?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation
Подождите немного. Документ загружается.
38
2 R
OC
HELLE LIEBER
c
ompounds is generally the internal argument of the verb, a generalization which
R
oeper and Siegel call the First Sister Princi
p
l
e
.
Both Selkirk
(
1982
)
and Lieber
(
1983
)
r
eplace syntactic transformations and
l
exical transformations with interpretive principles based on argument structure.
Selkirk’s First Order Pro
j
ection Condition (1982: 37) allows the direct internal
argument of the verb to be satisfied by the first stem of the compound, thus ruling
ou
t
bo
th f
o
rm
s
lik
e
*girl-swimming
in which the subject is the first stem, and forms
g
l
ik
e
*shelf-putting
in which the first stem is an internal argument of the verb, but not
g
t
he direct internal ar
g
ument. Lieber’s Ar
g
ument Linkin
g
Principle is similar to spirit
t
o Selkirk’s First Order Projection Condition, but attempts to account in a single
principle for the interpretation of both synthetic compounds and root compounds.
W
ith the advent of the Principles and Parameters model of syntax and more
r
ecently Minimalism, several theorists a
t
tempt a return to a movement-based
analysis of synthetic compounds. Roeper (1988) analyzes synthetic compounding in
English as a form of Head Movement, inspired by Baker’s (1988) analysis of noun-
i
ncorporation. In an attempt to inte
g
rate the anal
y
sis of word formation with an
anal
y
sis of s
y
ntax in which theta-role assi
g
nment in En
g
lish is strictl
y
to the ri
g
ht,
L
ieber (1992) also proposes a form of Head Movement for s
y
nthetic compounds.
L
ieber (2004) returns to an interpretive principle for synthetic compounding,
s
uggesting that the interpretation of synthetic compounds should be treated as a
m
atter of lexical conceptual structure, rathe
r
than as a matter of syntax. This is, of
r
c
ourse, only possible within a theoretical framework which allows discussion of the
s
emantics of word formation. Within the theory proposed in Lieber (2004), words of
all cate
g
ories, both simple and complex, have lexical conceptual structures, rou
g
hl
y
o
f the form developed b
y
Jackendoff (1990), but with the refinement that semantic
f
u
n
c
ti
o
n
s
ha
ve bee
n
b
r
o
k
e
n
dow
n int
o
a n
u
m
be
r
o
f f
e
at
u
r
es w
hi
c
h all
ow us
t
o
c
haracterize the core meanings of affixes as well as of stems. Roughly, both simplex
le
xi
c
al it
e
m
s
an
d
affix
es
ar
ec
hara
c
t
e
riz
ed
by a central core of meaning called the
skeleton
that consists of features like [+/– material] and [+/– dynamic], along with
t
heir arguments. For our purposes it is suffi
c
ient to note that
[
+material
]
items are
c
oncrete, and
[
–material
]
items abstract. Items which are [+dynamic] are events, and
t
hose which are [–dynamic] are states. The two features are not mutually exclusive,
h
owever, so that an eventive or processual noun like
sunset
can be both [–material]
t
and [d
y
namic]. Word formation consists of linkin
g
to
g
ether the skeletons of
m
orphemes accordin
g
to the followin
g
principle:
(
10
)
Principle of Coindexation:
I
n a configuration in which semantic skeletons are composed, coindex
t
he highest nonhead argument with the highest (preferably unindexed)
h
ead argument. Indexing must be consistent with semantic
c
onditions on the head ar
g
ument, if an
y
.
W
hen an affixal skeleton is composed with the skeleton of its base
,
the latter is
subordinated to the former. When two stems are composed in compounding, the two
s
k
e
l
e
t
o
n
s
ar
ecoo
r
d
inat
ed.
ENGLISH WORD
-
F
ORMATION PROCESSE
S
383
(11) is an example of the skeletal representation of the agent/instrument suffix
-
er
in English (see also below, section 2.2.1):
r
(
11
)
-
er
[+material, d
y
namic ([ ], <base>)]
L
ieber
(
2004
)
claims that a the suffix
–er
in English creates concrete dynamic
r
n
ouns. Further, its argument (the so-called ‘R’ argument, see Williams 1981,
Higginbotham 1985), is co-indexed and therefore identified with the highest
argument of its base. So a derived word like
driver
has the composed skeleton in
r
(
12
)
:
(
12) [+material, d
y
namic ([ ]
i,
[+d
y
namic ([ ]
i
, [ ]
)
]
)
]
-
e
r
d
r
ive
W
hat the representation in (12) says, in
e
ffect
,
is that the affix takes an eventive
base, in this case one corresponding to the activity verb
d
r
ive
,
and subordinates it
,
c
o-opting its highest argument. This is similar to the argument-structural analysis in
B
ooij (1986) or in work of Rappaport Hovav and Levin (1992), except that it
translates that analysis into lexical semantic terms, which Lieber (2004) argues to be
the a
pp
ro
p
riate terms.
W
e then inte
g
rate the skeletal re
p
resentation for the com
p
lex word
driver
in (12)
r
w
ith that
o
f th
es
t
em
t
r
uck
.
Since the affix –
er
has already has its “R” argument co-
r
i
ndexed with the highest argument of its base
d
r
ive
,
and therefore shares the subject
i
nterpretation with its referent, the “R” argument of
t
r
uck
is co-indexed with the next
k
available argument, namely
t
he internal argument of
d
r
ive
.
This is illustrated in
(
13
)
:
(
13
)
[+material
(
[ ]
j
)], [+material, dynamic ([ ]
i,
[
+dynamic ([ ]
i
,[]
j
)
]
)
]
t
r
uck
-
e
r
d
r
ive
The initial stem of the s
y
nthetic compound is interpreted as an internal ar
g
ument
because the hi
g
hest free ar
g
ument of the bas
e
happens to be the internal ar
g
ument of
t
h
eve
r
b
d
r
ive
.
3. DERIVATION
I leave the debate over the precise dividing line between inflection and derivation
t
o other authors. See Stump (this volume), as well as Anderson (1992) and
Haspelmath (1996) for discussion of this issu
e
. I will merely state at the outset that I
will not treat English participles in
–ing
or
g
–
en
in thi
ssec
ti
o
n
.
I
w
ill al
so
n
o
t mak
e
a
s
trict division in the section on
p
refixation between
p
refixes and elements that in
o
ther works have been termed combinin
g
forms or even compoundin
g
elements.
Derivation is an especiall
y
active means of word formation in En
g
lish. Amon
g
the many prefixes and suffixes in the language, some are quite productive, others
m
oderately so, and some exist only in lexicalized remnants of processes that no
384
R
OC
HELLE LIEBER
l
onger form new words in the language (see Kaisse, this volume). Some affixes can
be traced back to native stock
;
oth
e
r
s are borrowed from Latin, often through
French, or from Greek. Although the two stocks of affixes are diachronically
distinct, in modern English they do not always remain distinct, with etymologically
G
ermanic affixes often attaching freely to Latinate roots, and less often
e
tymologically Latinate affixes attaching to native stock. In fact, the partial
i
nte
g
ration of the historicall
y
different s
y
stems has
g
iven rise to much discussion
amon
gg
enerative morpholo
g
ists, providin
g
the basis for the theor
y
of Lexical
Phonolo
gy
and Morpholo
gy
(Kiparsk
y
1982, Mohanan 1986; Kaisse, this volume),
and for subse
q
uent criti
q
ues and refinements of that theory (Fabb 1988, Plag 1999,
G
iegerich 1999, Aronoff and Fuhrhop 2002).
A
s a way of summarizing the comple
x
i
ti
es
that
u
n
de
rli
e
th
e
th
eo
r
e
ti
c
al
discussion
,
we offer a number of tables which cull information from Marchand
(1969), Jespersen (1965), the OED, the above-mentioned sources, and my own
e
xplorations. Table 1 indicates what sorts of bases various suffixes prefer (root o
r
word, native or Latinate
)
, and whether these suffixes have stress or othe
r
phonolo
g
ical effects on their ba
ses.
Ta
b
l
e
2 in
d
i
c
at
es
f
o
r
e
a
c
h
su
ffix
w
h
e
th
e
r it
attaches to alread
y
affixed bases, or ca
n
itself under
g
o further affixation. Table 3
g
ives comparable information for prefixes
.
In all three tables
,
native affixes are
i
ndicated by boldface.
Affix
es
s
egmental
phonolo
g
ical
eff
ects
s
tress effects attaches to
roots
attaches to
wo
r
ds
attaches to
non
-
native
bases
attaches to
native bases
-
a
b
l
e
V
->A
as
–ible
:
p
erce
p
tible
so
m
e
tim
es:
c
om
p
arable
yes: capable,
p
robable
yes: washable yes yes
-
ag
e
N
->
N
n
o no no yes:
o
rphanag
e
yes yes
-a
l
V
->N
n
o no no y
e
s
:
arrival
,
re
f
us
al
yes occasionally:
betrothal
,
wi
t
hd
ra
w
a
l
-al
N->A
y
es: sacr
i
f
i
c
i
a
ly
es:
architectural
,
a
cc
i
de
ntal
y
es: m
i
n
i
ma
ly
es:
ar
c
hit
ec
t
u
ral
y
es occas
i
ona
lly
:
ti
d
al
-
an
,
-ian
N
,
A->N
yes: Christian yes:
c
ontrarian
,
Eliza
be
than
yes: pedestrian yes: Bostonian yes
o
ccasionally:
e
arthian
,
anti-
c
h
u
r
c
hian
-
anc
e,
-
e
nc
e
V
->N
y
e
s
:
i
n
duc
tan
ce
y
e
s
:
pre
f
erence
yes: obedience yes y
e
s
:
a
d
mittan
ce
yes: riddance,
f
o
r
be
aran
ce
-
ant
,
-ent
V
->N
y
es: resident
y
es: resident
y
es: liti
g
ant,
s
tim
u
lant
y
es:
a
ccou
ntant
y
es occasionall
y
:
coo
lant
-
at
e
s
tem->
V
n
o occasionally:
n
itrogenat
e
yes: litigate,
s
tim
u
lat
e
yes: fluorinate yes n
o
-
ati
on
y
es: yes:
reve
lati
o
n
y
es:
y
es: yes occasionall
y
:
ENGLISH WOR
D
-
FORMATION PROCESSES
385
A
ffix
es
s
egmental
phonological
effects
stress e
ff
ects attaches to
roots
attaches to
wo
r
ds
attaches to
non
-
native
bases
attaches to
native bases
V->N revolution
,
p
erce
p
tio
n
perception restoration flirtation
,
s
tar
v
ati
on
-
do
m
N
->
N
o
ccasionally:
w
i
sdom
no no yes:
k
ingdom,
p
o
p
edo
m
yes yes
-
ed
N->A
n
o no no yes:
i
ced
,
cu
lt
u
r
ed
yes yes
-
ee
V
->N
no
yes: absentee yes: evacuee yes: escapee yes: nominee yes: payee
-
en
N
->A
n
o no no
y
es: wooden
,
w
ax
en
n
o
y
es
-
en
A
->
V
n
o no no
ye
s
:
l
en
g
then,
b
la
c
k
en
o
ccasionall
y
:
c
hasten
,
mo
i
s
t
e
n
ye
s
-
er
V,
N->
N
n
o no yes:
a
stronomer
,
philosophe
r
yes: singer,
w
rit
er
yes yes
-
e
r
y
N
->N
n
o no
ye
s
:
m
onaster
y
,
s
ur
g
er
y
y
es: nunner
y
,
pi
gg
er
y
y
es
y
es
-
es
q
ue
N
->
A
no s
tr
ess
on
su
ffix
:
c
hivalresque
yes: grotesque,
c
hivalresque
yes: statuesque yes yes
-
ess
N
->N
n
o no y
e
s
:
abbess yes:
l
ioness yes yes
-
e
tt
e
N
->N
no s
tr
ess
on
su
ffix
?
ba
g
uette
y
es: featurette,
k
it
c
h
e
n
e
tt
e
y
es
y
es
-f
u
l
V
,N->A
n
o no occasionally:
v
engeful
yes: sorrowful yes yes
-h
ood
N->
N
n
o no no y
e
s
:
k
ni
g
hthood
y
e
s y
e
s
-i
c
N
->
A
y
es: parasitic
y
es: historic,
Germanic
y
es: electric,
g
eo
g
raphic
y
es: dramatic,
problematic
ye
s no
-if
y
N,
A-
V
no
n
o
y
es: pacif
yy
es: fishif
yy
es: pacif
y
,
terrify
y
es: beautif
y
-i
sh
N,A->
A
r
arely: wolvish no n
o
yes: mulish yes: modish,
s
tylish
yes
-
i
sm
N
->N
y
e
s
:
C
atholicis
m
y
e
s
:
c
ath
o
li
c
i
sm
yes: fascism,
p
la
g
iaris
m
yes: sexism,
escapis
m
y
e
s y
e
s
-
ist
y
es:
A
n
g
licist occasionall
y
:
y
es:
fa
sc
i
s
t
,
y
es: se
xi
s
t
,
y
e
s y
e
s
386
R
OC
HELLE LIEBER
Affix
es
s
egmental
phonological
e
ffects
stress effects attaches to
roots
attaches to
wo
r
ds
attaches to
non
-
native
bases
attaches to
native bases
N->N algebraist plagiarist
e
scapist
-
it
e
A->N
n
o y
e
s
:
c
osmopolite
yes: hematite,
c
osmopolite
y
e
s
:
subu
r
b
anit
e
y
e
s on
n
am
es
-i
t
y
A->N
y
es:
hi
stor
i
c
i
t
yy
es:
hi
stor
i
c
i
t
yy
es: atroc
i
t
y
,
h
ilarit
y
y
es: s
i
m
il
ar
i
t
y
,
purit
y
y
es occas
i
ona
lly
:
o
ddit
y
-
i
ve
V
->A
yes:
p
roductive,
all
us
i
ve
yes:
p
roductive
yes:
p
roductiv
e
yes:
s
uccessive
,
deco
rati
ve
yes onl
y
t
alkati
ve
-
iz
e
N
,
A->
V
y
es: historicize
y
es:
h
istoricize
,
i
mm
u
niz
e
y
es: baptize,
e
van
g
eliz
e
y
es: winterize,
s
tan
d
ar
d
iz
e
ye
s
ye
s
-l
ess
N
->
A
n
o no occasionally:
rec
kl
ess
yes: shoeless yes yes
-ly
N->
A
n
o no no yes: friendly yes: princely yes:
l
ordly
-
m
e
nt
V
->N
n
o no rarely?
t
es
tam
e
nt
y
e
s
:
development
y
e
s
:
d
evelopment
y
e
s
:
be
r
e
a
ve
m
e
nt
-
ness
A->
N
n
o no no
ye
s
ye
s
ye
s
-
or
y
V
->A
yes:
c
lassificatory,
s
atisfactory
yes: excretory yes:
perfunctory
yes:
c
ontradictory
yes no
-ous
N->A
yes: avaricious yes:
c
ourag
e
ous,
h
arm
o
n
ious
y
e
s
:
sonorous,
blasphemous
y
e
s
:
c
ourag
e
ous,
mu
r
de
r
ous
y
e
s y
e
s
-shi
p
N
->
N
n
o no no
y
es:
k
in
g
ship
y
es
y
es
-
so
m
e
A,N->
A
n
o no rarely:
g
ruesome
yes:
w
h
o
l
eso
m
e
o
ccasionally:
q
uarrelsome
yes
-u
r
e
V
->N
yes: closure no n
o
? yes: closure yes n
o
-
y
N->
A
se
l
do
m
:
w
orth
y
, pirac
y
no
r
arel
y
:
g
risl
y
?
y
es: sunn
y
,
s
alt
y
yes yes
Ta
b
l
e1
Suffixes and their bases
ENGLISH WORD
-
FORMATION PROCESSE
S
38
7
s
uffi
x
what this suffix attaches to what suffixes attach to it
-
a
bl
e Won’t atta
ch
to
–
le
wor
d
s. W
ill
attac
h
to
b
a
ses
in
–
en, -i
fy
, -ate.
-it
y
, -ness
-
a
g
e Tends to prefer unaffixed bases. Pla
g
(
1999, 76
)
notes that affixed Ns tend to
be abstract, to attaching another abstract
su
ffix
wou
l
dbe
r
edu
n
d
ant
.
-ize
?
-
al
(
N
)
L
ikes final stressed unaffixed
V
s. Note
t
hat verbal affixes are not stress-bearing.
-
al
(
A
)
Likes simplex Ns, but will also attach to
–ion, -ment, -or, -ure, -ant/-ent, -ance, -
or
y
, -ive.
-ize, -ism, -ist, -it
y
, -ness
-
an
,
-ian Prefers simplex bases. Also attaches to
–
ic
.
-ize, -ism, -it
y
, -ship
-
an
ce
A
ttaches to sim
p
lex Vs with final stress. -al
,
-less
-
ant
,
-ent Prefers sim
p
lex bases. -ize
,
-al
-
at
e
Atta
c
h
es
to
–ic
,
-ion
,
-ive -ion, -able, -ee, ive, -ory, -ous
-
ati
o
n Atta
c
h
es
t
o
affix
ed ve
r
bs
in –ize, -i
fy
. -al, -ize, -ist, -ism, -ate, -less
-
do
m
Atta
c
h
es
t
o
–er, -ess, -ee, -in
g
, -ist -ize?, -ish, -less
-
ed
Prefers sim
p
lex bases. -
ness
-ee
Prefers sim
p
lex bases. Will attach to
–
i
ze, -age.
-dom, -shi
p
, -hood, ism?, -ist?
-en (A)
Prefers sim
p
lex bases. -
ness
-en (V)
Attaches to monosyllabic bases ending
i
n an
obs
tr
ue
nt
.
-
er
Attaches quite freel
y
even to suffixed
f
o
rm
s
in –ize, -i
fy
, -en.
-ize, -dom, -ship, -hood, -ism, -ish, -less,
-l
y
-e
r
y
A
ttaches to consonant-final simplex
b
a
ses
-ish
?
-
es
q
u
e
A
ttaches to sim
p
lex bases. -ify
?
-ess
Attaches to sim
p
lex bases and –
er.
-dom, -hood, -less, -shi
p
-f
u
l
Attaches to monosyllabic Vs or
disyllabic Vs with final stress. Attaches
t
o simplex Ns or Ns in –th, -ing.
-
ness
-hoo
d
Prefers simplex bases, but will attach to
–ly, -y, -er, -ess, -ing
occasionally.
g
-
i
c
Atta
c
h
es
to
–ist
,
-ism
,
-or
.
-ist, -ism, -ize, -it
y
, -ness, -ian, -ate
-
if
y
Prefers monosyllabic bases or ones
ending in
y
.
C
an attach to –
ive
?
-ation
,
-able
-i
sh
Prefers sim
p
lex bases, but will attach to
-
y
-
o
r -
e
r
.
-
ness
-
i
s
m Atta
c
h
es
to
–ive, -ic, -al, -an, -ion, -er
.
-ic, -ize, -less
-i
s
t Atta
c
h
es
to
–ive
,
-ic
,
-al
,
-an
,
-ion
,
-
ment
.
-dom
,
-ic
,
-ize
-
it
y
D
oes
n
o
t atta
c
h t
o
–
or
y
r
.W
ill attach to
–
388
R
OC
HELLE LIEBER
suffi
x
w
hat this suffix attaches to
w
hat suffixes attach to i
t
i
ve, -ic, -al, -an, -ous, -able
.
Pr
efe
r
s
bases ending in non-glide consonants.
-
i
ve
Attacjes tp Latinate simplex bases
e
nding in t, d, s, or verbs in –
ate
.
-ication, -al, -ate, -ity, -ize, -ness
-
iz
e
W
ill not attach to
–ous
. Can attach to
simplex nouns and ad
j
ectives, as well as
th
ose
in
–
ive
,
-ic
,
-al
,
-an
,
-ism
,
-ist.
-ation
,
-ment
,
-able
,
-er
,
-ee
-l
ess
Avo
i
ds
–
ment
(Chapin 1967). Attaches
t
to
–
ion, -ing, -ism, -er, -ance, -dom, -
ess.
-ness
-ly
A
tta
c
h
es
t
o
–er
.
-ness
-
m
e
nt
A
ttaches to simplex Vs or Vs prefixed
w
ith
en-
or
be-
.
-
ness
A
ttaches freely to all ad
j
ective-forming
suffixes, as well as to sim
p
lex bases.
-
ory Attaches
to
–
ate
.
-al
-ous
Prefers simplex bases.
-
it
y
-sh
ip
A
tta
c
h
es
t
o
–
ee, -ian, -er, -y, -ess
-
some
Prefers simplex bases.
-u
r
e
Prefers simplex bases.
-al
-y
Prefers simplex bases.
-ness
Ta
b
l
e
2 Suffixes: what attaches to wha
t
Pre
f
i
x
Attac
h
es to non-native
b
ase?
A
ttac
h
es to native
b
ase?
a- yes: aseptic, amoral n
o
aft
e
r-
yes: afterimage yes: afterthought
ant
e
- yes: antechamber, anteorbital yes: anteroo
m
a
nti- yes: anti-hero, anti-communist yes: anti-clockwise, antibody
ar
c
h- yes: archenemy, archbishop n
o
a
u
t
o
-
y
es: auto
bi
o
g
rap
hy
rare
ly
: auto
h
arp
b
a
c
k-
r
are: backspace
y
es: backwater, backli
g
ht
be
-
y
es: bepowder, besie
g
e
y
es: befriend, bemoan
b
i- yes: bicentennial, bilingual yes: biyearly, biweekl
y
b
y-
yes: by-election yes: bystander, byword
c
ir
cu
m- yes: circumnavigate
r
ar
e: so
m
e
n
o
n
ce c
r
e
ati
o
n
s
in 1
7
th
c.
co
-
y
es: co-edit, co-defendant
no
cou
nt
e
r-
y
es: counterrevolution
y
es: counterwei
g
ht
de
-
y
es: dethrone
y
es: delouse
de
mi-
y
es: demiseason
r
are: demi
g
od
d
i
s
- yes: dishonour yes: distrust, dislike, dishearten
dow
n-
n
o
yes: downcast, downplay
e
n- yes: enchain yes: enshroud
E
NGLISH WORD
-
F
ORMATION PROCESSES
389
Prefi
x
A
ttaches to non-native base
?
A
ttaches to native base
?
e
x-
y
es: ex-pres
id
ent
y
es: ex-
h
us
b
an
d
extra-
y
es:
e
xtra
cu
rri
cu
lar n
o
fo
r
e-
y
es:
f
orecourt
y
es:
f
ore
g
o,
f
oreman
hy
per-
y
es: h
y
perph
y
sical, h
y
peractive n
o
in
- yes: intolerant, impossible
no
i
nt
e
r- yes: interchange, interact
r
ar
e:
int
e
rt
w
in
e
m
eta- yes:
m
etaphysics no
m
i
c
r
o
- yes: microbiology
r
ar
e:
mi
c
r
os
kirt
m
i
s
-
y
es: mismana
g
e,
y
es: misunderstand, mislike, misdeed
mu
lti-
y
es: multicellular
y
es: multi-armed
no
n-
y
es: non-resident
y
es: non-heathen
o
ff-
yes: offcolor yes: offhand
on-
r
are: onmarching yes: onset
ou
t
-
yes: outachieve, outpatient yes: outbid, outhouse
ove
r
-
yes: overcompensate yes: override
p
ost- yes: postclassical yes: postwa
r
pr
e
-
y
es: preclassical, pre-existence
y
es: prewar, preschool
pro-
y
es: proform, pro-abortion
y
es: pro-war, pro-life
p
seudo- yes: pseudo-intellectual yes: pseudo-book, pseudo-word
re
- yes: reassure yes: rewash
re
tr
o
- yes: retroanalysis, retrovision
no
se
mi- yes: semifluid, semiannual
r
are: semiweekl
y
sub
- yes: subhuman, subcommissioner rare: sublet
s
u
per
- yes: superhuman yes: superclean
s
upra-
ye
s
:
supraorbital no
t
ran
s
-
y
es: transpolar
y
es: tranship, trans-Mississippi
un
-
y
es: unintelli
g
ible, untenable
y
es: unhapp
y
u
n
de
r-
yes: underage, undergraduate yes: underlie, underhanded, undershirt
up
-
r
are: upcountry yes: uphold, upbeat, uprising, upland
Ta
b
l
e
3 Pre
f
ixes and their bases
B
efore we go on to look at cohorts of various affixes, some summary remarks
are in order. First, it appears that the distinction between native and Latinate is more
s
alient – has wider effects – in suffixes than in prefixes. The vast ma
j
ority of
prefixes in En
g
lish are of Latinate or Greek stock (see Bauer 2003). Native prefixes
are often prepositional in nature. Whatever their ori
g
in, En
g
lish prefixes tend to
attach to both native and Latinate bases. As prefixes
g
enerall
y
bear some stress
t
hemselves, they also tend not to perturb the stress pattern of their bases. And with
t
he exce
p
tion of
in-
an
de
n
-
,
which assimilate to the initial consonant of their base
,
prefixes neither change form themselves, nor have segmental phonological effects
o
n their bases. Prefixes, in addition, can often attach to other prefixes, and even to
themselves, provided their meanings are a
m
enable to scalar interpretations. So
390
ROC
HELLE LIEBER
f
o
rmati
o
n
s
lik
e
r
e
p
rehea
t
,
re
r
eheat
,
o
r
prepreheat
are reasonable formations.
t
Prefixes are rarely category-ch
a
n
ging in English; the prefixes
en-
an
d
de-
ar
e
n
o
ta
b
l
e
exce
p
tions here.
2
Generally, however, prefixes will attach to more than one category
o
f base and return a derived word of the same category as the base. The prefix
counte
r
-
,
for example, attaches to nouns (counterexample), adjectives
(
counte
r
intuitive
)
, and verbs
(
countersign), with no change of category.
W
ith suffixes, in contrast, those which are of Latinate ori
g
ins sometimes prefe
r
Latinate bases (e.
g
., -
acy
,
-
ation
), althou
g
h others are equall
y
at home on either
n
ative or Latinate bases (e.
g
., -
able
,
-
age
)
. Some attract stress from their bases, o
r
o
therwise change stress pattern, some hav
e
other phonological effects. Others affect
n
either stress nor base phonology. Native suffixes are stress and phonology neutral.
Latinate affixes almost always attach to roots
;
native affixes often do as well.
Suffixes, in general, are less amenable to iteration than prefixes. Although some of
t
h
e
Latinat
esu
ffix
es c
an atta
c
h t
o
e
a
c
h
o
th
e
r
so
that f
o
rmati
o
n
s
lik
e
organizationaliz
e
are possible, these derivational loops are restricted by pragmatic
c
onsiderations such as the ultimate interpretabilit
y
and usefulness of the resultin
g
w
ords. See Lieber
(
2004, cha
p
ter 6
)
for an extensive discussion of this issue. As
A
ronoff and Fuhrhop (2002) point out, there are a number of suffixes in En
g
lish that
attach only or almost exclusively to roots or unsuffixed words (e.g., verb-forming
-
al
,
-ant
/
-en
t
,
-
ance
.
Whether it is true, as they argue that English obeys the
‘Monosuffix Constraint’, namely that “[s]uffixes that select Germanic bases select
u
nsuffixed bases (2002: 473)”, is open to dispute, however. Suffixes, unlike
prefixes, are frequently (although not always) category-changing, and more often
t
han not attach onl
y
to a sin
g
le cate
g
or
y
of base.
A
ffixes can often be
g
rouped into what h
a
ve bee
n
c
all
ed co
h
o
rt
so
r ri
v
al
se
t
s
(
Van Marle 1985). Within these sets, affixes share s
y
ntactic and semantic
c
haracteristics, often rather neatly partitioning the set of available bases (see Plag
1999). We will first look at groups of prefixe
s
, grouped either by cohorts of this sort,
o
r by semantic categories like ‘locational’, and then go on to do the same fo
r
su
ffix
es.
3
.1 Pre
f
ixation
B
auer (2003) points out that it is often difficult in En
g
lish to distin
g
uish prefixes
from the first elements of compounds. Prefixes
g
enerall
y
bear stress in En
g
lish, and
o
ften behave as distinct phonological words. Increasingly, as Bauer points out,
e
l
e
m
e
nt
s suc
h a
s
agri-, eco-, Euro-, psycho-, techno
-
an
d
th
e
lik
e
ar
eused
a
s
c
ombining forms, and perhaps represent elements that should be classed
s
ynchronically with prefixes. Indeed, this seems to be an open question. Certainly,
so
m
e
r
ese
ar
c
h
e
r
s
ha
ve
tr
e
at
ed so
m
e
f
o
r
ms we will discuss below as compounding
elements. Here, I will not make too fine a distinction between prefixes and
c
ombinin
g
forms, althou
g
h for the sake of brevit
y
I will stick to t
y
pes that are
2
The prefix
be-
also changes category (
b
edew
,
becalm), although it is unproductive in present-day
English.
ENGLISH WOR
D
-
FORMATION PROCESSES
391
g
enerally or at least often classed as prefixes. See Lehrer (1995) as well fo
r
d
iscussion of the distinction between prefixes and combining forms in English.
G
enerally, prefixes attach to words, rather than roots, although Latinate forms
l
ik
e
s
uper-, supra-, trans-, circum-, inter-, intra-, sub-, an
d
th
e
lik
edo
atta
c
h t
o
Latinate roots
(
transport, circumvent, intervene, sub
j
ec
t
,
etc.
)
, albeit not with
c
ompositional meanin
g
s. In fact, it has lon
g
been a controvers
y
amon
g
m
orphologists whether to treat such forms as formally complex, since
synchronically they are not semantically complex (see, for example, Aronoff 1976).
H
ere I will confine myself to prefixation to words, leaving aside this problem.
3
.1.1 Negative prefixes (un
-
,
in-
,
non-
,
de-
,
d
is-
)
Negative prefixes have been discussed quite extensively in the literature over the
years; see for example Zimmer (1964), Marchand (1969), Algeo (1971), Mayno
r
(1979), Andrews (1986), Horn (1989/2001, 2002, forthcoming), and Lieber (2004).
Ea
c
h
o
f th
ese
affix
es
atta
c
h
es
t
o
a
t
least two lexical categories of base:
t
(
14
)
on A on N on V
de-
de
l
ouse
de
militariz
e
de
i
ce
deco
ntaminat
e
dis
-
d
i
scou
rt
eous d
i
sco
mf
o
rt
disloyal disrespect
in-
inaccurate incapacity
infinit
e
ina
c
ti
o
n
non-
n
o
n-m
o
ral n
o
n-
s
m
o
k
er
n
o
n-
v
i
o
l
e
nt n
o
n-
v
i
o
l
e
n
ce
O
f course
,
these affixes do not exhibit equal productivity on all these categories.
The
p
refixes
non-
an
d
de-
are both quite productive on nouns, the former bein
g
non-
c
ate
g
or
y
-chan
g
in
g
, the latter chan
g
in
g
nouns to verbs. Non
-
also attaches
q
uite
freel
y
to ad
j
ectives. Un
-
an
d
in-
are both productive on ad
j
ectives, with
un-
bein
g
t
he more productive of the two. As Horn (2002, forthcoming) has shown, however,
un
- exhibits a surprising productivity on nouns
i
n present-day English. As we will
s
ee below
,
un-
also attaches productively to verbs, albeit only within a certain
se
manti
cc
la
ss o
f
ve
r
bs.
D
e-
fa
vo
r
sde
ri
ved ve
r
bs
in -ize
,
-at
e
,
and -if
y
.
D
is-
may
attach to a wide variety of verbs.
The semantics of this cohort of prefixes has been of particular interest to
m
orpholo
g
ists. D
e-
is t
y
picall
y
character
i
zed as
p
rivative, un-
,
in-
,
an
d
non-
a
s
n
e
g
ative on ad
j
ectives and nouns,
dis-
as ne
g
ative or reversative on verbs. When
attached to verbs, the meaning of
un-
is generally described as reversative. At issue
un-
u
n
b
r
e
aka
b
l
eu
n
e
a
se
u
n
d
r
ess
unhappy uncola uncork