG concept "initial subject". Analogous to current theoretical proposals for derived "subject positions" (in the inflectional layer of the clause), RG theory recognized the distinct status of "final subjects" in triggering agreement and f�eding word order rules (among other things). For example, Bell (1976) proposed distinct grammatical properties for initial and final subjects in Cebuano, a proposal quite similar to that offered (in the current model) by Guilfoyle, Hung. and Travis for other Austronesian languages. 2.2 Objects and other objects Unlike subject, the definition of the GF notion direct object was less troublesome to Chomsky's transformational theory at the outset. It was, by and large, assumed that direct objects of simple clauses (such as her in (7a» are not derived. Subsequently, Chomsky'S (1981) adoption of the Projection Principle entailed that no surface direct object can be derived from any other position. This helped provide justification for the exceptional case marking (ECM) analysis of accusative infinitival subjects, by which him in (7b) is a deep and surface subject despite its accusative case.l . (7)a. b. James saw her. Garth wants [5 him to leave] Thus, the Projection Principle divided the class of apparent surface direct objects into true direct object complements (7a) and accusative infinitival subjects (7b). Under these assumptions, direct object is trivially defined as the NP sister ofV. This situation changed with the publication of I.-arson 1988, in which the position of a deep direct object was claimed to be dependent upon the valency of the verb. As shown in section 1 (example (2», the direct object of a transitive verb is analyzed as the NP sister of V, while the direct object of a ditransitive verb is claimed to occupy spec, VP (of a lowerVP). Thus. under a Larsonian approach to D-structure, object properties can no longer be attributed to a singular, configurational definition of the notion direct object.
- 28. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 9 While the Larson (and later. Pesetsky 1995) approach to lexical insertion did muddy the definition of deep-subject, it left intact the notion that direct objects are not derived (at least in contrast to the VP-intemal subject hypothesis which held that subjects occupy distinct thematic and inflectional positions). However, following Pollock's (1989) division of INFL into distinct AGR and Tense components, Chomsky (1991) proposed an inflectional position for objects parallel to that of subjects. Based on work by Kayne (1989), Chomsky takes Pollock's AGR to be a "subject-agreement element" (AgrS), and suggests that there is an "object-agreement element (AgrO) ... [that has] VP as its complement". On this view, (7a) would have the following representation at LF. As (8) illustrates, under the multiple Agr-projection hypothesis (as per Chomsky 1991) coupled with the VP-intemal subject hypothesis. all subjects and all direct objects are derived (at le�st by LF).4 Rounding out the picture, Bowers (1993) proposes a functional projection PredP which is instantiated inside the inflectional layer (IP) and outside the thematic layer (VP) and which mediates the relation between the subject and the VP. Unlike Larson's 1988 VP-shell analysis the instantiation of PredP is independent of the thematic requirements of the verb (which makes it distinct from the agentive vP proposed in Chomsky 1995, whose presence is triggered by the thematic structure of the verb). Thus, by 1993 we find proposals that collectively introduce functional projections corresponding to the three GFs (subject, direct object, predicate) that Chomsky rejected as being part of phrase structure in Aspects. Recent proposals (Franco 1993, Koizurni 1995. and Sportiche 1996) have gone further, and provided for a separate agreement element for indirect objects (AgrIO) which also projects a phrase. Taken all together, these proposals lead to important questions about the place of functional projections in Universal Grammar. One can ask whether there is a universal stock of functional projections instantiated in every language or whether a universal repertoire of such projections forms an inventory out of which individual languages select a subset. Alongside this question is the issue of whether functional projections are instantiated in some fixed and universal hierarchy or whether the relative dominance of individual functional projections might vary parametrically from one language to another. Indeed these questions have been the subject ofsome discussion in the literature and the positions taken on these issues vary across a spectrum (see Iatridou 1990, Nash & Rouveret 1997, and Tlmiinsson 1996 for discussion). On one end of this spectrum is the more restrictive view that the set of functional categories and their relative dominance is fixed by UG. At the other is the suggestion that the inventory of functional projections and their relative dominance varies from language to language. As an example of the kind of debate that surrounds the instantiation of particular functional projections, one might compare Koizurni 1995 (which maintains that English has an AgrIO projection, in addition to AgrO) with Montrul 1998 (which assumes that AgrIO is specific to languages that have dative clitics, such as French and Spanish). In this volume
- 29. 10 WILLIAM D. DAVIES AND STANLEY DUBINSKY (chapter 6), Travis claims that Malagasy has no derived objects (and presumably no equivalent of AgrO). There is even some question as to whether a particular functional projection is uniformly instantiated in all construction in a given language. As an example of this, Lasnik (chapter 5) proposes that AgrO is o f.tional in English, leading to the possibility of both derived and underived objects. There are various proposals regarding the relative dominance of functional projections (especially AgrS and T). Pollock t989 posits TP above AgrP, while many subsequent analyses reverse this. The relative dominance ofAgrS and T may in fact correlate with other aspects of language structure, such as word order. We find that proposals for V-initial languages converge on the view that T dominates AgrS (McCloskey, chapter 7; Massam, chapter 9). In chapter 10, we propose a language typology that reflects, among other things, the parametric variation of TIAgr dominance. 3. THE STATE OF FUNCTIONAL ART In the ensuing discussion, we will point to a range of converging opinions on the status of subjects and objects cross-linguistically, as revealed in the chapters that follow. We will also indicate where the evidence either leads to divergent opinions about functional structure, or is simply inconclusive. Finally, we will sketch out what we believe to be an overarching set of functional parameters, which lead to a predictive typology of possible language classes and which may provide explanations for attested typological variation. In what might be viewed as a relatively surprising result, two of the chapters in section one point to ways in which P&P theory has come to incorporate grammatical functions as primitives, albeit not always in the most straightforward manner and through rather different subparts of the theory. Baker (in chapter 2) suggests that phrase structure representations have evolved in such a way that at the thematic layer phrase structure is a representation of grammatical functions, where grammatical functions are encoded through configurational superiority. As these phrase structures are determined neither by traditional constituency tests nor by any language-particular properties. current Chomskyan theory essentially shares with explicitly relational theories the notion that grammatical relations are primitives. Taking a far different tack. Alsina (in chapter 4) investigates the application of the notions structural Case and inherent Case to distinguishing objects in multiple object constructions. He notes that various properties of objects in languages (e.g., pronoun incorporation, lexical binding) consistently accrue to structurally Case marked objects as opposed to inherently Case-marked objects but that the thematic role ofthe argument targeted for inherent Case is not consistent between languages. On this basis, he suggests that the structural vs. inherent Case distinction has been used as another means ofencoding distinctions between "object" GFs. In chapter 2. Baker shows that configurational prominence can uniformly represent both intraclausal relational prominence and interclausal "embedding" prominence. Thus, the subject of a matrix clause has configurational prominence over its same-clause object and over its embedded clause subject. In an effort to
- 30. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX II achieve greater uniformity of representation, Baker suggests reducing all grammatical prominence (inter- and intra- clausal) to embedding prominence. He implements this through the adoption of a lexical decomposition strategy that requires at least one syntactic predicate for each subcategorized argument. By assuming, for instance. that a subject is in the specifier ofa vP and that an object is in spec,VP, the intraclausal superiority of subject over object can be treated as arising out ofan embedding asymmetry. While lexical decomposition can clearly be used as a vehicle for mapping out relational prominence, Newmeyer (in chapter 3) points out that lexical semantic representations (LSR) cannot take the place ofsyntactic grammatical functions. He notes that there are important reasons for representing relational prominence syntactically, and reminds us that syntactically unaccusative structures can arise out of a varied set of lexical semantic representations (a point made early on in Rosen 1984). Since the set of possible argument structures is much smaller than the set of possible semantic representations, and since the mapping from semantic representations to argument structures is not fixed (as documented in Grimshaw 1979), there is still a need for an autonomous level of deep syntactic structure in which "grammatical functions, thematic roles, and constituent structure are not [necessarily] in alignment". 3. 1 Sub ject attributes through the clause There are a number of syntactic properties that are characteristic of a single nominal in English. These properties include being the controllee in control structures, bearing nominative case and triggering agreement on the verb. and obligatorily undergoing raising in appropriate contexts. (9)a. h. (10)a. b. Vanessa tried to phone Andrew. *Andrew tried Vanessa to phone. Amy appears to be out ofthe office. *Appears Amy to be out ofthe office. Such properties have come to be known as "subject" properties inasmuch as they accrue to a single constituent referred to as the Subject in English. However, looking over a range of languages one finds that these properties can be distributed among distinct constituents in different languages. In order to understand the mechanism by which these properties are concentrated in a single constituent or divided among several. one must understand them to be reflexes of formally distinct syntactic attributes. Reviewing the list ofproperties just presented, we can ascribe controlleehood as arising from configurational or thematic salience, nominative case and agreement as arising from the interaction of an argument and a functional head
- 31. 12 WILLIAM D. DAVIES AND STANLEY DUBINSKY or feature, and obligatory raising as being driven by an EPP (D) feature of the clause. Taking these three syntactic attributes to be primitive, we find that in English the thematically/configurationally highest argument is the orie which checks nominative case and triggers agreement and is also the same argument which responds to the D feature in T. Accordingly, the full range of "subject" properties is mapped ta a single argument in English. However, this is not the case cross-linguistically. In Spanish, for example, the three syntactic attributes just described are not necessarily localized in the same argument. As Goodall shows in Chapter 8, arguments other than the thematic subject may satisfy the EPP, checking offa D-f eature in T. (t 1) A quien habia visto la madre de Juan? who had seen the mother of 'Who had Juan's mother seen?' According to Goodall, in (1 1), a quien checks off the D-f eature as wen as a Q feature in T, satisfying the EPP. In languages such as Irish and Niuean, we find that the EPP feature does not interact with any argument. As McCloskey (Chapter 7) and Massam (Chapter 9) show, there is no EPP position in the clause. Consequently, thematic subjects only display a subset of the English "subject" properties described above (raising is optional in these languages). This is illustrated in (12) for Niuean (with data from Seiter 1983), where lama 'boy', the subject ofthe embedded clause, has raised to be subject ofthe matrix clause (12b). (12)a. Kua kamata ke hala he tama e akau. PERF begin SBJ cut ERG boy ABS tree 'The boy has begun to cut down the tree. • b. Kua kamata e tama ke hala e akau. PERF begin ASS boy SBJ. cut ABS tree 'The boy has begun to cut down the tree.' Finally. we find that a single one of these attributes may be associated with more than one argument. Massam proposes that in Niuean the subject and object of a transitive clause occupy multiple specifier positions of the same functional projection. Since they have the same configurational salience, they are equally able to undergo optional raising. as illustrated in (13), where the object of-the embedded clause, akau 'tree', has raised to be subject ofthe matrix clause. (13) Kua kamata e akau ke hala he tama. PERF begin ABS tree SBJ cut ERG boy 'The tree was begun to be cut by the boy.'
- 32. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 1 3 Thus it is the many-to-one mapping from syntactic attributes to a single argument, which creates the illusion ofa single set of"subject" properties in English (and other languages like it). The cross-linguistic evidence presented in this volume shows that the notion ofsubject is epiphenomenal rather than primitive. It is also the case that the syntactic attributes which contribute to the epiphenomenon of subject are themselves parameterized rather than universal. For instance, we. find that agreement heads (or features) can be obligatory or optional. Irish clauses do not require the presence of a nominal argument, thereby allowing f or unaccusative predicates whose sole argument is prepositional. (14) Bhreathaigh at: ab aimsir. became-fine on the weather 'The weather became nice.• (McCloskey, ch. 7) In Niuean, on the other hand, the absolutive argument is obligatory. Another parameterized syntactic attribute that plays a central role in the determination of clausal . behavior is the EPP. We find, from cross-linguistic evidence, that the presence of an EPP f eature appears to be universal. That is, the head ofTP (or its equivalent) is always found to bear a category feature that induces movement into the inflectional layer ofthe clause. In this regard, though, languages vary according to whether this f eature is D or V. In D-prominent languages, such as English, French, and Spanish. the EPP f eature in T f orces the appearance of a DP in spec,TP. In V-prominent languages, such as Bulgarian, Irish, and Niuean, the EPP feature induces the movement of V (or some projection of V) to T and yields (in most instances) V-initial word order. In V-prominent languages. EPP properties normally associated with the thematic subject arc absent, since the EPP does not aff ect the distribution of any nominal element. Accordingly, in these languages, raising is optional as in (12), subject expletives are absent, and subject island effects are frequently missing. For instance, Bulgarian allows extraction from sentential sub jects, as in (15) (Davies and Dubinsky, Chapter 1 0). (15) na kakvol mislis [�e fda otide tl ] bese vaZno za nego] to what you.think that to go was important for him 'To what do you think that to go was important f or him?' The choice of V- or D- prominence also interacts with another parameter that detennines the relative superiority of T (or the EPP f eature) and Agr/Case. This parameter, as f ar as we can determine, is only active in D-prominent languages. Irish and Niuean, in addition to being V-prominent, also f eature a T(-like) projection higher than Agr/Case. In contrast, English is D-prominent and has Agr above T. Spanish, while D-prominent like English, is like Irish in having T above Agr. This difference between Spanish and English results in certain key differences between these two languages. Unlike English, the thematic subject in Spanish need not occupy the preverbal position, since it only need check Agr/Case features lower in the clause. At the same time, unlike Irish, some DP element must check the 0-
- 33. 1 4 WILLIAM D. DAVIES AND STANLEY DUBINSKY feature in T higher up. This results in a obligatory "subject'.'-like position, which is sometimes filled by a nominal locative or by the trace of a moved wh-operator. The choice ofEPP feature coupled with the relative height of T and Agr leads to a range oflanguage types, as expressed in the foHowing table. D-prominent EPP V-prominent EPP T/Agr Spanish Irish Agr/T English -- 3.2 Ob jectproperties Lasnik (chapter 5) proposes that, while overt movement to check features is in general obligatory, AgrO in English is optionally present. Because of this, he argues, direct objects in English can be underived. As an illustration of this, he shows first that extraction from object NPs is ruled out when the object does raise to spec,AgrOP in order to bind a reflexive pronoun, as in (16). ( 1 6) ??Which senatordid the special prosecutor question [ two friends of_ ] during each other's trials. When derivation is not motivated by binding requirements, the object may remain underived, thereby f acilitating extraction, as in (1 7). ( 1 7) Which senator did the special prosecutor question [ two friends of_ ] during the president's trial. Lasnik concludes from this and other evidence that, at least for English, the AgrO can be distinguished from AgrS by its optionality. Similarly, Travis (chapter 6) proposes for Malagasy that the analog of the AgrO projection (AspP) is inert. She provides evidence that a number of constructions, previously assumed to involve derived objects, in fact involve derived subjects instead. She shows that Malagasy fails to have the equivalent of AgrO in its clause structure. Thus, both Lasnik and Travis agree that objects may be distinguished functionally from subjects, in at least some languages, in that the former may or must remain in their base-generated position within a basic transitive clause. This state ofaffairs is comparable to that of Niuean, for which Massam (chapter 9) claims that the Ergative argument is underived. In their account of objects, both Lasnik and Travis take the lower Agr/Asp projection to be intermediate between the two verbal projections (vP and VP). in keeping with recent proposals by Koizumi (1995). According to this view of clause structure the inflectional and thematic space cannot be divided into distinct layers, since an inflectional projection intervenes between two lexical ones, as in (18).
- 34. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX (I 8) [AgrP • • • Agr [vp ... V [AgrP • • • Agr [vp ... V .. . Jm 1 5 While this perspective erases a useful distinction between the inflectional and thematic space within the clause, it does simplify the representation of Agr. With this representation it is not necessary to distinguish between AgrS and AgrO, since the internal and external core arguments of the clause each have their own immediately dominating Agr projection. In examining Irish, McCloskey provides evidence that all inflectional projections are higher than vP so that there appears to be, at least in some languages, a strict separation of thematic and inflectional layers. At the same time, there also seems to be no cJear consensus regarding the interpretability of functional heads and features. For Lasnik, agreement is mediated by the purely inflectional head Agr, while Travis suggests that the relevant projection is an Asp(ect) phrase. If Agr (and other similar f eatures) are in f act borne by interpretable heads, then the motivation is reduced f or drawing a clear division between inflectional and thematic layers of structure. At this point, though. the divergence of opinions on this point reduces the motivation f or further speculation on this issue. 3.3 Functionalparameters and typological variation It is possible to identify a number of parameters of functional projections that account for some of the typological variation presented above. One such parameter is the nature of the EPP feature that resides in T. We suggested above (consistent with proposals by Massam & Smallwood 1 997, Massaro 2000b, and Rackowski & Travis 2000) that this can be characterized as V-prominence vs. D-prominence; that is, this feature on T will force either V-movement or D-movement. Languages such as English, French and Spanish are D-prominent, while languages such as Bulgarian, Irish, Malagasy. and Niuean are V-prominent. A second, cross-cutting parameter is the relative height of the functional projections T and Agr. While we propose (in Chapter 10) that T always commands Agr projections (or features) in V-prominent languages, in D-prominent languages there is evidence of a split. In his proposal f or Spanish, Goodall provides evidence that T is higher than Agr, showing that a non-nominative DP argument will satisfy the strong D-feature in T when the nominative thematic subject is postverbal. Conversely, in our account of English EPP eff ects, we provide evidence f or the current view that one Agr pro jection is above T. In our account of English. the superiority of T over Agr (coupled with D-prominence) is shown to explain the concentration of "sub ject" properties in a single argument. A final parameter of functional projections is the obligatory vs. optional nature of Agr heads or f eatures. Starting with the assumption that there are two possible Agr heads or sets of f eatures ass�ciated with transitive predicates-a high Agr f eature (commonly ref erred to as AgrS) and a low Agr f eature (AgrO)-the possibility exists f or onc or the other or both ofthese features to be optional in a language. For instance, Lasnik proposes that the lower Agr in English is only optionally instantiated. Putting a slightly diff erent f ace on this, only the higher Agr f eature in
- 35. , I 1 6 WILLIAM D. DAVmS AND STANLEY DUBINSKY English is obligatory. What this means is that ALL English clauses must have the higher Agr feature. that which has been associated with nominative Case and agreement. Thus, regardless of transitivity, there must always be a nominative argument in the clause. When the higher Agr feature is obligatory, the same Agr feature associated with the thematic subject ofa transitive clause will need to also be checked by the single argument of an intransitive clause (whether unergative or unaccusative). This is schematized in (I9). (19)a. Transitive clause: [ Thematically-high-arg AgrX [ Thematically-Iow-arg (AgrY) ... JJ b. Intransitive clause: [ Single-argument AgrX ... ]] This state of affairs results in a case/agreement system traditionally characterized as nominative/accusative, in which AgrX is AgrS (or Nominative) and AgrY is AgrO (or Accusative). The need, in English, for a nominative argument in each clause might be seen as one side of the "coin" that motivated the Final I Law and other similar proposals (with the D-prominence of the EPP being the other side of that "coin"). In some ways, what Massam proposes for Niuean is the inverse of Lasnik's proposal for English. Citing the optionality of the Ergative argument and the obligatory nature of the Absolutive, Massam proposes an obligatory Absolutive feature in vP. In Massam's analysis then, the obligatory Agr feature is the configurationally lower one (in that it is checked by the lower oftwo arguments). In this circumstance, the same Agr feature associated with the thematic object of a transitive clause will need to also be checked by the single argument of an intransitive clause (whether unergative or unaccusative). It is this same Agr feature that is checked by the thematic subject of a transitive clause when the thematic object undergoes noun incorporation with the verb. This case system is schematized in (20). (20)a. Transitive clause: [ Thematically-high-ar8x [Thematically-Iow-argy AgrX AgrY ... ]] b. Transitive NI clause: [ Thematically-high-arg AgrY [ Thematically-low-arg . .. ]] c. Intransitive clause: [ Single-argument AgrY ... ]] The result of this is, of course, a case/agreement system characterized as ergative/absolutive, in which AgrX is Ergative and AgrY is Absolutive. The view adopted here unifies derivation of Accusative objects in English and Absolutives in Niuean, in that both reference the same (i.e. lower) Agr feature. The difference
- 36. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 17 bet,ween them, then, has simply to do with optionality/obligatoriness o fthe relevant Agr f eatures. This approach to case/agreement is similar to that developed by Bobaljik (1992), but consistent with proposals by Levin and Massam 1985 and Massam 1985. Characterized as the "absolutive-as-accusative" analysis, the agreement f eatures of absolutive arguments are checked in spec,AgrO and those of ergative arguments are checked in spec,AgrS.6 It should be clear from the discussion so far how distinct Agr realization requirements in these languages lead naturally to the traditional nominative/accusative and ergativeJabsolutive characterizations. A third possible parameterization for agreement f eatures is seen in McCloskey's analysis of Irish clause structure. Irish, like English and Niuean, instantiates two distinct Agr heads (or f eatures), with the higher being characterized as Nominative (since it can mark intransitive arguments). However, according to McCloskey, both Agr heads are optional. The result of this is that some Irish clauses have only a single prepositional argument. and involve no checking ofcase/agreement (I4). The optionality of Agr f eatures, which can be distinguished as ''high'' and "low", gives rise to the f ollowing range ofclause structure types. High Agr optional High Agr obJigatory Low Agr optional Irish English (nominative/accusative) Low Agr obligatory Niuean (ergative/absolutive) -- Irish, then. is a language which exhibits an extreme case of "subjectlessness". Irish contrasts most strongly with English, which has a D-oriented EPP, an obligatory (nominative) Agr feature, AND a requirement that both of these reference the same argument. It contrasts with Spanish, which has D-prominence in EPP and obligatory (nominative) Agr. And. it contrasts with Niuean. which has an obligatory (absolutive) Agr feature. A V-prominent EPP coupled with an optional (nominative) Agr f eature leads Irish to exhibit no clausal requirements suggestive of having to have subjects. 4. CONCLUSION This discussion leads us naturally to the question of where does this discussion leave us. On the one hand, individual grammatical functions (such as subject) appear not to be as much unanalyzable primitives as has been imagined in theories such as RG, LFG, and HPSG. On the other hand, subjects (or particular subject properties) are clearly distinct from objects (or their analogous object properties) in ways that are not fully understood and which suggest the existence of some primitive (possibly feature based) distinction. The papers that follow do not provide definitive answers to all the questions that arc addressed in this volume, and in f act raise a number of additional questions about the status of grammatical functions in syntactic representations. They do. however, exhibit significant agreement on a number of
- 37. , 18 WILLIAM D. DAVIES AND STANLEY DUBINSKY issues, such that the resulting convergences dearly provide a path for future research in this area. For example, the EPP, however it may ultimately be characterized, is clearly associated with subjects and not objects. It is also clear from some of the chapters in this volume that the derivation ofsubjects and objects out of the thematic layer into the inflectional layer ofa clause is detennined independently for each relation in the grammar of a given language. Thus, to the extent that objects are optionally derived in English, for example (Lasnik, chapter 5), their grammatical status is distinct from subjects in ways that have yet to be determined. The resulting situation appears to be as follows: (i) specific properties generally related to grammatical functions (such as agreement) have autonomous reflexes in the grammar and are often implemented here as distinct functional heads (e.g., Agr); (ii) at the same time, the relation between a particular property type (such as agreement) and its associated grammatical functions (such as subject and object) is not unifonn across all grammatical functions. So, while (i) suggests that there is no discrete grammatical reflex for any particular grammatical function, (ii) suggests that there is more to the distinction between particular grammatical functions than "all and only" the functional properties that have been taken account of thus far. This then surely argues for further exploration into the underlying role of GFs as determinants of grammatical structure. The sort of conclusion that the work in this volume points to was in many respects anticipated in Keenan's 1976b 'Towards a universal definition of "subject"'. In that paper, Keenan attempts to arrive at a definition of the notion subject by testing 30 some properties (some of which had been widely used as traditional diagnostics for identifying subjects) against data from a variety of languages. With respect to these properties he notes that no one of them is "both necessary and sufficient for an NP in any sentence in any L [language] to be the subject of that sentence." Keenan concludes that '''subject' does not represent a single dimension of linguistic reality. It is rather a cluster concept, or as we shall say, a multi-factor concept." Appealing to the metaphor of 'intelligence', which is recognized as "a combination of abilities", he draws an analogy between it and the properties that he attributes to the subject relation. He goes on to say, "Being a subject is...more like being intelligent than...like being a prime number." At the same time he suggests (hopefully) that "the concept of subject might coincide with...groupings of [these] properties, though in the worst of cases each [property]...would be an independent factor." ENDNOTES I Despite the f act that Chomsky uses thc tcons "logical sub jecf' and "grammatical sub ject" in his discussion of grammatical functions, the structural definitions that he supplies for GFs are most directly applicablc to deep structure configurations. As he notes (in note 32, pp. 220-221). the existence of sentences containing PP sub jects or two initial NPs may necessitate a distinct set of OF definitions relevant to swface structure. 1 In the glosses, AT stands for AGENT TOPIC and 1T stands for THEME TOPIC. The other abbreviations are: TOPOC), ACC(USATIVE), OBL(lQUE), and GEN{lTIVE).
- 38. GRAMMATICAL FUNCTIONS IN TRANSFORMATIONAL SYNTAX 19 Ofcourse, the assertion that him is a surface subject (8b) is ofdubious status given the wealth of evidence to the contraty in Postal 1974, which has since been rerecognized as compelling beginning with Lasnik and Saito 1991. Empirical evidence f or such a move is obvious in languages with overt object agreement (as in many languages of Afiica and the Americas) as well as in participial agreement in Romance languages, as triggered by leftward movement of the object, as in the French example in (ib), as pointed out to us by Liliane Haegeman.. (i)a. Jean a detroit Ispreuve. •Jean has destroyed the evidence.' b. Jean l'a detrWte. •Jean has destroyed it.' The notion "derived", as it pertains to movement ITom lexical to inflectional pOSitions, remains somewhat unclear as used in the current literature. One might take uderived object" to mean"any object that moves from a base position into an inflectional one". One might alternatively take it to mean "an object that moves ITom a base position into an inflectional one be f orespell-out(i.e. overtly)". An alternative view (as represented by Bittner 1988, Campana 1992, and Murasugi 1992) is what Campana characterizes as the "ergative-as-accusative" analysis, in which, under Minimalist assumptions, absolutive DPs have their f eatures checked in spec,AgrS while ergative DPs have their f eatures checked in spec,AgJO.
- 39. • 1
- 40. MARK C. BAKER PHRASE SlRUCTURE AS A REPRESENTATION OF "PRIMITIVE" GR.AMJ.AATICAL RELAnONS One ofthe defining properties ofthe nan:owly Chomskian approach to syntax' over the years has been its corrunitment to the idea that grammatical relations such as subject and object are not primitives of grammatical theory. Rather, these notions, inherited from traditional grammar, are to be understood in terms of more basic syntactic relationships, in particular phrase structure configurations. This view gives phrase structure a kind ofprimacy in Chomskian theorythat it does not have in other frameworks. In this conceptually oriented paper, I want to reexamine this distinctive claim, reviewing the old controversy between Chomskian theory and Relational Grammar (RG) and related frameworks that that claim was part of. In the course of this, I will consider what the ongoing legacy of these debates is in the "post-RG" syntactic theories of the 2000s. This issue is roughly equivalent to the question of what the characteristic phrase structure representations of Chomskian theory actually mean within the network of assumptions that they are now embedded in. My basic claim is that the meaning of the old slogan has changed somewhat over time, as ideas about phrase structure have changed, in ways that have not been fully realized. In particular, the Chomskian notion ofphrase structure has come partially unhinged from its origins as a representation ofbasic constituency facts. As a result, it is now f air to say that phrase structure is essentially a representation of grammatical function relationships, not fundamentally different in kind from the representations posited by the Relational Grammarians and others. However, I will argue that Chomskian phrase structure is a particularly good representation of grammatical relations. In particular, it is superior to relational nets or the f structures of Lexical Functional Grammar (LFG) in that it captures certain basic properties of linguistic prominence that are not captured by these more general and flexible-looking representational schemes. . Thus, while the meaning of the claim that grammatical relations are derived from phrase structure has shifted over time� that claim is still meaningful. Finally, I will ask why it is that phrase structure is such a successful representation, showing how this relates to lexical semantic work on the decomposition ofverbs into more basic components. . 21 William D. Davies and Stanley Dubinsky (eds.), Objects cznd other sub jects: Grammaticalf unctions,f unctional categories, and configurationality, 21-5 1. e 2001 Kluwer Academic Publishers. Printed in the NetherlandS.
- 41. , 22 MARK C. BAKER 1. THE ORIGINS OF THE CHOMSKIAN VIEW To put these issues into context, let us begin by reviewing how the claim that grammatical relations are expressed in phrase structure arose. In f act, it emerged rather naturally out of the sequence of topics that were investigated and came to be partially understood by generative grammar. However, my f ocus will be on the idealized conceptual order of the ideas, rather than trying to trace their actual historical order in detail. This conceptual order also reflects the way that Chomskian syntax is often presented to students in textbooks and syntax courses, and this may be of more current significance than the history itself. The first step is motivating that there is such a thing as phrase structure f or languages like English, showing how this phrase structure can be established using traditional constituency tests. For example, one of the less obvious f eatures of English syntax is that there is a verb phrase that contains the verb and the direct object, together with some particles and PPs, but not the subject. (It is my inf onnal experience that when introductory linguistics students are asked out of the blue which is more closely related to the verb, the subject or the object, they have no clear intuitions: 50% chose one and 50% the other prior to seeing any arguments.) However, one can show that the verb and the NP that f ollows it f onn a unit with a familiar battery oftests, including movement (la), replacement with a pronoun (l b), deletion (lc), and coordination ( I d). (1)a. Sue said she would eat the octopus, and [vp eat the octopus] she did - . b. Sue will [vp eat the octopus], and John will [vp do so] too. c. Sue will [vp eat the octopus], and John will [vp - ] too. d. Chris will [vp chop the vegetables] and [vp eat the octopus]. Once this notion of phrase structure is firmly in place, it is a rather trivial observation that sub jects and ob jects always come in a fixed position in the phrase structure. In particular, the direct object is the only noun phrase that is contained in the VP but not in any smaller phrase. The subject is consistently contained in the clause but not inside the VP. Indirect objects in a sentence like Chris gave the book to Pat are uniquely contained in a prepositional phrase headed by to. And so on.2 Since there is such a close relationship between the traditional grammatical functions and phrase structure positions, it seems ulUlccessary to ref er to both in one's syntactic theory. Now, some representation of constituency is needed in the grammar anyway, to characterize which sets ofwords can be moved as a unit, or can be deleted, or can be replaced with a pronoun. Theref ore, Chomsky (1965) eliminated subject and object with the f ollowing definitions in terms of phrase structure:
- 42. PHRASE STRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 23 (2)a. Subject is "NP immediately dominated by S" ([NP, S]). b. Object is ''NP immediately dominated by VP" ([NP, VPD. c. Indirect ob ject is "NP immediately dominated by PP headed by to" ([NP, PPD· Logically speaking, the alternative reduction would have been to eliminate phrase structure in f avor ofgrammatical relations. However, that makes little sense. For example, one wants to say that asflat as a pancake is a phrase in (3), even though it has little to do with the traditional grammatical relations. (3) Chris pounded the clay [as flat as a pancake]. Thus, phrase structure is more general and easier to motivate independently than grammatical relations. So if one can only keep one, it should be phrase structure. Basically, Chomsky's suggestion was an Occam's razor-style argument par exceIIance. Next, investigation into other f ancier phenomena reveals that there are many ways in which SUbjects, objects, and other phrases behave diff erently from one another. At the level of basic description, the operative generalizations can be stated in terms of grammatical relations. The f ollowing are f amiliar examples from the history ofthe field: (4)a Only subjects can be controlled in nonfinite clauses. b. The antecedent ofa reflexive must be a subject. c. The direct object becomes the subject ofa passive verb. d. Only the direct object can incorporate into the verb, etc., etc. But using the equivalencies in (2), these statements can be rephrased in terms of phrase structure, as follows. . (5)a. Only the highest NP in a nonfinite clause can be controlled. b. The antecedent of a reflexive must be the NP immediately dominated by S. c. The NP in VP moves to the Specifier of IP ifthe verb is passive. d. Only the nominal that is the sister ofthe verb can incorporate into it, etc., etc. , : . = ",
- 43. 24 MARK C. BAKER In terms of the discussion so far, clearly nothing is lost by this translation. One might argue that nothing is gained by it either, except parsimony. But parsimony is something, the proper Chomskian would reply. For better or worse, the choice to develop one's principles of grammar around statements like those in (5}--or more basic statements from which these can be derived-accounts for the characteristic phrase-structure-centric cast of Chomskian theory. In this theory, establishing the structure is paramount to doing an analysis, and phrase structural relations such as "sisterhood", "government", "containment" and "c-command" usually figure prominently in the account. 2. THE RELATIONAL GRAMMAR CRITIQUE A cogent and influential critique of this structurally oriented approach to syntactic theory was mounted in the 1970s and early 1980s. This critique was presented first and most prominently by the Relational Grammarians in articles like those in Perlmutter (1983c), but it was also adopted and extended by the papers in Bresnan (1982b) within Lexical Functional Grammar. It has also influenced other, less generatively-minded linguists to various degrees. sometimes consciously. and sometimes not. As a result. most other theories ofgrammar are much less concerned with establishing the phrase structure ofa clause than narrowly Chomskian work is. Often. the matter ofphrase structure does not even come up in these works. The RG critique starts with the observation that traditional constituency tests like those in (1) for English often do not apply in other languages, or ifthey do they give quite different results. Indeed, the tests that reveal a VP constituent in English are actually somewhat rare crosslinguistically. There is nothing exactly like VP fronting, or VP ellipsis, or VP conjunction in Warlpiri (Simpson 1 991) or Malayalam (Mohanan 1982) or Mohawk, tor example. The weak interpretation of these results would be that there is no evidence for a VP node in these languages. Typically, however, people took the stronger interpretation, that these observations showed that there was no VP in these languages. This conclusion was motivated by a tacit positivist-style assumption that any phrase that exists in a language should be detectable by this kind ofevidence. Now iflanguages differ in substantial ways in their phrase structures, this affects the feasibility of defining grammatical relations in terms of phrase structure, as in (2). For example, ifthere is no VP in Warlpiri or Malayalam or Mohawk, then one cannot distinguish the object from the subject in terms of elementary phrase structure configurations. There may simply be no phrase that contains the object but not the subject in these languages. This could have turned out to be a positive result ifthe grammatical relations of these languages turned out to be substantially different from those ofEnglish. Thus, the next step is to look at grammatical phenomena other than simple constituency tests, to see ifgeneralizations like those in (4) are valid crosslinguistically. Suppose, for example, that both the "subject" and the "object" in Warlpiri or Malayalam could be the antecedents of simple reflexives, and that both could be" controlled by matrix clause constituents when in a nonfinite embedded clause. This would fit beautifully
- 44. It· ,. t., I PHRASE STRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 25 with the fact that these languages seem to have no VP, so both the subject and the object are immediate constituents of the clause. Phrase structure would vary across languages, and this would be a puzzle f or strong views about Universal Grammar. But grammatical relations would also vary across languages, in a way that is correlated with the variations of phrase structure. This would be striking crosslinguistic evidence that phrase structure is central to syntactic theory. However, this was not generally what was f ound (although see fil. 3 for some possible cases of this type). On the contrary, the Relational Grammarians f ound a rather impressive array ofevidencethat OR-related generalizations like (4) are more stable and robust across languages than basic constituency is. For example, Perlmutter and Postal (1977) f ocused on the passive, comparing (4c) to (Sc). They showed that there are many languages in which it is not so attractive to think of passive as involving the movement ofan NP from inside the VP to outside the VP as in (5c). In part, this is because it might be dubious whether there is a VP at all in the language in question. Also, on a more surface level, there are many languages in which passive sentences do not diff er from active sentences in word order, but only in the case marking ofnoun phrases, or in the agreement inflections on the verb. For such languages, "movement" does not seem like a good way to think about the passive. (6) is one of Perlmutter and Postal's examples from Cebuano. (6)a Magluto' ang babaye ug bugas. ACTIVE cook-ACT NOM woman rice 'The woman will cook rice.• b. Luto'on sa babaye ang bugas. PASSIVE Cook-PASS OEN woman NOM rice 'The rice will be cooked by the woman.' However, if one thought of passive in terms of the traditional grammatical relations, then one could recognize a fundamental unity between (6b) and its English equivalent, a unity that has little to do with surf ace wOTd order or constituency. In both languages the normal object becomes the subject, displacing the usual subject and creating an intransitive clause. Described in these terms, the passive is consistent across languages, whereas the observed differences are purely a matter of how the two languages encode subjects and objects. In English, the subject is realized as the first NP in the sentence (the NP dominated only by the clause), whereas in Cebuano it is realized as a postverbal NP preceded by the nominative case particle ang. Crucially, these differences are observable even in simple, active clauses. Thus, they are independent ofthe passive per se, and passive is better stated in a f ormat that abstracts away from these matters. In other words, passive should be stated purely in terms of "primitive" grammatical functions. And it is not only passive that works like this, but a host of syntactic phenomena, including control, reflexivization, dative shift phenomena, incorporation, complex predicate f orming processes, and so on.
- 45. 26 MARK C. BAKER With these empirical observations in hand, the Relational Grammarians and related approaches off ered the f ollowing deduction. Grammatical relation phenomena are relatively consistent across languages. Phrase structure is not consistent across languages. Therefore. it is wrong to derive grammatical relation phenomena from phrase structure. Rather, they should be retained as elements of syntactic analysis in their own right, as "primitives" of syntactic theory. It is merely a coincidence that in English there happen to be one-to-one correspondences between structure and grammatical relations that make a reductive approach conceivable. When the theory is developed in this way, phrase structure is much less important than on the Chomskian approach and can be left to the periphery of the theory. The urgent task is to determine what the grammatical relations are in any given clause. In practice, the Relational Grammarians rarely said much about phrase structure issues, beyond the basic f acts that concerned the embedding of clauses and NPs. In particular, it paid little attention to the details of word order and its relationship to constituency. On this point, LFG illustrates the basic logic of the situation more clearly. LFG holds that every sentence has at least two representations that are quasi-independent of each other: a functional structure that is established by looking at GR-sensitive phenomena, and a constituent structure that is established by basic word order considerations. Phrase structure exists on this view-it is just not very interesting, not very relevant to other syntactic phenomena, and certainly not very stable across languages. 3. THE CHOMSKIAN REACTION With this historical perspective in mind, the question of primary interest here is how has Chomskian theory responded in practice to the RG critique and the empirical f acts that motivated it? How does Chomskian theory now compare to the theory bef ore as a result ofthese issues being raised? As is often the case on matters of high-level theoretical comparison. direct engagement over these issues was relatively light. Perhaps the most prominent direct responses to the RG critique were some passages of Chomsky (1981). In section 2.7, he questioned whether there was really a unified phenomenon of passive in the sense that Perlmutter and Postal (1977) assume. He claimed instead that there were simply various language specific constructions in various languages that happened to resemble the English passive to a greater or lesser degree. One of the interesting examples that he cites was the yi-Ibi- alternation in Nava jo. which does not fit comf ortably under the usual passive rubric, but yet has some similarities to passive, both structurally and functionally (see Speas 1990 f or more data and a Chomskian analysis). For example. (7b) is like a passive in that involves preposing and highlighting the thematic object, but it is unlike the passive in that the agent is not marked as oblique. nor can it be omitted in the context ofan overt theme.
- 46. PHRASE STRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 27 (7)a. Ashkii at'eed yiyiHts!. ACTIVE Boy girl saw 'The boy saw the girl.' b. Ashkii at'eed biBst!. PASSIVE? Boy girl saw 'The boy was seen by the girl. ' Whether we call this a passive or not is an uninteresting tenninological question. Chomsky suggests; it has the properties it does, whatever we call it. There is no universal passive to have a universal theory of. he argued,just particular passives to have particular theories of. Then in section 2.8, Chomsky proceeded to sketch a way of generalizing his theory to the passive in Japanese, which was at the time thought to be a nonconfigurational language, lacking a VP. However, these relatively direct responses had fairly little impact on the subsequent developmept of the theory. in practice. The real influence of Relational Grammar came tacitly as various people tried to incorporate the impressive data and empirical generalizations that the Relational Grammarians discovered and described into their Chomskian theories (see, f or example, Marantz (I984), Baker (1988), Speas (1990), among others). This proved interesting when it converged with considerations that arose internal to the Chomskian program itself as that program tried to pursue a finer-grained understanding of English and a greater degree of crosslinguistic generality in its own tenns. The key shift of perspective was learning to be distrustful of the traditional constituency tests. It is a cruciaJ (partly tacit) assumption of the RGILFG critique that these traditionaJ constituency tests are reliable methods f or establishing phrase structure relationships. They reason that since constituency tests are not available or give different results in other languages, phrase structure does not exist or is fundamentally different in other languages. But that conclusion does not nece�sarily f ollow. It could be that constituency tests give different results not because phrase structure is diff erent, but because the internal workings of the tests themselves are different. In the ultimate scheme of things, there should be a substantive theory of VP preposing, a theory ofellipsis, and a theory of what can be a pro-f orm, as well as a theory ofpassivization and a theory ofcontrol. There is no a priori reason to think that these theories will be simple or trivial. Once we understand clefting better, there may be excellent reasons why (say) a VP can be clefted in some languages and constructions but not in others. In the earliest transformational syntax - often recapitUlated in the earliest syntax teaching - the theories of movement. deletion. pronominalization, and con junction are indeed rather triviaJ, offering a way to break into the system. If there were nothing else to say about them, then the simple claim thatphrases move, delete. are replaced, and are combined as units might indeed lead to the stronger claim that all phrases should be recognizable in this way. But this is surely an oversimplification, mistaking the first draft of an understanding of these phenomena f or a mature one. Thus, it is quite reasonable to concede to the Relational Grammar critique that traditional constituency tests are not reliable
- 47. " 1 28 MARK. C. BAKER crosslinguistically without conceding that phrase structure itself is not present crosslinguistically. Now suppose that English-like phrase structure exists in a language like Malayalam even though it cannot be detected by English-like movements and deletions. Then the grammatical relations ofsubject and object can still be defined in terms ofthat phrase structure, and passive, control, and reflexive binding in Malayalam can be analyzed in terms ofnotions like c-command. It is instructive to see how these concerns about constituency tests arose internal to Chomskian syntax itself, before considering further how they affect the comparison with RG and its allies. For various reasons, Chomskian" theory in the 1980s and 1990s was led to enrich its stock of phrases beyond what was accepted previously. For example. it was argued that bare adjectival predicates form a "small clause" constituent with their NP subjects, at least in some contexts (Chomsky 1981. Stowell 1983). Later, Abney (1987) argued the determiner in English heads its own phrase, DP, distinct from the NP that is headed by the noun itself. As a third example, Larson (1988) argued that the two complements of a triadic verb in English constitute a phrase oftheir own, specifically a VP headed by a trace ofthe overt verb. (8)a. Mary considers [John overly proud ofhimself7*herseIf]. b. Mary bought lop that fNP picture ofJohn n. c. Mary gave [each worker tv his paycheck]. Small Clauses Functional heads Larsonian shells But if one accepts these kinds of phrase structures, then one is clearly forced to acknowledge that not all phrases can be revealed by traditional phrase structure tests in any simple way. There is sporadic constituency evidence in favor of these innovations, but many ofthe traditional phrase structure tests do not apply to these new phrases. For example. many phrase types in English can be clefted, including DPs, PPs, APs, and CPs. However, one cannot cleft a small clause, the NP complement ofa determiner, or a Larsonian shell: (9)a. *It's [John overly proud ofhimselt] that Mary considers -- . b. "'It's [picture ofJohn] that Mary Dought that --. c. "'It's [each worker tv his paycheck] that Mary gave -. Small Clauses Functional heads Larsonian shells
- 48. PHRASE SlRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 29 If these phrase structure proposals are at all on the right track, then there are many phrases that are not readily detected by movement, even in highly configurational English. The same holds true for pronominalization and ellipsis: these phrases are not readily omitted or replaced by a pronoun. ( lO)a. b. *Mary considers [him smart]. and Bill considers (it) too. *Mary thinks that [John is nice]. and Bill wonders if(it). *Mary answered the third question correctly, and Bill answered the (it) too. Small Clauses Functional heads c. *Mary threw [John tv his jacket] and Bill handed (it). Larsonian shells Why does one get these patterns of facts, with some phrases being cleftable (or elidable, or pronominalizable) and others not? In many cases, this is still not well understood, but it is assumed that there should be an explanation internal to the workings of(say) clefiing. In some cases reasonable proposals have been put on the table. For example. (9a) and (9c) might be out because there is no source of Case checking for the NP at the beginning of the clefied constituent within that constituent; this could lead to a Case theory violation. On the other hand, (9b) might be out because the functional head D or C is not strong enough to license a trace as its complement, according to the Empty Category Principle or its descendents in more recent theory. Similarly, Loebeck (1995) among others proposes substantive syntactic conditions on which phrases can be elided and which cannot. In general. these gaps in the distribution ofclefting and ellipsis are interesting syntactic puzzles, but they do not look especially mysterious or problematic f or the overall framework. Rather. they should be amenable to the usual kind ofsyntactic inquiry. These kinds of issues would need to be f aced by Chomskian theory anyway in the course of pursuing its own agenda, even if there were no language other than configurational English. From this perspective, when the RG critique points out that second position clitics in Warlpiri (Simpson 1991) and clefiing in Malayalam (Mohanan 1982) do not reveal a VP constituent, there is no particular surprise. One does not automatically assume that Warlpiri and Malayalam do not have a VP, but only that the structure of phrases in those languages somehow interacts with the nature ofcliticization or clefting to have this eff ect. This could have as much to do with the nature ofclefting in these languages as it does with phrase structure proper. Now, if we no longer trust traditional constituency tests to reveal to us the basic phrase structure ofa construction in a language in any straightf orward way, how can that phrase structure be determined? There could be many ways, some easier to apply in one language and others easier to apply in another language. But if one is committed to the idea that phenomena like passive. control, and reflexives are structurally determined along the lines summarized in (5), then it is legitimate to
- 49. 30 MARK C. BAKER infer from these phenomenathe phrase structures that are needed to make them work properly. If c1efting and cliticization and word order do not work smoothly from these structures, those are problems to be solved, but not necessarily reasons to hold back. In short, Chomskian syntax plunges ahead and assumes that the structures it needs to work properly are present. Many particular examples of this characteristic approach could be cited. An early example is Chomsky's (1981) discussion ofJapanese, in which he uses [NP, S] terminology that implies a certain kind ofphrase structure for Japanese grammatical functions, even though these do not correspond to a directly observable "D structure" or "S-structure." Hale's (1 983) approach to Warlpiri is similar: he argues that a configurational "L-structure" exists for Warlpiri alongside the nonconfigurational "C-structure" that determines word order and clitic placement. This is similar to the "two independent levels" view of LFG, but with the significant diff erence that f or Hale the representation of grammatical functions (his L-structure) has the f ormal properties of a phrase marker, over which predicates like c-command are defined. Similarly. when I studied noun incorporation in languages like Mohawk in Baker ( l988), I simply assumed that direct objects were the only NPs that were structural sisters of the verb, and used that as the basis f or a structural explanation of why only direct objects can be incorporated. I then took the overall success ofthis account to be evidence that languages like Mohawk do in f act have phrase structures comparable to English, not worrying about how this fit with other issues about word order and constituency in the language (until later: see Baker (1991, 1996c» . Another landmark in this changing attitude toward phrase structure is Larson's (1988) study of double object constructions in English. In these constructions, phrase structure does not seem to reveal any difference between the positions ofthe Prime Minister and a letter in a sentence like I sent the Prime Minister a letter. Both seem to be objects in the sense of being NPs immediately dominated by VP. However, Barss and Lasnik (1986) showed that there are differences between the two for most binding phenomena: the first object can bind the second but not vice versa. Barss and Lasnik look at this apparent conflict between phrase structure and binding theory and ask whether pure c-command is really the right condition on binding. However, Larson takes a markedly different approach: he takes it as established that c-command is the right condition on binding, and uses the binding facts to infer a novel phrase structure for English double object constructions. Thus, for him, GR-related phenomena are (almost) enough to establish phrase structure. In fact, Larson does make an eff ort to show that there is converging evidence for the structures he proposes from traditional phrase structure tests, such as conjunction. But his independent evidence is slender and debatable (see Jackendoff (1 990) vs. Larson (I 990», and it is clear that this is not what is driving his account. Nor did anyone ever arrive at these phrase structures purely by traditional constituency tests. Conceptually rather similar is the work on scrambling and word order in languages like Japanese and ' Hindi inspired by Saito (1 985) and Hoji (I 985), in which configurational structure is inferred indirectly and freedom of word order is derived from movement processes that fit within the internal logic ofthe system.
- 50. PHRASE STRUCTURE AS A REPRESENTATlON OF GRAMMATICAL RELATrONS 3 1 . While many other examples could be given, the last one I will mention is Speas (1990), who considers nonconfigurationality in general, and Navajo in particular. She discusses the basis for inferring that a familiar phrase structure is present in languages in which it is not obvious. Among other things, she clearly makes the point that absence ofevidence for configurational phrase structure is not evidence of absence of configurational phrase structure in the Chomskian ethos. She also proposes some analyses of nonconfigurationality as it is f ound in particular languages. This work might strike the W1sympathetic outside observer as a circular reaction to the RG critique: it assumes that phrase structure is fundamental, rather than grammatical relations, but in practice its evidence f or phrase structure in many languages comes primarily from grammatical relation phenomena. But there is a more useful way to look at the comparison. How did Relational Grammarians establish in practice that a certain NP was a subject at some level ofa clause, and that another NP was an object? Since for them grammatical relations were primitive, there was no automatic way to decide this that was guaranteed to be reliable. One could not be certain, for example, that the subject would be in a particular position in the clause, or that it would have a certain theta role, or that it would bear a certain Case ending. Any ofthese facts might be relevant to identifying the subject in particular languages, depending on the basic linking rules ofth� language, but none ofthem was decisive a priori. Rather, in the best cases, the relational analysis ofthe clause was established by the convergence ofa variety of kinds ofevidence, as an inference to the best overall explanation. How does this compare to the way that contemporary Chomskian approaches establish the phrase structure analysis of a clause? I see the two methodologies as fundamentally the same. Also f or the Chomskian there is no single source of evidence that is guaranteed in advance to work. In particular, data from traditional constituency tests may weJl be relevant, but they are not a priori decisive until one knows how those tests work in the particular language under study. Like the relational analysis, the phrase structural analysis is established by the general pattern of evidence, as an inf erence to the best overall explanation. And that is as it should be in any theory. The best theory is not the one that brings everything into line with its one favorite fact, but the one that finds the greatest degree of harmony and convergence among all the facts. If all this is so, then the fundamental diff erence between contemporary Chomskian theory and classical Relational Grammar is not exactly where one might think, based on casual exposure to the old slogans. Relational Grammar assumed that grammatical functions were universal, foundational to syntax, and could be identified in particular cases (only) by understanding the system as a whole. That was the practical content of their claim that grammatical relations were primitive. Chomskian theory does not differ markedly from RG in these respects. Therefore, I think it is reasonable to say that grammatical relations are, in the relevant sense, "primitive" in Chomskian theory too. The real difference between the two approaches is not so much in the status of grammatical relations as primitive or not, but rather in the kind of representational system that is used to represent grammatical relations. For the relational grammarians, the representation of choice
- 51. 32 MARK C. BAKER was a relational network, whereas f or Chomskian theory the proper representation of "primitive" grammatical relations is a phrase marker. 4. IS PHRASE STRUCTURE A GOOD REPRESENTATION FOR GRAMMATICAL RELATIONS? This does not mean that the diff erences between RG style approaches and current Chomskian theory are small or insignificant. The f act that their main difference is in their representational schemes does not mean that they are mere notational variants. If we look at phrase structures not as replacing grammatical relations but as representing them, the next logical question is whether they constitute a good representation or not. Do the system's important f ormal properties accurately reflect the basic qualities of the thing being represented? If so, then it is a good representation. Or is it an arcane device, the basic topology of which has no theoretical relevance, interfering with the ability to capture generalizations rather than enhancing it? In that case, it is a bad representation system-a relic that preserves some of the idiosyncratic history of the field but is now more confusing and distracting than it is worth. Then it might be a good idea to get rid of it, replacing it with a "cleaned up" representation that was designed to present grammatical relations in language-neutral terms from the start, such as a relational network or an f -structure. One can make a case that phrase structure does in fact happen to be a very good representational system f or grammatical relations, better at capturing their basic nature than its competitors. To see how, compare the RG representation f or a simple English sentence like I persuaded John thai Mary will cheal Bill. These are compared in (I I). (1 1)a. RG relational network P 2 3(?) persuade I John p 2 cheat Mary Bill
- 52. PHRASE SlRUCTIJRE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 33 b. Chomskian phrase structure e IP � I' � INFL vP � NP v' I � I v VP � NP V' I � John V CP I � persuade C IP I � that INFL vP I � will NP v' I � Mary v VP � NP V I I Bill cheat To a substantial extent, these two representations say the same things: that Mary is the sub ject ofthe embedded clause and John its ob ject; that this embedded clause as a whole bears some grammatical relation (indirect ob ject?) to the matrix verb; that I and John are the sub ject and ob ject ofthe matrix clause, respectively. However, they express these relationships in representations that have diff erent topologies. The relational net expresses subjecthood by writing a number l over the nominal that is on the same level as some P symbol that marks the predicate. The Chomskian phrase structure expresses subjecthood by (in this version) putting the nominal in the specifier of a v node that selects a verb phrase. To this degree, the two representations say the same things. But do they say them equally well? There are some relatively minor diff erences worth pointing out immediately. The most obvious one is that the RG representation is (in this case) more succinct and easier to type. All things being equal, that is an advantage to the RG approach. (Although ease of typing and reading was not overall a ma jor priority in the RG f amily of theories, as one can see by looking at the Arc-Pair grammar representations in Johnson and Postal 1980.) The RG representation also clearly
- 53. , 34 MARK C. BAKER abstracts away from language particular details such as word order and morphological marking. As part ofthis. it does not represent various elements ofthe clause that have no obvious significance f or core grammatical relations phenomena, including the tense particles and complementizers. On the other hand, these "minor categories" are present in the Chomskian representation. Indeed, they are full partners in the representation, entering into the same basic structural relations of head, complement, and specifier as any other element. Which is the better choice here could be debated. Proponents ofthe relational net (or LFG's f -structure) could argue that these representations factor the linguistic inf ormation into modular parts, where the elements represented in the relational net have extensive interactions with each other but do not interact extensively with the linguistic elements not represented at this level. On the other hand. suppose that there is no clear modularity in this respect. but rather there are significant interactions between subjects and objects on the one hand and tense and complementizers on the other hand. This state of affairs would tell in favor of the Chomskian representation. which has all of these elements present in the same representation on the same terms. While there are things to be said in favor of both views, I suspect that the implicit claim of the Chomskian representation is more correct on this point. For example, it is clear that the nature of the Inft and Complementizer has important eff ects on how the subject is Case-marked, whether it is controlled or not, and even whether it can raise into the matrix clause, as shown by the f amiliar data in (12). (l2)a. b. c. d. e. I predict that she will win. 1 want PRO to win. *1 predict that PRO will win. I am expected - to win. *1 am expected that - will win. Thus, there do seem to be the kinds of interactions between the "minor" categories of C and IntI and grammatical function phenomena that justify including them in the same representation. If nothing else. there is a psychological/methodological advantage to this: if syntacticians are always required to write these minor categories into their representations, they do not have the luxury of f orgetting that syntactic theory ultimately needs to account f or their properties too. But perhaps the most important property of the Chomskian mode of representation-and the one that I will f ocus on-is the distinctive way that it represents the prominence of one nominal over another. Languages are f ull of asymmetrical relationships, in which element A can be somehow dependent ·on element B but not vice versa. The most theory-neutral word f or this phenomena that
- 54. PHRASE SlRUCTURE AS A REPRESENTATION OF GRAMMATICAL RELATIONS 35 is more or less accepted by all generative approaches is "prominence": we say that B has'prominence over A. Now all the theories that accept the Relational Grammar critique (RO, LFG, I-IPSG, etc.) end up distinguishing two distinct kinds of prominence. First there is prominence internal to a single clause, the prominence that subjects have over objects and objects have over indirect objects. Keenan and Comrie's (1977) Accessibility Hiererachy was one important view ofthis kind. Relational Granunar spoke of the Relational Hierarchy; HPSG calls it the Obliqueness Hierarchy. It is also closely related to the Thematic Hierarchy used in LFG and many other approaches. Whatever the tenninoJogy, the relevant hierarchy expresses the kind of prominence that the matrix subject I has over the matrix object John, and that the embedded subject Mary has over the embedded object Bill in the sentence represented in (1 1). RG in particular built this hierarchy deeply into its tenninology, calling subjects Is, objects 2s, and indirect objects 3s. They also ref erred to it as an "advancement" when passive made an object into a subject or dative shift made an indirect object into a direct object. I will call this phenomenon "Relational Prominence", to coin a tenn that is relatively neutral amongthe variants. These theories also have a second kind of prominence. which I will call "Embedding Prominence." This is a kind of prominence that comes as a result of embedding one clause (or other linguistically complex expression) inside another, which all theories do in one f orm or another to capture the basic recursiveness of natural language. This is the kind ofprominence that the subject ofa matrix clause has over the subject of the clausal complement in a sentence like (1 1), or that the object of a matrix clause has over the object of the clausal complement in such a sentence. Relational and Thematic Hierarchies say nothing about this kind of prominence by their basic nature. Both I and Mary are subjects in ( I I). and both bear an agent thematic role, albeit in different clauses. Theref ore, the two cannot be distinguished by Relationalffhematic Hierarcbies; they are not different in their Relational Prominence. Similarly, both John and Bill are direct objects bearing a theme/patient role in (1 1), so they too tie in terms of Relational Prominence. Any prominence that matrix clause constituents have over embedded clause constituents must be a different kind of prominence altogether. Again RG's tenninology is instructive: the subject ofan embedded clause could become the subject ofa matrix clause as in the RG analysis of raising, but this was not represented by a change of number. Rather, it was a change in what the number was connected to; it was called an "ascension", not a "advancement". These two logically distinct kinds of promi.nence are summarized in (13). (13)a. Relational Prominence, ofsubjects over objects over indirect objects. 1>2>3>. . ., and/or Agent>Theme>Goal>Oblique.... (the prominence ofl over John, Mary overBill in ( I I)) b. Embedding Prominence, ofelements in a matrix clause over elements in embedded clauses. (the prominence ofI and John overMary and Bill in (1 1).)
- 55. Random documents with unrelated content Scribd suggests to you:
- 56. destination. Nous faisons ce que pourrait faire, dans la ville dont nous parlions tout à l'heure, une réunion de citoyens qui viendrait s'opposer aux nouvelles prétentions du conseil municipal. Il nous semble qu'ils pourraient fort bien, et sans inconséquence, formuler ainsi le but précis et limité de leur association: «Tant qu'un droit modéré sur les légumes a fait entrer 20,000 fr. dans la caisse municipale, c'était une question de savoir si ces 20,000 fr. n'auraient pas pu être recouvrés de quelque autre manière moins onéreuse à la communauté. «Cette question est toujours pendante, s'étend à tous les impôts, et aucun de nous n'entend aliéner, à cet égard, la liberté de son opinion. «Mais voici que quelques propriétaires de jardins veulent systématiquement empêcher l'entrée des légumes afin de mieux vendre les leurs; voici que, pour justifier cette prétention, ils émettent une bizarre théorie de l'échange, qui représente ce fondement de toute société comme funeste en soi; voici que cette théorie envahit les convictions de nos concitoyens et que nous sommes menacés de la voir appliquée successivement à tous les articles du tarif de l'octroi; voici que, grâce à cette théorie qui décrédite les importations, les arrivages vont diminuer, jusqu'à affaiblir les recettes de l'octroi, en sorte que nous verrons accroître dans la même proportion les autres impôts: nous nous associons pour combattre cette théorie, pour la ruiner dans les intelligences, afin que la force de l'opinion fasse cesser l'influence qu'elle a exercée et qu'elle menace d'exercer encore sur nos tarifs.» 20.—LE MONDE RENVERSÉ. 18 Avril 1847.
- 57. Un navire arriva au Havre, ces jours-ci, après un long voyage. Un jeune officier, quelque peu démocrate, débarque, et rencontrant un de ses amis: Oh! des nouvelles, des nouvelles! lui dit- il, j'en suis affamé. —Et nous, nous sommes affamés aussi. Le pain est hors de prix. Chacun emploie à s'en procurer tout ce qu'il gagne; l'énorme dépense qui en résulte arrête la consommation de tout ce qui n'est pas subsistance, en sorte que l'industrie souffre, les ateliers se ferment, et les ouvriers voient baisser leurs salaires en même temps que le pain renchérit. —Et que disent les journaux? —Ils ne sont pas d'accord. Les uns veulent laisser entrer le blé et la viande afin que le peuple soit soulagé, que les aliments baissent de prix, que toutes les autres consommations reprennent, que le travail soit ranimé et que la prospérité générale renaisse; les autres font à la libre entrée des subsistances une guerre ouverte ou sourde, mais toujours acharnée. —Et quels sont les journaux pour et contre? —Devine. —Parbleu! le journal des Débats défend les gros propriétaires, et le National le peuple. —Non, les Débats réclament la liberté et le National la combat. —Qu'entends-je? que s'est-il donc passé? —Les mariages espagnols. —Qu'est-ce que les mariages espagnols, et quel rapport ont-ils avec les souffrances du peuple?
- 58. —Un prince français a épousé une princesse espagnole. Cela a déplu à un homme qui s'appelle lord Palmerston. Or, le National accuse les Débats de vouloir ruiner tous les propriétaires français pour apaiser le courroux de ce lord.—Et le National, qui est très- patriote, veut que le peuple de France paye le pain et la viande cher pour faire pièce au peuple d'Angleterre. —Quoi! c'est ainsi qu'on traite la question des subsistances? —C'est ainsi que, depuis ton départ, on traite toutes les questions. 21.—SUR L'EXPORTATION DU NUMÉRAIRE. 11 Décembre 1847. À l'occasion de la situation financière et commerciale de la Grande-Bretagne, le National s'exprime ainsi: «La crise a dû être d'autant plus violente, que les produits étrangers, les céréales, ne s'échangeaient pas contre des produits anglais. La balance entre les importations et les exportations était toute au désavantage de la Grande-Bretagne, et la différence se soldait en or. Il y aurait lieu, à cette occasion, d'examiner la part de responsabilité qui revient au libre-échange dans ce résultat; mais nous nous réservons de le faire plus tard. Contentons-nous de constater aujourd'hui que cette vieillerie qu'on appelle la balance du commerce, si dédaignée, si méprisée, du reste, par certaine école économiste, mérite cependant qu'on y prenne garde; et la Grande-Bretagne, en comparant ce qu'elle a reçu à ce qu'elle a envoyé depuis un an, doit s'apercevoir que les plus belles théories ne peuvent rien contre ce fait très-simple: quand on achète du blé en Russie, et que la Russie ne prend pas en échange du calicot anglais, il faut payer bel et bien ce blé en argent. Or, le blé consommé, l'argent exporté, que reste-t-il à
- 59. l'acheteur? Son calicot, peut-être, c'est-à-dire une valeur dont il ne sait que faire et qui dépérit entre ses mains.» Nous serions curieux de savoir si le National regarde en effet la balance du commerce comme une vieillerie, ou si cette expression, prise dans un sens ironique, a pour objet de railler une certaine école qui se permet de regarder, en effet, la balance du commerce comme une vieillerie. «La question vaut la peine qu'on y prenne garde,» dit le National. Oui, certes, elle en vaut la peine, et c'est pour cela que nous aurions voulu que cette feuille fût un peu plus explicite. Il est de fait que chaque négociant, pris isolément, fort attentif à sa propre balance, ne se préoccupe pas le moins du monde de la balance générale du commerce. Or, il est à remarquer que ces deux balances apprécient les choses d'une manière si opposée, que ce que l'une nomme perte, l'autre l'appelle profit, et vice versâ. Ainsi, le négociant qui a acheté en France pour 10,000 fr. de vin, et l'a vendu pour le double de cette somme aux États-Unis, recevant en payement et faisant entrer en France 20,000 fr. de coton, croit avoir fait une bonne affaire.—Et la balance du commerce enseigne qu'il a perdu son capital tout entier. On conçoit combien il importe de savoir à quoi s'en tenir sur cette doctrine; car, si elle est juste, les négociants tendent invinciblement à se ruiner, à ruiner le pays, et l'État doit s'empresser de les mettre tous en tutelle,—ce qu'il fait. Ce n'est pas le seul motif qui oblige tout publiciste digne de ce nom à se faire une opinion sur cette fameuse balance du commerce; car, selon qu'il y croit ou non, il est conduit nécessairement à une politique toute différente. Si la théorie de la balance du commerce est vraie, si le profit national consiste à augmenter la masse du numéraire, il faut peu acheter au dehors, afin de ne pas laisser sortir des métaux précieux,
- 60. et beaucoup vendre, afin d'en faire entrer. Pour cela, il faut empêcher, restreindre et prohiber. Donc, point de liberté au dedans; —et comme chaque peuple adopte les mêmes mesures, il n'y a d'espoir que dans la force pour réduire l'étranger à la dure condition de consommateur ou tributaire. De là les conquêtes, les colonies, la violence, la guerre, les grandes armées, les puissantes marines, etc. Si, au contraire, la balance du négociant est un thermomètre plus fidèle que la balance du commerce,—pour toute valeur donnée sortie de France,—il est à désirer qu'il entre la plus grande valeur possible, c'est-à-dire que le chiffre des importations surpasse le plus possible dans les états de douane, le chiffre des exportations. Or, comme tous les efforts des négociants ont ce résultat en vue,—dès qu'il est conforme au bien général, il n'y a qu'à les laisser faire. La liberté et la paix sont les conséquences nécessaires de cette doctrine. L'opinion que l'exportation du numéraire constitue une perte étant très-répandue, et selon nous très-funeste, qu'il nous soit permis de saisir cette occasion d'en dire un mot. Un homme qui a un métier, par exemple un chapelier, rend des services effectifs à ses pratiques. Il garantit leur tête du soleil et de la pluie, et, en récompense, il entend bien recevoir à son tour des services effectifs en aliments, vêtements, logements, etc. Tant qu'il garde les écus qui lui ont été donnés en payement, il n'a pas encore reçu ces services effectifs. Il n'a entre les mains pour ainsi dire que des bons qui lui donnent droit à recevoir ces services. La preuve en est que s'il était condamné, dans sa personne et sa postérité, à ne jamais se servir de ces écus, il ne se donnerait certes pas la peine de faire des chapeaux pour les autres. Il appliquerait son propre travail à ses propres besoins. Par où l'on voit que, par l'intervention de la monnaie, le troc de service contre service se décompose en deux échanges. On rend d'abord un service contre lequel on reçoit de l'argent, et l'on donne ensuite l'argent contre lequel on reçoit un service. Ce n'est qu'alors que le troc est consommé.
- 61. Il en est ainsi pour les peuples. Quand il n'y a pas de mines d'or et d'argent dans un pays, comme c'est le cas pour la France et l'Angleterre, il faut nécessairement rendre des services effectifs aux étrangers pour recevoir leur numéraire. On les nourrit, on les abreuve, on les meuble, etc.; mais tant qu'on n'a que leur numéraire, on n'a pas encore reçu d'eux les services effectifs auxquels on a droit. Il faut bien en arriver à la satisfaction des besoins réels, en vue de laquelle on a travaillé. La présence même de cet or prouve que la nation a satisfait au dehors des besoins réels et qu'elle est créancière de services équivalant à ceux qu'elle a rendus. Ce n'est donc qu'en exportant cet or contre des produits consommables qu'elle est efficacement payée de ses travaux. (V. tome V, p. 64 et suiv.) En définitive, les nations entre elles, comme les individus entre eux, se rendent des services réciproques. Le numéraire n'est qu'un moyen ingénieux de faciliter ces trocs de services. Entraver directement ou indirectement l'exportation de l'or, c'est traiter le peuple comme on traiterait ce chapelier à qui l'on défendrait de jamais retirer de la société, en dépensant son argent, des services aussi efficaces que ceux qu'il lui a rendus. Le National nous oppose la crise actuelle de l'Angleterre; mais le National tombe dans la même erreur que la Presse, en parlant de l'exportation du numéraire, sans tenir compte de la perte des récoltes, sans même la mentionner. Le jour où les Anglais, après avoir labouré, hersé, ensemencé leurs champs, ont vu leurs blés détruits et leurs pommes de terre pourries, ce jour-là, il a été décidé qu'ils devaient souffrir d'une manière ou d'une autre. La forme sous laquelle cette souffrance devait naturellement se présenter, vu la nature du phénomène, c'était l'inanition. Heureusement pour eux, ils avaient autrefois rendu des services aux peuples contre ces bons, qu'on appelle monnaies, et qui donnent droit à recevoir, en temps opportun, l'équivalent de
- 62. ces services. Ils en ont profité dans cette circonstance. Ils ont rendu l'or et reçu du blé; et la souffrance, au lieu de se manifester sous forme d'inanition, s'est manifestée sous forme d'appauvrissement, ce qui est moins dur. Mais cet appauvrissement, ce n'est pas l'exportation du numéraire qui en est cause, c'est la perte des récoltes. C'est absolument comme le chapelier dont nous parlions tout à l'heure. Il vendait beaucoup de chapeaux, et, se soumettant à des privations, il réussit à accumuler de l'or. Sa maison brûla. Il fut bien obligé de se défaire de son or pour la reconstruire. Il en resta plus pauvre. Fut-ce parce qu'il s'était défait de son or? Non, mais parce que sa maison avait brûlé.—Un fléau est un fléau. Il ne le serait pas si l'on était aussi riche après qu'avant. «Le blé consommé, l'argent exporté, que reste-t-il à l'acheteur?» demande le National.—Il lui reste de n'être pas mort de faim, ce qui est quelque chose. Nous demanderons à notre tour: Si l'Angleterre n'eût consommé ce blé et exporté cet argent, que lui resterait-il? des cadavres[26]. 22.—DU COMMUNISME. 27 Juin 1847. Les préjugés économiques ne sont peut-être pas le plus grand obstacle que rencontrera la liberté commerciale. Entre hommes qui diffèrent d'opinion sur un point, à la vérité fort important, d'économie politique, la discussion est possible, et la vérité finit toujours par jaillir de la discussion. Mais il est des systèmes si complétement étrangers à toutes les notions reçues, qu'entre eux et la science il ne se trouve pas un terrain commun qui puisse servir de point de départ au débat.
- 63. Tel est le communisme, tels sont les systèmes qui n'admettent pas la propriété, et ceux qui reposent sur cette donnée: que la société est un arrangement artificiel imaginé et imposé par un homme qu'on appelle législateur, fondateur des États, père des nations, etc. Sur ces systèmes, l'observation des faits et l'expérience du passé n'ont pas de prise. L'inventeur s'enferme dans son cabinet, ferme les rideaux des croisées et donne libre carrière à son imagination. Il commence par admettre que tous les hommes, sans exception, s'empresseront de se soumettre à la combinaison sociale qui sortira de son cerveau, et, ce point admis, rien ne l'arrête. On conçoit que le nombre de ces combinaisons doit être égal au nombre des inventeurs, tot capita, tot sensus. On conçoit encore qu'elles doivent présenter entre elles des différences infinies. Elles ont cependant un point commun. Comme toutes supposent l'acquiescement universel, toutes visent aussi à réaliser la perfection idéale. Elles promettent à tous les hommes, sans distinction, un lot égal de richesses, de bonheur et même de force et de santé. Il est donc assez naturel que les hommes, qui ont bu à la coupe de ces rêves illusoires, repoussent les réformes partielles et successives, dédaignent cette action incessante que la société exerce sur elle- même pour se délivrer de ses erreurs et de ses maux. Rien ne peut les contenter de ce qui laisse aux générations futures quelque chose à faire. Notre époque est fertile en inventions de ce genre. Chaque matin en voit éclore, chaque soir en voit mourir. Elles sont trop irréalisables pour être dangereuses en elles-mêmes; leur plus grand tort est de détourner des saines études sociales une somme énorme d'intelligences. Pourtant, parmi ces systèmes, il en est un qui menace véritablement l'ordre social, car il est d'une grande simplicité apparente, et, à cause de cette simplicité même, il envahit les esprits
- 64. dans les classes que le travail manuel détourne de la méditation; nous voulons parler du communisme[27]. On voit des hommes qui ont du superflu, d'autres qui n'ont pas le nécessaire, et l'on dit: «Si l'on mettait toutes ces richesses en commun, tout le monde serait heureux.» Quoi de plus simple et de plus séduisant, surtout pour ceux qu'affligent des privations réelles; et c'est le grand nombre? Ce n'est pas notre intention de réfuter ici ce système, de montrer qu'il paralyserait complétement dans l'homme le mobile qui le détermine au travail, et tarirait ainsi pour tous la source du bien-être et du progrès; mais nous croyons devoir prendre acte de la réfutation décisive qui en a été faite, dans le dernier numéro de l'Atelier, par des hommes qui appartiennent aux classes ouvrières. C'est certainement un symptôme consolant de voir des systèmes subversifs repoussés et anéantis, avec une grande force de logique, par des hommes que le sort a placés dans une position telle qu'ils seraient plus excusables que d'autres s'ils s'en laissaient séduire. Cela prouve non-seulement leur sincérité, mais encore que l'intelligence, quand on l'exerce, ne perd jamais le noble privilége de tendre vers la vérité. Pour beaucoup de gens, le communisme n'est pas seulement une doctrine, c'est encore et surtout un moyen d'irriter et de remuer les classes souffrantes. En lisant l'article auquel nous faisons allusion, nous ne pouvions nous empêcher de nous rappeler avoir entendu un fougueux démocrate, appartenant à ce qu'on nomme la classe élevée, dire: «Je ne crois pas au communisme, mais je le prêche parce que c'est le levier qui soulèvera les masses.» Quel contraste! Une chose nous surprend de la part des rédacteurs de l'Atelier, c'est de les voir s'éloigner de plus en plus de la doctrine de la liberté en matière d'échanges.
- 65. Ils repoussent le communisme, donc ils admettent la propriété et la libre disposition de la propriété, qui constitue la propriété elle- même. Ce n'est pas posséder que de ne pouvoir troquer ce qu'on possède. L'Atelier le dit en ces termes: «Ce que nous prétendons, c'est que la liberté veut et la possession individuelle et la concurrence. Il est absolument impossible de sacrifier ces deux conditions de la liberté sans sacrifier la liberté même.» Il est vrai que l'Atelier ajoute: «Mais est-il possible de limiter les droits de la propriété? Est-il quelque institution qui puisse ôter à la propriété les facultés abusives qu'elle a aujourd'hui? Nous le croyons, nous sommes certains de cette possibilité, comme aussi nous sommes convaincus que la concurrence peut être disciplinée et ramenée à des termes tels qu'elle ait beaucoup plus le caractère de l'émulation que celui de la lutte.» Dans ce cercle, il nous semble que l'Atelier et le Libre-Échange ne sont pas loin de s'entendre, et que ce qui les divise, c'est plutôt des questions d'application que des questions de principe. Nous croyons devoir soumettre à ce journal les réflexions suivantes: On peut abuser de tout et même des meilleures choses, de la propriété, de la liberté, de la philanthropie, de la charité, de la religion, de la presse, de la parole. Nous croyons que le gouvernement ou la force collective est institué principalement, et presque exclusivement, pour prévenir et réprimer les abus. Nous disons presque exclusivement, parce que c'est du moins là sa tâche principale, et il la remplirait d'autant mieux, sans doute,
- 66. qu'il serait débarrassé d'une foule d'autres attributions, lesquelles peuvent être abandonnées à l'activité privée. Quand nous parlons de propriété, de liberté, nous n'en voulons pas plus que l'Atelier les abus, et comme lui nous reconnaissons en principe à la force collective le droit et le devoir de les prévenir et de les réprimer. D'un autre côté, l'Atelier voudra bien reconnaître qu'en fait les mesures répressives, et plus encore les mesures préventives, sont inséparables de dépenses, d'impôts, d'une certaine dose de vexations, de dérangements, d'arbitraire même, et qu'après tout la force publique n'acquiert pas certains développements sans devenir elle-même un danger. Dans chaque cas particulier, il y a donc ce calcul à faire: les inconvénients inséparables des mesures préventives et répressives sont-ils plus grands que les inconvénients de l'abus qu'il s'agit de prévenir ou de réprimer? Ceci ne touche pas au droit de la communauté agissant collectivement, c'est une question d'opportunité, de convenance et non de principe. Elle se résout par la statistique et l'expérience et non par la théorie du droit. Or, il arrive, et c'est sur ce point que nous appelons l'attention du lecteur, qu'il y a beaucoup d'abus qui portent en eux-mêmes, par une admirable dispensation providentielle, une telle force de répression et de prévention, que la prévention et la répression gouvernementales n'y ajoutent presque rien, et ne se manifestent dès lors que par leurs inconvénients. Telle est, par exemple, la paresse. Certainement, il serait à désirer qu'il n'y eût pas de paresseux au monde. Mais si le Gouvernement voulait extirper ce vice, il serait forcé de pénétrer dans les familles, de surveiller incessamment les actions individuelles, de multiplier à l'infini le nombre de ses agents, d'ouvrir la porte à un arbitraire
- 67. inévitable; en sorte que ce qu'il ajouterait à l'activité nationale pourrait bien n'être pas une compensation suffisante des maux sans nombre dont il accablerait les citoyens, y compris ceux qui n'ont pas besoin, pour être laborieux, de cette intervention. (V. Harmonies, chap. XX.) Et remarquez qu'elle est d'autant moins indispensable qu'il y a, dans le cœur humain, des stimulants,—dans l'enchaînement des causes et des effets, des récompenses pour l'activité, des châtiments pour la paresse, qui agissent avec une force à laquelle l'action du pouvoir n'ajouterait que peu de chose. Ce sont ces stimulants, c'est cette rétribution naturelle dont ne nous paraissent pas tenir assez compte les écoles qui, faisant bon marché de la liberté, veulent tout réformer par l'interférence du Gouvernement. Ce n'est pas seulement contre les vices dont les conséquences retombent sur ceux qui s'y livrent que la nature a préparé des moyens de prévention et de répression, mais aussi contre les vices qui affectent les personnes qui en sont innocentes. Dans l'ordre social, outre la loi de responsabilité, il y a une loi de solidarité. Les vices de cette catégorie, par exemple la mauvaise foi, ont la propriété d'exciter une forte réaction de la part de ceux qui en souffrent contre ceux qui en sont atteints, réaction qui a certainement une vertu préventive et répressive, toujours exactement proportionnelle au degré de lumière d'un peuple. Ce n'est point à dire que le Gouvernement ne puisse concourir aussi à punir ces vices, à prévenir ces abus. Tout ce que nous prétendons, et nous ne pensons pas que cela puisse nous être contesté, c'est que cette pression gouvernementale doit s'arrêter et laisser agir les forces naturelles, au point où elle-même a, pour la communauté, plus d'inconvénients que d'avantages. Nous ajouterons qu'un des inconvénients de la trop grande intervention du pouvoir en ces matières, est de paralyser la réaction des forces naturelles, en affaiblissant les motifs et l'expérience de
- 68. cette police que la société exerce sur elle-même. Là où les citoyens comptent trop sur les autorités, ils finissent par ne pas assez compter sur eux-mêmes, et la cause la plus efficace du progrès en est certainement neutralisée[28]. Si ces idées se rapprochent de celles que l'Atelier a développées dans l'article que nous avons en vue, nous ne devons pas être peu surpris du ton d'irritation avec lequel il persiste à s'exprimer sur la liberté du commerce et ce qu'il nomme l'école économique anglaise. L'Atelier est plein de douceur pour les communistes, qu'il vient de combattre et même de terrasser, mais il conserve envers nous les allures les plus hostiles. C'est une inconséquence que nous ne nous chargeons pas d'expliquer, car il est évidemment beaucoup plus loin du communisme que de la liberté du travail et de l'échange. L'Atelier croit la protection plus nécessaire que la liberté à la prospérité nationale. Nous croyons le contraire, et il conviendra du moins que les doctrines sur la propriété et la liberté, qu'il a opposées aux communistes, mettent la présomption de notre côté. Si la propriété est un droit, si la liberté d'en disposer en est la conséquence, la tâche de prouver la supériorité des restrictions, l'onus probandi, incombe exclusivement à celui qui les réclame. Nous n'abandonnerons pas le sujet du communisme sans adresser quelques réflexions aux classes qui tiennent de notre constitution le pouvoir législatif, c'est-à-dire aux classes riches. Le communisme, il ne faut pas se le dissimuler, c'est la guerre de ceux qui ne possèdent pas, ou le grand nombre, contre ceux qui possèdent ou le petit nombre. Partant, les idées communistes sont toujours un danger social pour tout le monde, et surtout pour les classes aisées. Or ces classes ne jettent-elles pas de nouveaux aliments à la flamme communiste quand elles font en leur propre faveur des lois partiales? Quoi de plus propre que de telles lois à semer l'irritation
- 69. au sein du peuple, à faire que, dans son esprit, ses souffrances ont leur cause dans une injustice; à lui suggérer l'idée que la ligne de démarcation entre le pauvre et le riche est l'œuvre d'une volonté perverse, et qu'une aristocratie nouvelle, sous le nom de bourgeoisie, s'est élevée sur les ruines de l'ancienne aristocratie? De telles lois ne le disposent-elles pas à embrasser les doctrines les plus chimériques, surtout si elles se présentent avec le cachet d'une simplicité trompeuse; en un mot ne le poussent-elles pas fatalement vers le communisme? Contre le communisme, il n'y a que deux préservatifs. L'un, c'est la diffusion au sein des masses des connaissances économiques; l'autre, c'est la parfaite équité des lois émanées de la bourgeoisie. Oh! puisque, dans l'état actuel des choses, nous voyons des ouvriers eux-mêmes se retourner contre le communisme et faire obstacle à ses progrès, combien la bourgeoisie serait forte contre ce dangereux système si elle pouvait dire aux classes laborieuses: «De quoi vous plaignez-vous? De ce que nous jouissons de quelque bien-être; mais nous l'avons acquis par le travail, l'ordre, l'économie, la privation, la persévérance. Pouvez-vous l'attribuer à d'autres causes? Examinez nos lois. Vous n'en trouverez pas une qui stipule pour nous des faveurs. Le travail y est traité avec la même impartialité que le capital. L'un et l'autre sont soumis, sans restriction, à la loi de la concurrence. Nous n'avons rien fait pour donner à nos produits une valeur artificielle et exagérée. Les transactions sont libres, et si nous pouvons employer des ouvriers étrangers, de votre côté vous avez la faculté d'échanger vos salaires contre des aliments, des vêtements, du combustible, venus du dehors, quand il arrive que nous tenons les nôtres à un taux élevé.» La bourgeoisie pourrait-elle aujourd'hui tenir ce langage? Ne l'a-t- on pas vue, il n'y a pas plus de huit jours, décréter, en face d'une disette éventuelle, que les lois qui font obstacle à l'entrée des substances alimentaires animales n'en seraient pas moins
- 70. maintenues? Ne l'a-t-on pas vue prendre une telle résolution, sans admettre même le débat, comme si elle avait eu peur de la lumière, là où elle ne pouvait éclairer qu'un acte d'injuste égoïsme? La bourgeoisie persévère dans cette voie, parce qu'elle voit le peuple, impatient de beaucoup d'injustices chimériques, méconnaître la véritable injustice qui lui est faite. Pour le moment, les journaux démocratiques, abandonnant la cause sacrée de la liberté, sont parvenus à égarer ses sympathies et à les concilier à des restrictions dont il n'est victime qu'à son insu. Mais la vérité ne perd pas ses droits; l'erreur est de nature essentiellement éphémère; et le jour où le peuple ouvrira les yeux n'est peut-être pas éloigné. Pour le repos de notre pays, puisse-t-il n'apercevoir alors qu'une législation équitable[29]! 23.—RÉPONSE AU JOURNAL L'ATELIER. 12 Septembre 1847. (Écrite en voyage et adressée à l'éditeur du Journal des Économistes.) Si j'ai eu quelquefois la prétention de faire de la bonne économie politique pour les autres, je dois au moins renoncer à faire de la bonne économie privée pour moi-même. Comment est-il arrivé que, voulant aller de Paris à Lyon, je me trouve dans un cabaret par delà les Vosges? Cela pourra vous surprendre, mais ne me surprend pas, moi qui ne vais jamais de la rue Choiseul au Palais-Royal sans me tromper. Enfin me voici arrêté pour quelques heures, et je vais en profiter pour répondre au violent article que l'Atelier a dirigé contre le Libre- Échange dans son dernier numéro. Si j'y réponds, ce n'est pas parce qu'il est violent, mais parce que cette polémique peut donner lieu à quelques remarques utiles et surtout opportunes.
- 71. Dans un précédent numéro de ce journal, nous avions remarqué cette phrase: «Ce que nous prétendons, c'est que la liberté veut et la possession individuelle et la concurrence. Il est absolument impossible de sacrifier ces deux conditions de la liberté sans sacrifier la liberté elle- même.» Cette phrase étant l'expression de notre pensée, posant nettement les principes dont nous nous bornons à réclamer les conséquences, il nous semblait que l'Atelier était infiniment plus rapproché de l'Économie politique, qui admet, comme lui, ces trois choses: Propriété, liberté, concurrence, que du Communisme, qui les exclut formellement toutes trois. C'est pourquoi nous nous étonnions de ce que l'Atelier se montrât plein de douceur pour le communisme et de fiel pour l'économie politique. Cela nous semblait une inconséquence. Car enfin, à supposer que l'Atelier et le Libre-Échange diffèrent d'avis sur quelques-unes des occasions où l'un peut trouver bon et l'autre mauvais que la loi restreigne la propriété, la liberté et la concurrence; en admettant que nous ne posions pas exactement à la même place la limite qui sépare l'usage de l'abus, toujours est-il que nous sommes d'accord sur les principes, et que nous différons seulement sur des nuances qu'il s'agit de discuter dans chaque cas particulier, tandis que, entre l'Atelier et le Populaire, il y a autant d'incompatibilité qu'entre un oui universel et un non absolu. Comment donc expliquer les cajoleries de l'Atelier envers le communisme, et son attitude toujours hostile à l'économie politique? À cet égard, nous avons préféré nous abstenir que de hasarder des conjectures. Mais l'Atelier nous donne lui-même les motifs de sa sympathie et de son antipathie.
- 72. Ils sont au nombre de trois. 1o Notre doctrine est en cours d'expérience, tandis que celle des communistes est inappliquée et inapplicable; 2o Les économistes appartiennent à la classe riche et lettrée, tandis que les communistes appartiennent à la classe pauvre et illettrée; 3o L'économie politique est l'expression du côté inférieur de l'homme et est inspirée par l'égoïsme, tandis que le communisme n'est que l'exagération d'un bon sentiment, du sentiment de la justice. Voilà pourquoi l'Atelier, fort doucereux envers les communistes, se croit obligé de tirer sur nous, comme il le dit, à boulets rouges et aussi rouges que possible. Examinons rapidement ces trois chefs d'accusation. Notre doctrine est en cours d'expérience! L'Atelier veut-il dire qu'il y a quelque part des possessions individuelles reconnues, et que toute liberté n'est pas détruite? Mais comment en fait-il une objection contre nous, lui qui veut et la propriété, et la liberté? Veut- il insinuer que la propriété est trop bien garantie, la liberté trop absolue, et qu'on a laissé prendre à ces deux principes, bons en eux- mêmes, de trop grands développements? Au point de vue spécial des échanges, nous nous plaignons, il est vrai, du contraire. Nous soutenons que la prohibition est une atteinte à la liberté, une violation de la propriété, et principalement de la propriété du travail et des bras; d'où il suit que c'est un système de spoliation réciproque, des avantages duquel un grand nombre est néanmoins exclu. Quiconque se déclare à cet égard notre adversaire, est tenu de prouver une de ces choses: ou que la prohibition d'échanger ne restreint pas la propriété, aux dépens des uns et à l'avantage des
- 73. autres (ce qui est bien spoliation), ou que la spoliation, au moins sous cette forme, est juste en principe et utile à la société. Ainsi, quant à l'échange, notre doctrine n'est pas appliquée. Et elle ne l'est pas davantage, si l'Atelier veut parler de l'économie politique en général. Non, certes, elle ne l'est pas, de bien s'en faut;—pour qu'on puisse dire qu'elle a reçu la sanction de l'expérience, attendons qu'il n'y ait ni priviléges, ni monopoles d'aucune espèce; attendons que la propriété de l'intelligence, des facultés et des bras soit aussi sacrée que celle du champ et des meulières. Attendons que la loi, égale pour tous, règle le prix de toutes choses, y compris les salaires, ou plutôt qu'elle laisse le prix de toutes choses s'établir naturellement; attendons qu'on sache quel est le domaine de la loi et qu'on ne confonde pas le gouvernement avec la société; attendons qu'une grande nation de 36 millions de citoyens, renonçant à menacer jamais l'indépendance des autres peuples, ne croie pas avoir besoin, pour conserver la sienne, de transformer cinq cent mille laboureurs et ouvriers en cinq cent mille soldats; attendons qu'une énorme réduction dans notre état militaire et naval, la liberté réelle de conscience et d'enseignement, et la circonscription du pouvoir dans ses véritables attributions permettent de réduire le budget d'une bonne moitié; que, par suite, des taxes faciles à prélever et à répartir avec justice suffisent aux dépenses publiques; qu'on puisse alors supprimer les plus onéreuses, celles qui, comme l'impôt du sel et de la poste, retombent d'un poids accablant sur les classes le moins en état de les supporter, et celles surtout qui, comme l'octroi, la douane, les droits de mouvement et de circulation, gênent les relations des hommes et entravent l'action du travail; alors vous pourrez dire que notre doctrine est expérimentée.—Et pourtant, nous ne prédisons pas à la société, comme font beaucoup d'écoles modernes, qu'elle sera exempte de toutes souffrances; car nous croyons à une rétribution naturelle et nécessaire, établie par Dieu même, et qui fait que, tant qu'il y aura des erreurs et des fautes dans ce monde, elles porteront avec elles les conséquences
- 74. destinées précisément à châtier et réprimer ces fautes et ces erreurs. Il y a quelque chose de profondément triste dans le second grief articulé contre nous, tiré de ce que nous appartenons, dit-on, à la classe riche et lettrée. Nous n'aimons pas cette nomenclature de la société en classe riche et classe pauvre. Nous comprenons qu'on oppose la classe privilégiée à la classe opprimée partout où la force ou la ruse, transformées en loi, ont fondé cette distinction. Mais sous un régime où la carrière du travail serait loyalement ouverte à tous, où la propriété et la liberté, ces deux principes proclamés par l'Atelier, seraient respectées, nous voyons des hommes de fortunes diverses, comme de taille et de santé différentes; nous ne voyons pas de classes riche et pauvre. Encore moins pouvons-nous admettre que les riches soient un objet de haine pour les pauvres. Si l'économie politique a rendu à la société un service, c'est bien lorsqu'elle a démontré qu'entre la richesse due au travail et celle due à la rapine, légale ou non, il y a cette différence radicale que celle-ci est toujours et celle-là n'est jamais acquise aux dépens d'autrui. Le travail est vraiment créateur, et les avantages qu'il confère aux uns ne sont pas plus soustraits aux autres que s'ils fussent sortis du néant. Au contraire, il me serait facile de démontrer qu'ils tendent à se répartir sur tous. Et voyez les conséquences du sentiment exprimé par l'Atelier. Il ne va à rien moins qu'à condamner la plupart des vertus humaines. L'artisan honnête, laborieux, économe, ordonné, est sur la route de la fortune; et il faudrait donc dire qu'en vertu de ses qualités mêmes il court se ranger dans la classe maudite! La distinction entre classes riches et classes pauvres donne lieu, de nos jours, à tant de déclamations que nous croyons devoir nous expliquer à ce sujet. Dans l'état actuel de la société, et pour nous en tenir à notre sujet, sous l'empire du régime restrictif, nous croyons qu'il y a une
- 75. classe privilégiée et une classe opprimée. La loi confère à certaines natures de propriété des monopoles qu'elle ne confère pas au travail, qui est aussi une propriété. On dit bien que le travail profite par ricochet de ces monopoles, et la société qui s'est formée pour les maintenir a été jusqu'à prendre ce titre: Association pour la défense du travail national, titre dont le mensonge éclatera bientôt à tous les yeux. Une circonstance aggravante de cet ordre de choses, c'est que la propriété privilégiée par la loi est entre les mains de ceux qui font la loi. C'est même une condition, pour être admis à faire la loi, qu'on ait une certaine mesure de propriété de cette espèce. La propriété opprimée au contraire, celle du travail, n'a voix ni délibérative ni consultative. On pourrait conclure de là que le privilége dont nous parlons est tout simplement la loi du plus fort. Mais il faut être juste; ce privilége est plutôt le fruit de l'erreur que d'un dessein prémédité. La classe qui vit de salaires ne paraît pas se douter qu'elle en souffre; elle fait cause commune contre nous avec ses oppresseurs, et il est permis de croire que, fût-elle admise à voter les lois, elle voterait des lois restrictives. Les journaux démocratiques, ceux en qui la classe ouvrière a mis sa confiance, la maintiennent soigneusement, nous ne savons pourquoi, dans cette erreur déplorable. S'ils agissent en aveugles, nous n'avons rien à dire; s'ils la trompent sciemment, comme il est permis de le soupçonner, puisqu'ils disent que nous avons raison en principe, ce sont certainement les plus exécrables imposteurs qui aient jamais cherché à égarer le peuple. Toujours est-il que la classe ouvrière ne sait pas qu'elle est opprimée et ce qui l'opprime. Aussi, tout en défendant ses droits, comme nous l'avons fait jusqu'ici et comme nous continuerons à le faire, nous ne pouvons nous associer à ses plaintes contre les riches, puisque ces plaintes, portant à faux, ne sont que de dangereuses et stériles déclamations.
- 76. Nous le disons hautement: ce que nous réclamons pour toutes les classes, dans l'intérêt de toutes les classes, c'est la justice, l'impartialité de la loi; en un mot, la propriété et la liberté. À cette condition, nous ne voyons pas des classes, mais une nation. Malgré la mode du jour, notre esprit se refuse à admettre que toutes les vertus, toutes les perfections, toutes les pensées généreuses, tous les nobles dévouements résident parmi les pauvres, et qu'il n'y ait parmi les riches que vices, intentions perverses et instincts égoïstes. S'il en était ainsi, si le bien-être, le loisir, la culture de l'esprit pervertissaient nécessairement notre nature, il en faudrait conclure que l'éternel effort de l'humanité, pour vaincre la misère par le travail, est la manifestation d'un mobile à la fois dépravé et indestructible. Il faudrait condamner à jamais le dessein de Dieu sur sa créature de prédilection[30]. Il ne me reste pas d'espace pour réfuter la troisième accusation formulée contre l'économie politique, celle fondée sur cette assertion, qu'elle est l'expression du côté inférieur de l'homme. C'est, du reste, un vaste sujet sur lequel j'aurai l'occasion de revenir. Parce que l'économie politique circonscrit le champ de ses investigations, on suppose qu'elle dédaigne tout ce qu'elle ne fait pas rentrer dans sa sphère. Mais, sur ce fondement, quelle science ne devrait-on pas condamner? L'économie politique, il est vrai, n'embrasse pas l'homme tout entier; elle laisse leur part de cet inépuisable sujet à l'anatomie, à la physiologie, à la métaphysique, à la politique, à la morale, à la religion. Elle considère surtout l'action des hommes sur les choses, des choses sur les hommes, et des hommes entre eux, en tant qu'elle concerne leurs moyens d'exister et de se développer. Exister, se développer, cela peut paraître aux rédacteurs de l'Atelier chose secondaire et inférieure, même en y comprenant, comme on doit le faire, le développement intellectuel et moral aussi bien que le développement matériel. Pour nous, après ce qui se rapporte aux intérêts d'une autre vie, nous ne savons rien de plus important; et ce qui prouve que nous n'avons pas tout à fait tort, c'est que tous les hommes, sans exception, ne s'occupent guère
- 77. d'autre chose. Après tout, il ne peut jamais y avoir contradiction entre ce que les sciences diverses renferment de vérité. Si l'économiste et le moraliste ne sont pas toujours d'accord, c'est que l'un ou l'autre se trompe indubitablement. On peut réfuter tel économiste, comme tel moraliste, comme tel anatomiste; mais la guerre déclarée à l'économie politique me paraît aussi insensée que celle que l'on ferait à l'anatomie ou à la morale[31]. 24.—RÉPONSE À DIVERS. 1er Janvier 1848. Un journal émané de la classe laborieuse, la Ruche populaire, fait remonter au travail l'origine de la propriété. On est propriétaire de son œuvre. Nous pensons absolument comme ce journal. En même temps, il attaque la liberté d'échanger. Nous l'adjurons de dire, la main sur la conscience, s'il ne se sent pas en contradiction avec lui-même. Est-ce être propriétaire de son œuvre que de ne la pouvoir échanger sans blesser l'honnêteté et en payant l'impôt à l'État? Suis-je propriétaire de mon vin, si je ne le puis céder à un Belge contre du drap, parce qu'il déplaît à M. Grandin que j'use du drap belge? Il est vrai que la Ruche populaire ne donne pas d'autre raison de son opposition au libre-échange, si ce n'est qu'il se produit dans notre pays à l'encontre des journaux indépendants. En cela, fait-elle preuve elle-même d'indépendance? L'indépendance, selon nous, consiste à penser pour soi-même, et à oser défendre la liberté, même à l'encontre des journaux dits indépendants. La même considération paraît avoir décidé une feuille de Lyon et une autre de Bayonne à se mettre du côté du privilége. «Comment ne serions-nous pas pour le privilége, disent-elles, quand nous le voyons attaquer par les journaux ministériels?» Donc, si le ministère
- 78. s'avisait de réformer les contributions indirectes, ces journaux se croiraient tenus de les défendre? Il est triste de voir les abonnés se laisser traiter avec un tel mépris. Mais laissons parler le Courrier de Vasconie: «Il est très-vrai que le Libre-Échange a trouvé pour prôneurs tous les journaux ministériels de France et de Navarre, ce qui prouve, pour nous, une impulsion partie de haut lieu.» Ce qu'il y a de pire dans ces assertions, c'est que ceux qui se les permettent n'en croient pas un mot eux-mêmes. Ils savent bien, et Bayonne en fournit de nombreux exemples, que l'on peut être partisan de la liberté sans être nécessairement ministériel, sans recevoir l'impulsion de haut lieu. Ils savent bien que la liberté commerciale, comme les autres, est la cause du peuple, et le sera toujours jusqu'à ce qu'on nous montre un article du tarif qui protége directement le travail des bras; car, quant à cette protection par ricochet dont on berce le peuple, pourquoi les manufacturiers ne la prennent-ils pas pour eux? pourquoi ne font-ils pas une loi qui double les salaires, en vue du bien qu'il leur en reviendra par ricochet? Les journaux, auxquels nous répondons ici, savent bien que toutes les démocraties du monde sont pour le libre-échange; qu'en Angleterre la lutte est entre l'aristocratie et la démocratie; que la Suisse démocratique n'a pas de douanes; que l'Italie révolutionnaire proclame la liberté; que le triomphe de la démocratie aux États-Unis a fait tomber la protection; que 89 et 93 décrétèrent le droit d'échanger, et que la Chambre du double vote le confisqua. Ils savent cela, et ce sera l'éternelle honte de nos journaux indépendants d'avoir déserté la cause du peuple. Un jour viendra, et il n'est pas loin, où on leur demandera compte de leur alliance avec le privilége, surtout à ceux d'entre eux qui ont commencé par déclarer que la cause du Libre-Échange était vraie, juste et sainte en principe.
- 79. Quant à l'accusation, ou conjecture du Courrier de Vasconie, nous lui déclarons qu'elle est fausse. Le signataire du Libre-Échange affirme sur l'honneur qu'il n'a jamais été en haut lieu, qu'il ne connaît aucun ministre, même de vue, qu'il n'a eu avec aucun d'entre eux la moindre relation directe ou indirecte, que ses impulsions ne partent que de ses convictions et de sa conscience. 25.—À MONSIEUR F. BASTIAT, RÉDACTEUR EN CHEF DU LIBRE-ÉCHANGE. Paris, 25 décembre 1847. Monsieur, Voulez-vous me permettre de répondre quelques mots à l'Avis charitable à la Démocratie pacifique, que vous avez inséré dans votre numéro du 12 de ce mois? «Nous avons toujours été surpris, dit l'auteur en débutant, de rencontrer les disciples de Fourier parmi les membres de la coalition qui s'est formée en France contre la liberté des échanges.» Quelques lignes plus loin, l'auteur cite un fragment d'une brochure que j'ai publiée en 1840, et il veut bien en faire précéder la reproduction des mots suivants: «On a rarement écrit des choses plus fortes, plus pressantes contre le système actuel des douanes.» Après la citation, il ajoute: «Laissons à part la définition de ce que M. Considérant appelle la protection directe..... Le régime des douanes est déclaré anti-social, impolitique, ruineux, vexatoire. L'abolition de ce système fait partie de ce qui, selon le chef des phalanstériens, doit être l'âme de la politique française. On a donc lieu d'être surpris de voir M. Considérant et ses amis se ranger de fait parmi les défenseurs de ce régime; car toutes les fois qu'ils parlent de la liberté des échanges, n'est-ce pas pour la combattre ou
- 80. la travestir? Comment des hommes intelligents peuvent-ils ainsi briser un de leurs plus beaux titres, etc.?» Permettez-moi, monsieur, de vous faire observer que la personne charitable qui voudrait nous tirer de l'abîme de contradiction où elle nous croit tombés, tombe elle-même dans une étrange méprise. Son erreur vient d'une confusion que j'ai vraiment peine à m'expliquer. Il y a, monsieur, trois choses: La question de la protection, celle des douanes et celle de la liberté des échanges. Dans le passage cité de ma brochure, je montre de mon mieux la nécessité d'un système de protection, et j'indique à quelles conditions, à mon tour, ce système peut être bon. Je cherche à prouver que le système douanier est un détestable procédé de protection; j'expose enfin un système de protection directe qui remplacerait très-avantageusement, suivant moi, celui des douanes. Ce système, dont l'auteur de l'avis charitable «laisse à part la définition,» tout en protégeant les industries qui, toujours suivant moi, doivent être protégées, satisfait à toutes les conditions de la liberté des échanges, puisqu'il enlève toute entrave à l'introduction des produits étrangers. Nous reconnaissons donc: 1o La nécessité de protéger le développement de beaucoup d'industries nationales, que la concurrence étrangère anéantirait dans leur marche au travail net; 2o La barbarie du système douanier, au moyen duquel cette protection s'exerce aujourd'hui; 3o L'excellence du système qui protégerait efficacement et directement les industries qu'il convient de soutenir, sans arrêter par des entraves de douane à la frontière les produits étrangers.
- 81. Vous, monsieur, vous ne voulez pas de protection, et vous ne vous élevez pas contre le système douanier. Vous acceptez les douanes, seulement vous voulez qu'elles fonctionnent comme instrument fiscal jusqu'à 20 p. 100, mais non comme instrument protecteur. Nous, nous voulons la protection; mais nous ne la voulons pas par des douanes. Tant que l'on n'entrera pas dans le système de protection directe, nous admettons la douane, en vue de la protection qu'elle exerce. Dès qu'on protégera directement avec une efficacité suffisante, nous demanderons la suppression absolue des douanes, que vous voulez conserver à condition qu'elles ne prélèvent pas plus de 20 p. 100. Vous voyez bien, monsieur, que nous n'avons jamais été d'accord, pas plus en 1840 qu'aujourd'hui. Nous sommes et nous avons toujours été protectionnistes: vous êtes anti-protectionniste. Nous trouvons barbare et détestable le système douanier; nous ne le souffrons que temporairement, provisoirement, comme instrument d'une protection dont vous ne voulez pas, mais à laquelle nous tenons beaucoup.—Vous, vous ne repoussez les douanes qu'autant qu'elles font de la protection au-dessus de 20 p. 100; vous les maintenez pour donner des revenus au trésor. En résumé, nous sommes plus libre-échangistes que vous, puisque nous ne voulons pas même de la douane pour cause de fiscalité; et nous sommes, en même temps, protectionnistes. Vous, monsieur, et vos amis, vous êtes purement et simplement anti-protectionnistes. Les choses ainsi rétablies dans leur sincérité, vous reconnaîtrez, je l'espère, monsieur, que si nous ne sommes pas d'accord avec vous, nous avons du moins toujours été parfaitement d'accord avec nous- mêmes. Agréez, etc.
- 82. Victor Considérant. À MONSIEUR CONSIDÉRANT, DIRECTEUR DE LA DÉMOCRATIE PACIFIQUE, MEMBRE DU CONSEIL GÉNÉRAL DE LA SEINE. Monsieur, Il est certainement à désirer que les hommes sincères, qui ont le malheur de différer d'opinions sur un sujet grave, n'altèrent pas la lettre ou l'esprit de ce qu'il leur convient de citer; sans quoi le public assiste à un tournoi d'esprit au lieu de prendre une part utile à une discussion qui l'intéresse. Ainsi, nous aurions tort, si, en citant le passage où vous flétrissez, avec tant de force et de bon sens, la protection par la douane, où vous faites une analyse si complète des dommages sans nombre que ce système inflige au pays, nous avions dissimulé que vous étiez partisan d'une protection directe, d'une distribution de primes et de secours aux industries qu'il importe d'acclimater dans le pays. Mais nous ne sommes pas coupable d'une telle omission. Il suffit, pour s'en assurer, de jeter un coup d'œil sur l'article de notre numéro du 12 décembre, qui a donné lieu à votre réclamation. D'un autre côté, monsieur, permettez-moi de dire que vous interprétez mal la pensée de notre association, quand vous dites QU'ELLE VEUT la douane fiscale. Elle ne la veut pas, mais elle ne l'attaque pas. Elle a cru ne devoir se donner qu'une mission simple et spéciale, qui est de montrer l'injustice et les mauvais effets de la protection. Elle n'a pas pensé qu'elle pût agir efficacement dans ce sens, si elle entreprenait en même temps la refonte de notre système contributif. Chaque membre de notre association réserve son opinion sur la préférence à donner à tel ou tel mode de percevoir l'impôt. Supposez, monsieur, que certains propriétaires des hôtels du faubourg Saint-Honoré ou de la rue de Lille, s'emparant du Conseil municipal de la Seine, où vous ont appelé votre mérite et les
- 83. suffrages de vos concitoyens, fassent subir à l'octroi un grave changement; qu'ils fassent voter la prohibition du bois à brûler et des légumes, afin de donner plus de valeur aux jardins de ces hôtels. Est-il donc si difficile de comprendre qu'une association pourrait se former ayant pour but de combattre cette énormité, ce fungus parasite enté sur l'octroi, sans néanmoins demander la suppression de l'octroi lui-même, chaque membre de l'association réservant à cet égard son opinion? N'est-il pas sensible qu'il y a là deux questions fort différentes? Supprimer l'octroi, c'est s'engager à supprimer des dépenses ou bien à imaginer d'autres impôts. Cela peut faire naître des opinions fort diverses, parmi des hommes parfaitement d'accord, d'ailleurs, pour repousser l'injustice de messieurs les propriétaires de jardins. Demander, comme vous le faites, la suppression de la douane, c'est demander la suppression de 160 millions de recettes. Si toutes les dépenses actuelles de l'État sont utiles et légitimes, il faudrait donc que nous indiquassions une autre source de contributions; et quoique notre Association compte dans son sein des hommes d'une imagination très-fertile, je doute beaucoup qu'ils pussent trouver une nouvelle matière imposable. À cet égard le champ de l'invention est épuisé. C'est donc à la diminution des dépenses qu'il faudrait avoir recours; mais s'il y a des dépenses superflues dans notre budget pour 160 millions, à supposer que nous réussissions à les éliminer, la question qui se présenterait est celle-ci: Quels sont les impôts les plus vexatoires, les plus onéreux, les plus inégaux? car, évidemment, c'est ceux-là qu'il faudrait d'abord supprimer. Or, quels que soient les inconvénients de la douane fiscale, il y a peut-être en France des impôts pires encore; et quant à moi, je vous avoue que je donne la préférence (j'entends préférence d'antipathie) à l'octroi et à l'impôt des boissons tel qu'il est établi. Nous comprenons que l'État soit réduit à restreindre la liberté, la propriété, l'échange dans un but légitime, tel qu'est la perception de
- 84. l'impôt. Ce que nous combattons, c'est la restriction pour la restriction, en vue d'avantages qu'on suppose à la restriction même. Évidemment, quand on prohibe le drap étranger, non-seulement sans profit pour le fisc, mais aux dépens du fisc, c'est qu'on se figure que la prohibition en elle-même a plus d'avantages que d'inconvénients. J'arrive à la protection directe. Mais avant, permettez-moi encore une réflexion. Vous proposez de supprimer la douane, c'est-à-dire de priver le trésor d'une recette de 160 millions. En même temps vous voulez que le trésor fasse des largesses à l'industrie, et apparemment ces largesses ne seront pas petites, car, pour peu que vous ne mettiez pas de côté l'agriculture, comme il y a plus de 2 millions de propriétaires en France, à 50 fr. chacun, cela passera vite cent millions. Monsieur, il est par trop facile de mettre la popularité de son côté, et de s'attirer les préventions bienveillantes du public inattentif quand on vient lui dire: «Je vais commencer par te dégréver de toutes les taxes, et quand j'aurai mis le trésor à sec, j'en tirerai encore de grosses sommes pour en faire une distribution gratuite.» Ce langage peut flatter la cupidité; mais est-il sérieux? Dans votre système, je vois bien qui puise au trésor, mais je ne vois pas qui l'alimente. (V. tome IV, pages 327 à 329.) Vous croyez indispensable que l'État favorise, par des largesses, certaines industries afin qu'elles se développent. Mais d'où l'État tirera-t-il de quoi faire ces largesses? C'est ce que vous ne dites pas. Du contribuable? Mais c'est lui que vous prétendez soulager. Ensuite, quelles sont les industries qu'il faudra soutenir aux dépens du public? Apparemment celles qui donnent de la perte. Car vous ne voulez pas sans doute que l'État prenne de l'argent dans la poche du menuisier, du maçon, du charpentier, de l'artisan, de
- 85. l'ouvrier, pour le distribuer aux gens dont l'industrie prospère, aux maîtres de forges, aux actionnaires d'Anzin, etc. Mais alors, ces industries ruineuses (devenues lucratives par des largesses du public), je vous demanderai avec quoi elles se développeront. Avec du capital, sans doute. Et d'où sortira ce capital? Des autres canaux de l'industrie où il gagnait sans mettre la main au budget. Ce que vous proposez revient donc à ceci: Décourager les bonnes industries pour encourager les mauvaises; faire sortir le capital d'une carrière où il s'accroît pour le faire entrer dans une voie où il se détruit, et faire supporter la destruction, non par l'industriel maladroit et malavisé, mais par le contribuable. N'est-ce pas exactement les mêmes injustices, les mêmes désastres que vous reprochez avec tant de vigueur à la protection indirecte, quand vous dites: «Chose incroyable que les industries vigoureuses soient toutes immolées aux industries débiles, rachitiques ou parasites!» Entre la protection directe et la protection indirecte, la similitude est telle, quant aux effets, que souvent nous avons cru démasquer celle-ci en exposant celle-là. Permettez-moi de vous rappeler ce que j'en ai dit moi-même dans un petit volume intitulé: Sophismes économiques. Ce passage commence ainsi (V. tome IV, pages 49 et suiv.): «Il me semble que la protection, sans changer de nature et d'effets, aurait pu prendre la forme d'une taxe directe prélevée par l'État et distribuée en primes indemnitaires aux industries privilégiées.» Et, après avoir analysé les effets de ce mode de protection, j'ajoute: «J'avoue franchement ma prédilection pour le second système (la protection directe). Il me semble plus juste, plus économique et plus loyal. Plus juste, car si la société veut faire des largesses à quelques-
- 86. uns de ses membres, il faut que tous y contribuent; plus économique, parce qu'il épargnerait beaucoup de frais de perception et ferait disparaître beaucoup d'entraves; plus loyal, enfin, parce que le public verrait clair dans l'opération et saurait ce qu'on lui fait faire.» Vous voyez, monsieur, que je n'ai pas attendu la lettre dont vous avez bien voulu m'honorer pour reconnaître tous les mérites de la protection directe. Oui, comme vous, et par d'autres motifs, il me tarde qu'on nous prenne notre argent sous une forme qui nous permette de voir où il passe. Il me tarde que chacun de nous puisse lire sur son bulletin de contribution à combien se monte la redevance que nous imposent MM. tels ou tels[32]. Veuillez recevoir, monsieur, l'expression de mes sentiments de considération et d'estime. Frédéric Bastiat.
- 87. 26.—LA LIGUE ANGLAISE ET LA LIGUE ALLEMANDE. Réponse à la Presse. (Journal des Économistes.) Décembre 1845. La Ligue anglaise représente la liberté, la Ligue allemande la restriction. Nous ne devons pas être surpris que toutes les sympathies de la Presse soient acquises à la Ligue allemande. «Les États, dit-elle qui composent aujourd'hui l'association allemande, ont-ils à se féliciter du système qu'ils ont adopté en commun?... Si les résultats sont d'une nature telle que l'Allemagne, encouragée par les succès déjà obtenus, ne puisse que persévérer dans la voie où elle est entrée, alors nécessairement le système de la Ligue anglaise repose sur de grandes illusions... «Or, voyez les résultats financiers... D'année en année le progrès est sensible et doublement satisfaisant: les frais diminuent, les recettes augmentent;... la masse de la population est soulagée,... etc. «Les résultats économiques ne sont pas moins significatifs. De grandes industries ont été fondées; de nombreux emplois ont été créés pour les facultés physiques et pour l'intelligence des classes pauvres; d'abondantes sources de salaires se sont ouvertes; la population s'est accrue; la valeur de la propriété foncière s'est élevée; etc. «Enfin, les résultats politiques se manifestent à tous les yeux,... etc.» Après ce dithyrambe, la conclusion ne pouvait être douteuse.
- 88. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com