?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation
Подождите немного. Документ загружается.
86
A
NDRE
WS
PEN
C
ER
a. *John so speedy repainting the fence
b. *John repainting the fence so speedy
Although not perhaps especially common, constructions such as (22b) are
perfectly natural and idiomatic, especially with pronoun sub
j
ects (We were surprised
at you doing such a thin
g
). The puzzling case is that of (22c), a construction to
which a considerable amount of discussion has been addressed. The problem is that
t
he expression
John’s repainting
seems to be lexically headed by a noun
g
r
epaintin
g
,
while the expression
r
epaintin
g
the
f
ence seems to be lexicall
y
headed b
y
a verb
r
epaintin
g
.
It i
sco
mm
o
n t
o
r
e
f
e
r t
osuc
h a
c
o
n
s
tr
uc
ti
o
n a
s
an in
s
tan
ce o
f a ‘mix
ed
c
ategory’.
W
e only find mixed constructions of the type (22c) with
-ing
nominals, because
g
t
h
e
latinat
e
n
o
minalizati
o
n
s
in -
(
at
)
ion
,
-
ment
and so on cannot take direct objects in
t
any case. However, in other respects we see modest categorial mixture. The eventive
s
emantics underlying the nominalization is sufficiently strong to warrant certain
ty
pes of adverbial, under certain cond
i
tions (thou
g
h
j
ud
g
ements reported in the
l
iterature tend to be contradictor
y
). Thus, for man
y
speakers the examples in (26) are
acceptable
:
(26) a. the destruction of the city, quickly and with great brutality
b. the departure of the guests, rather more suddenly than we expected
c.
th
e
r
e
m
ov
al
o
f th
e
m
u
mmified remains, very carefully and
m
ethodically
As can be seen, for me at least, an adverbial is onl
y
permitted to modif
y
a
n
o
minalizati
o
n
o
f thi
sso
rt
w
h
e
n it i
sa
ppositive. An example such as (27a) I find
a
a
v
ery awkward at best compared to (27b):
(27) a. *We expected the departure of the guests less suddenly
b. We expected a less sudden departure of the guests
Solutions to the descriptive dilemma posed by this construction and related
c
onstructions in other lan
g
ua
g
es depend
o
n the s
y
ntactic framework chosen. A
s
amplin
g
of these is found in Abne
y
(1987), Baker (1985), Bierwisch (1989),
B
levins (1994), Bresnan (1997), Fu, Roeper and Borer (2001), Grimshaw (1990),
Hudson
(
2003
)
, La
p
ointe
(
1993
)
, Malouf
(
2000a, 2000b
)
, Pullum
(
1991
)
, Roe
p
e
r
(1993), Rozwadowska (1997), Spencer (1999), Yoon (1996). A useful summary of a
number of these proposals is given in Malouf (2000b).
In addition to the category mixing in
–ing
nominals we observe constructions in
g
English in which a sub
j
ect role appears to be expressed by a prenominal modifier. In
(28) the prenominal is itself a noun, but in (29) it would appear that the ad
j
ective
A
me
r
ican
realizes the sub
j
ect ar
g
ument of the underl
y
in
g
verb
invade
:
(28) government proposals to reduce taxes
(
29
)
the American invasion of Ira
q
W
ORD
-
FORMATION AND SYNTA
X
8
7
There has been brief discussion of such constructions in the literature
(
fo
r
i
nstance, Radford 1997 discusses examples such as (28), while Grimshaw 1990
m
entions cases like (29)) but the
y
remain somewhat puzzlin
g
.
4.1.2 Nominals denoting grammatical functions
Another type of deverbal nominalization that has attracted attention is the sub
j
ect
n
ominalization represented b
y
-
er
affixation as in
r
d
r
ive
r. In this t
y
pe of
no
minalizati
o
n th
e
r
e
f
e
r
e
nt
o
f th
ede
ri
v
ed noun is the sub
j
ect ar
g
ument (‘external
ar
g
ument’) of the verb (Levin and
R
appaport Hovav 1988, Rappaport Hovav and
L
evin 1992
)
. However, the di
r
ect object argument is still s
y
n
tactically ‘active’, in
t
he sense that it can be expressed by a PP complement: the driver of the train
,
a
d
river of trains
.
Other subject nominalization affixes do not generally permit the
expression of the base verb’s ob
j
ect or co
m
plement (30), though this is occasionally
possible (31):
(
30) a. claimant (*of the prize)
b. exorcist
(
*of the demon
)
c. t
y
pist (*of the report)
(
31) a. applicant for the job
b. critic of the plan
U
nlike action nominals, sub
j
ect nominals fail to exhibit the syntax of verbs at
an
y
level:
(
32) a. *a driver expensive cars
b. *a driver of expensive cars quickly
c. *a driver of ex
p
ensive cars to im
p
ress women
Handbooks on word formation generally mention the suffix -
ee
as denoting the
o
b
j
ect of a transitive verb:
e
mployee
,
paye
e
.
This is not quite the right
c
haracterization, since certain sorts of intransitives also permit -
ee
su
ffixati
o
n
:
escapee
.
These derived nouns show essentiall
y
no verbal properties:
(
33) The
y
emplo
y
ed hi
m
a. ... intermittently
b. ... as a messenger boy
c. ... to satisfy their disabled work force quota
(
34) an employee
a. *... intermittently
b. *... as a messen
g
er bo
y
c. *... to satisf
y
the disabled work force quota
88
A
NDRE
WS
PEN
C
ER
A thorough and compelling analysis of the -
ee
co
n
s
tr
uc
ti
o
n
c
an
be
f
ou
n
d
in
Barker (1998), who argues that the salient properties of
-ee
wo
r
ds
ar
ede
ri
ved
fr
om
t
heir semantics rather than their syntax.
4.1.3 -able adjectives
Th
e
final
se
t
o
f
deve
r
b
al
de
ri
v
at
es
that has occasionally attracted attention
(Randall 1982, Roeper 1987, this volume) is those formed by affixation o
f
-
able
t
o
yield an adjective with the meaning ‘(entity) such that one can
V
erb it’
.
Thi
s
c
onstruction applies productively to tra
n
sitive verbs, effectively creating an
adjective with the meaning of passive potential (see Beard 1995: 197), though
w
ithout recourse to the normal passive morpholo
gy
of En
g
lish. A handful of
-able
ad
j
ectives betra
y
passive-like verbal ori
g
ins in that the
y
seem mar
g
inall
y
to permit
expression of the suppressed sub
j
ect ar
g
ument b
y
a
by
-p
hrase or ex
p
ression of othe
r
v
erbal arguments/satellites:
(
35) a. This grammar is learnable by children
b. This expression is analysable as a kind of passive
c. The disease is treatable by non-invasive methods
H
owever, this is not
g
enerall
y
true of -
able
ad
j
ectives, and is prett
y
well
universall
y
excluded in the ne
g
ative form with
un-
(m
yj
ud
g
ements):
(
36) a. *This grammar is unlearnable by children
b. *This expression is unanalysable as a kind of passive
c. *The disease is untreatable by non-invasive methods
(
37) a. *The game is playable by children
b. *The s
y
mphon
y
is performable b
y
an amateur orchestra
c
. *The device is repairable onl
y
b
y
a qualified mechanic
4.2 Synthetic compounds and noun incorporation
A
number of investigators have follo
w
ed Marchand
(
1969
)
and others in
d
istinguishing two types of noun-noun compound in English:
r
oot compounds
suc
h
a
s
coffee tabl
e
an
d
v
erbal nexus compounds
or
s
ynthetic compounds
,
in which the
l
exical head is derived from a verb. Some investigators limit such synthetic
c
ompounds to those derived from action nominals and sub
j
ect nominalizations (
t
r
ain
d
rivin
g
,
t
r
ain d
r
ive
r) while others (e.
g
. Selkirk 1982) would include examples such
a
s
hand
-
made
,
derived from a
p
assive
p
artici
p
le. Some authors also include cases
suc
h a
s
machine
-r
eadable
(
Roe
p
er 1987
)
, and
p
erha
p
s even government employe
e
.
The
p
oint about these constructions is that the non-head of the com
p
ound seems to
bear a syntactic dependency to the head, realizing its direct object or some othe
r
g
rammatical function. Roeper and Siegel (1978) claim that the incorporated element
i
s the nearest argument/adjunct to the verb in canonical representation, the F
i
r
st
W
OR
D
-
F
ORMATION AND SYNTA
X
89
S
ister Principl
e
.
There is thus a clear prima facie case for the involvement of syntax
at some level of representation and, indeed, synthetic compounds bear some
r
esemblance to noun incorporation structures, which some take to be a classic case
o
f s
y
ntactic word formation (e.
g
. Baker 1988).
True noun incorporation, in which the verb head can behave
j
ust like an
y
othe
r
finite verb form, is not a
g
enerall
y
observed feature of Indo-European, thou
g
h
Scandinavian languages are developing it a
n
d Frisian seems to have a fully-fledged
s
ystem of noun incorporation (Dijk 1997). For this reason, synthetic compounding is
always found with some kind of deverbal word formation. A minor exception to this
i
s provided by the progressive aspect c
o
nstruction. Thus, we have the paradig
m
s
hown in
(
38
)
:
(
38) a. Horse-ridin
g
is fun
b. Mar
y
en
j
o
y
s horse-ridin
g
c. Mar
y
went horse-ridin
gy
esterda
y
d. Mary was horse-riding yesterday
e. *Mary horse-rides every day
f. *Mary horse-rode yesterday
g. *Mary has horse-ridden today
I
n
(
38a, b, c
)
horse-riding
is essentiall
y
a noun, which means that its head,
g
r
idin
g
,
is a noun
,
not a verb. For this reason
t
he compound is perfectl
y
acceptable. In
(
38e, f
)
, however,
rides
,
rode
are finite verb forms. En
g
lish does not permit
c
ompounding with finite verb forms (i.e. true noun incorporation). Moreover, as
s
een in (38g) English does not permit compounding with participles when they are
part of a finite construction (though I conf
e
ss that (38g) sounds slightly better than
(
38e, f)). The surprising construction, then, is (38d), in which we have a finite
c
onstruction, the progressive aspect, but a synthetic compound is possible. This
s
eems to be an instance of not-quit
e
-
completed grammaticalization. The
-ing
form of
g
t
he verb retains
j
ust sufficient of its nominal past to permit the compound (perhaps
h
elped b
y
the fact that in other uses the
-ing
form is clearly still a noun).
g
U
nfortunatel
y
, the paradi
g
m in (38) i
s
scarcel
y
remarked upon in the literature.
5
. THEORETICAL APPROACHES TO WORD FORMATION
There are essentially two ways of thinking of the constructions described in
Section 4
,
which I shall label
lexical
and
l
r
adically syntacti
c
f
o
r th
es
ak
eo
f
argument. Under the lexical treatment the derived word is formed in the morphology
but it can ‘inherit’ certain syntactic p
r
o
perties, such as argument structure
r
ealization. Under the radically syntacti
c
account word formation takes place in the
sy
ntax. I shall briefl
y
summarize the cent
r
al
p
oints of theoretical treatments of
sy
nthetic compounds and action nominals, these bein
g
the most ‘s
y
ntactic’ of the
co
n
s
tr
uc
ti
o
n
su
n
de
r
d
i
scuss
i
o
n
.
One of the first treatments of synthetic compounding in the generative
framework was essentially a combination of the lexical and the radically syntactic.
90
A
NDRE
WS
PEN
C
ER
R
oeper and Siegel (1978) proposed a set of transformational (syntactic) rules
o
perating over lexical entries which have the effect of (a) turning a verb into an
adjective/noun and creating a preverbal
slot
for the non-head, (b) inserting the non-
t
h
ead noun in the position of the syntactic
c
omplement of the original verb, (c)
m
oving the complement noun to the preverba
l
non-head position. A variant of this
ide
a r
esu
rfa
ces
in th
e
a
ccou
nt
o
f
v
an H
out
and Roeper (1998) (see Roeper, this
t
v
olume), except that for them the entire derivation is now, apparentl
y
, ‘in the
s
y
ntax’. The
y
observe certain differences i
n
interpretation between examples such as
(
39a, b
)
:
(
39
)
a. lawn mowe
r
b.
a
/
th
e
m
owe
r
o
f th
e
la
w
n
/o
f la
w
n
s
I
n
(
39b
)
mower
denotes a person who at some time must have mown at least one
r
l
awn, while (39a) could denote simply a person whose duty it is to mow lawns even
i
f he has never actuall
y
done this. More plausible examples are those such as
Rappaport Hovav and Levin’s (1992)
life saver
vs.
r
s
aver o
f
lives. In addition
,
lawn
mower
,
unlike mower o
f
lawns
,
is
g
enerall
y
interpreted as an instrument (
g
arden
t
ool) rather than an agent. In the theoretical framework adopted by van Hout and
R
oe
p
er as
p
ectual and event-related
p
ro
p
e
r
t
i
es
ar
e
a r
e
fl
ec
ti
o
n
o
f
v
ari
ous
f
u
n
c
ti
o
nal
h
eads at clause level, above the level of VP (AspP, TP, Voice-EventP and so on).
They therefore argue that synthetic compounds lack such functional pro
j
ections,
while they are present in the structures underlying expressions such as the mower o
f
the lawn
(which therefore functions essentiall
y
as a s
y
non
y
m of the person who
mowed the lawn
). Van Hout and Roeper account for s
y
nthetic compounds b
y
g
eneratin
g
the non-head noun in the s
y
ntactic position of an ob
j
ect and then movin
g
i
t to the pre-verb stem position by means of a syntactic incorporation (head-to-head
m
ovement
)p
rocess. Since Van Hout and Roe
p
er assume that all clausal functional
i
nformation is lacking from the synthetic compound, for them the incorporated noun
fails to pass through any of the various functional heads which might provide it with
an interpretation as a direct ob
j
ect (in this framework a DP is ‘licensed’ as an ob
j
ect
by passing through a functional pro
j
ection such as AgrOP, AspP or some such, none
o
f which are selected b
y
the -
er
suffix). The interpretation of
r
lawn
as the ob
j
ect of
mow
therefore comes about b
y
virtue of some modification relationship between
lawn
an
d
mow
,
restrictin
g
the (implicit) mowin
g
activit
y
to mowin
g
s of lawns.
I
n some lexical accounts it is typically proposed that the derived word ‘inherits’
c
ertain features of the argument structure of the root verb. Thus, on the model of Di
Sciullo and Williams
(
1987
)
the affix -
er
has in its argument structure grid a
r
s
emantic role <R> normally found on nouns. However, this argument is a functor,
w
hi
c
h m
e
an
s
that it
c
an
co
m
b
i
ne with another argument st
r
ucture grid, for instance,
t
hat of a transitive verb, to form a more complex
g
rid. Fo
r
driver
the complex grid
r
would take the form (schematicall
y
) <<A
g
, Th>R>
,
such that
t
he a
g
ent role is
i
dentified with the <R> role and the theme role can be satisfied b
y
a PP complement
(
d
river of trucks
)
or com
p
ounded noun
(
t
r
uck d
r
ive
r). On this type of account it is
possible to say that the constituent structure of the com
pound is [truck [driver]].
m
m
W
ORD
-
FORMATION AND SYNTA
X
91
Other lexicalist accounts are closer to the radically syntactic models in that they
propose that the compound is verb-headed and that this compound then serves as the
i
nput to the derivational morphology. Thus, we first form the compound [
V
tr
uc
k
drive
]
, in which
t
r
uck
satisfies the internal argument (direct object) position of the
k
ve
r
b
d
r
ive
. This compound then under
g
oes -
er
affixation, to give [
r
N
[
V
truck drive
]
er]. At the morphophonolo
g
ical level this representation has to correspond to a
bracketing (truck (driver)).
A
set of proposals for handling morphology such as
truck driver
has recently
r
been developed by Ackema and Neeleman (2004). They reject the notion of building
up sentence structures by com
b
i
ning ‘lexical entries’ or by ‘lexical insertion’.
Rather, they argue for a modular picture of language structure. A sentence has a
s
emantico-syntactic representation and a morpho-phonological representation and
the two levels are mapped to each other by
means of correspondence principles (cf
y
also Jackendoff 1997
,
2002b and Culicover and Jackendoff 2005 for a detailed
defence of a related view
p
oint
)
.
Ac
k
e
ma an
d
N
ee
l
e
man
w
i
s
h t
o
r
e
fl
ec
t i
n
th
e
ar
c
hit
ec
t
u
r
eo
f th
e
ir m
ode
l th
e
i
de
a
that there tends to be a systematic relationship between linear ordering of affixes and
their semantic scope. However, they also wish to adopt a version of ‘Separationism’,
under which morphemes are not Saussurean signs, that is, lexically listed pairings of
form and meaning. They therefore assume that there are two notions of ‘affix’.
O
ne
i
s a purel
y
morphophonolo
g
ical notion, /affix/, while the other, ‘AFFIX’, is a
s
emantico-s
y
ntactic notion. The wa
y
that th
e
AFFIX an
d
th
e/
affix
/
ar
e
atta
c
h
ed
t
o
appropriate components of the representation is
g
overned b
y
sets of mappin
g
p
rinci
p
les. In the default case, the AFFIX-/affix/ pairing is effectively a classical
m
orpheme and the linear ordering of the /
a
ffix/-es with res
p
ect to each other and
with respect to the root corresponds to the ordering of the AFFIX-es and the head.
T
his is guaranteed by a Linear Correspondence principl
e
.
D
ev
iati
o
n
s
fr
om
agglutination are handled by special spell-out rules, such as those in (40):
(
40
)
[TYPE ER]
/type/ /ist
/
[
COOK ER
]
/coo
k
/
[STEAL
ER]
/
thi
e
f
/
On the other hand when the base of affixation is itself com
p
lex
(
for instance, a
c
om
p
ound
)
we will need instructions as to
w
hi
c
h
e
l
e
m
e
nt t
o
atta
c
h th
e/
affix
/
c
om
p
onent to. Ackema and Neeleman assume an
I
n
p
ut Corres
p
ondence
p
rinci
p
l
e
,
under which the /affix/ is attached to the morphophonological exponent of the head
o
f the host. Thus, in English synthetic compounds we see the parallel structures
s
hown in
(
41
)
:
(
41) semantico-syntactic representation [[TRUCK DRIVE] ER]
morphophonological representation (/truck/ (/drive/ /er/))
T
here remain a number of interesting empirical and conceptual questions with
this approach. For instance, why is
truck driver
more or less synonymous with
r
9
2
A
NDRE
WS
PEN
C
ER
d
river of trucks when there is no ex
p
ression
*
cook of pastr
y
corresponding to pastry
cook
?
k
k
Action nominalizations
p
ose a more difficult descri
p
tive and theoretical
c
hallenge because they are much more likely to preserve a wide variety of verbal
properties (see Kopt
j
evska
j
a-Tamm 1993, 2003 for a detailed typological survey o
f
t
he possibilities). This is clearly seen in the examples of
-ing
nominals given above.
g
One solution to this problem is to assume that these nominals are formed in the
sy
ntax, for instance, b
y
movin
g
the verb root to a hi
g
her functional head, sa
y
, NOM.
This effectivel
y
makes the action nominal into an instance of cate
g
or
y
-chan
g
in
g
i
nflection (see Abney 1987, Baker 1985, Sproat 1985 for variants of this idea).
Morphologists refer to such processes as trans
p
ositions
(
see Beard 1995 for detailed
discussion
)
. The nominal retains verbal
p
r
o
perties because it remains essentially
a
v
erb. However, it acquires the distributi
o
nal properties of a noun phrase because its
h
ead is categorially a noun.
A very robust finding about such nominalizations is that they may mark thei
r
ar
g
uments solel
y
in the manner of a noun (22a), or solel
y
as a verb (22b), or the
y
c
an mark the ob
j
ect in the manner of a verb and the sub
j
ect in the manner of a noun
(
22c
)
. What seems never to be
f
ound is a construction corresponding to *
f
f
John
r
epainting of the fenc
e
,
in which
repainting
seems to be a noun with respect to its
g
o
bject and a verb with respect to its subject. A popular explanation for this
behaviour centres around the idea that the verb is more closely associated with its
o
b
j
ect (‘internal argument’) than its sub
j
ect
(
‘exernal argument’). If we assume that
t
he verb-ob
j
ect relation has priority in some sense over the verb-sub
j
ect (or verb
phrase-sub
j
ect) relation we can sa
y
that the nominalization switches to bein
g
(essentiall
y
) a noun but once it has made this switch it cannot
g
o back to bein
g
a
v
erb. In other words, if the verb’s ob
j
ect is marked as the ob
j
ect of a noun, then the
s
ubject must be so marked too. Variations on this theme couched in various
t
heoretical models are given in Bor
s
l
ey and Kornfilt (2000), Bresnan (1997),
L
a
p
ointe
(
1993
)
, S
p
encer
(
1999
)
.
A somewhat different approach to the problem of the gerund is to treat it as in
s
ome sense a member of both the N and the V category simultaneously. Variations
o
n this idea can be found in Lapoint (1993), Pullum (1991), Spencer (1999) and a
r
ecent variant has been proposed b
y
M
a
l
ouf
(
2000a, b
)
. Malouf
(
2000b: 31-32
)
s
ummarizes the crucial facts shown in section 4.1.1 as in
(
42
)
:
(42) a. A verbal gerund takes the same complements as the verb fro
m
w
hi
c
h it i
sde
ri
ved.
b. Verbal gerunds are modified by adverbs and not by adjectives
c. The entire verbal gerund phrase has the external distribution of an
N
P
d. The sub
j
ect of the
g
erund is optional and, if present, can be eithe
r
a
g
enitive or an accusative NP
W
OR
D
-
F
ORMATION AND SYNTAX
93
Malouf’s analysis is presented within the framework of HPSG. In that model
syntactic categories are expressed by means of a multiple inheritance hierarchy
.
4
Such a hierarchy is a state
m
ent of the relationships that hold between subcategories.
Malouf’s (2000b: 65) hierarch
y
is shown in (43).
Malouf’s (2000b) inheritance hierarch
y
:
(
43
)
head
n
ou
n r
e
lati
o
nal
p-noun c-noun
g
erund verb ad
j
ective
This hierarchy distinguishes three types of noun and three types of ‘relational’
c
ategory. The gerund is defined as being simultaneously a noun and a relational, but
d
istinct from a verb. For this reason the gerund phrase has the distribution of a noun
phrase, but not that of a verb phrase (property (42c). Malouf claims that adjectives
m
a
y
onl
y
modif
y
common nouns. Therefore, a
g
erund has to be modified in the
m
anner of a relational, that is, b
y
adver
b
s. This
g
ives propert
y
(42b). To account for
p
ro
p
erties
(
42a, d
)
Malouf
(
2000b: 66
)
assumes a lexical rule which creates a lexical
entry for gerund in which the verb’s complements are preserved but the subject
argument is expressed as the specifier (and hence subjects and specifiers cannot
c
ooccur in the gerund.
6
.
SU
MMARY AND AFTER
WO
RD
There are two ways in which syntax can be said to be relevant for the study of
word formation. On the one hand, theoreticians developing models of syntax have
sometimes ar
g
ued that most or all word formation can or should be treated as a
species of syntax. Followers of the word
syntax approach see words as having a
d
co
n
s
tit
ue
nt
s
tr
uc
t
u
r
ew
hi
c
h i
s
in
so
m
e
de
g
ree homolo
g
ous to that of phrases. Fo
r
so
m
e
th
eo
ri
s
t
s
th
e
fa
c
t
so
f
wo
r
d
f
o
rmati
o
n ha
ve bee
n tak
e
n a
sev
i
de
n
ce
in fa
vou
r
o
f
o
ne or other specific syntactic approach, and some seem to be claiming that the only
i
nteresting aspects of word formation are to be found in syntax. I have surveyed
some of these claims, concluding that th
e
evidence for a ‘synta
x
-all-the-way-down’
approach to word formation is at best sca
n
t
y. However, it remains true that there are
i
mportant ways in which syntax impinges on word formation, whatever one’s
t
h
eo
r
e
ti
c
al
s
tan
ce.
I ha
ve su
r
ve
y
ed the most important
o
f these as
p
ects, lexical
4
N
o
t
e
that thi
s
i
s
a
d
iff
e
r
e
nt
se
n
se o
f ‘inh
e
ritan
ce’
from that used in the expression ‘argument structure
i
nh
e
ritan
ce
’
.
94
ANDRE
WS
PEN
C
ER
i
ntegrity, phrase-based word formation and the realization of argument structure,
from this
p
ers
p
ective.
I
conclude on a methodological note.
O
ther things being equal it would
undoubtedly be simpler and hence methodologically superior to assume a single
o
verarching model encompassing both sentence structure and word structure (though
whether the central model should be syntax, as in minimalism, or morphology, as
assumed b
y
Harris, 1951, is a moot point). However, other thin
g
s are rarel
y
equal. If
we assume that morpholo
gy
is
g
overned b
y
principles which are partiall
y
distinct
from s
y
ntax, and it should turn out that actuall
y
all we need is s
y
ntactic (o
r
m
orphological) principles to cover both domains
,
then no harm will be done. We
will simply discover as our understandi
n
g progresses that the syntactic and
m
orphological principles come to converge
o
n each other. In other words, ‘splitting’
i
s a perfectly reasonable research strategy for the field as a whole to adopt. On the
o
ther hand, suppose there really are differ
e
n
ces between morphology and syntax and
we adopt one or other versi
o
n of the monolithic approach to research. If we assume
a ‘s
y
ntax-all-the-wa
y
-down’ approach we will never h
y
pothesize possible
i
ndependent morpholo
g
i
c
al principles and therefore we will never find them. The
m
onolithic strate
gy
, in contrast to the ‘splittin
g
’ strate
gy
, uniquel
y
brin
g
s with it the
r
isk that it will seriously impede progress. For this reason it would be, in a literal
s
ense, irrational for the linguistics community as a whole to encourage the ‘syntax-
all-the-way-down’ approach. Even those syntacticians who deep in their souls
believe that syntax provides a Theory Of Everything, should, if they are to behave
r
ationally, encourage morphologists to seek morphology-specific principles and
s
hould discoura
g
e under-motivated attempts to reduce all morpholo
gy
to s
y
ntax.
A
ndrew Spencer
Department of Language and Linguistics
University of Essex
Colchester CO4 3S
Q
U
K
e
-
mail
:
s
pena
@
essex.ac.u
k
R
EFERENCES
A
bne
y
, Stephen. 1987.
T
he En
g
lish Noun Phrase in its Sentential Aspec
t
.
Unpublished PhD dissertation,
Massachusetts Institute of Technolo
gy
.
A
ckema
,
Peter and Neeleman
,
Ad. 2004.
B
eyond Morphology: Interface Conditions on Word Formation
.
O
xford:
O
xford
Un
i
versity Press.
A
ckerman, Farrell and LeSourd, Phili
p
. 1997. “Toward a l
e
x
ical re
p
resentation of
p
hrasal
p
redicates.” In:
A. Alsina et al.
(
eds.
)
, 67-106.
Ackerman, Farrell and Webelhuth, Gert. 1998.
A
Theory of Predicates. Stanford University: Center fo
r
the Study of Language and Information.
A
lsina, Alex; Bresnan, Joan; and Sells, Peter
(
eds.
)
. 1997. Com
p
lex Predicates
.
Stanford University:
C
enter for the Study of Language and Information.
A
nderson, Stephen R. 1992. A-morphous Morphology. Cambridge: Cambridge University Press.
W
OR
D
-
FORMATION AND SYNTAX
95
B
aker, Mark C. 1985. Syntactic affixation and English gerunds. Proceedings of the 4th West Coas
t
Con
f
erence on Formal Lin
g
uistics
,1
-
11
.
B
aker
,
Mark C. 1988.
I
ncorporation. A Theor
y
o
f
Grammatical Function Chan
g
in
g
. Chica
g
o: Universit
y
o
f Chicago Press.
Barker, Christo
p
her. 1998. “E
p
isodic
-
ee in En
g
lish: a thematic role constraint on a new word formation.”
Lan
g
ua
ge
74
,
695-727.
Bauer, Laurie. 1990. “Be-headin
g
the word.”
J
ournal o
f
Lin
g
uistics 26
,
1-31.
Beard
,
Robert. 1995. Lexeme Morpheme Base Morpholog
y
. Stony Brook, NY: SUNY Press.
Bierwisch
,
Manfred. 1989.
E
vent nominalization:
p
ro
p
osals and
p
roblems
.
Akademie der Wissenschaften
der DDR
,
Berlin
,
1-73.
B
levins
,
James P. 1994. „A lexicalist analysis of gerundive nominals in English.” Australian Journal of
Linguistics
14, 1
-
38
.
Bloomfield
,
Leonard. 1933. Languag
e
.
N
ew
Y
o
rk
:
H
o
lt
.
Booi
j
, Geert 2002a. Dutch Morpholo
gy
.O
xford:
O
xford
Un
i
versit
y
Press.
B
ooi
j
, Geert 2002b. “Separable complex verbs in Du
t
ch: A case of periphrastic word formation.” In:
Dehé et al.
(
eds.
)
, 21-42.
B
orsley, Robert D. and Kornfilt,
J
acklin. 2000. “Mixed extended projecti
o
ns.” In: R. D. Borsley (ed.),
T
he Nature and Function o
f
S
y
ntactic Cate
g
ories. S
y
ntax and Semantics vol. 3
2
.
San Die
g
o:
A
cademic Press
,
101-131.
B
resnan, Joan. 1997. “Mixed categories as head sharing constructions.” In: M. Butt and T. H. King (eds.),
P
roceedings of the LFG97 Conferenc
e
,
CSLI
p
ublications online: htt
p
://www-
c
sli.stanford.edu/publications/LFG2/lfg97.html.
B
resnan
,
Joan.
2001
. Lexical Functional Synta
x
.
O
xford: Blackwell Publishers.
C
homsky, Noam. 1957.
S
yntactic Structures
.
The Hague: Mouton.
C
homsky, Noam. 1970. “Remarks on Nominalization.” In: R. Jacobs and P. Rosenbaum (eds.), Readings
i
n English Transformational Grammar
.
Waltham
,
MA: Blaisdell
,
184-221.
C
homsk
y
, Noam. 1995. “Bare phrase structure.” In: G. Webelhuth (ed.), Government and Bindin
g
Theor
y
and the Minimalist Pro
g
am
.
Oxford: Blackwell Publishers
,
383-439.
C
ulicover, Peter and Jackendoff, Ra
y
S. 2005. Simple(r) S
y
nta
x
.O
xford:
O
xford
Un
i
versit
y
Press.
D
ehé, Nicole; Jackendoff, Ray; Macintyre, Andrew; and Urban, Silke
(
eds.
)
. 2002. Verb
p
article
e
x
p
lorations. Berlin: Mouton de Gruyter.
D
i Sciullo
,
Anna-Maria and Williams
,
Edwin. 1987.
On the Definition of Word
.
Cambrid
g
e,
Ma
ss
a
c
h
use
tt
s:
MIT Pr
ess.
Dijk, Stefan. 1997. Noun Incorporation in Frisian. Ljouwert: Fryske Akademy.
D
ixon, Robert M. W. and Aikhenvald, Alexandra Y. 2002. Word: a typological framework. In: R. M. W.
Dixon and A. Y. Aikhenvald
(
eds
)
, Word. A cross-linguistic typolog
y
.
Cambridge: Cambridge
U
niversity Press, 1-41.
Don, Jan. 2004. “Categories in the lexicon.” Linguistics
42, 931
-
9
5
6
.
Downin
g
, Pamela. 1977. “On the creation and use of En
g
lish nominal compounds.” Lan
g
ua
g
e 55
,
810-
8
42
.
Fabb, Ni
g
el. 1998. “Compoundin
g
.” In: A. Spencer and A. Zwick
y
(eds.), 66-83.
Fu, Jingqi; Roeper, Thomas; and Borer, Hagit. 2001.
“
The VP within
p
rocess nominals: Evidence fro
m
adverbs and the VP ana
p
hor Do-so.” Natural Language and Linguistic Theor
y
19
,
549-582.
Gleason, Henry Allan Jr. 1969.
An Introduction to Descriptive Lin
g
uistics. London : Holt
,
Rinehart and
W
inston.
Grimshaw
,
Jane. 1990.
A
rgument Structure
.
Cambridge, Massachusetts: MIT Press.
Hale, Kenneth and Keyser, S. Jay (eds.). 1993. The View From Building 20: Essays in Linguistics in
H
onor of Sylvain Bromberger
.
Cambridge, MA: MIT Press.
Hale, Kenneth and Keyser, S. Jay. 1993. „On argument structure and the lexical expression of syntactic
r
elations.” In: K. Hale and S. J. Keyser (eds.), 53-110.
Hale, Kenneth and Keyser, S. Jay. 2002.
P
rolegomenon to a Theory of Argument Structur
e
.
Cambridge,
MA
:
MIT Pr
ess.
Halle
,
Morris and Marantz
,
Alec. 1993. “Distributed
m
orpholo
gy
and the pieces of inflection.” In: K.
H
ale and S. J. Keyser (eds), 111-176.
Harris, Zelli
g
S. 1951.
S
tructural Lin
g
uistics
.
Chica
g
o: Universit
y
of Chica
g
o Press.
Hockett, Charles F. 1954. “Two models of
g
rammatical description.”
Word
10, 210-231 [Reprinted in
d
Martin Joos
(
ed.
)
. 1958
(
2nd edition
)
. Readin
g
s in Lin
g
uistics
.
Chica
g
o, Chica
g
o Universit
y
Press].