?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation
Подождите немного. Документ загружается.
W
OR
D
-
F
ORMATION AND INFLECTIONAL MORPHOLOGY
65
derivative whose sole si
g
n of causativization is its inflection as a member of the
t
enth con
j
u
g
ation; that is, the morpholo
g
ical difference between
‘
s/
h
e
hat
es
’
an
d
‘s/he causes to hate’ is purely an effect of their contrasting positions
i
n the system of inflection classes in Table 9.
7
. INFLECTIONAL PARADIGMS AND WORD-FORMATION PARADIGMS
Modern research on inflectional morphology (e.g. Matthews 1972, Zwicky 1985,
A
nderson 1992) suggests that a language’s inflectional rules are realizational
,
in the
s
ense that they apply to the pairing
¢
X
,
ı
²
of a stem X with a morphosyntactic
property set
ı
to yield an inflected word for
m
w
,
the
R
EALIZATION
of
¢
X
,
ı
²
;
on this
v
iew, English has, for example, an inflectional rule which applies to the pairing
¢
walk
,
{
3
rd
singular present indicative}
d
²
to yield the realization
walks
.
If th
ece
ll
s
in a
l
exeme’s paradigm are regarded as pairings of the type
¢
X
,
ı
²
,
then a language’s
i
nflectional morphology can be seen as a system of rules for assigning realizations to
t
he cells in inflectional paradigms. (See Stump 2001 for an extensive justification of
t
his realizational a
pp
roach to inflectional
m
orpholo
gy
and a formal elaboration of its
p
rinci
p
les.
)
This conception of inflectional morpholo
gy
is quite distant from the
‘morpheme-based’ conception inherited fr
o
m Am
e
ri
c
an
s
tr
uc
t
u
rali
s
m
:
in a
morpheme-based approach
,
a word’s morphosyntactic properties are built up
incrementally through the addition of
its component morphemes; in a
f
reali
z
ational
approach, by contrast, a word’s morphosyntactic properties serve to determine the
s
equence of operations spelling out its morphological markings. Of the two
approaches, only the latter affords a natural and parsimonious account of certain
widel
y
observable phenomena, includin
g
extended exponence, nonconcatenative
exponence, and the underdetermination of a word’s morphos
y
ntactic properties b
y
i
ts inflectional markin
g
s (Stump 2001: 3-12).
I
f a language’s patterns of inflection are defined by a rules realizing a
paradigmatic system of cells, what of its
p
atterns of word-formation? Recent work
(e.g. Bauer 1997, Booij 1997) has raised
the possibility that word-formation
d
(
specifically, derivation) involves a paradi
g
matic system of organization as well.
I
ndeed, many of the arguments that motivate the postulation of paradigms in the
i
nflectional domain have straightforward analogues in the domain of derivation. As
t
wo cases in point, consider a
g
ain the phenomena of head markin
g
and blockin
g
.
7.1 Paradigms and head marking in inflection and derivation
T
h
e
n
o
ti
o
n
o
f
head marking
ultimately rests on a funda
g
m
ental distinction among
word-formation rules – the distinction between category-preserving and
c
ategory-changing rules. A category-preserving
rule
o
f
wo
r
d
-f
o
rmati
o
n i
s
a r
u
l
e
which applies to base
b
to produce value
d
and in so doing allows one or more of
d
b
’
s
m
orphos
y
ntactic properties to persist to
d
; a
cate
g
or
y
-chan
g
in
g
rul
e
,
b
y
contrast,
all
ows
n
o
n
eo
f
b
’s properties to persist, but instead simpl
y
imposes all of
d
’
s
dves
ѽ
t
ѽ
i
dves
́
ayati
66
G
REGORY
T
.
ST
U
MP
properties. Given this distinction, a
headed
expression can then be defined as one
d
arising through the application of a category-preserving rule; in particular, the
head
o
f a headed expression
d
is the base from which one or more of
d
d
’s properties
persist. Cate
g
or
y
-preservin
g
rules are themselves of two sorts: those that appl
y
e
ith
e
r t
o
an
u
ninfl
ec
t
ed s
t
em
or to a fully inflected word, and those that apply only to
m
an uninflected root
;
rules of the f
o
rm
e
r
so
rt ha
ve bee
n t
e
rm
ed
wo
r
d
-
to
-
wo
r
d
rules
.
I
n the inflectional domain
,
h
e
a
d
-infl
ec
t
ed
f
o
rm
ssuc
h a
s
mothe
r
s
-
in
-
law
present
an important problem: does
mothe
r
s
-
in
-
law
ari
se
fr
om
mothe
r
-
in
-
law
t
hrough the
application of a special “head operation” which suffixes -
s
to a complex noun’s
h
ead
,
or does
mothe
r
s
-
in
-
law
ari
se
fr
om
mothe
r
s
through the application of a
word-to-word rule of compounding? In formal terms, the issue here is whethe
r
he
a
d
-infl
ec
t
ed
f
o
rm
s
ari
se b
ecause certain inflectional o
p
erations are sti
p
ulated as
bein
g
head operations or because certa
i
n word-formation operations are, b
y
s
tipulation, allowed to appl
y
to inflected forms. The head-operation approach
(
assumed by Hoeksema 1984, Anderson 1992, and others) entails that if an
i
nflectional rule functions as a head operati
o
n in
o
n
ec
a
se
th
e
n it
s
h
ou
l
ddoso
in all
c
ases; the word-to-word approach, by contrast, entails that the same inflectional rule
m
ay sometimes give rise to head marking and sometimes not, depending on whethe
r
th
ewo
r
d
-f
o
rmati
o
n r
u
l
ew
ith
w
hi
c
h it int
e
ra
c
t
s
i
so
r i
s
n
o
t a
wo
r
d
-t
o
-
wo
r
d
r
u
l
e.
Moreover, the word-to-word approach entails that if a given stem exhibits head
m
arkin
g
an
y
where in its paradi
g
m, it shoul
d
do so ever
y
where in its paradi
g
m, and
that stems arisin
g
throu
g
h the application of the same rule of word-formation should
be alike in either exhibitin
g
or failin
g
to exhibit head markin
g
.
Em
p
irical evidence confirms the word-to-word a
pp
roach
(
Stum
p
2001: 112-119
)
.
T
hus
,
consider first the Breton inflectional rule which suffixes -
(
i
)
où to
p
lural nouns:
this rule gives rise to head marking in loose compounds (e.g.
TOK
-
SIVI ‘strawberry
h
ull’ [‘hat-strawberries’], pl.
tokoù
-
sivi
) but not in derivatives of temporal duration
i
n -
vez
(e.g.
z
N
OZVEZ ‘a night’s duration’, pl.
no
z
ve
z
ioù
/*
no
z
ioùve
z
,
c
f
.
no
z
ioù
‘nights’)
.
1
0
This contrast is inexplicable under the head-operation approach,
accordin
g
to which the inflectional rule of -
(
i
)
où suffixation should either alwa
y
s
function as a head operation or never do so. Under the word-to-word approach, b
y
c
ontrast
,
one can account for the contrast between
tokoù
-
sivi
an
d
no
z
ve
z
ioù
by
assuming that the rule of loose compounding is a word-to-word rule but that the rule
of
-
vez
suffixation is not. Consider in addition the third-person singular imperfect
z
a
c
ti
ve
f
o
r
m
ny
-
a
-
patat
-
of the Sanskrit verb
t
N
I
-
PA
T
‘fly down’: this form clearly
exhibits head marking, since its preterite prefix is positioned on the head of the ste
m
ni
-
pat
-
-
rather than at its periphery. This means that under the word-to-word
approach, the rule of preverb+verb compoundin
g
must be a word-to-word rule; this
i
n t
u
rn
e
ntail
s
that all f
o
rm
so
f
N
I
-
PA
T
should exhibit head markin
g
(which the
y
do)
and that all other verbs arisin
g
, like
N
I
-
PA
T
,
throu
g
h the application of the rule of
preverb+verb compounding should likewise exhibit head marking (which they do).
1
0
A
de
ri
v
ati
ve
in -
vez
is headed by its base noun because the base noun’s gender persists to the
z
derivative
;
for instance
,
no
z
ve
z
‘duration of night’ inher
i
ts the feminine gender of
noz
‘night’ (Stump
z
2001: 113
)
.
W
ORD
-
FORMATION AND INFLECTIONAL MORPHOLOGY
6
7
These facts su
gg
est that morpholo
g
ical theor
y
need not incorporate a special
c
ate
g
or
y
of head operations. Moreover, the
y
su
gg
est that the inflectional realizations
o
f a stem which exhibits head markin
g
are
projected from those of
its head by means
f
o
f a universal principle having the following formulation:
(3) Head-Application Principle (Stump 2001: 118):
W
here ste
m
d
arises from stem
d
b
through the application of a
wo
r
d
-t
o
-
wo
r
d
r
u
l
e
r
,
then for each cell ¢
b
,
ı
²
in
b
’s paradigm, if
¢
b
,
ı
²
h
a
s
r
e
alizati
o
n
x
,
then the corresponding cell
¢
d
,
ı
²
in
d
’s paradigm has
re
alizati
on
r
(
x
(
(
).
I
f derivation involves a paradigmatic system of organization, then one would
expect the Head-Application Principle to apply in the domain of derivation as well
as in that of inflection. In fact, it does. Suppose the cells in a derivational paradig
m
are pairings of the for
m
¢
d
,
ı
²
,
where
ı
i
s
a l
e
xi
cose
manti
c
c
ategory rather than a
m
orphosyntactic property set. In that case, the derivational paradigm of the verb
PASS might contain the cell
¢
pass
¢
¢
,p
ersonal nou
n
² having the realizatio
n
p
asser.
On
t
he assumption that the English rule of verb+particle compounding is a word-to-word
r
ule, the Head-Application Principle then correctly predicts that the cell
¢
pass by
¢
¢
,
p
ersonal nou
n
²
in the derivational paradigm of the derivative verb
PASS B
Y
should
Y
h
a
ve
passer by
as its realization. Thus, on the assumption that the Head-Application
y
Principle regulates the phenomenon of head
marking, its applicability in both the
d
i
nfl
ec
ti
o
nal an
dde
ri
v
ati
o
nal
do
main
s
fa
vo
r
s
t
h
eco
n
c
l
us
i
o
n that
bo
th
do
main
s
in
vo
l
ve
a paradi
g
matic s
y
stem of or
g
anization.
7.2 Paradigms and blocking in inflection and derivation
In instances of blocking, a lexically
l
i
s
t
ed e
l
e
m
e
nt
e
x
c
l
udes
th
e use o
f an
e
quivalent competitor. If derivation, like inflection, is assumed to possess a
paradi
g
matic s
y
stem of or
g
anization, then the fact (cf. Table 5) that blockin
g
re
lati
o
n
s
ar
e
f
ou
n
d
in
bo
th th
e
infl
ec
ti
o
na
l
an
d
th
ede
ri
v
ati
o
nal
do
main
sc
an
be
attributed to this paradi
g
matic or
g
anization
.
In
p
articular, one can assume that if two
forms compete to realize the same paradig
m
atic cell, principle determines
t
he outcome of this competition. Thus, because the lexical stipulation that the cell
¢
go
¢
¢
,{
past
}
² is realized b
y
went
applies more narrowly than
t
t
h
e
r
u
l
e
that r
e
aliz
es
a
ce
ll ¢X
V
,
{
past
}
²
a
sX
ed
,
went
blocks *
t
goed
;
in the same wa
y
,
j
ud
ge
(
n.
)
blocks
*
judger
*
because the lexical sti
p
ulation that the cell
r
¢
judge
¢
¢
(
v.
)
,
p
ersonal nou
n
²
i
s
r
ealized b
y
j
ud
ge
(
n.
)
is narrower than the
ru
l
e
that r
e
aliz
es
a
ce
ll ¢
X
V
,p
ersonal
n
oun
² a
sX
er
.
The notion that blocking rela
t
ions are regulated by principle receives
c
l
e
ar
co
nfirmati
o
n in th
ose
in
s
tan
ces
i
n which a word blocks a competing
combination of words. As an example of
this sort, consider the periphrastic
f
future-tense paradigm of Sanskrit
D
Ɩ
‘give’ in Table 10. Most of the forms in this
paradigm are periphrastic combinations consisting of the nominative singular for
m
ii
P
Ɨ
n
Ɨ
ѽ
ini’s
n
68
G
REGORY
T
.
ST
U
MP
of
D
Ɩ
’
s agentive derivativ
e
t
ogether with a present indicative form of
A
S ‘
be
’
i
nflected for the appropriate person and number; compare Tables 10-12. The
t
hird-person forms, however, are not periphrastic: they are simply nominative forms
o
f the a
g
ent noun, appropriatel
y
inflected for number. If the third-person plural for
m
followed the more
g
eneral pattern, it would be the periphrase
*
bu
t thi
s
i
s
blocked b
y
the actuall
y
occurrin
g
for
m
(
= the nominative
p
lural form of DâTè
)
.
Some attribute such blocking to a universal principle that favors synthetic
expressions over analytic expressions of
t
he same content
(
cf. Andrews 1990, Sells
1998, Bresnan 2000, and others
)
, but while such a
p
rinci
p
le could account for the
ove
rri
de o
f
*
b
y
,
it would also wrongly predict (for example) that
v
aporiz
e
s
h
ou
l
d ove
rri
de
t
urn
t
o vapor
.
Singula
r
D
u
al Pl
u
ral
1
s
t
2
nd
3
rd
asmi
asi
svas
st
has
t
t
ràu
smas
stha
ras
Ta
b
l
e
10 Periphrastic future paradigm of the Sanskrit verb
D
Ɩ
D
D
‘give’
Ɩ
Singula
r
Du
al Pl
u
ral
N
o
minati
ve
ru ras
V
ocativ
e
dtar
ru
ras
Accus
ati
ve
ram
I
n
s
tr
u
m
e
ntal
dtr
D
ati
ve
Ab
lati
ve
G
enitive
L
oc
ati
ve
Ta
b
l
e
11
Paradigm of the Sanskrit noun ‘giver’
The Pàõinian approach, by
c
ontrast, makes
j
ust the ri
g
ht predictions. Because the
r
ule realizing the cell
¢
d
Ɨ
d
d
,
{
PER
:3
rd
,
NU
M:
p
lural, TNS:second future
}
²
a
s
i
s
narrower than the default rule realizing a cell
¢
d
Ɨ
d
d
,{
PER
:
Į
,
N
U
M
:
ȕ
,
TNS
:seco
n
d
plural
}
²
a
s
th
eco
m
b
inati
o
n
o
f
w
ith th
e
r
e
alizati
o
n
o
f
¢
as
,{
PER
:
Į
,
NU
M
:
ȕ
,
TNS
:p
resent, MOO
D
:
indicative
}
²
,
the Pàõinian a
pp
roac
h
p
redicts that
s
h
ou
l
d
ove
rri
de *
(
Stump 2001: 230ff). On the other hand, the phrase
tu
r
nto
vapor
(unlike
r
,
etc.) is not a periphrase according to any of criteria in
(F)-(H); that is, it isn’t defined by the realizational morphology of English, but rathe
r
b
y
its s
y
ntax. As a consequence,
turn to vapor
doesn’t compete with
r
v
a
p
oriz
e
t
o
r
ealize the paradigmatic cell
¢
v
apor
,
inchoativ
e
²
; neither expression is capable of
blockin
g
the other.
d
Ɨ
d
d
t
Ɨ
t
t
santi
;
d
Ɨ
d
d
t
Ɨ
t
t
쟤
santi
;
d
Ɨ
d
d
t
Ɨ
t
t
r
as
Ɨ
d
Ɨ
d
d
t
Ɨ
t
t
r
as
Ɨ
d
Ɨ
d
d
t
Ɨ
쟤
d
Ɨ
d
d
t
Ɨ
tt
쟤
d
Ɨ
d
d
t
Ɨ
t
t
쟤
d
Ɨ
d
d
t
Ɨ
t
t
d
Ɨ
d
d
t
Ɨ
tt
쟤
d
Ɨ
d
d
t
Ɨ
tt
쟤
d
Ɨ
d
d
t
Ɨ
t
t
쟤
d
Ɨ
d
d
t
Ɨ
d
Ɨ
d
d
t
Ɨ
d
Ɨ
d
d
t
Ɨ
t
t
Ɨ
d
Ɨ
d
d
t
Ɨ
t
t
Ɨ
r
r
d
Ɨ
d
d
t
r
é
d
Ɨ
d
d
tú
r
d
Ɨ
d
d
tú
r
d
Ɨ
d
d
tá
r
i
d
Ɨ
d
d
t
Ě́
tt
s
Ě
Ě
́
u
d
Ɨ
d
d
t
Ɨ
t
t
쟤
d
Ɨ
d
d
t
Ɨ
tt
d
Ɨ
d
d
t
Ě́
tt
bhy
Ě
Ɨ
y
y
m
Ɨ
Ɨ
d
Ɨ
d
d
t
Ě́
t
t
bhy
Ě
Ɨ
y
y
m
Ɨ
Ɨ
d
Ɨ
d
d
t
Ě́
t
t
bhy
Ě
Ɨ
y
y
m
Ɨ
Ɨ
d
Ɨ
d
d
t
r
ós
d
Ɨ
d
d
t
r
ós
d
Ɨ
d
t
Ɨ
t
d
Ɨ
d
d
t
Ɨ
tt
d
Ɨ
d
d
t
Ě́
t
t
n
Ě
d
Ɨ
d
d
t
Ě́
t
t
bhis
Ě
Ě
d
Ɨ
d
d
t
Ě́
t
t
bhyas
Ě́
Ě
d
Ɨ
d
d
t
Ě́
t
t
bhyas
Ě́
Ě
d
Ɨ
d
d
t
r
́
n
r
m
Ɨ
d
Ɨ
d
d
t
Ɨ
tt
쟤
d
Ɨ
d
d
t
Ɨ
t
t
santi
D
Ɩ
T
R
ѽ
Ɨ
r
r
Ɩ
ras
d
Ɨ
d
d
t
Ɨ
t
t
asmi
d
Ɨ
d
d
t
Ɨ
ru
d
Ɨ
d
d
t
Ɨ
tt
쟤
Ɨ
r
r
dt
dt
Ɨ
r
r
ras
d
Ɨ
d
d
t
Ɨ
W
OR
D
-
FORMATION AND INFLECTIONAL MORPHOLOGY
69
Singular Dual Plural
1
s
t
2
nd
3
rd
ásmi
ási
ásti
svás
sthás
stás
smás
sthá
sánti
Ta
b
l
e
12 Present indicative paradigm of the Sanskrit verb
AS
‘be’
S
H
ead marking and blocking are only two of the phenomena that have been cited
as evidence of parallelism between infl
e
ction and derivation; other confirming
evidence exists as well (see e.
g
. Bauer 1997, Booi
j
1997). Facts such as these see
m
l
ikel
y
to stimulate a thorou
g
h rethinkin
g
of the relation of inflection to
w
ord-formation within the coming decade. Th
e
th
eo
r
e
ti
c
al ar
c
hit
ec
t
u
r
es
n
ecess
itat
ed
by these two types of morphology may u
l
timately be found to possess a much highe
r
d
egree of parallelism than they are curre
n
t
ly accorded in contemporary theoretical
mode
l
s.
Gregory Stump
Department o
f
En
g
lish
Universit
y
o
f
Kentuck
y
1
215 Patterson O
ff
ice Tower
Lexington, KY 4
0
5
06
-
0027
USA
e-mail: gstump@pop.uky.edu
R
EFERENCES
A
ckerman, Farrell and Stump, Gregory (2004). “Parad
i
gms and periphrastic expression: A study in
re
alizati
o
n-
b
a
sed
l
e
xi
c
ali
s
m
.
” In
:
L
.
Sadler and A. S
p
encer
(
eds.
)
,
P
rojecting morpholog
y
.S
tanford:
C
SLI Publications
,
111-157.
A
itchison
,
Jean. 1994.
W
ords in the mind
[
2
nd
edition]. Oxford UK & Cambridge USA: Blackwell.
A
nderson, Stephen R. 1982
.
“Where’s morphology?” Linguistic Inquir
y
13
,
571-612.
A
nderson, Stephen R. 1985. “Inflectional morphology.” In: T. Shopen (ed.), Language typology and
s
yntactic description
,
volume III. Cambridge: Cambridg
e
University Press, 150-201.
A
nderson, Stephen R. 1992, A-morphous morpholog
y
. Cambridge: Cambridge University Press.
A
ndrews, Avery D. 1990. “Unific
a
tion and morphological blocking.” Natural Language & Linguistic
T
heor
y
8,
5
0
7-557.
B
auer, Laurie. 1997. “Derivational paradi
g
ms.
”
In: G. Booi
j
and J. van Marle (eds.),
Y
earbook o
f
M
orpholo
gy
199
6
. Dordrecht: Kluwer
,
243-256.
B
ochner, Harr
y
. 1984. “Infl
ec
ti
o
n
w
ithin
de
ri
v
ati
o
n
.
” The Lin
g
uistic Review 3
,
411-421.
B
ochner, Harry. 1992. Simplicity in Generative Morpholog
y
.
Berlin and New York: Mouton de Gruyter.
B
ooij, Geert. 1996. “Inherent vers
us co
nt
e
xt
u
al infl
ec
ti
o
n an
d
th
e
split morphology hypothesis.” In: G.
B
ooij and J. van Marle (eds.),
Y
earbook of Morphology 1995. Dordrecht: Kluwer
,
1-16.
B
ooij, Geert. 1997. “Autonomous morphology and paradig
m
atic relations.” In: G. Booij and J. van Marle,
(
eds.
)
, Yearbook of Morphology 199
6
.
Dordrecht: Kluwer
,
35-53.
B
ooij, Geert. 2000. “Inflection and derivation.” In:
G
. Booij, Ch. Lehmann, and Joachim Mugdan (eds.),
Morphology: An international handbook
on inflection and word-formation
k
.
Berlin
&
New York:
W
alter de Gru
y
ter, 360-369.
70
GREGORY T
.
ST
U
M
P
Börjars, Kersti; Vincent, Nigel; and Chapman, Ca
r
ol. 1997. “Paradigms, p
e
r
i
p
hrases and
p
ronominal
i
nflection: a feature-based account.” In: G. Booi
j
and J. van Marle (eds.), Yearbook of Morphology
1996
.
Dordrecht: Kluwer
,
155-180.
Bresnan, Joan. 2000. “Optimal syntax.” In: J. Dekker
s
,
F. van der Leeuw, and J. van de Weijer (eds.),
O
ptimalit
y
Theor
y
: Phonolo
gy
, s
y
ntax and acquisition
.
Oxford: Oxford Universit
y
Press, 334-385.
C
súcs, Sándor. 1988. “Die wot
j
akische Sprache.” In: D. Sinor (ed.),
T
he Uralic lan
g
ua
g
es: Description,
histor
y
and
f
orei
g
n in
f
luences
.
Leiden: E. J. Brill
,
131-146.
Di Sciullo
,
Anna Maria and Williams
,
Edwin. 1987.
O
n the definition of word
.
Cambridge: MIT Press.
D
ressler, Wolfgang U. 1989. “Prototypical diff
e
re
n
ces be
t
wee
n infl
ec
ti
o
n an
dde
ri
v
ati
o
n
.
” Zeitschrift für
Phonetik, S
p
rachwissensch
a
ft und Kommunikationsforschung
a
42
,
3-10.
g
Ettinger, Stefan. 1974. Form und Funktion in der Wortbildung. Die Diminutiv- und
Augmentativmodifikation im Lateinischen
,
Deutschen und Romanischen. Ein kritischer
Forschungsbericht 1900-1970
.
Tübinger Beiträge zur Linguistik 47. Tübingen: Gunter Narr
.
Halpern, Aaron L. 1992.
T
opics in the placement and morphology of clitics
.
D
oc
t
o
ral
d
i
sse
rtati
o
n
.
Stanford Universit
y
.
Haspelmath, Martin. 1993. “The diach
r
o
ni
ce
xt
e
rnalizati
o
n
o
f infl
ec
ti
o
n
.
” Lin
g
uistics 31
,
279-309.
Hoeksema
,
Jacob. 1984. Cate
g
orial morpholo
gy
.
Doctoral dissertation, Universit
y
of Gronin
g
en [New
Y
ork: Garland, 1985
]
.
L
a
p
ointe, Steven G. 1990. EDGE features in GPSG. In: M. Ziolkowski, M. Noske, and K. Deaton
(
eds.
)
,
P
apers from the 26th Regional Meeting of the Chicago Linguistic Societ
y
, volume 1. Chicago:
C
hicago Linguistic Society, 221-235.
L
ieber
,
Rochelle. 1980.
O
n the Organization of the Lexicon
.
Doctoral dissertation, MIT. [Reproduced b
y
Indiana University Linguistics Club, 1981.]
Marantz, Alec. 1988. “Clitics, morpholo
g
ical mer
g
e
r
,
and the mappin
g
to phonolo
g
ical structure.” In: M.
Hammond and M. Noonan
(
eds.
)
,
T
heoretical morpholo
gy
: Approa
c
hes in modern lin
g
uistics
.S
an
Diego: Academic Press, 253-270.
Matthews
,
Peter. H. 1972.
I
n
f
lectional morpholo
gy
: A theoretical
s
tud
y
based on aspects o
f
Latin verb
con
j
u
g
ation
.
Cambrid
g
e: Cambrid
g
e Universit
y
Press.
Matthews
,
Peter. H. 1991.
M
orpholo
gy
[2nd edn.]. Cambrid
g
e: Cambrid
g
e Universit
y
Press.
Miller, Phili
p
H. 1991. Clitics and constituents in Phrase Structure
G
rammar. Doctoral dissertation
,
U
niversiteit te
U
trecht.
Mufwene, Salikoko. 1980. “Bantu class pref
ixes: inflectional or derivational?” In
f
f
J
. Kr
e
iman an
d
A
.
E
.
Ojeda (eds.)
,
Papers from the Sixteenth Regional Meeting of the Chicago Linguistic Societ
y
.
C
hicago: Chicago Linguistic Society, 246-58.
Perlmutter, David. 1988. “The sp
l
it morpholo
gy
h
y
pothesis: evide
n
ce
fr
o
m Yi
dd
i
s
h
.
” In
:
M
.
Hamm
o
n
d
and M. Noonan
(
eds.
)
, Theoretical morpholo
gy
: Approa
c
hes in modern lin
g
uistics. San Die
g
o:
A
cademic Press
,
79-100.
Pesetsky, David. 1985. “Mor
phology and logical form.”
r
r
Linguistic Inquir
y
16
,
193-246.
Sadler, Louisa and Spencer, Andrew. 2001. “Syntax a
s
an exponent of morphological features.” In: G.
B
ooi
j
and J. van Marle (eds.), Yearbook o
f
Morpholo
gy
2000
.
Dordrecht: Kluwer
,
71-97.
Sells, Peter. 1998. “Optimality and ec
onomy of expression in Ja
panese and Korean.” In: N. Akatsuka et al.
a
a
(
eds.
)
,
J
apanese/Korean Linguistics 7. CSLI, Stanford Linguistics Association, 499-514.
S
p
roat, Richard. 1988. “Bracketi
ng paradoxes, cliticization and ot
her topics: the mapping between
t
t
syntactic and phonological structure.” In: M. Everaert, A. Evers, R. Huybregts, and M. Trommelen
(
eds.
)
,
M
orphology and modularit
y
. Dordrecht: Foris
,
339-360.
Stump, Gregory T. 1990a. “Breton i
n
flection and the split morphology hypot
h
esis.” In: R. Hendrick
(
ed.
)
,
S
yntax and semantics, volume 23: The
s
yntax of the modern Celtic languages
.
San Diego: Academic
Press
,
97-119.
Stump, Gregory T. 1990b. “La morphologie bretonne et la
frontière entre la flexion et la derivation.”
a
L
a
Breta
g
ne lin
g
uistique 6
,
185-237.
Stump, Gregory T. 1991. “A paradigm-based theory of morphosemantic mismatches.” Languag
e
67
,
6
75-725.
Stump, Gregory T. 1993. “How peculiar is evaluative morphology?”
J
ournal of Linguistics 29
,
1-36.
Stump, Gregory T. 1995. “The uniformity of head ma
r
king in inflectional morphology.” In: G. Booij and
J
. van Marle
(
eds.
)
, Yearbook of Morphology 199
4
.
Dordrecht: Kluwer
,
245-296.
Stump, Gregory T. 1998. “Inflection.” In: A. Spencer and A. M. Zwicky (eds.),
T
he handbook of
morphology. Oxford & Malden
,
MA: Blackwell
,
13-43.
W
ORD
-
FORMATION AND INFLECTIONAL MORPHOLOGY
71
Stump, Gre
g
or
y
T. 2001.
I
n
f
lectional morpholo
gy
.
Cambrid
g
e: Cambrid
g
e Universit
y
Press.
Stump, Gregory T. 2002. “Morphologi
c
al blocking and Pàõini’s princi
p
le.” Pa
p
er
p
resented at the
Periphrasis and Paradigms Workshop, University of California, San Diego.
Stump, Gregory T. 2004. “Delineating the boundary b
e
tween inflection-class marking and derivational
m
arking: The case of Sanskrit -aya.” Paper presented at the 11th I
n
t
ernational Morphology Meeting,
U
niversity of Vienna
.
Z
wicky, Arnold M. 1985. “How to describe inflec
t
ion.” In: M. Niepokuj, M. van Clay, V. Nikiforidou
and D. Feder
(
eds.
)
,
P
roceedin
g
s o
f
the Eleventh Annual Meetin
g
o
f
the Berkele
y
Lin
g
uistics Societ
y
.
B
erkele
y
Lin
g
uistics Societ
y
, 372-386.
73
Štekauer, P. and R. Lieber (eds.), Handbook of Word-Formation, 73—97.
©
2005 Sprin
g
er. Printed in the Netherlands.
WORD-FORMATION AND SYNTAX
A
NDRE
WS
PEN
C
ER
1. INTROD
U
CTION
In this chapter we consider the relationship between word formation and phrase
formation (syntax). Parallels and differe
n
ces
ha
ve bee
n n
o
t
ed be
t
wee
n th
e
t
wo
fr
om
the earliest studies on word formation
(
see Štekauer 1998: 33f for detailed
discussion in the context of English). The most striking parallels between word
s
tructure and phrase structure are found in compounding structures. Compounding
w
ill n
o
t
be
th
ece
ntral f
ocu
s of this chapter, though I will have some remarks to
m
ake about so-called synthetic compounds. (A cross-linguistic survey o
f
c
ompounding can be found in Fabb 1998).
There are considerable difficulties in discussing the relationship between word
formation and syntax in a theory-neutral fashion. The problem is that different
theoretical models take substantially different approaches with respect to the
r
elationship between word structure and phrase structure. Syntactic models such as
Head-Driven Phrase Structure Grammar
(HPSG, Pollard and Sag 1994) or
r
L
exical
Functional
G
rammar (LFG, Bresnan 2001) respect a principle of lexical inte
g
rit
y
,
under which s
y
ntactic rules or principles
h
a
ve
n
o
a
ccess
t
o
int
e
rnal
wo
r
ds
tr
uc
t
u
r
e
and hence cannot create words. On the
o
th
e
r han
d
th
e
r
e
i
s
an
u
n
b
r
o
k
e
n tra
d
iti
o
n
s
tretching from the original
G
enerative
S
emantics model to its contemporary
i
ncarnations in transformational grammar (minimalism) in which word structure is
t
reated as a species of phrase or clause st
r
ucture. For syntacticians such as Kayne
(1994) it would seem that inflection is essentially a form of syntax, while for authors
s
uch as Hale and Keyser (1993) even word formation is syntax. Such an approach
l
eaves no room for a dedicated morpholo
gy
component and renders the relationship
between word formation and s
y
ntax essentiall
y
trivial.
It is lo
g
icall
y
impossible to investi
g
ate the relationship between word formation
and syntax if you do not believe in words and hence word formation. Except where
o
therwise stated
,
therefore
,
I shall make the assumption that there is a category of
word distinct from that of morph(eme) or phrase, even if the boundary between
t
hese categories is sometimes difficult to d
r
aw.
G
ranted the exist
e
nce o
f
wo
r
ds
th
e
r
e
are several respects in which we can i
n
vestigate the relationship between word
formation and syntax. First, we examine the extent to which syntactic principles can
h
a
ve
a
ccess
t
o
th
e
int
e
rnal
s
tr
uc
t
u
r
eo
f
wo
r
ds (‘lexical inte
g
rit
y
’). Next, we ask to
what extent s
y
ntactic constructions can be incorporated into word
s
. Then
,
we ask to
what extent properties of newl
y
formed words show up in their s
y
ntactic behaviour,
e
specially in argument structure realization. Against this background I conclude by
asking how various types of syntactic model propose to handle some of the more
s
ali
e
nt fa
c
t
sd
i
scussed.
74
A
NDRE
WS
PEN
C
ER
2. LEXICAL RELATEDNESS AND SYNTAX
2
.1 Morphotactics in classical US structuralism
I
n classical structuralist theories of the kind associated with Bloomfield
(
1933
)
,
H
arris (1951), Hockett (1958), s
y
ntax was
g
enerall
y
thou
g
ht of as the concatenation
o
f morphemes. Gleason (1969) was able to discuss word order patterns in lan
g
ua
g
es
such as English in terms of ‘position classes’, essentially the same notion as that
u
sed to describe the complex morphologies of American polysynthetic languages.
This meant that there was no obvious difference between a word and a
p
hrase, which
some linguists took to mean that there was no principled difference between these
t
wo constructs
(
see Dixon and Aikhenvald 2002 for a review of these issues
)
.
2
.2 Morpholo
gy
as s
y
ntax
There are abundant instances of s
y
ntactic constructions occurrin
g
inside
productive, or at least common, morphological constructions. In Dutch, whole
phrases can be freely incorporated into noun-headed compounds (Booij 2002a: 146;
see also, for example, Lieber 1989). In Romance languages we often find
c
ompounds consisting of a finite (usually 3sg present tense) verb form followed by a
n
oun denoting the verb’s ob
j
ect, e.g. Italian portaletter
e
‘postman, mailman’,
l
iterally ‘carries-letters’ (Scalise 1984). More generally, word formation processes in
m
an
y
American lan
g
ua
g
es ma
y
involve rather elaborated sentence constructions (see
Spencer 2000: 317 for examples of Nava
j
o ‘descriptive nouns’). Facts such as these
alert us to the possibilit
y
that derived words
g
enerall
y
mi
g
ht reflect the s
y
ntactic
s
tructure of a language.
I
n its simplest form the syntactic approach to morphology reduces to the clai
m
t
hat there exists a phenomenon of syntactic affixation. This essentially means that an
i
nfl
ec
ti
o
nal
o
r
de
ri
v
ati
o
nal f
o
rmati
ve c
an be represented as a syntactic node and
h
ence subtend syntactic relations with stru
c
tures which surface as words or parts of
words. The anal
y
sis of the tense s
y
stem of En
g
lish provided in Chomsk
y
(1957) is
an example of this mode of reasonin
g
. A tense suffix such as
-ed
is represented as a
d
sy
ntactic terminal which is then ad
j
oined to the verb stem in the s
y
ntax. Chomsk
y
(
1970), however, argued that we can distinguish idiosyncratic types of
n
ominalization
,
such as
dest
r
uction
,
from regular nominalizations such as shootin
g
by assuming that
dest
r
uction
i
s
f
o
rm
ed
‘in th
e
l
e
xi
co
n’
w
hil
e
shooting
is formed ‘in
g
t
he syntax’. Thus, we can think of (one interpretation of) an expression such as
the
s
hooting of the hunters as arising from the application of
-ing
to the VP
g
shoot the
hunte
r
s
.
This then captures the intuition that such nominalizations nominalize the
entire VP or even clause rather than an individual verb. Lieber (1992) explores the
i
dea of s
y
ntactic affixation in some detail.
W
ithin
g
enerative
g
rammar the claim that words have essentiall
y
the same
s
yntactic structure as phrases was first
a
dvanced in Toman (1983) and developed by
v
arious authors including Selkirk (1982), Di Sciullo and Williams
(
1987
)
and Liebe
r
W
OR
D
-
F
ORMATION AND SYNTA
X
75
(
1992) (see Toman, 1998 for a survey, and Scalise & Guevara, this volume). It is
c
onvenient to refer to these proposals as word synta
x
.
A
ce
ntral f
e
at
u
r
eo
f th
ewo
r
d
s
yntax approach is that word
s
ar
ee
n
doce
ntri
cco
n
s
tr
uc
ti
o
n
s
,
that is, they are headed,
j
ust like phrases (note that this is a diffe
r
ent sense of ‘head’ from that introduced b
y
Stump, this volume). In the case of derivational morpholo
gy
the head is identified
with the affix which realizes the derivation, and it is responsible for the cate
g
or
y
c
hange associated with derivation. This can be illustrated with the word
i
ndecipherabilit
y
shown in
(
1
)
:
(
1
)
N
A
A
V
V
N
in
de
c
ipher abil
i
t
y
T
he node labels are defined either by phrase structure rules (Selkirk 1982, Di Sciullo
and Williams 1987) or b
y
percolation of features from the terminals
(
Lieber 1983,
1992
)
.
For inflectional morpholo
gy
the notion of
headedness
i
sd
iffi
cu
lt t
ode
f
e
n
d
without resort to theory-internal justification, and different word syntax models of
i
nflection ado
p
t somewhat different a
pp
roach
e
s to this
q
uestion of headedness. Fo
r
i
nstance, Di Sciullo and Williams
(
1987
)
ado
p
t Selkirk’s
(
1982
)
notion of a
‘relativized head’
,
under which different affixes can serve as the head of an inflected
word with respect to different features, while Lieber (1992) adopts a percolation-
based approach which generalizes this idea
.
1
Th
e
n
o
ti
o
n
o
f
wo
r
d
h
e
a
ded
n
ess
i
s
c
ontroversial, even for derivation
(
see the discussion in Bauer
,
1990
,
Hudson
,
1987
,
Zwick
y
, 1985). In derivation Lieber (1983) notes that man
y
prefixes do not seem to
d
etermine the lexical category of th
ede
ri
ved wo
r
d.
F
o
r in
s
tan
ce
in counter-analysis
,
1
Di Sciullo and Williams
(
1987
)
also retain Williams’ id
e
a that th
e
h
e
a
do
f a
wo
r
d
is always to the right
o
f the stem, the ‘Right-hand Head Rule’. I confess that I have always been mystified by this proposal,
w
hi
c
h am
ou
nt
s
t
o
th
ec
laim that th
e
r
e
ar
e
n
o
infl
ec
ti
o
n
a
l
or derivational prefixes (or, for that matter,
c
ircumfixes
)
, so I leave it without further comment.