<< . .

( 19)

. . >>

(see, among many other references, the discussion in Friedemann and
Rizzi (2000), Rizzi (2000), Wexler (1994, 1998) and the references
quoted there; on the connections between language acquisition, lan-
guage change and creolization in terms of the parametric approach,
see Degraff (1999)).

5 Parametric models and linguistic uniformity
The development of parametric models was made possible by an im-
portant empirical discovery: human languages are much more uniform
than was previously thought. Let us illustrate this point through some
simple examples.

5.1 Overt vs. covert movement
Consider first question formation. Human languages generally take
one of two options to form constituent questions. The option taken by
English (Italian, Hungarian, etc.) consists of moving the interrogative
phrase (who, etc.) to the front, to a position in the left periphery of the
clause; the option taken by Chinese (Japanese, Turkish, etc.) consists of
leaving the interrogative phrase in situ, in the clause-internal argument
position in which it is interpreted (e.g. in (18) as the internal argument
of love):

(17) Who did you meet ?
(18) Ni xihuan shei?
You love who?

On nature and language

Colloquial French allows both options in main clauses:
(19) a. Tu as vu qui?
You have seen who?
b. Qui as-tu vu ?
Who have you seen?
The very existence of only two major options is already an indication of
uniformity. In no known language, for instance, is the question formed
by moving the interrogative phrase to a lower structural position in the
syntactic tree, say from the main clause to an embedded complemen-
tizer position. Moreover, there are good reasons to believe that the
uniformity is even deeper. At Logical Form, an abstract level of mental
representation at the interface with thought systems (on which see May
(1985), Hornstein (1984)), movement seems always to be required, also
in Chinese and colloquial French, giving rise to structures in which the
interrogative phrase binds a clause-internal variable:
(20) For what x, you met/saw/love x?
Important empirical evidence for the idea that movement applies
covertly in these systems was provided by Huang™s (1982) observa-
tion that certain locality constraints hold uniformly across languages.
For instance, an interrogative adverb cannot be extracted from an in-
direct question in English-type interrogatives, a property related to
the operation of a fundamental locality principle, giving rise to viola-
tions which are much more severe and linguistically invariable than
the extraction cases discussed in connection with (11) and (12):

(21) How do you wonder [who solved the problem ]?
For instance, the equivalent of (21) is also strongly excluded in Italian, a
language which rather freely allows extraction of argumental material
from indirect questions, as we have seen:

Editors™ introduction

(22) Come ti domandi [chi ha risolto il problema ]?
How do you wonder who solved the problem?
The constraint violated in (21) and (22) is, according to Huang™s orig-
inal approach, the Empty Category Principle (ECP), a principle giving
rise to stronger and cross-linguistically invariant violations than Sub-
jacency: in a nutshell, the Wh adverb cannot be connected to the em-
bedded clause across another Wh element; see, among many other ref-
erences, Lasnik and Saito (1992), Rizzi (1990, 2000, 2001a,b), Cinque
(1990), Starke (2001) on the different behavior of argument and adjunct
extraction in this environment, and the discussion of locality below.
In parallel with (21) and (22), an interrogative adverb within an
indirect question cannot be interpreted as a main question element in
Chinese-type languages, Huang showed. The parallel is immediately
shown by French: starting from a structure like (23a), a main interrog-
ative bearing on the embedded adverb is excluded, whether the adverb
is moved or not (NB these judgments hold with normal stress contour;
if the interrogative element in situ is heavily stressed the acceptability
improves: see Starke (2001) for a discussion of the relevance of the
stress contour in these cases):
(23) a. Tu te demandes qui a resolu le probl`me de cette mani`re
´ e e
You wonder who solved the problem in this way

b. Comment te demandes-tu qui a resolu le probl`me
´ e ?
How do you wonder who solved the problem?

c. Tu te demandes qui a resolu le probl`me comment?
´ e
You wonder who solved the problem how?
This is immediately explained if speakers of Chinese, colloquial
French, etc. assign Logical Forms like (20) to in situ interrogatives
through covert movement of the interrogative phrase. The same
locality principles apply that are operative in cases like (21) and (22),

On nature and language

barring overt and “mental” movement on a par. So, it appears that, in
abstract mental representations, questions are represented uniformly,
in a format akin to (20); what varies is whether movement to the front
has audible consequences, as in English, or is covert as in Chinese,
etc., a difference expressible through a straightforward parametriza-
tion (e.g. in the feature system of Chomsky (1995a)). A single locality
principle applying on uniform Logical Forms accounts for the ill-
formedness of overt extraction in the English and Italian structures
and for the absence of main clause interpretation in the Chinese
structure, with French instantiating both cases. Analogous arguments
for covert Wh movement can be based on the uniform behavior of
moved and in situ interrogative elements with respect to the possibility
of binding a pronoun (Weak Crossover Effects), an extension of the
classical argument for covert movement in Chomsky (1977: ch. 1).
(See also Pollock and Poletto (2001), who reinterpret certain apparent
in situ cases as involving leftward movement of the Wh element,
followed by “remnant movement” of the rest of the clause to an even
higher position, in terms of Kayne™s (1994) approach; and Watanabe
(1992), Reinhart (1995), Fox and Nissenbaum (1999) for alternative
approaches to covert movement.)
The syntax of questions already looks rather uniform on a su-
perficial analysis, but other aspects of syntax seem to vary considerably
across languages at first glance. What the work of recent years consis-
tently shows is that, as soon as the domain is studied in detail and with
appropriate theoretical tools, much of the variability dissolves and we
are left with a residue of few elementary parameters.

5.2 Adverbs and functional heads
One aspect with respect to which natural languages seem to vary a lot
has to do with the position of adverbials. For instance, certain low

Editors™ introduction

adverbs typically intervene between the verb and the direct object in
French and other Romance languages, while they appear between the
subject and the inflected verb in English:

(24) Jean voit souvent Marie
Jean sees often Marie
(25) John often sees Mary

An elegant and far-reaching approach to this problem was inspired
again by an intuition of uniformity. Perhaps the adverb occupies the
same position in both languages, as is strongly suggested by the fact
that it occurs in a cross-linguistically fixed order with respect to other
adverbs: it must be preceded by negative adverbs like not, must precede
adverbs like completely, etc. What can vary is the position of the verb in
a constant structural configuration: if the sentence contains a T(ense)
specification in between the subject and the predicate VP, in languages
like French the verb moves to T across the adverb (giving rise to a
representation like (26b) derived from underlying structure (26a)),
while in English it remains in its base position (Emonds 1978, Pollock
1989) or undergoes only minimal movement to a lower functional head
(Johnson 1991):

(26) a. Jean T [souvent voit Marie]
b. Jean voit+T [souvent Marie]
(27) John T [often sees Mary]

Once this mode of explanation is adopted in simple cases, it imme-
diately extends to more complex patterns. For instance, the following
paradigm shows that the verb can occupy at least four distinct posi-
tions in French, depending on whether it is inflected or not and on
other properties of the construction (the three positions not occupied
by the verb in a specific example are designated by X):

On nature and language

(28) a. X ne X pas X compl`ment comprendre la theorie (c™est decevant)
e ´ ´
X ˜ne™ X not X completely understand the theory is disappointing

b. X ne X pas comprendre compl`tement X la theorie (c™est decevant)
e ´ ´

c. X il ne comprend pas X compl`tement X la theorie
e ´
X he ˜ne™ understands not X completely X the theory

d. Ne comprend-il X pas X compl`tement X la theorie?
e ´
˜Ne™ understands he X not X completely X the theory?

Under the influential research trend established by Jean-Yves Pollock™s
theory of verb movement (Pollock 1989), all these cases are reducible
to a unique underlying structure, with the lexical verb VP- internal and
adjacent to the direct object it selects, as in (28a). The clausal struc-
ture is conceived of as an array of hierarchically organized functional
heads, the positions indicated by X in (28). These heads may express
tense and other properties of the morphosyntax of clauses, such as
agreement with the subject (following traditional terminology, the
head where agreement is checked is referred to as AGR, but it may
also express other interpretively relevant properties, such as mood,
etc., if “pure” agreement heads are barred, as in Chomsky (1995a)),
and the declarative or interrogative force in the left-peripheral head
C(omplementizer). A general process of head-to-head movement may
or must raise the verb to a higher functional head depending on its
morphological shape and other properties of the structure:

(29) C il ne+AGR pas T compl`tement comprend la theorie
e ´
C he ˜ne+AGR™ not T completely understand the theory

So, in French a non-finite verb may remain in the position of head of
the VP, as in (28)a, or optionally move to a functional head expressing
tense higher than certain adverbs like completely but lower than

Editors™ introduction

negation, as in (28b); a finite verb must raise to the AGR head higher
than negation to pick up agreement morphology, as in (28c) (we fol-
low here the ordering argued for in Belletti (1990)); in questions, the
verb continues its trip to the next higher functional head, the comple-
mentizer (C), to fulfill certain construction-specific well-formedness
requirements, as in (28d).
Different languages exploit the head movement mechanism in
different ways: some never raise the lexical verb out of the VP (English),
others raise finite and non-finite verbs on a par to higher functional
heads (Italian), others systematically exploit the verb movement pos-
sibility to C in a wider range of cases (Verb Second languages), etc. The
patterns are many, varying across constructions and languages, but
they are all reducible to extremely elementary computational mech-
anisms and parameters: a phrase structure consisting of lexical and
functional heads and their phrasal projections, head-to-head move-
ment (also covering different types of incorporation, as in Mark
Baker™s (1988) approach), certain parametrized principles determin-
ing the (partly language-specific) morphosyntactic conditions trigger-
ing head movement.
A major development of this research trend is Cinque™s (1999)
systematic analysis of adverbial positions, leading to a strict universal
hierarchy, which matches the universal hierarchy of functional heads
expressing properties of tense, mood, aspect, and voice. Cinque™s
result also strongly supports the view of a fundamental cross-linguistic
uniformity in this domain up to a very fine-grained level of analysis:
languages vary in the morphological marking of temporal, aspec-
tual and modal properties on the verb, but the rich clausal structure
expressing such properties and hosting adverbial positions is strictly

On nature and language

5.3 Arguments and functional heads
Once this scheme of explanation is adopted to explain diverse and
subtle cross-linguistic properties involving adverbial positions with
respect to verbs, it is natural to extend it to more salient types of varia-
tion, such as the order of verbs with respect to arguments, a classical
topic of typological studies. Consider, for instance, the existence
of languages in which Verb“Subject“Object (VSO) is the dominant
ordering pattern, such as Irish and other Celtic languages (examples
from McCloskey (1996)):

(30) a. Cheannaigh siad teach anuraidh
Bought they a house last year
b. Chuala Roise go minic an t-amharan sin
Heard Roise often this song

The existence of VSO languages has often been regarded as raising
a major theoretical puzzle. In general, a direct object shows a closer
relation to the verb than the subject, which gives rise, for instance, to
frequent V“O idioms (kick the bucket, etc.), to the fact that the subject
is structurally higher than the object, so that a subject can bind a re-
flexive in object position but not vice-versa, etc. These properties are
immediately expressed by the assumption that the verb and the object
form a constituent, the VP, which excludes the subject, the “external
argument” of Williams (1981) (or, in terms of the VP-internal subject
hypothesis of Kuroda (1988), Koopman and Sportiche (1991), these
properties follow from the assumption that the subject is higher than
the object VP-internally). This can be expressed straightforwardly in
S[VO] and S[OV] languages, but what about VSO languages? How can
they fail to express the structural asymmetry between subjects and ob-
jects, and the VP node? By adopting the head movement paradigm, the

Editors™ introduction

VSO order is naturally amenable to standard VP structures, with the
verb adjacent to the direct object in underlying representations, plus
independently motivated movement of the verb to a higher functional
head (Emonds (1980), McCloskey (1996) and the references quoted
there). If the functional head is already filled by an autonomous func-
tional verb, like the auxiliary in (31b) in Welsh, the lexical verb remains
in its VP-internal position (or anyhow in a position lower than the
subject; examples from Roberts (2000)):

(31) a. Cana i yfory
Will-sing I tomorrow

b. Bydda i ™n canu yfory
Will-be I singing tomorrow

Along somewhat analogous lines, Koopman (1983) had ana-
lyzed the word-order alternations in the West African language Vata
(SVO; SAuxOV) in terms of a V-final VP and an I-medial IP, with move-
ment of V to I when the inflection is not expressed by an auxiliary,
determining the SVO order.
This mode of explanation was quickly extended to different lan-
guage families, e.g. to the detailed analysis of the clausal structure in
Semitic (Borer 1995, Shlonsky 1997). Examples of this sort easily mul-
tiply. Even basic variations in head-complement order turned out to be
plausibly reducible to a fixed underlying order plus possible rearrange-
ments (e.g. OV derived by VO plus leftward movement of the object),
an analysis enforced by Kayne™s (1994) Antisymmetry approach.

5.4 Left periphery, DP, and other extensions
Analogous developments were possible in the analysis of the higher
layers of clausal structures, the left periphery of the clause. A variety

On nature and language

of inversion phenomena in main interrogatives (Subject“Auxiliary in-
version in English, Subject“clitic inversion and complex inversion in
French, etc.: see different essays in Belletti and Rizzi (1996)) was
amenable to the same fundamental ingredients: the postulation of
an essentially uniform structure across languages with movement
of the inflected verb to a head position in the C system and movement
of the interrogative phrase to a Specifier position; such cases were then
reduced to construction-specific residues of generalized Verb Second,
a process still fully active in Germanic root clauses, with the notable
exception of Modern English. The study of the left periphery also led to
detailed investigations of dedicated positions for Topic and Focus (Kiss
(1995), Rizzi (1997b), among many other references), preposed adver-
bials and the positions of various types of left-peripheral operators,
again with the uncovering of important elements of cross-linguistic
A parallel trend characterized the analysis of nominal struc-
tures under the DP hypothesis. Originally thought of as the projection
of the lexical head N, ever since the mid 1980s (see Abney™s (1987)
dissertation), the NP started being regarded as the complement of a
functional head, the determiner D, generating its own projection, the
DP. Subsequent studies (Ritter (1991) and references cited there) have
further enriched the functional structure of nominal expressions, with
the identification of several independent layers dominating the lexi-
cal projection NP. The noun phrase then became a complex structural
entity, sharing crucial properties with the functional structure of the
clause. The DP projection could be seen as the periphery of the noun
phrase, a structural zone parallel to the CP projection with respect to the
clause proper (Szabolcsi 1994, Siloni 1997); agreement-related func-
tional projections matched the agreement-related functional skeleton

Editors™ introduction

of the clause. A substantial parallelism between clauses and nominal
expressions emerges, thus embodying intuitions of cross-categorial
uniformity which went back to the very origin of transformational
grammar, but were now expressible within a much more constrained
setting (see Lees™s (1960) approach to nominalization and the critique
in Chomsky (1970)).
Under the DP analysis, various types of cross-linguistic varia-
tion in the nominal system found a natural interpretation: different
distributional properties of adjectival modifiers in different languages
could be partly related to the different scope of N movement, in a way
which significantly paralleled the study of V-Adv orders in the clause
as a function of V movement. The AN order of Germanic languages
and the (prevalent) NA order of Romance languages with the same
class of adjectives could be partly reduced to the lack or shorter scope
of N movement in the former languages (Cinque 1996; see also
Longobardi 1994, Giorgi and Longobardi 1991):

(32) a. The Italian invasion of Somalia
b. L™invasione italiana della Somalia
(33) [ L™[ invasione+X [ italiana t della Somalia ]]]

If the NA order is determined by N movement to a functional head inter-
mediate between N and D (designated by X in (33)), along similar lines,
the ND order of certain languages (Romanian portret-ul “portrait-the”)
plausibly manifests further movement of N all the way to (affix-like)
D (see Giusti 1993, Dobrovie-Sorin 1988 for discussion).
The DP hypothesis also suggests a natural analysis of Romance
pronominal clitics as DPs lacking the lexical restriction, thus captur-
ing the close morphological correspondence to the definite determiner

On nature and language

(for third-person accusative clitics). Clitic constructions therefore may
not involve a peculiar, language-specific category, but, rather, spe-
cial distributional properties (V-relatedness, for Romance clitics) of

<< . .

( 19)

. . >>

Copyright Design by: Sunlight webdesign