Working With Syntax.: A Generative Approach
Working With Syntax.: A Generative Approach
Working With Syntax.: A Generative Approach
Part 2
UNED
1. Preliminaries
Although the examples illustrate different constituents, all of them include the same
nucleus: a noun. The other constituents correspond to other category labels, a tentative
inventory follows:
In order to incorporate all these elements within the nominal phrase, it has been
proposed that the nominal projection is dominated by functional projections. According
to Abney (1987), the first such projection is a Determiner Phrase (DP) projection. The
distribution of the noun with respect to other constituents of the nominal projection
within the DP phrase has been interpreted in terms of overt raising of N to D. Consider
the following examples:
As the examples in (3) illustrate, the common noun “casa” in Italian seems to be in
complementary distribution with the definite article in the presence of a post-
pronominal possessor. The grammaticality of the examples (3a) and (3b) can then be
explained by assuming that the common noun occupies the D position in these two
cases. Since the D position is not the canonical position of Nouns but of Determiners,
the noun ‘casa” is assumed to move into empty Determiner position from its N position,
in these two examples.
On the other hand, it has been claimed that languages may differ with respect to the
movement of N to D. Consider the English equivalent examples in (4).
Unlike “casa” in the previous example (3b), the English constituent “house” cannot
occupy the same D position in its equivalent example in (4b), although the D position is
empty. This shows that there is no similar N to D movement in this language. In fact,
the N-to-D movement strategy seems to be a parametrical issue (Longobardi 1994,
1996).
(10) XP
X’
X’ ADJUNCT
X COMPLEMENT
In fact we can extend the structure in (10) to noun phrases as in the structure in (11):
(11) NP
SPEC N’
N’ ADJUNCT
N COMPLEMENT
In the syntactic structure in (11), the NP is a single endocenctric projection with one
unique head: N. According to this phrase structure, Determiners will occupy the SPEC
position. However, this analysis cannot account for other examples like those mentioned
above, now again in (12), in which there are two constituents preceding the Noun head,
as in (12a), or other genitive NPs preceding the noun head, as in (12b):
Furthermore, if we analysed demonstratives and articles like pure Specifiers, they would
be optional elements of the nominal projection. However, it is well known that bare
nouns are not always possible, as Alexiadout et al (2007) argue providing the following
examples (Alexiadout et al, 2007: 11, p.63):
The ungrammaticality of the examples in (13) shows that bare nouns cannot occupy
thematic positions. In contrast, they can be subjects or objects of verbs or of
prepositions provided that they appear together with a definite or indefinite article:
Provided these facts, Alexiadout et al (2007) argue that “the presence of an article
appears to be indispensable with noun phrases that function as arguments of verbs”.
Some apparent counterexamples however exist, as these authors also point out with the
following examples (Alexiadout et al, 2007: 12, p.64):
Alexiadout et al, 2007 observe that nouns like wine, beer and cheese belong to the
subclass of mass nouns, whereas flowers and cats are plural nouns. As an exception,
then, both mass nouns and bare plurals can appear in thematic positions without being
introduced by an article.
The role of the article introducing a singular noun is to denote the existence of that
particular noun in the world. In this sense, the article provides a particular referential
reading of the noun it introduces. That is there is just one referent for each determiner.
Abney (1987:77) claims:
“The function of the determiner is to specify the reference of a noun phrase. The noun
provides a predicate, and the determiner picks out a particular number of that
predicate’s extension. Apparently, the (definite) article enjoys a privileged position
among all those elements that constitute the class of determiners, as we shall see
shortly. However, the key question is how from this fact, namely the ability of the
article to pass referentiality onto the noun it modifies, the conception of the category
DP, as a superordinate category to NP, emerged.”
Abney (1987) puts forward the functional phrase DP as responsible for the referential
status of the article. In this view, the article plays both a syntactic and a semantic
function and is analyzed as a head that projects its own (functional) category DP.
---EJERCICIOS ---
3. The DP hypothesis
(1) NP
The N’
wet N’
red N’
soft N’
spongy ... N’
N (complement)
ball
1
This example has been extracted from the paper "Introduction to Layers in DP" by
Roberto Zamparelli, May 3, 1996 (see website:
http://www.cogsci.ed.ac.uk/~roberto/layers/dp-hyp.html).
The main reason to make a distinction between the position of all different
elements to the left of N was the fact that in English they cannot co-occur provided the
ungrammaticality of the following examples (Alexiadou et al, 2007: 46):
(3) DP
Spec D’
D NP
the Spec N’
N (Complement)
ball
In this new configuration, we have at least three different positions for elements to
the left of N. In the following sections we will discuss the linguistic data in English and
other languages that support this analysis.
First of all, we may distinguish between articles and demonstratives according to their
semantic interpretation: whereas the former may be definite or indefinite as we saw
before. Demonstratives, as Lyons (1999) points out, are always definite:
As for their syntactic distribution, in contrast to what the English data may suggest,
provided the example in (2b) above, there is cross-linguistic evidence in favour of the
fact that the demonstrative and the determiner do not occupy the same structural
position in many languages where both elements may co-occur (Alexiadou et al 2007:
59):
Alexiadou et al 2007 also discuss further syntactic facts in English in order to make an
additional distinction between demonstratives, on the one hand, and articles, on the
other hand. First, there is a similar semantic interpretation between the demonstrative
“this” and the modifier “such” in examples such as the ones provided in (5)
(Alexiadou et al 2007: 63a):
In order to account for the differences between the demonstrative and the article one
may conclude that they occupy two different positions. Then we may assume that
demonstratives occupy the same position as degree modifiers like “such” or “so” in
SpecDP, whereas articles like “a” occupy the D position as in Alexiadou et al 2007: 65:
(7) DP
Spec D’
such D NP
this a Spec N’
that N (Complement)
reaction
In the following sections of this chapter, we will examine the position of the other
different elements to the left of the noun illustrated in the examples provided in the
introduction: quantifiers, empty determiners, and genitive possessive NPs.
Given our analysis so far, we know that demonstratives occupy the SpecDP position,
the question we want to address in this section is the exact position for English
quantifiers in this type of examples.
On the assumption that Quantifiers project their own functional category the first
question we want to address is whether QP occupies a higher or lower position within
the DP. Consider the following examples:
All the parcels Susan received
The many parcels Susan received
Whereas the quantifier “all” appears to the left of the Determiner in the first example,
the quantifier “many” appears to the right of the Determiner in the second example.
Each syntactic distribution could be derived from the following two phrase
representations:
QP
Spec Q’
Q DP
All Spec D’
D NP
the parcels
DP
Spec D’
D QP
The Spec Q’
Q NP
many parcels
One could argue that both structures exist in one language provided that the lexicon of
this language contains two types of quantifiers that project in a very different fashion.
However, as Giusti (1991) argues a more detailed analysis of “many” in an example like
“The many parcels Susan received” shows that it is not a quantifier but an adjective and
as such it can be analyzed as an adjunct modifier, sister to N’ to the left or to the right.
Recall our previous discussion on the difference between complements and adjuncts,
only the latter are able to show recursively.
In fact, Giusti (1991) provides two solid arguments to suggest that many/few and
numerals can function either as adjectives or as quantifiers and can accordingly appear
to the left or to the right of D.
The second argument in favour of a similar behavior of many/few and numerals with
respect to adjectives when they follow a Determiner lies on the following contrast of
grammaticality (Giusti 1991, 15):
a *I had already met the many you introduced to me last night
b *I had already met the nice you introduced to me last night
c I have already met many
d *I have alredy met nice
On the assumption that “many” is a real quantifier when it is not preceded by D
explains why it doesn’t behave like an adjective phrase in the examples above.
The process of N-movement onto the functional head occupied by the affix-article can
be clearly adopted under the DP-hypothesis as follows:
DP
D’
D NP
-en
N’
N
gutt
For Italian, Longobardi (1994) extends the N-To D analysis to account for proper
names, on the basis of the following data:
a) il mio Gianni
the my John
b) Gianni mio
John my
‘my John’
c) *mio (il) Gianni
In effect, in example a) the article il (`the`) and the proper name Gianni co-occur, the
possessive adjective mío (‘my’) intervening between the two. In contrast there is no
definite article in example b) and the proper noun Gianni appears to the left of the
possessive adjuective mio. Note that an initial possessive adjective is in complementary
distribution with N in example c). As we have argued in our previous discussion on
adjectives and quantifiers, possessive adjectives are required to appear adjoined to N,
not to D. The generalization we can draw from this observation is that in the absence of
the article, D is still there but it is not realised phonologically. According to this
analysis, the grammatical examples above undergo the following analyses:
a) DP
D’
D NP
il
AP N’
mio
N’
N
Gianni
b) DP
D’
D NP
Gianni
AP N’
mio
N’
In all the English examples above are used generically. However, as has been observed
in the literature ( Krifka et al, 1995; Carlson 1977; Alexiadou et al 2007 and all
references therein) there is considerable crosslinguistic variation of how genericity is
displayed in the nominal system.
In Spanish, for example, the equivalent of a) does not exist. The generic reading can
exclusively be given through the presence of a determiner (d vs –e,f):
d) *Paquetes son regalos bienvenidos
e) El paquete es un regalo bienvenido
f) Un paquete es un regalo bienvenido
In Spanish, bare plurals have a more restricted distribution than in English. The same
holds for other languages like Greek where generic noun phrases require the presence of
a definite article for both the plural and the singular form (Alexiadou et al , 2007,
p.177:16 ) .
a) I ghates ine aneksartita zoa
“The cats are independent animals”
b) I ghata ine aneksartito zoo
“The cat is an independent animal”
c) *Mia ghata ine aneksarito zoo
“A cat is an independent animal’
Since the determiner in these languages seem to satisfy the Referential character of the
noun we can understand why it occupies the D position, which as we have seen in our
previous subsection can also be occupied by proper names which are referential
expressions by definition. In contrast, when the D position is empty the indefinite
reading of bare noun phrases follows straightforwardly, on the assumption that they stay
in their canonical position in N, in cases such as the following.
a) *Susan likes present
b) *Present arrived
On the other hand, we have argued that in English the definite article does not normally
occur before proper names and generic nouns. Recall our previous example:
Parcels are welcome presents
So the question is how bare plurals get their generic reading in this type of examples.
Diesing (1992) following Wilkinson (1986) and Gerstner & Krifka (1987) assumes that
the generic reading associated with a nominal expression is due to the presence of an
abstract generic operator at the sentential level. Likewise, we can assume that the
generic operator is realised by empty D. So, when bare plurals trigger the generic
reading they undergo an N-to-D-movement operation, similar to the one undergone by
proper names, in the absence of a phonological article. In this sense, we can conclude
following Alexiadou et al (2007: 183) that ‘proper names are like generic nouns in that
both noun types are used to denote directly; the former directly denote an individual, the
latter directly denote a class.’ This assumption also follows Longobardi’s (1994) theory
since this linguist assumes that a language that exhibits bare proper names like English
or Spanish necessarily allows generic readings for bare plurals or mass nouns
(Longobardi 2001) .
On the other hand, we also have to account for mass nouns that trigger an
existential reading, not a generic one. Consider the following examples both in English
and in Spanish:
In both examples above the Determiner is phonologically empty and we cannot trigger
any generic reading in contrast to the bare plural nouns we have recently examined with
our previous examples. Before we assumed that bare nouns like proper nouns undergo
N-to-D movement to trigger the generic reading. Following Longobardi’s theory we
propose that null determiners are interpreted existentially:
[D 0] has a default existential interpretation (Longobardi 1994: 64)
Then, there is one way to distinguish between bare plurals triggering a generic or an
existential interpretation provided their syntactic distribution inside the DP.
(3) DP
Spec D’
D NP
‘s
Since the D position is already filled with ‘s, we can explain why articles are
impossible, as expected:
Just as we analysed the Norwegian and Romanian data, we could assume that the proper
name undergoes N-to-D movement, and as a result the ‘s appears enclitic to the proper
noun, as follows:
DP
D’
D NP
‘s
N’
N
Susan
(4) DP
John’ D’
s books
Since the morpheme ‘s and determiners are in complementary distribution, they should
be represented under the same node (D).
Chapter 4
Verb Phrase Structure
4.1 Introduction
The existence of a VP constituent in English was prefigured in the immediate
constituent analysis of Bloomfield (1933), and became explicit in early generative work
(Chomsky 1965), The VP constituent, along with NP, became a model for constituent
structure in X-bar theory (Chomsky 1970, Jackendoff 1977) and has been a cornerstone
of most analyses of English to this day: Minimalist Program (Chomsky 1995), Lexical
Functional Grammar (Bresnan 2001), Head-Driven
Phrase Structure Grammar (Pollard & Sag 1994), The Cambridge Grammar of the
English Language (Huddleston & Pullum et al 2002). Standard arguments for the
constituency of VP (ellipsis, pro-forms, displacement, coordination) recur in most
syntax textbooks alongside those for the constituency of noun phrases.
As explained above, any phrase is headed by a nucleus in the generative
grammar framework. In the case of the verb phrase structure, the verb phrase is headed
by a verb. However the other elements that appear inside the verb phrase may vary. If
the verb phrase may be constructed from a single verb; the verb phrase will, actually,
often consist of various combinations of the main verb and any auxiliary verbs, plus
optional specifiers, complements, and adjuncts. For example, consider the following
sentences:
Example (1a) contains the verb phrase start practicing for the upcoming season.
Example (1b) contains the main verb paint, the determiner phrase (DP) complement the
house, and the prepositional phrase (PP) adjunct on Tuesday. Additionally, example
(1c) contains the main verb gave, and two determiner phrases Peter and his father, both
selected by the verb in this case.
Note that according to this definition, the verb phrase corresponds to what is commonly
called the predicate.
Up to the mid/late 1980s, it was thought that some languages lacked a verb phrase.
These included languages with extremely free word order (so-called non-
configurational languages, such as Japanese, Hungarian, or Australian aboriginal
languages), and languages with a default VSO order (several Celtic and Oceanic
languages). The current view in some varieties of generative grammar (such as
Principles and Parameters) is that all languages have a verb phrase, while others (such
as Lexical Functional Grammar) take the view that at least some of these languages do
lack a verb phrase constituent.
This more narrow definition is often applied in functionalist frameworks and traditional
reference grammars. However this view goes against the generative theory of the verb
phrase, since the underlined strings are not constituents independent form their
complements under standard generative analyses. For example, consider the analysis of
the example (1b) above as follows:
VP
V’ PP
V NP P NP
paint on
DET N’ N’
the
N N
house Tuesday
This sentence is ambiguous depending on the status of the PP with guns. Under one
interpretation, the VP consists of an DP complement (the fugitives) and a PP (with
guns), while under a second interpretation both the DP and the PP can function as a
complement forming a single constituent. Look at the two corresponding
representations in (5a) and (5b) below:
(5) a. VP
V’ PP
V NP P’
shoot
the fugitives P NP
with
N’
N
guns
b. VP
V’
V NP
shoot
N’ PP
the fugitives
P’
P NP
with
N’
guns
In other words, we need to find out a mechanism that can block out the type of
structures as in (7).
Firstly, in the lexicon we can restrict the number of arguments of a verb in the
same way we may restrict the pronunciation, the syntactic category, and the meaning of
any word. Although language allows recursion, the lexicon fixes the number of
syntactic arguments for each verb and when one learns a new word one should also
memorize this. Some linguists have used this information to establish different classes
of verbs matching one specific theta grid. This is, for example, much of the research
work found in Levin (1993) and Faber & Mairal (1999). We can assume that theta-grid
information is encoded as part of the lexical entry of the predicate. Laugh and receive in
the examples above would then be given the lexical representations in (8a) and (8b)
respectively:
(8) a) laugh: verb
Agent
i
b) receive: verb
Benefactive Theme
i J
As marked by the slot distinction, unlike receive, laugh lacks a theme argument.
This explains why the sentences *John laughed a present and *He received above are
both ungrammatical.
In addition to theta-grid, this chapter also deals with the semantic relations
existing between the constituents of a sentence. The discussion begins with the analysis
of two main units of sentences: the predicate and the arguments. As Castillo (2003: 163)
puts it:
The term predicate has a syntactic meaning, and a semantic or interpretive meaning;
predicate is the central part or head of a proposition. As for the arguments of a
proposition, these are the participants in the event expressed by the predicate itself.
For a brief explanation of most common thematic relations found with the
arguments of a verb, consider the following:
Agent: Animate, volitional. Initiates action.
Patient: Animate or inanimate. Undergoes (and is affected by) action
initiated by Agent.
Experiencer: Animate. Undergoes a perceptual experience.
Theme: Animate or inanimate. Undergoes motion, or an action that does
not affect it significantly.
Recipient: Generally animate: one who receives something.
Benefactive: Generally animate: one who benefits from the event.
Goal: Animate or inanimate. The endpoint of the action.
Location: The place where the event occurs.
Source: Animate or inanimate. The starting point of an action.
Instrument: Often inanimate. Used in carrying out an action.
At first sight, thematic roles usually have a regular mapping with respect to
structure. On an active sentence agents are usually linked to a subject position, and
patients are linked to an object position, but this is not always the case. As stated in the
book, there is no one-to-one correspondence between thematic roles and arguments.
That is the thematic role theme may either be related to the subject or to the object.
In any case, arguments always require to be marked with a theta role. Adjuncts, on
the other hand, don’t, and therefore they are not listed in any verb’s lexical entry. As a
common notational device, adjuncts are represented within brackets:
Other lexical restrictions that affect the syntactic configurations are reflected in
the so-called Projection Principle which assures that all lexical requirements be
satisfied in the syntax at all levels. The Projection Principle entails that sentence (10)
cannot have the structure in (11).
The DP him is the subject of the embedded sentence at the level of D-structure, so
it has to be analyzed as its subject at all syntactic levels of representation, even though it
has objective case.
In addition, one has to capture the fact that there are special classes of predicates
which have no lexical (semantic) arguments, like most weather verbs, e.g. snow, rain ,
and those predicates that optionally take an embedded sentential subject, e.g be likely.
For both types of predicates the proposal is that they take what is called a pleonastic
pronoun (also called expletives) which exceptionally don’t take any theta role. In
connection with this, the theory calls for the Extended Projection Principle (EPP)
which extends the Projection Principle discussed above with the requirement that
clauses have subjects (cf. Chomsky 1981 and Chomsky 1986 for a detailed discussion
of these two principles).
The EPP then turns out to be a mechanism which is exclusively internal to the
syntax of English. In fact, this principle guarantees the well-formedness of certain
structures, otherwise ruled out. In the case under study, the lack of an expletive pronoun
turns the example (12) into an ungrammatical sentence:
One should be careful when dealing with expletive or pleonastic it and distinguish
it from the neuter pronoun it which is always associated with a theta role. Contrast the
example in (12) with the examples in (14).
In these new examples, the neuter pronoun appears in the subject, as in (14a), in
the object, as in (14b), and in the ndirect object position, as in (14c), with three different
classes of verbs above. A way of encoding the lexical entries for the intransitive,
transitive and ditransite verbs in (14) is given in (15) as follows:
In sum, at this point we can affirm that the lexicon is in charge of establishing a
number of important lexical distinctions and, in addition, the computational component
is responsible for the trigger of special mechanisms to guarantee the grammaticality of
certain configurations. Thus, we may conclude that the lexicon and the computational
component cannot go separately.
To begin with, we will have to explain why verbs may be separated form each other by
intervening adverbials as in the example we are discussing in (16), which is not
expected if the cluster simply consists of stacked V heads.
(16) The new theory certainly may possibly have indeed been badly formulated
(Quirk et al. 1986: § 8.20, p. 495)
Then we will have to look at a strict adjacency requirement in English for the verbs in
the cluster with respect to their selected complements provided the following minimal
pair:
The simplest way of giving an account of the grammatical auxiliaries preceding the
“adverb-V” cluster in examples such as in (19) is to assume the presence of an
independent functional head dominating V. This functional head is responsible for the
checking of the inflectional features of the verb (has/is vs. have/are) and therefore is
assumed to project an Inflection Phrase (IP)
(21) TP
DP T’
She
T VP
is/has
V’
V’ PP
V DP at night
watching/watched
TV
In the analysis above, the functional head I takes a VP complement headed by the verb
form. When the syntactic structures are passed on to the morphological component, the
tense-verb combinations are simply spelled out appropriately according to the
corresponding morphological rules.
The verb “gave” in example (22) is associated with three semantic arguments (agent,
recipient, theme). In traditional grammar, the terms “transitive”, “intransitive” and
“ditransitive” make reference to the possibility that each verb has with respect to the
number of arguments they may have. Previously, we saw that the theme and the
recipient, for example, are taken to be the verb's direct and indirect object, respectively.
In this chapter, we argue in favour of a VP-shell as a way to make double-object
sentences consistent with the binary-branching hypothesis. The proposal hinges on the
fact that ditransitive verbs like “give” can be semantically decomposed into a causative
part and a remainder whose meaning differs according to the verb in question. More
examples follow:
The decomposability of the verbs in (23) suggests deriving sentences like (22) from
schematic structures of VP-shells like (24), where there are tw abstract verbal heads (V)
that represent two abstract relations: CAUSE and GET, respectively.
(24) VP
DP V’
agent
V VP
CAUSE
DP V’
recipient V DP
GET
theme
In effect, according to the structure in (24) the three arguments, as in the original
sentence in (22), are related between temselves in the same VP thanks to the presence of
two V heads, neither of which needs to be associated with more than one complement,
just as required by the binary-branching hypothesis. The assumption is that the verb
“gave” is based generated in the lower Verb node and further raises into the higher Verb
node as an instance of abstract verb movement.
The structure DP can then be well accommodated for examples such as (22). However,
we know that there are other important cases in English, where there are not only DPs
but PPs functioning as arguments. Consider the case where the recipient argument is
expressed by a PP as follows:
Besides its consistency with the binary-branching hypothesis, the VP-shell hypotheis
can also be extended to account for this second word-order possibility of double-
complement structures as we will be arguing in the following section.
Apparently, direct and indirect objects freely alternate inside the English VP phrase.
Interestingly, however, recipients preceding themes seem to be constrained to be
animate, given the following minimal pairs:
On the assumption that verbs like “give” and “send” underlie a VP-shell, this second
puzzle can be solved.
Suppose that double-complement sentences with double DPs are not completely
synonymous with their double-object counterparts with a DP (object) and PP (indirect
object), then we could derive both word orders independently. In fact there is
independent evidence to believe that there is no a one-to-one correspondence between
thematic DPs and thematic PPs, given the ungrammaticality of the examples (28b) and
(29b). On the assumption that the first thematic DP has to be a recipient, the contrast of
grammaticality follows straightforward since recipients are by definition animate.
On the other hand, if thematic PPs in the previous double-object configuration are
recipients or locations, the previous data are well captured. In this analysis, the tree
representation for previous examples (26a), (27a), (28a) and (29a) is as follows:
(30) a. VP
DP V’
them
V PP
GO
P’
P DP
to
recipient
b. VP
DP V’
them
V PP
GO
P’
P DP
to
location
In effect, if we look at the argument arrays for the abstract predicate GO we see that
there is a double possibility for its arguments:
Given the two structures for recipients as a DP in (24) above and as a PP, now in (30a),
the acceptability of both object-complement configurations in (26) and (27) follows
straightforwardly. Moreover, since a single syntactic structure can be associated with
more than one argument array, the recipient argument in a {theme, recipient} argument
array can occupy the same structural position as the location, given the two
configurations in (30a) and (30b). Hence the grammaticality of the examples in (28a)
and (29a), in contrast to the locative Note that a locative adverb like there
EXERCISES