N ull Constituents

4.0 Ovcrvicw
So far, our discLL�sion of syntactic structure has tacilly assumed that
cünstituents in a given structure are overt, in the sense that they carry not only
grammatical and semantic features, but also audible phonetic features (unless
they undergo ellipsis and so receive a silent spellout in the PF component: see
§1.4). Tn this chapter, however, we will see that syntactic strnctures may also
cüntain null consliluents (also known as empty categories), that is, constituents
which have grnmmatical and semantic features but lack audible phonetic features
(and so are 'silenl', 'inaudible' or 'unpronounced'). We will look al null subjects
in§4.1, null T constituents in§4.2 and §4.3, null C Cünstituents in§4.4 and§4.5,
and null D, Q and P constituents in§4.6.

4.1 Null Subjccts

Tn this section, we will see that certain types of clause allow (or
require) a null subject. We are already familiar with one kind of null subject
fi-om the discussion of the Null Sub_ject Parameter in§1.5. There, we saw that
alongside finite clauses like that produced by speaker A in the dialogue in ( 1)
below with an overt subject like Maria, Ttalian also has finite clauses like that
produced by speaker B, with a null subject pronoun conventionally designated as
pro (and referred to affectionately a� 'little pro')

(1) SPEAKER A: Maria e3_sc tomalaF.sc?

Maria is retumed?
'Has Maria retumed?'
SPEAKER B: Si, pro C:3 .SG tomatai:.SG
Yes, pro is retun1cd
'Yes, shc has rerumed.'

(The subscripts in the glosses mark the following features: 3 = third person;
SG = singular number; F = feminine gender.) One reason for positingpro in (1b)
is that it captures the intuition that the sentence has an 'understood' subject (as is
clear rrom the fact that its English Lranslation Cüntains the subject pronoun she). A
secünd rea�on relates to the agreement morphology carried by the auxiliary e 'is'

152
 4.5 Null C in lnfinilivc Clau.scs 195

Under the TI' analysis ofsmalll clauses proposed in §4.3, small clauses are Tl'+Vl'
strnctures in which TI' is headed by a null variant ofinflnitival to and VI' is headed
by a null counterpart ofthe verh be. Jf(as claimed here) small clauses are defec1ive
clauses, Lhey will contain no CI' )ayer and thus pn¡ject only as far as TI'. On Lhis
view, (124) has the structure in (126) below:

(126) CP
�
C TP
0
�
DP T'
u,ey --------
T VP
M ---------------
V
· n•
consider
�
DP ·1•
him ---------
T VP
te
�
V AP
h uosuit.able

Under Lhc analysis in (126), the small clause subjecl him will be assigned accusative
case by the immediately adjacent u-msitive verb amsider which c-command� Íl, in
acconJance
: with lhe accusative case assignmenl condition (103i). Since the small
clause is a Tr (and not a Cl'), iL� sub_ject can fl<L"-�ivise and thereby move fn>m being
the sub_ject oíthe complement claLL�e TI' Lo becoming Lhe subjecl oíthe main clause
TI', in Lhe manner shown by the arrow below:

(127)
-----------
cr

e
o -----------
11'

--------
Dr 1•
he
----------
T vr
is
V Ti>
cortSidercd �

--------------------- 1
DI'
' ---------
T vr
te
-------------
V i\P
Be unsuírnble
 4.6 Null Hcads in Nominals 197

(131) CI'

e TP

-------------
0

DP ·r

D NP T VP
" Jolrn Af
--------------
V DP
admire
�
D NP
" Mary

The analysis in ( 131) is consisten! with the view 'lhat ali definile noun expressions
are DPs, including lhose not containing any overt detem1iner. A DP analysis ofbare
definile noun expressions is plausible from a semantic perspective in Lhat a name
like Jo/111 is a referring expression which denotes a specific/definite individual in
much Lhe same way as a DP such as this hoy dcies.
One piece of empirical evidence in support of analysing bare nouns as DPs
Cúmes from sentences like:

(132) Jo/111 and [thc chairman] are ancnding a meeting

The fact thal a bare noun like./0/111 can be coordinated wilh a determiner phrase/
DP like the chairman provides us with empirical evidence thaL such bare nouns
are DPs, given that only like constituents can be coordinatcd.
Tf(as suggested here) English has a null D constiluenl, we should expecl this not
only to have identifiable semanlic properties (in marking definiteness/specificity)
bul also to have identifiable gr'dmmatic-dl properties. And indee<l there is evidence
Lhat (like definite detem1iners such as this/these), Lhe null D constituent carries
person properties. In this respecL, consider sentences such as:

(133) a. Wc linguists takc oursclvcsr-yoursclvcs/ºthcmsclvcs too scriou.�ly, don'!

wet•ymú• //rey?
b. You linguists takc yoursclvcst•oursclvcs/•thcmsclvcs too scriously, don't
yuul• wel• lhey?
c. John takcs himsclU•oursclvcsJ•yourselvcs 100 scriously, docsn't l,e/"'don'I
we/ºdon'l you?

(133a) shows that a first persún expression such as we lin?;ttisL� can only bind (i.e.
serve as the anteceden! o() a firsl person rellexive like ourselves, and can only be
tagged by a first person pronoun like we. ( 133b) shows thal a second person
expression like you linguisl� can only bind a second person rellexive like
yourselves, and can only be lagged by a second person pronoun like you.
(133c) shows thal a bare noun like ./0/111 can only bind a third person rellexive
like /rimse/fand can only be tagged by a Lhird person pronoun like he. One way to
accounl for lhe relevant íacL� is to suppose lhat !he nominals we linguists/you
(as in 142).
206 NIJLI, CONSTITIJENTS

(103) Case Assignmcnt Conditions

A (pro)nominal constituent is case-marked by an immediately adjacent case­
assigner that c-command� it, and the case a�signed is:
(i) accusalive if the case assign er is transitive

Jf the case assigner is a compleme,uise·r with a TP complement, the ca�e

assigned is accusative (by 103i) i f the complementiser is transitive, but
otherwisc, the ca�e assigned is:
(ii) 11omi11ative ifthe case assigner is finite
(iii) 11111/ Clise if the case assigner is nonfinite

(taO) Itcration Filtcr

Structures involving iteration of likes are filtcred out at PP where one
immediately precedes and c-commands the other

(120) lmpcnctrability Condition

A constituent c-commanded by a complementiser C is impenetrable to
(so cannot undergo a syntactic opcration involving) a constituent which
c-commands the CP headed by C
(128) CP Hypothcsis
Ali non-defective clauses are CPs, but defective clauses (i.e. bare nonfinite
complement clauscs with passivisable subjects) lack the CP projection found
in other clauses
(140) (i) OP llypotbcsis
Dcfinite nominal argumenis are DPs (headed by an oven or null D)
(ii) QP llypothcsis
lndcfinitc nominal argw11cnts are QPs (headcd by an overt or null Q)

(149) PP 1-lypot-hcsis
Adverbial (pro)nominals are PPs headed by a null preposition

4.8 Bibliographical Notes

For a range of ac.-coU11ts of the llalian-slyle null pro subjecls discussed

in §4.1, see Chomsky (1981), Rizzi (1982, 1986, 1997), Jaeggli (1982, 1984),
Huang (1984), Monlalbelli (1984), Salir (1984), Suñer (1984), Hyams (1986),
Jaeggli & Salir (1989), Roberts (1993), Barbosa (1995, 2000, 2007), Alexiadou &
Amtgnoslopoulou (1998), Kal(> (1999, 2000), Barbosa, Duarte & Kato (2005),
Holmberg (2005), Neeleman & Szendrííi (2005) ami Tamburelli (2006, 2007). On
tnmcated null subjects in English, see Thrasher (1977), Napoli (1982), Haegeman
(1990, 1997, 2000a, 2000b, 2008, 2013, 2017), Haegernan & Thsane (1999, 2002),
Rizzi (1994, 2000), Franks (2005: ín. 5) and Weir (2008, 2012). 0n impera tives in
English, see Potsdam (1998b), Han (2000, 2001), Flagg (2002), and Rupp (2003).
The idea Lhal control infinilives have a null PRO subject dates back 10 Chomsky