?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation

Подождите немного. Документ загружается.

2 R

HELLE LIEBER

ieber argues that

re-

is instead a quantitative prefix, doing derivationally what

lik

totter

umme

do intrinsically – that is, creating an iterated action.

Further, Lieber points out that

re-

attaches only to verbs whose results are not fixed

r permanent. So it does not attach to verbs like yawn which have no result

nor will

t atta

h t

lik

explode

ose

esu

lt i

t r

eve

3.1.4 Quantitative prefixes

English has absorbed a variety of quantitative prefixes from Latin, Greek, and

rench, including

di-

demi

multi-

poly

and

semi-

No particular theoretical

attention has been paid to these in the literature, so the reader is referred to

Marchand (1969) or the OED for historical description.

3.1.5 Verbal prefixes

ith the exce

tion of

rivative

de-

, En

lish has no productive verb-formin

refixes. The

refix

- is relativel

strai

htforward, formin

privative or ablative

erbs from nouns

(

elouse, debu

, deice, dethrone, deplan

)

or from other verbs,

n th

ose

in -

-if

ate

(

eregister

demilita

denazif

decontaminate

n nouns, the prefix adds a causative meaning, as well as a privative one: to

delouse

omething is to cause it to be without lice. On verbs which are already causative in

eaning, e.g.,

milita

only the privative meaning is added.

According to the OED and Marchand (1969), the prefix

be-

derives historically

from the preposition

and attaches to verbs adding a meaning of ‘around’ or ‘all

around’ (e.

besti

besmea

r) or b

extension an intensive meanin

(

bedew

becrowd

). It can attach to nouns and adjectives as well, often with causative force

(e.g.,

befriend

befoul

). It is not, however, productive in present day English.

The final verb-forming prefix of English is

en-

which

be-

is barely if at all

roductive. It attaches to nouns

(

enth

one

entomb

) or adjectives (

enra

ennoble

)

forming causatives of them; so to

entomb

something is to cause it to go into a tomb,

enrag

someone is to cause them to be in a rage.

These three prefixes have principally

f int

t in th

fram

ewo

enerative morphology, because they for

potential counter-examples to the Right-

and Head Rule of Williams

(

1981

)

or its

ariants (see, for example, Selkirk 1982,

Di Sciullo and Williams 1987). The Right-hand

ead Rule

(

RHR

)

states that it is the

htmost morpheme in a word that determines the cate

and morphos

ntactic

properties of that word. This rule accounts nicel

for much of En

lish word

formation, as suffixes frequentl

(but not alwa

s) determine the cate

of the

derived word (see below), and prefixes most often do not. But it is certainly not a

niversal rule; as Lieber (1980) points out, some languages like Vietnamese and

Tagalog have robustly leftha

ded derivation, and even in English there are suffixes

hat are not category-changing, and of course prefixes like

and

- that

appear to determine the category of the resulting word. Granted that only one of

these three prefixes has any current productivity, the prefix

de-

till

tit

roblem for the RHR.

ENGLISH WOR

FORMATION PROCESSE

403

One attempt to explain cases such as these can be found in Walinska de Hackbeil

(

1986), who argues that the prefix

- is not a prefix at all, but rather a positional

allomorph of the preposition

. According to her analysis, the preposition

en-

tak

a noun or an ad

ective as its complement, and a zero affix then forms a verb from the

P+N or P+A collocation. While this anal

sis cannot be dismissed out of hand, it

oes suffer from a number of defects. First, it forces us to posit a phonolo

icall

null

affix for which there is no independent evidence. And further, the analysis does not

easily extend to

- which has no

ositional counter

art, and which indeed is the

nly one of these three prefixes that is at all productive.

.2 Su

ixation

Suffixation in En

lish has received far mo

att

nti

n in th

lit

rat

formation than prefixation has. It is be

ond the scope of this article to cover ever

suffix in En

lish, or even ever

suffix listed in Tables 1-2. Here, I will confine

yself again either to cohorts of suffixes or semantic groupings that have been

idely discussed or raise interesting theoretical questions.

1 P

sonal nouns

Suffixes which are used to form personal nouns include -

(

ite

talke

ondone

)

, -ant

-en

(

accountant

claimant

ist

(

guitarist, Marxist

(

), and

(

mployee

nominee

biographe

)

. We will concentrate on the four suffixes above,

althou

h -an

-ian

(

African

(

ust

alian

)

and -

ite

(

anaanit

talinit

) mi

ht be

added to them. This cohort of affixes, and especiall

and

ve bee

n m

iscussed in the literature, both within the generative framework (e.g., Levin and

ort 1988, Ra

ort Hovav and L

in 1992

Barker 1998

Bauer 1987

Lieber 2004, Booij and Lieber 2004), as w

l as within Cogniti

Grammar

(

Panthe

and Thornburg 1998, Ryder 1999) and from a functionalist perspective (Heyvaert

2001). The chief reason for this interest is that this cohort of affixes displays quite a

ange of polysemy; indeed, the same polysemy is largely replicated in other

anguages (see Booij 1986 and Booij and Lieber 2004 on Dutch, for example). The

ange of polysemy of these affixes is illustrated in (34), which is taken from Liebe

(

2004: 17

(

)

-e

ent writer, driver, thinker, walke

instrument opener, printer, page

eriencer heare

stimulus

leaser, thrille

patient/theme fryer, keeper, looker, sinker, loane

denominal noun Londoner

villa

er, carpetbagger, freighte

There are authors

of course

who conside

hat the meaning distinctions in affixes like -

constitute

ases of homon

rather than pol

sem

. For example, within Lexeme Morpheme Base

Morpholo

, the a

ent and instrument readin

onstitute quite different semantic functions.

404

ROC

HELLE LIEBER

ver

ati

patient/theme employee, nominee, deportee

overned preposition

xperimentee, lau

hee, e

ectee

no ar

ument amputee

denominal noun biographee, mastectomee, asylee, aggresse

-ant

agent accountant, claimant, servant

instrument evacuant

adulterant

irritant

experiencer dependent, destant, discernant

patient/theme confidant, insurant, descendant

-i

denominal noun guitarist, Marxist

deadjectival noun purist, fatalist

Perhaps the most curious part of this pattern of polysemy is that these affixes

verlap in meaning: while -er

-ant/-ent

ist

most often form personal agent

ouns

and -

most often forms patient/the

e nouns, not infrequentl

we find

precisel

the opposite situation, where -

and its cohort form patient nouns and

ent or at least sub

ect-oriented nouns. R

der (1999)

ives a wealth of data on -

hat shows that

atient forms in -

are not just lexicalized oddities, but rather are

formed with surprising productivity in present day English. Barker (1998) shows as

well that agent or subject forms in

-ee

are not infre

uent either. The

uestion then

arises why such overlap should exist.

number of analyses have been proposed

n recent years which address parts of

this problem. Probably best known is the analysis of Rappaport Hovav and Levin

(

1992), which is similar in man

s to Booi

’s (1986) anal

sis of Dutch facts.

est that much of the behavior of -

can be explained by saying that

binds the external ar

ument of a verb, the

fore takin

on whatever thematic role the

external argument can carry. This analysis accounts nicely for agentives, of course,

r in

nt n

in -

r. Some of the

atient/theme -

forms can be

analyzed as deriving from middle forms of

the base verb, where the external

argument bears the patient role: for example,

sinker

his ship sinks easil

t n

all -

forms can be explained this way; some patient

-er

forms like

looker

are not

asil

related to middles, and noun-based -

forms are neglected as well.

vaerts (2001) uses a functionalist approach in her anal

sis of -

. She ar

ues

(

2001: 7

)

that nouns in -

“realize an active, passive or middle voice relationship

between an entity and a process,” and that -

forms function as subjects: “like

Subjects, -

nominals will be argued to profile entities that are

peech functionall

ather than causally responsible fo

instantiating

the process type specification that

agent/sub

ect escapee, attendee, standee, arrive, resignee,

indirect ob

ect addressee

dedicatee

offeree

ENGLISH WORD

FORMATION PROCESSE

405

s realized by the verb [emphasis in original].” In many ways, this analysis is close

to that of Rappaport and Levin (1992); Heyvaerts is concerned only with deverbal

forms, however, and does not try to extend her analysis to denominal forms.

der (1999), in contrast, attempts to account for the full ran

e of -

data.

orkin

within the framework of

nitive Grammar, R

der (1999: 278) has

ested that the breadth of usa

e for the -

suffix stems from a lack of inherent

emantic content for the affix

“which in Present-day English indicates only that the

whole word is a noun.” The suffix refers to the event-schemas evoked by its base,

and the narrower the range of event sch

as evoked by its base, the likelier the -

form is to be interpretable. Since ver

evoke a narrower range of event schemas

than nouns, nouns a narrower range than ad

ectives

and so on

the suffix is more

ikely to attach to verbs than to oth

categories. As for the preference of -

for

reatin

ent, as opposed to patient or other t

pes of nouns, R

der assumes that the

f th

noun is constrained by two things, the ‘salience’ of the referent

(

“the de

ree to which somethin

is noticeable in comparison with its surroundin

s”

(

1999: 285)), and its ‘identifiability’ (“the e

ent to which a

icipant is readily

dentifiable by mention of the event alone” (1999: 285)). Unless overridden by

ontext, agents are more salient and identifiable than patients. Ryder’s analysis is

omprehensive, but its reliance on vague notions like salience and identifiability

educes its appeal.

esu

ffix -

ece

ved so

hat l

ess

att

ion in the theoretical literature

but the anal

sis of Barker (1998) is both thorou

h and insi

htful.

Like Ra

ort

Hovav and Levin

(

1992

)

on -

Barker ar

ues that -

binds an ar

ument of its base.

nlike them, he argues that the conditions on the binding of the argument are

emantic, rather than argument-structural: the argument that -

‘

episodically linked’ to the verb (it must be a participant in the event denoted by the

erb), and it must be sentient, but not volitional. Generally, the argument that is

bound, then, is the patient/theme, as it generally is this argument that lacks

olitionality (for example, in the verb emplo

). But if the patient/theme argument is

onsentient, another ar

ument can suit as well,

ust as lon

as it fulfils the

equirements of sentience and non-

olitionalit

. So with a verb like

add

ess

the -

form denotes the recipient, rather tha

the patient/theme. And if there is no

patient/theme argument, a subject argument

an be bound, as long as it is not too

olitional

(

standee

esca

ee). Nouns can function as bases if they have some sort of

eventive meaning.

ieber (2004) is an attempt to account for the full range of polysemy in this

ohort of affixes

and to make sense of them in relation to one another. Building on

he analyses of Rappaport Hovav and Levin (1992) and Barker (1998), and using a

framework of lexical semantic anal

sis that attributes broad semantic content to

affixes, Lieber su

ests that all affixes

n this cohort form concrete processual

ouns; that is, in her system, they bear the semantic features [+material, dynamic]

(

cf. section 1.4 above

)

. The affixes differ from one another

however

in their

argument-binding properties. The suffix -

(as well as

-ant

-en

)

binds the highest

argument of its base, whatever the sema

hara

cs of

hat argument. The

See also Bauer

(

1987, 1993

)

for useful data.

406

HELLE LIEBER

ffi

binds the first argument that is

nti

t n

liti

nal

esu

ffix -

ist

attaches only to nouns and adjectives, and it binds the first

volitional

argument of its

t that -

and

-ant

-en

denote concrete nouns without specifying

olitionality accounts for their ability to c

ate instrument as well as agent nouns.

t that all

f th

ese

affix

es be

emantic content accounts for the processual

nature of the resulting words, even when they attach to non-verbal bases. And the

fact that their ar

ument-bindin

properti

are not alwa

s mutuall

exclusive

accounts for their occasional overlap in meani

: if the hi

hest ar

ument of a verb is

sufficientl

nonvolitional, either -

can attach, giving rise to doublets like

esca

ee an

esca

er, both attested in a

iece about an esca

e from a WWII

risoner-

f-war cam

in which the

lanners are referred to as the esca

ers

and the

accompanying prisoners as the esca

ees

(

Lieber 2004: 66

)

ooi

and Lieber (2004) push this analysis a bit further, by analyzing difficult

ses suc

h a

loaner

and

keepe

as paradigmatic extensions of

forms. They point

ut that English lacks a dedicated affix which forms nouns meaning ‘thing which

ne Xes’, that is, a concrete, theme-oriented thin

noun. The suffix -

is usuall

theme-oriented, b

virtue of seekin

o bind a nonvolitional ar

ument, but because

f its requirement of sentience for this ar

ument, it cannot form thin

-nouns as -

an. What happens, Booij and Lieber argue, is that given a real-world need for a

word which designates a ‘thing which one Xes’, speakers will extend the use of the

losest productive affix in coining a word. As -

is productive and has no

equirement of sentience on its associated argument, it fits the bill best, and therefore

lik

loaner

and

keepe

can be coined, although clearly only under pragmatic

pressure, that is, the real-w

rld need to coin a word to fit a special situation.

2 A

bst

act nouns

mong these affixes are the nominalizers -

(

refusal

-ance

-enc

(

admittance

)

ation

(

and its allomor

)(

resto

ation

)

, -

ment

(

amusement

), and

(

closu

)

, all

f which attach to verbs

;

the affixes -it

(

imilarit

)

and -

ness

(

appiness

)

which

attach to ad

ectives; and the affixes -

dom

(

kin

dom

)

-hood

(

knighthood

)

and

shi

(

kinshi

)

, which attach to nouns. I will touc

on each of these cohorts in turn.

eve

bal nominali

edeve

al n

minaliz

ance

ation

ment

and -

all f

rm a

tra

ouns from verbs. In Beard’s (1995) terms, they in fact are alternative phonological

anifestations of the same morphological relationship; all are transposers of nouns

o verbs. In semantic terms, a si

nificant number of nominalizations formed with

hese suffixes have lexicalized meanin

s: a

ansmission

is a part of a car, a

development

can be a group of new houses, and

o on. Nevertheless, compositional

eanings are also frequent and with them individual formations can often have both

a result meaning (e.g., a long examination) and a process/event meaning (

the

doctor’s examination of the patient

)

. Much of the theoretical discussion surrounding

ese

affix

sco

lati

hip between these semantic nuances and the

yntactic behavior of nominalizing affixes.

ENGLISH WOR

FORMATION PROCESSES

This cohort of suffixes also bears the distinction of being the motivating force

behind the birth of generative morphology in the mid-seventies. In some of the

earliest work on word formation in the generative tradition. Lees (1961) generated

derived nominals transformationall

; a noun like

dest

uction

wou

ve bee

enerated via a series of transformations from a sentential deep structure somethin

The Romans destro

ed the cit

In response to such anal

ses, Chomsk

(1970)

n ‘Remarks on Nominalizations’ argues that nominalizations like

estruction

transmission

refusal

are idiosyncratic in meaning, and that the suffixes which

erive them attach only to selected verbs. Given their lack of full productivity and

endency towards noncompositional meaning,

uch forms should not be generated

ia syntactic rule. (See chapters by Roeper

and Scalise and Guevara, this volume.

)

principal motivation for the transformational analysis of nominalizations had

been the resemblance between the ar

uments manifested b

the derived nouns and

heir correspondin

verbs. It has lon

een noted that derived nouns frequentl

exhibit the same ar

ument structure as

heir base verbs; however, the ar

ument

structures of derived nouns are always optional, even where the arguments of the

orresponding verb are obligatory (Booij and van Haaften 1988, Grimshaw 1990,

among others):

(35) a. The enemy destroyed the city.

b. *The enem

destro

ed.

c. The enem

’s destruction of the cit

d. The destruction of the enem

e. The enemy’s destruction.

edes

The question then arises within current

theoretical frameworks what the proper

analysis is for these correspondences. The facts are notoriously complex. First, it

appears that neither ‘sub

ect’ nor ‘ob

ect’ argument is obligatory in this cohort of

erived nouns. Both ar

uments can, of course, be present, as in (35c). But it is also

possible for onl

one of the ar

uments to appear, with either ‘sub

ect’ or ‘ob

ect’

nterpretation bein

available for the sin

le ar

ument; so in (35e), the enem

’s

dest

uction

can either be interpreted as the enemy doing the destroying, or being

estroyed, as can (35d) the destruction of the enem

urther, it was pointed out at least as early as M. Anderson (1979), that only

‘affected’ ob

ect arguments can appear in the possessive position of the noun phrase.

o whereas the city’s destruction is fine

(

where cit

is an affected ob

ect),

the

book

’

s discussion

is not. In the latter case, the object

is not affected by the action.

See Comrie and Thompson (1985), Zubiza

etta

(

1987

)

, and Zubizaretta and van

aaften (1988) for discussion. To m

knowled

e, althou

h these facts have been

noted, no satisfactory explanation has been offered to date.

hat has been

rominent in the literature has been the attem

t to ex

lain the

ariant com

ositional inter

retations of the deverbal nominalizations. Grimshaw

(1990) attempts to link interpretation to the presence or absence of arguments. With

all arguments expressed, for example, abst

act nominalizations have what

rimshaw

(1990: 45) calls the ‘complex event’ interpretation (otherwise referred to as the

408

ROC

HELLE LIEBER

‘process’ or ‘event’ reading); this is the interpretation of (35c). With arguments

issing or completely absent, only a result interpretation is available. She argues

hat only complex event nominals inherit and project the argument structures of thei

base verbs; other argument-like interpretations that one finds in result or simple

event nominalizations (like the simplex noun tri

)

arise fro

Lexical Conceptual

tructure rather than from syntactic argument

ntrast

Bierwisch

(

1989: 38-39), discussin

similar issues in German, ar

ues that nominalizations

alwa

s inherit the ar

ument structures of

heir base verbs

;

the alternation between

event and result readin

s is not to be attr

buted to presence or absence of ar

ument

tructure, but rather to an “overall phenomenon of polysemy” that is exhibited in

implex lexical items as well. See Booij and van Haaften (1988) and Lieber and

aayen (1999), and Harley and Noyer (2000) for other views.

More recently, there has been a move to return to a syntactic analysis of deverbal

ominalizations. Fu, Roeper, and Borer (2001) argue on the basis of facts such as

hose in (36) that there must be a VP pro

ection inside the nominal pro

ection of

ved

minal

(

36) a. The occurrence of the accident suddenl

The verbal base is generated as head of a VP pro

ection and then raises to ad

oin

o the nominalizing affix. The presence of the VP pro

ection explains, they say, the

possibilit

of usin

VP adverbs like suddenl

or proforms like

do so

ith

eve

ominalizations like those above, but not with simplex event nominals. It remains to

be seen whether this is the last word in the anal

sis of nominalizations, or whether a

emantic analysis of the

ed.

his cohort of affixes has also been of interest from the

oint of view of

productivity: individual affixes vary widely

in their productivity in present day

English, with various morphological, etymological, and phonological factors

entering into their productivity vis a vis one another. Bauer (2001: 180) argues on

he basis of a number of measures that the abstract nominalizing suffixes -al

-ance/-

ence

and -

ure

have all had periods of productivit

in the histor

of En

lish, but that

are no lon

er productive in present da

lish. The suffix -

ment

still has some

ree of productivit

, and most productive of all is -

ation

(

and its various

allomorphs); as Plag (1999) points out, one reason for the productivity of

-ation

hat it i

minaliz

ed w

ith -

-if

both of which

are reasonably productive in present day English.

Plag (1999) generally gives an

nsightful analysis of the way in which this cohort of affixes divides up the range of

available bases. For example, it follows from the observation that verbs derived with

ffix

-ate

-ize

-i

are nominalized usin

ation

(or one of its allomorphs)

hat the remainin

affixes are restricted to simplex bases. Of these, -

and

-ance/

ence

are confined for the most part to simplex verbs which have final stress (e.

efusal

attendance

)

. The suffix -

ment

as well favors disyllabic bases with final

This, of course, is a

ood example of potentiation.

b. Sue’s exploration of Easter Island was impressive, then Amy’s

doing so was a real surprise.

ENGLISH WOR

FORMATION PROCESSE

409

tress (e.g.,

amusement

), although this is a tendency rather than a hard and fast

res

tri

esu

ffix -

ment

also favors prefixed bases,

especially those prefixed

ith

(

endearment

), and to a lesser extent

dis

(

disfigurement

)

. The suffix

ation

and its allomorphs attach to simplex bases as well, but almost exclusivel

to Latinate

es.

Dead

ectival nominalizers

lthough Riddle (1985) attempts to question whether the deadjectival

minaliz

ness

-it

are in fact synonymous, they are most frequently treated

as a rival pair. They have been discussed in the generative literature primarily as test

ases for various theories of productivity. Generally speaking, it is clear that -

ness

ighly productive and gives rise to forms which are compositional in meaning. The

ffi

-it

is less productive, althou

h it has a measure of productivit

in some

domains, attachin

to Latinate bases, and especiall

to those endin

in the suffixes

ile.

In contrast

ness

attaches freel

either to nati

ve o

r t

Latinat

ses.

The main interest in this pair is that they divide up the range of certain Latinate-

suffixed forms in an interesting way. Aronoff (1976: 37-45) notes that both -

ness

-it

can attach to bases ending in the suffix -

ous

in -

ousness

predictable in meaning than those in -osit

; the latter frequently have lexicalized

eanings. Aronoff correlates the productivity o

-ness

in thi

sdo

main

ith th

ransparenc

of meanin

, a

d h

pothesizes that forms in -

ousness

t li

he mental lexicon

whereas forms in -(os)it

frequentl

Listin

allows semantic

rift to take place. Forms in -(os)it

are blocked, however, if a simplex abstract noun

exists (e.g., glory ~ glorious

but

gloriosit

)

; since items in

–ness

are not listed

blocking cannot occur, and therefore nothing rules out gloriousness as a

ossible

ord alongside glor

. Aronoff also shows that adjectives which end in -

acious

always take -it

(

eracious ~ veracity, capacious ~ capacit

), and are rarely

exicalized in meaning, whereas ad

ectives in -

ulous

are less productive with -it

and less regular in meaning (e.g, credulous ~ credulit

meticulous ~

meticulousness

). Further discussion of the productivit

of these suffixes can be

found in Romaine

(

1983

)

and Bauer

(

2001

)

enominal nominali

esu

ffix

-dom

-hood

-shi

attach to nouns and form nouns. Although

argely neglected in the generative literat

these suffixes have received increased

attention in recent years as generative mor

phologists have turned

ir int

t t

emantic constraints on affixation. For example, Aronoff and Cho (2001) propose

hat -shi

-hood

differ in the semantic class of base nouns to which they attach:

the

claim, attaches onl

to sta

e-level nouns (that is, those that denote

temporary characteristics of

their referents), while

-hood

can attach either to stage-

r individual-level nouns. So words like

riendshi

are possible, the

claim, but

arentshi

are not. Lieber (2004) argues agains

this constraint, however, citing

lik

kinshi

, uncleshi

, an

twinshi

all listed in Lehnert

(

1971

)

, and all

presumably based on individual-level nouns.

HELLE LIEBER

Plag (1999) suggests a more general restriction on the attachment of

-hood

Discussing this suffix, he proposes that abstract-noun-forming suffixes should not

attach to already abstract nouns. This appears generally to be true of

-hood

; we find

lik

puppyhood

t n

gloryhood

3.2.3 Verb-forming suffixes

English has several verb-forming suffixes: -

-if

are the most

roductive,

both forming verbs from nouns and ad

ectives. The suffix -

ate

is less productive,

ften attaching to roots of indeterminate category, but also to nouns and ad

ectives.

The dead

ectival verb-forming suffix -

is rather unproductive, according to Plag

(1999), attachin

predominantl

to ad

ectives, althou

h it does appear on some

in -

The

roductive affixes have been of some

nterest over the years, both because of

heir semantic and argument-structural p

perties, and because of morphological and

phonological restrictions on their use.

f primary interest has been the complex polysemy shared by this cohort of

uffixes. Indeed, Beard

(

1995

)

treats -

-i

alizati

f th

ransposition, alon

with noun to verb conversion. Detailed semantic anal

ses can be

found in Lieber

(

1998

)

Pla

(1999) for all three productive suffixes, and

ieber

(

2004

)

-if

In all of these works

an effort is made first to

untangle the semantic nuances of the suffix

and then to exa

nt t

which they can be said to have a unitary core of meaning. Lieber (2004: 77)

llustrates the variety of meanings found with

–i

–if

ows:

(

)

on N on A on root

-ize unionize legalize evangelize

-ify personify purify magnify

-ate oxygenate activate vaccinate

-en lengthen blacken

(

)

(

’s

)

‘make x

’ ‘cause to become x’ causative

tandardize

velarize

esu

ltati

crystallize, unionize

‘make x go to/in/on something’ ornative apologize, texturiz

‘make something go to/in/on x’ locative

ospitalize,

containe

‘

do/

mak

in th

mann

imilati

oswellize, despotize

ENGLISH WOR

FORMATION PROCESSES

411

r lik

x’

(

)

-if

(Plag’s terminology)

‘make x

’ ‘cause to become x’ causative purify, acidif

esu

ltati

yuppif

‘make x go to/in/on something’ ornative glorif

‘make somethin

o to/in/on x’ locative s

llabi

, codi

‘do x’

erformative speechify, boozif

‘do/act/make in the manner of similative no exam

les

r lik

x’

‘

beco

x’ in

ati

acidi

, calci

Lieber

(

2004

)

does not discuss the semantics of -

ate

but Pla

(1999: 205)

haracterizes

roductive -

ate

formations as either ornative (e.g., fluoridat

use

fluoride to go into something) or resultative (e.g., gelate = to cause to become gel).

Plag and Lieber are generally agreed

hat at th

eco

f th

ese

manti

epresentation of -

-if

is a causative frame, which Plag, using Jackendovian

(

1990) Lexical Conceptual Structures (LCSs) represents as in (40) (1999: 79):

(40) CAUSE ([ ]

[GO

(

[

Property, Thin

]

;

[

Property,Thin

]

se/

])

]

)

The causative or resultative meanin

arise

when the base noun or ad

ective is

inked to the lowest open argument position (so

standa

is decom

osed to ‘cause

something to go to standard’). The ornative meaning results when the base is linked

o the argument of GO (so

lorif

is decomposed as ‘cause glory to go to

something’). Inchoatives are dealt with by making the outer CAUSE argument

ptional, represented by underlining in (40). Plag attempts further to derive what he

alls the performative and similative meanings o

verbs from the same L

hile this is a plausible enou

h anal

sis of a similative like

(which Pla

decomposes as ‘cause the doctrines of Marx to

o to somethin

’), Lieber (2004)

argues that it is less plausible for performatives like

otiz

ooliganiz

which

are not easily interpretable as causatives

(

otiz

ann

be s

an ‘

use

despot to go to something’ or ‘cause something to go to a despot’).

Lieber (2004), using her own framework of lexical semantic representation (see

above), suggests the representation in (41):

‘do x’

erformative

hiloso

hize, theoriz

“

beco

x” in

ati

oxidize

aerosoliz