3    X' theory: Lexical categories

Before proceeding, let us clarify the meaning of the term 'lexical,' which is used with several different, potentially confusing meanings in syntactic theory. In the collocation lexical item, 'lexical' means 'pertaining to the (mental) lexicon,' the collection of morphemes in a language. In the collocation lexical projection, 'lexical' is a synomyn for 'zero-level.' In this sense, the term stands in opposition to 'intermediate' and 'maximal/phrasal,' as discussed in Chapter 2. The term 'lexical' also occurs in the collocation lexical category, where it has two meanings. First, it may be opposed to 'functional.' In this sense, the lexical categories include N, V, A, and P, as just mentioned, whereas the functional categories, which are discussed in Chapter 4, include D(eterminer), I(nflection), and C(omplementizer), among others. The distinction between lexical and functional categories roughly corresponds to the distinction in traditional grammar between content words and function words and to the distinction in psychology between open-class items and closed-class items. Second, 'lexical' in the collocation 'lexical category' may be opposed to 'silent' or '(phonologically) empty.' In this sense, lexical categories are syntactic categories with phonological content (that is, they are pronounced). Empty categories, by contrast, are necessary from the point of view of syntactic structure, as we will see later on in the course, but they are not pronounced.

The X' theory of phrase structure

(1)	a.		The army destroyed the city.
	b.		the army's destruction of the city

If we treat destroyed and destruction as the heads of (1a) and (1b), then in both cases, the expressions preceding and following the head (the army('s)) and (of) the city) express the agent and patient, respectively, of the destruction.

The parallelism in (1) is supported by the correspondence between the passive sentence in (2a) and its passive-like noun phrase counterpart in (2b).

(2)	a.		The city was destroyed by the army.
	b.		The city's destruction by the army

In both examples in (2), it is the pre-head expression the city('s) that expresses the patient of the destruction and the post-head expression by the army that expresses the agent.

The classic X' schema

(3)

As with verbs, three projection levels are distinguished: heads, intermediate projections, and maximal projections. The daughter of XP that is not part of the spine (here, ZP) is the specifier of XP, which is standardly abbreviated as Spec(XP) (read as 'speck XP'). The daughter of X' that is not part of the spine (here, YP) is the complement of X. As mentioned earlier in connection with verbs, elementary trees have at most one specifier. Specifiers are obligatory for verbs, but optional for the other lexical categories. In principle, any elementary tree can contain more than one complement, but verbs are the only lexical category for which this has been argued to be the case. The other lexical categories have at most one complement, and it has been argued that phrase structure is constrained to be binary-branching in general (Kayne 1984).

Given the general schema in (3), the intuitively parallel expressions in (1) can now be represented by the topologically parallel trees in (4). (For the moment, we ignore (2), because (2a) involves a complication that we don't cover until we discuss functional categories in Chapter 4.)

(4)

a.

b.

Although the two structures in (4) differ in the category labels of the individual nodes, the important point is that they exhibit identical configurational relations between the heads and their arguments. Specifically, the head and the complement c-command each other, whereas the specifier asymmetrically c-commands both the head and the complement.

Adjuncts can be integrated into elementary trees like (3) in the usual way, by adjunction. The resulting general X' schema of phrase structure is shown in (5), where W is a variable ranging over syntactic categories.

(5)

More evidence concerning the phrase structure of noun phrases

Consider the noun phrase in (6), where the italicized phrase in the first conjunct may be replaced by the pro-form one.

(6)

this book on the shelf, and that one

As in the case of do so substitution, we assume that a subsequence of a noun phrase can be replaced only if it is a constituent. The availability of one substitution in (6) then shows that there is a level of structure within the noun phrase that is distinct from the entire noun phrase (this book on the shelf) as well as from its head (book). Given the assumption that noun phrases are projections of N, we are led to posit an N' projection that exhaustively dominates the sequence book on the shelf. At first glance, the structure in (7a) might seem satisfactory. But since the PP on the shelf is clearly a modifier in (6), it must be represented not as a sister of the head, as in (7a), but as a sister of the intermediate projection, as in (7b).

(7)

a.

*

b.

If one substitutes for N', just as do so substitutes for V', then it should be able to substitute not only for book on the shelf, but for book as well, since both sequences are dominated by N'. As (8) shows, this expectation is indeed borne out.

(8)

this book on the shelf, and that one on the table

The grammaticality of (8) is another argument against (7a) (assuming that one is constrained to substitute for N').

As in the case of V', adjunction to N' can apply more than once, yielding recursive structures like (9).

(9)

As expected, one is able to replace any of the three instances of N' in (9), as shown in (10).

(10)	a.		this book on the shelf with a blue cover, and that one (one substitutes for the highest N' and is interpreted as 'book on the shelf with a blue cover')
	b.		this book on the shelf with a blue cover, and that one with a red cover (one substitutes for the middle N' and is interpreted as 'book on the shelf')
	c.		this book on the shelf with a blue cover, and that one on the table with a red cover (one substitutes for the lowest N' and is interpreted as 'book')

The discussion so far assumes that one substitutes for N', just as do so substitutes for V'. It is worth pointing out that the pattern of substitution in (6), (8), and (10) is equally consistent with an alternative assumption, namely that one, unlike do so, is able to substitute either for intermediate projections or for heads. The first alternative is preferable on conceptual grounds, since it allows us to maintain cross-categorial parallelism to the fullest extent possible. There is also an empirical consideration in favor of it. Consider the contrast in (11).

(11)	a.		this author of murder mysteries, and that one
	b.	*	this author of murder mysteries, and that one of nature guides

If we make the reasonable assumption that the PPs in (11) are complements rather than adjuncts (cf. He authors murder mysteries, and she authors nature guides), the structure of the first conjunct is given in (12).

(12)

Given this structure, the first alternative, according to which one substitutes for N', but not for N, is consistent with the contrast in (11), whereas the second alternative, according to which one can substitute for either N or N', leaves it unexplained.

It is worth noting that some speakers do not completely reject (11b). For such speakers, it is possible to maintain one of two hypotheses. Unlike speakers who report a clear contrast in (11), these speakers might in fact allow one to substitute for heads as well as intermediate projections. Alternatively, they might restrict one substitution to N', but might differ from the speakers reporting the clear contrast in allowing the PP in (11) to attach as an adjunct, as in (13).

(13)

I know of no evidence to decide between these two alternatives. The conceptual consideration, mentioned earlier, of maintaining cross-categorial parallelism between nouns and verbs would favor the second alternative.

Extending the X' schema to the remaining lexical categories

(14)	a.	*	They are very fond.
	b.		They are very fond of their grandson.

The contrast in (15) then shows that so can substitute for an adjective and its complement, but not for the adjective without the complement.

(15)	a.		I am fond of Annabelle, and they are so, too.
	b.	*	I am fond of Annabelle, and they are so of Roberto.

However, as we would expect on the basis of the other pro-forms, so can substitute for the adjective alone without an adjunct, as shown in (16).

(16)	a.		I am happy about my trip, and they are so, too.
	b.		I am happy about my trip, and they are so about theirs.

For concreteness, the structures of the adjective phrases fond of Annabelle and happy about my trip are shown in (17).

(17)

a.

b.

There is no pro-form comparable to do so, one, and so for prepositional phrases. Therefore, the extension of the X' schema to prepositional phrases is based purely on the conceptual consideration of cross-categorial parallelism.

Phrase structure variation across languages

(18)	a.	V		pursue their objective
	b.			submit her application
(19)	a.	N		pursuit of their objective
	b.			submission of her application
	c.			pride in her work
(20)	a.	A		proud of her work
	b.			fond of his children
(21)	a.	P		over the next five years
	b.			with great fanfare

But Universal Grammar by no means prescribes head-initial phrase structure. Rather, many languages exhibit consistently head-final phrase structure; two such languages are Japanese or Korean. Mixed phrase structure is also possible, as in Dutch, German, or Farsi. As shown in (22)–(25), verbs and adjectives in German are head-final, whereas nouns and prepositions are head-initial.

(22)		V		ihr	Ziel	verfolgen
				their	objective	pursue
				'pursue their objective'

(23)		A		seinen	Prinzipien	treu
				to his	principles	loyal
				'loyal to his principles'

(24)	a.	N		Verfolgung	ihres	Ziels
				pursuit	of their	objective
				'pursuit of their objective'
	b.			Treue	zu	seinen	Prinzipien
				loyalty	to	his	principles
				'loyalty to his principles'

(25)		P		auf	dem	Tisch
				on	the	table
				'on the table'

To complicate matters, Dutch and German also allow postpositions, as illustrated for German in (26).

(26)			den	Fluss	entlang (German)
			the	river	along
			'along the river'

And finally, to really liven things up, certain adpositions in these languages can either precede or follow their complements, as illustrated, again for German, in (27) and (28).

(27)	a.		wegen	des	Wetters
			because	of the	weather
			'because of the weather'
	b.		des Wetters wegen
(28)	a.		gegenüber	der	Kirche
			across from	the	church
			'across from the church'
	b.		der Kirche gegenüber

Dutch, too, allows such variation between head-initial and head-final adpositions, and in that language, it is even accompanied by a systematic meaning difference. Specifically, when adpositions with variable headedness are postpositions, their meaning is always directional, but when they are prepositions, their meaning is generally locative. This is illustrated in (29) and (30) (Kroch 1994).

(29)	a.		Ik	fiets	in	de	straat.
			I	bike	in	the	street
			'I bike in the street.' (locative)
	b.		Ik	fiets	de	straat	in.
			I	bike	the	street	in
			'I bike into the street.' (directional)

(30)	a.		Ik	klim	in	de	boom.
			I	climb	in	the	tree
			'I climb in the tree.' (locative) 'I climb into the tree.' (directional)
	b.		Ik	klim	de	boom	in.
			I	climb	the	tree	in
			'I climb into the tree.' (only directional)

In case the German and Dutch examples just discussed sound exotic, it is worth noting that English exhibits two postpositions of its own, as illustrated in (31) (for more information, consult the OED under the entries over, senses 2.II.7d and 2.V.17b, and through, sense 7.B.I.4b).

(31)	a.		They searched the whole world over.
	b.		They work the whole week through.

It is not uncommon, incidentally, for languages to undergo phrase structure change over time. For instance, the phrase structure of Old English (ca. 800–ca. 1100) was reminiscent of that of modern German. In particular, verbs were head-final in the oldest attested records. In late Old English, we find the first instances of verb-initial phrase structure, and the change from head-final to head-initial verb phrases was essentially complete in all dialects of Middle English by ca. 1350 (Ann Taylor, personal communication). Among the relics of the old verb-final phrase structure in the modern language are the saying Indictments do not a conviction make and perhaps also expressions like in so doing.

In order to better understand the relation between the universal and the parochial (= language-particular) aspects of phrase structure, it is useful to think of syntactic trees as mobiles. The important property of mobiles for our purposes is that their configuration with respect to the vertical axis is fixed, but that their branches are free to rotate in the remaining two dimensions of space. Universal Grammar resembles a mobile in that it fixes hierarchical phrase structure relations (domination and relations defined in terms of domination such as c-command), but not linear precedence relations. We can compare the language-particular grammars to the many possible particular states that a mobile can enter into.

The mobile metaphor is also helpful in understanding the central importance that modern syntactic theory accords to Universal Grammar. When we watch a mobile, what catches our eye and entertains us is the rotation of its branches, but that is also what is most ephemeral about it. For instance, if we wanted to replicate a mobile, we would focus on replicating the dominance relations among the nodes, which remain constant. Having replicated these, the shifting relations in the other two spatial dimensions would come along for free.

Two consequences of structure dependence

In the absence of recursion, a language would consist of a finite set of sentences, with the cardinality of the set a function of the number of elementary trees in the language. It is recursion that allows speakers of human languages, in a famous formulation, "to make infinite use of finite means" (Humboldt 1836, cited in Chomsky 1964:17). The type of recursion brought up in Chapter 2 is associated with the process of adjunction. Because of the way that adjunction is defined, a recursive node in an adjunction structure immediately dominates another instance of the same category. A simple example illustrating the power of recursion is shown in (32), where an adjective phrase is recursively adjoined as a nominal modifier and where the recursive nodes are highlighted by boxes.

(32)

a.

b.

c.

d.

e.

Since there is no reason to believe that the grammar of English imposes any limit on how often adjunction can be performed, it follows that an infinite set of noun phrases can be derived by the process illustrated in (32).

Recursion need not involve immediate domination. A recursive node may dominate, rather than immediately dominate, another instance of the same category. Nor is adjunction the only source of recursion in language. Consider a noun phrase containing a PP complement. Since the PP itself contains a noun phrase complement, the containing noun phrase is a recursive node. More precisely, in a structure of this type, both NP and N' are recursive nodes, as shown in (33).

(33)

A concluding note concerning recursion is in order. It is noteworthy that researchers have had some success with teaching chimpanzees, bonobos, and gorillas certain aspects of human language (Premack and Premack 1983). For instance, chimpanzees can be taught the use of symbols (that is, of signs that are both arbitrary and conventional) for objects and properties of objects such as color. They can also be taught to evaluate statements containing logical connectives such as 'and,' 'not,' and 'if … then.' To date, however, no nonhuman primate has shown any evidence of producing recursive linguistic structures.

Structural ambiguity