?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation
Подождите немного. Документ загружается.
392
R
OC
HELLE LIEBER
i
n the literature, however, is whether these meanings are different and distinct, or
whether they are the same, with differe
n
c
es arising from the category and semantics
o
f the particular bases to which they attach. Marchand (1969) treats different
n
uances of meaning separately in each of
t
hese affixes, and Dowty (1979) argues in
e
ffect that negative and reversative
un
- are separate homophonous prefixes. Horn
(
1989/2001, 2002, forthcoming) and Lieber (2004) follow earlier work of Mayno
r
(
1979) and Andrews (1986) in ar
g
uin
g
that this cohort of affixes shares a central
c
ore of meanin
g
, with the extensive pol
y
sem
y
that the
y
displa
y
followin
g
from the
s
emantics of the base to which the
y
attach
.
3
I believe that a good case can be made for polysemy in this cohort of prefixes,
r
ather than homophony. Let us look first at the notion of negativity, which itself is
n
ot necessarily a unitary concept. As Horn (1989/2001) points out, we must
distinguish contradictory negation
f
r
om
contrary negation
.
In contradictories
,
s
omething is either X or not-X; the pos
i
tive and negative counterparts admit of no
m
iddle ground. For example something may be eithe
r
b
r
eakable
or
unb
r
eakable
,
and
so
m
eo
n
ee
ith
er
pregnant
or
t
n
on-pre
g
nan
t
, but there is nothin
g
in between. In
c
ontrast, someone ma
y
be neithe
r
h
app
y
n
or
unhapp
y
,
and a movie ma
y
be neithe
r
violent
nor
t
non
-
violent
(at least for me). These latter negatives are contraries. In
t
terms of logic, contradictories obey the Law of the Excluded Middle
,
while
c
ontraries do not. In Lieber (2004) I argue that negative prefixes may deliver either a
c
ontradictory meaning or a contrary meaning, depending on the meaning of thei
r
base. Which interpretation arises depends on whether the base ad
j
ective can be
i
nterpreted as gradable or scalar. If a
n
ad
j
ective is amenab
l
e to a gradable
i
nterpretation (as happ
y
an
d
violent
are for me), the resulting negative-prefixed form
t
will have a contrar
y
meanin
g
; if the ad
j
ective is not scalar or
g
radable in
i
nterpretation, the ne
g
ative-prefixed form will be interpreted as contradictor
y
. This
i
s true regardless of which negative prefi
x
we use. Speakers may of course differ on
which adjectives they feel to be gradable, and even whether a given adjective is
g
radable depending on nuances of meaning. For example, the adjective
Ame
r
ican
i
s
n
on-gradable to me if it merely denotes
n
ationality: someone either holds American
c
itizenship or not. But to the extent that
so
m
eo
n
e
’
sbe
ha
v
i
o
r
c
an
be c
hara
c
t
e
riz
ed
a
s
‘
American’, that concept is
g
radable for me, and hence
un
-A
me
r
ican
in that
se
n
se
h
as a contrar
y
rather than a contradictor
y
interpretation.
It has been noted in the literature
(e
.g
., Al
g
eo 1971, Horn 1989, 2001) that some
ad
j
ectives can be prefixed b
y
un-
(
o
r
in-
)
and
non-
,
with the ne
g
ated ad
j
ective
h
aving slightly different nuances of meaning. When this occurs,
non-
generally
t
argets a non-emotive or non-affective sense of the base, while
un-
or
in
-
p
icks out a
m
ore emotion-laden sense. So, if we say that someone is
non-professional
we
l
m
erely say that they do not work as a doctor, lawyer, professor, or the like. But if we
c
all th
em
u
nprofessional
,
we are accusing them of questionable conduct in the
workplace
.
3
Of the references mentioned, onl
y
Lieber (2004) treat
s
all
o
f th
ese
affix
es.
An
d
r
ews co
n
ce
ntrat
es o
n
un-
an
d
de-
,
and Horn on
in-
,
un
-
,
an
d
non-
.
ENGLISH WOR
D
-
FORMATION PROCESSES
393
Regarding the prefix
un-
on nouns, Horn (forthcoming 1-2) points out that there
are a number of special nuances tha
t
can arise. Specifically, while an
t
uncola
an
d
an
unhotel
are both in some sense not the nouns denoted by their bases (respectively a
l
c
ola or a hotel), the
y
require sli
g
htl
y
different interpretations: “… a
g
iven un-noun
r
efers either to an element
j
ust outside a
g
iven cate
g
or
y
with whose members it
s
hares a salient function (e.
g
.
un
-
cola
) or to a peripheral member of a
g
iven cate
g
or
y
(
a
n
unhotel
is a hotel but not a good exemplar of the category…).”
l
U
n- n
ou
n
sof
both sorts appear to be having a vogue in present-day English.
W
e may now discuss whether a distinction
needs
t
obe
ma
de be
t
wee
n r
eve
r
s
ati
ve
and negative
un
-
.
W
e have seen that we can characterize
un-
as negative when it
attaches to ad
j
ectives and nouns. With respect to verbs,
un-
shows a special
preference for causative/inchoative ver
b
s, and specifically those causative
/
i
nchoative verbs which impl
y
a result which is not permanent. So
unbutton
or
unco
r
k
are possible
k
un-
verbs
,
but
*
unexplode is not. As has been pointed out before,
i
t is possible to interpret these verbs as involvin
g
ne
g
ation of an impermanent result,
so
that
unbutton
in effect means ‘cause to be not buttoned’. Note, interestingly, that
t
he
p
refix
dis
-
,
which typically returns a negative interpretation when attached to
s
tative verbs
(
dislike
means ‘not like’
)
, returns a r
e
v
ersative meaning when attached
t
o a verb which implies an impermanent result. So
dis
r
obe
m
e
an
s
‘
c
a
use
t
obe
n
o
t
r
obed’,
j
ust as
und
r
ess
means ‘cause to be not dressed’. This suggests that the
r
eversative meanin
g
is not meanin
g
-specific to the verbal affix
un-
,
but rather arises
from a
g
eneral ne
g
ative affix when attached to the ri
g
ht sort of verbal base.
The final meanin
g
expressed in this cohort of prefixes is the privative meanin
g
of
de-
.V
erbs in
de-
can be interpreted in some sense as negative causatives. So to
debug
is to ‘cause to not have bugs’, and to
g
debone
is to ‘cause to not have bones’
,
and so on. The latter exam
p
le is of some interest for other reasons: in this form the
prefix
de-
is actually redundant, as to
bone
it
se
lf m
e
an
s
‘
c
a
use
t
o
n
o
t ha
ve bo
n
es
’
.
This does not
,
I believe
,
mean that
de-
has no meaning here, but only that its
addition to the word does no more than intensify the negative causative meaning
alread
y
inherent in
bone
. In an
y
case, if privatives can be classed as ne
g
ative
c
ausatives, the
y
clearl
y
belon
g
to the same
g
eneral cohort of prefixes as
in-
,
un-
,
non
-
,
an
d
dis-
.
O
ne last issue has figured prominently in the literature on negative prefixes. At
l
east since Jespersen (1965) and Zimmer (1964), it has been noted that negative
p
refixes tend not to attach to bases whic
h
already have negative connotations. So we
h
a
ve
unhapp
y
bu
t n
o
t
unsad
,
unatt
r
active
,
but not
u
nugl
y
. As Zimmer points out,
h
owever, this is only a tendency, albeit a fairly strong one. There are a significant
n
umber of exceptions to this observation, if one looks for them, among them words
l
ik
e
unhostile, unvul
g
ar, an
d
incorrupt
that Zimmer himself cites (1964: 30, 35-37).
t
3
.1.2 Locational prefixes
A
mong the locational prefixes are the following:
over-
,
unde
r
-
,
out-
,
back-
,
down-
,
up-
,
o
ff-
,
on-
,
f
ore
-
,
inte
r-
,
sub-
,
supra
-
,
t
r
ans
-
.
Th
ese
ha
ve
n
o
t r
ece
i
ved
a
s
m
uch attention in the literature as have ne
g
ative affixes. To the extent that this
c
ohort of affixes has attracted some attention
,
that attention has been focused
394
R
OC
HELLE LIEBER
primarily on two issues: polysemy and the effects of some locational prefixes on
argument structure. See Fraser (1976), Bresnan (1982), Walinska de Hackbeil
(1986), Lakoff (1987), Williams (1992), Lehrer (1995), Roeper (1999), Tyler and
Evans (2001, 2004), and Lieber (2004) for different approaches.
A
ll of these affixes attach to multiple categories, as the chart in (15) illustrates:
(
15
)
on N on V on A
dow
n-
dow
n
be
at
dow
n
s
hift
dow
n
c
a
s
t
up- upcountry uplift, upgrade
off- offshoot, offprint offload offwhite
f
ore-
f
oretaste
,f
oremast
f
oresee
b
y
- b
y
stander, b
y
wa
y
b
yg
one
ant
e
- ant
ec
ham
ber
circum- circumnavigate circumpola
r
e
xtra-
e
xtrat
e
rr
es
trial
inter- interface intermarry interdependent
sub
-
subb
a
se
m
e
nt
sub
a
cu
t
e
super- superstructure supersaturate supersensual
supra- supracouncil supramolecula
r
trans- tranship transpola
r
All of the locational
p
refixes that corres
p
ond to
p
re
p
ositions exhibit
s
ubstantial polysemy, as do most of the Latinate locational prefixes. It has been
pointed out many times that prepositions in English are multiply polysemous.
W
ithin the literature on Cognitive Linguistics, for example, extensive explorations
o
f prepositional polysemy can be found in Lakoff (1987), and Tyler and Evans
(
2001, 2004
)
, and within other frameworks discussions can be found in Walinska de
H
ackbeil
(
1986
)
, Bierwisch
(
1988, 1996
)
,
J
ackendoff (1996), Landau (1996), Talm
y
(2000) and elsewhere. The pol
y
sem
y
of th
e
prepositional prefixes lo
g
icall
y
stems
from this generally property of polyse
m
y among prepositions, although most of the
prefixes exhibit a somewhat smaller range of polysemy than do their prepositional
c
ounter
p
arts.
This pattern can be well-illlustrated with the case of
ove
r. As a preposition,
ove
r
h
as purely locational uses (
T
he hummingbird hovered over the flower
)
, as well as
l
ocational uses that add a notion of ‘completion’ (
W
e threw a blanket over her
)
. The
preposition is extended as well to mean ‘to excess’ (
O
ur article is over the pa
g
e
limit
). In addition, as Tyler and Evans (2004) point out, there are a variety of
t
extended meanings including repetition (
Do it ove
r
(
(
)
, transfer
(
We handed the money
ove
r
)
, and
p
ower
(
T
he guard kept control over the prisoners
)
. The first three of the
prepositional meanings occur in prefixal uses as well, as Lieber (2004) points out:
over- overcoat
,
overlord ov
e
r
eat, overfly overproud,overfull
under- underarm
,
undershirt
u
ndercut underrip
e
out- outhouse
,
outfield outdo
,
outrun
back- backlash, backstitch backstroke backbreakin
g
EN
G
LI
S
H
WO
RD
-
F
O
RMATI
O
N PR
OC
E
SS
E
S
395
(
16
)
over- locational on N overlord, overarm
on V overfly, overarch
l
oc
ati
o
nal an
d
on N overcoat
,
overshoe
contemplative on V overrun, overpowe
r
excess on N overconfidence
,
overdose
on V overdevelop, overcharge
on A overfond, overgenerous
Parall
e
l t
o
ove
r
-
is the
p
refix
unde
r- which shares with its
p
re
p
ositional mate
l
ocational (lower than) and quantitative (less than) meanings:
(
17) under- locational on N underarm, understory
on V underthrust
,
underscore
on A undersigned, underhanded
less than on N underem
p
hasis
on V underbid
,
underachieve
on A underprivileged
W
hat is odd, however, is that the prefix
out-
,
while exhibiting meanings that
g
enerally parallel those of its prepositiona
l
counterpart, exhibits one meaning – its
predominant one in fact – that is not
f
ound in the preposition. The former meanings
ff
i
nclude a purely locational sense of exteriority (
S
he walked out the door
)
, as well as
a sense of non-existence
(
We’re out of milk
), and an exclusion sense (
kk
We voted the
Republicans out
); the prefixal counterparts are illust
t
t
r
ated in (18). With re
g
ard to the
u
niquel
y
prefixal sense illustrated in (19), the OED states, “It is not ver
y
clear how
t
his use arose
,
or to what sense of
out
it is to be referred”:
t
In its comparative sense,
out-
attaches onl
y
to verbs, and is accompanied b
y
the
s
ystematic addition of an argument to the argument structure of its base; we will
r
eturn to this latter
p
henomenon below.
(
18
)
out- locational on N outback, outhouse
on V outbreak
,
outflow
on A outdoor, out
g
oin
g
non-existence on
V
outblot
exclusion on N outparty
(
19) out- comparative on V outhit, outdebate
396
R
OC
HELLE LIEBER
Th
e
f
o
rm
s
back-
,
up-
,
and
down-
again parallel their prepositions when they
a
pp
ear as
p
refixes. As a
p
re
p
osition,
back
has its central locational sense (
k
W
e turned
back
), as well as a temporal sense (
k
k
We were happy back then
)
, and a sense of
r
eversal
(
We had to put it back
)
. These three senses appear
k
k
i
n prefixed forms as well,
although only the locational sense appears with any frequency:
(
20
)
back- location
(
rear
)
on N backchamber, backdraft
on V backslide
,
backfire
on A backlookin
g
temporal on N backpayment
r
eve
r
se
o
n N
b
a
c
kfill
on
V
backfill
Similarl
y
,
up
an
d
down
both have central locational senses that appear frequentl
y
i
n their prefixal form, as well as a quantitative meanin
g
(
T
urn up/down the volume
on the
r
adio
) that appears onl
y
infrequentl
y
in prefixed forms:
(
21
)
u
p
- location on N u
p
land, u
p
curve
on V uplift
quantity on V upgrade, uprate
(22) down- location on N downcountr
y
, downbeat
on V downturn
,
downshift
quantity on V downgrade, downplay
The preposition
off
exhibits a variety of senses, among them its chief directional
f
sense
(
He walked off the field
)
, an ‘abnormal’ sense
(
d
d
This milk smells a bit off
), a
f
f
separation sense (We cut off the dead flowers), and an interruption sense (
The
interviewer cut her off
). The first three of these can be found in prefixed forms,
f
f
althou
g
h onl
y
the first two appear with an
y
frequenc
y
:
(
23) off- direction on N offglide
on V offload
,
offset
on A offputting
abnormal on A off-white
,
off-flavo
r
separation on N offline
I
t might be thought that
on
would occur prefixally, as does
off
, and the other
f
f
p
re
p
ositional forms detailed above. Howev
e
r
, exce
p
t for a few formations like
ENGLISH WORD
-
FORMATION PROCESSE
S
39
7
oncoming
or
g
onlooke
r
,
this preposition seems hardly to have been productive as a
prefix. Marchand (1969: 116) mentions it only as ‘a weak formative’.
The remaining locational prefixes
–
b
y-
,
ci
r
cum
-
,
ext
r
a
-
,
ante
-
,
fore
-
,
inte
r
-
,
sub-
,
supra
-
,
super
-
,
and
t
r
ans-
– sometimes exhibit as man
y
nuances in meanin
g
as
prepositional prefixes, but locational meanin
g
s usuall
y
seem to be central. The
L
atinate prefix
sub-
,
for example, exhibits the
e
xpected locational sense (
s
ub
f
loor
)
,
as well as extended senses of subordinacy (
subdeacon
)
, ‘
p
art of’
(
subcommittee
)
,
bordering (
subu
r
ban
), and smaller quantity
(
subaudible
)
. The native
p
refix fore
-
occasionally has temporal as well as locational force (
forecast
(
(
), as does
t
t
inte
r
-
(
inte
r
wa
r
)
and
ante-
(
antedawn
)
.
S
upra
-
has quantitative extensions (
s
uprarational,
s
upracentarian) in addition to its locational uses, and in fact is used in many ways as
a synonym of the prefix
s
uper
-
which has largely extended itself beyond locational
uses; so in addition to locational examples like
s
uperstructur
e
,
we find an excess
m
eanin
g
(
s
upersensitiv
e
), and a superlative meanin
g
(
supermom
). The prefixes
t
r
ans
-
,
b
y-
,
ci
r
cum
-
,
an
d
ext
r
a-
seem the least pol
y
semous of this cohort.
For the most
p
art, when locational
p
refixes attach to verbs, they have no effect
o
n the argument-taking possibility of their bases:
(
24) under- We estimated the damage
We underestimated the damage
back- She pedalled (the bic
y
cle)
She backpedalled (the bic
y
cle)
down- I shifted (the gears)
I downshifted (the gears)
up- We graded the goods
We upgraded the goods
off- The
y
loaded the ship (with wood)
/
The
y
loaded wood onto the ship
They offloaded the ship (of wood)
/
They offloaded wood from the ship
f
o
r
e
- N
oo
n
ec
an
see
th
e
tr
ee.
N
oo
n
ec
an f
o
r
esee
th
e
f
u
t
u
r
e
super- We saturated the cloth
We supersaturated the solution
trans- They shipped the goods
They transhipped the goods
T
he prefixes
out-
an
d
ove
r
-
have a somewhat more complicated effect on the
argument structures of verbs to which they attach. It has been noted in the literature
398
R
OC
HELLE LIEBER
as early as Bresnan (1982) that comparative
out-
systematically adds an argument to
i
ts base verb if that verb is intransitive to begin with. For verbs that are normally
transitive, the nature of the internal argument changes:
Similarly, as first noted in Walinska de Hackbeil (1986), in its locational, o
r
l
ocational/competive meaning,
ove
r
-
often (but not always) adds an argument:
(
26) over- *The plane flew the field (locational)
The plane overflew the field
*The enem
y
ran the battlefield (locational/completive)
The enemy overran the battlefield
I
n this meaning,
ove
r
-
prefers intransitive bases (e.g.,
a
r
ch
,
flow
,
hand
,
fl
y
,
r
ise
,
l
a
y
,
soa
r). Lieber (2004) suggests that
ove
r
-
comes with an argument which is added
to that of its base, making transitives of intransitives. Where it does attach to
transitive verbs, Lieber su
gg
ests (2004: 131)
t
hat the prefixal ar
g
ument is linked to
the second ar
g
ument of the verb, imposin
g
its semantic characteristics on that
ar
g
ument. It is likel
y
that comparative
out-
could be anal
y
zed in a similar wa
y
.
W
ith the excess sense
,
ove
r
-
most frequently has no effect on the argument
s
tructure of its base verb
;
it attaches to either transitive or intransitive verbs with the
s
ame sort of verb resulting:
(27) over- We slept (excess)
We overslept
The
y
developed the area
The
y
overdeveloped the area
The one verb that behaves oddly in this respect is
eat
;
although
eat
can occur
t
with an object or not,
ove
r
eat
cannot have an object:
t
(
28
)
We ate
W
e overate
We ate pickles
*We overate pickles
L
ieber
(
2004: 132
)
ex
p
lains this case as follows:
(25) out- Fenster ran (quickly)
*Fenster outran (quickly)
F
e
n
s
t
e
r ran th
e
B
os
t
o
n Marath
o
n
*F
e
n
s
t
e
r
ou
tran th
e
B
os
t
o
n Marath
o
n
*F
e
n
s
t
e
r ran L
e
titia
F
e
n
s
t
e
r
ou
tran L
e
titia
ENGLISH WORD
-
F
ORMATION PROCESSES
399
I
t is interesting that the internal argument
i
s disallowed only when the “excess” part of
t
he semantic bod
y
of
ove
r- would implicitl
y
have to tak
e
scope over that ar
g
ument. If
s
he overate pickles were to be acceptable, it would ha
v
e to have the meanin
g
“she ate
t
oo man
y
pickles.” On the other hand, in the more t
y
pical cases where
ove
r- ha
s
n
o
e
ffect on ar
g
ument structure, the “excess”
m
eanin
g
added b
y
the prefix is t
y
picall
y
i
nterpreted as havin
g
scope over the verb phrase. That is, the
y
overworked the peasants
c
an be paraphrased as “the
y
wo
r
ked the peasants too hard” and not as “the
y
worked too
m
any peasants.” …although I am not sure how to formalize this observation, it seems
plausible to say that
over-
does not actually delete an argument when it attaches to the
ve
r
bb
a
se
eat
,
but rather chooses as its base the i
n
transitive form of the verb, avoiding,
i
f it possibly can, taking scope over an overt object.
Another case in affixation of a locational
p
refix that can im
p
ose restrictions on
i
ts ar
g
uments is the prefix
inte
r
-
:
(28) The villagers intermarried.
The villagers and the city-dwellers intermarried.
Th
e
tri
be
int
e
rmarri
ed.
*The villager intermarried.
F
e
n
s
t
e
r marri
ed
L
e
titia
.
*F
e
n
s
t
e
r int
e
rmarri
ed
L
e
titia
.
These effects stem not so much from the addition or deletion of an argument, but
from the quantificational nature of the arguments. Specifically,
inte
r
-
re
q
uires
p
lural,
c
onjoined, or collective arguments.
A final issue concerning locational prefixes is raised in Roeper. He notes (1999:
36-37) that structures of the sort illustrated in (29a) are generally preferred to those
i
n
(
29b
)
:
(
29
)
a. the outbreak of disease
th
eou
tfl
ow o
f f
u
n
ds
b. *the handout of good examples
*the carryout of food
Grammaticality
j
udgments in these examples are as indicated in Roeper (1999).
W
orking from a minimalist perspective that privileges anti-symmetrical structures
(Ka
y
ne 1994), he attributes the putative difference in
g
ra
m
maticalit
y
to a structural
constraint; prepositional prefixes (or what
he calls ‘leftward-moved particles’)
o
ccup
y
a Specifier position, which c-commands a complement position. Particles
which follow are incor
p
orated into their base and cannot themselves c-command a
c
omplement. The anti-symmetrical structure that Roeper has in mind is given in
(
30
).
The difficulty with this analysis is that
j
udgments on the data are not at all solid.
For example, it is possible to find many attestations on the web of phrases like
the
carr
y
out o
f
(X
)
or
the sleepover o
f
(Y
)
– precisel
y
the forms that this anal
y
sis rules
o
ut. It remains to be seen, then, if minimalist treatments of prefixed words are in an
y
wa
y
illuminatin
g
.
400
R
OC
HELLE LIEBER
(
30
)
Word
S
p
ec Head
|
out break PP-complement
o
f
d
i
se
a
se
3.1.3 Temporal and aspectual prefixes
English has a number of prefixes that might be categorized as primarily temporal
o
r aspectual, all of them from the Latinate sector of the morpholo
gy
. The prefixes
p
re
-
,
p
ost
-
,
ex
-
,
and
ret
r
o-
clearl
y
add a temporal
m
eanin
g
to their bases:
p
re
-
u
suall
y
means ‘before’,
p
ost
-
‘after’
,
ex-
‘formerl
y
’, and
ret
r
o-
‘
b
a
c
k
w
ar
ds
’
.
O
ccasionally, pre
-
,
pos
t-
,
an
d
ret
r
o
- ha
ve
l
oc
ati
o
nal
uses
a
swe
ll
.
(
31
)
on N on V on A
pre- preschool, prerinse preheat, presift prehistoric
ex- ex-ma
y
or, ex-president
retro- retrochoir retrofit
,
retroact retro-ocula
r
W
hile these prefixes have not received too much attention in the literature
;
Southerland
(
1994
)
discusses the disc
o
urse function of new coinages in
pre-
,
and a
number of these prefixes are mentioned in Lehrer (1995) and Keyser and Roepe
r
(
1992
).
This is not to sa
y
that the
y
are uninterestin
g
. Writin
g
about the Catalan and
B
as
q
ue e
q
uivalents of
pre-
,
Gracia and Azkarate
(
1998
)
make a number of
o
bservations that are potentially of interest in the lexical semantic analysis of words
formed with this cohort of prefixes. They note that words in pre
-
an
d
post
-
in
R
omance (and hence in English, since these prefixes belong to the Latinate sector o
f
t
he morphology) are all right-headed, in the sense that it is their base, rather than the
prefix that determines the category (and the morphosyntactic class) of the derived
w
ords. Nevertheless, only some forms in
pre-
an
d
pos
t-
seem to be semantically
h
eaded b
y
their bases. That is, a
preschool
is a kind of school, and the
l
p
ost-abdomen
i
s a
p
art of the abdomen. Prehistor
y,
however, is not a period of histor
y
, but rather a
period of time before histor
y
. Similarl
y
, music that is
preclassical
is not classical.
l
G
racia and Azkarate (1998) suggest that the latter type should be analyzed as
syntactically endocentric, but semantically exocentric.
The
p
refix
ex-
is also not without interest. First, it a
pp
ears that
ex-
i
s
far m
o
r
e
c
omfortable attaching to stage-level nouns (
mayor
,
president
)
than to individual-
tt
l
evel nouns
(
car
). For example, an
ex-mayor
is someone who was formerly mayor,
r
bu
t an
ex
-
car
,
if it’s an
y
thin
g
at all, can onl
y
be interpreted as a twisted pile of scrap
post- posteternity postdate, postpose post-adolescent
ENGLISH WOR
D
-
FORMATION PROCESSES
401
m
etal. Oddly, however, the prefix
ex-
can attach quite comfortably to an individual-
l
evel noun as long as it can take scope over something stage-level other than the
n
oun; that is, it is perfectly normal to talk about my ex-car, with the prefix taking
s
cope over the possessive pronoun. Gracia and Lieber (2000) discuss these facts
briefl
y
, without arrivin
g
at a satisfactor
y
formal anal
y
sis, however.
F
ar more attention has been devoted to the prefix
re-
,
both in terms of its s
y
ntax
and in terms of its semantics. With regard to syntax, Keyser and Roeper (1992)
argue on the basis of facts concerning this prefix that it is generated postverbally in
what they call the ‘abstract clitic’ position, and then moved to a position before the
v
erbal head. They cite as evidence for this analysis that the prefix
re-
appears to
o
ccur in complementary distribution with items like dative nouns, particles, and
i
diomatic nouns, all of which occur post-verbally:
(
32) a. We
g
ave him mone
y
.
b. *We re
g
ave him mone
y
.
c. *W regave ourselves up.
d.
W
e lost touch.
e. *
W
e relost touch.
Lehrer (1995: 147) points out a number of problems with this analysis. The most
o
bvious is that it is possible to use the prefix
re-
recursivel
y
; formations like
re-
reread
are not at all unusual. Keyser and Roeper’s analysis depends on the
d
assumption that onl
y
one item can occup
y
the Abstract Clitic position; recursion of
the prefix should therefore not be possible. Further, Keyser and Roeper suggest that
ve
r
b
al
un-
might also be argued to occupy the Abstract Clitic position. If so, it also
s
h
ou
l
d
n
o
t
co
-
occu
r
w
ith
re-
. However, given the right pragmatic circumstances, it
s
eems possible to combine the two; for example to
unrefill
something seems
l
possible.
Others have been more concerned with the semantics of this prefix than with its
sy
ntactic properties. Marchand (1969: 189-90) su
gg
ests that “
re-
does not express
m
ere repetition of an action; it connotes
t
he idea of repetition onl
y
with actions
c
onnected with an ob
j
ect.” He
g
oes on further to su
gg
est that
re-
r
ec
tifi
es
a
previously imperfect or inadequate result, as in
rew
r
ite
,
or reverses a
f
ormer state
,
as
in
r
e
p
ossess
.
Smith (1997: 179) suggests that
re-
is a ‘telic’
p
refix:
The verbal prefix
re-
(to do over a
g
ain) is telic: it appears in constellations with verbs
suc
h a
s
r
eo
p
en, reevaluate, reassembl
e
,
etc. Stative , Activity and Semelfactive verbs
do not take this
p
refix;
*
rebelieve
,
*reunderstand
,
*r
esneeze, *relaugh, *reknock
ar
e
k
i
m
p
ossible, nor do verbs with
re-
a
pp
ear in atelic constellations.
Lieber (2004: 146-7) argues against the characterization of
re-
as a telic
p
refix,
h
owever. She points out that
re-
c
an atta
c
h t
oso
m
eve
r
bs
which are atelic
,
with the
r
esult still being atelic:
(
33
)
I stocked the shelves for hours.
I r
es
t
oc
k
ed
th
es
h
e
l
ves
f
o
r h
ou
r
s.