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

THE LEXICALIST APPROACH TO WORD

-

FORMATION

15

7

4.2 Arono

ff

(1976)

A

ronoff (1976) was the first comprehensive mono

g

raph on morpholo

gy

in the

framework of generative grammar. Its mai

n

contributions consisted in refining the

n

otion of rule in morphology, providing it with a system of restrictions on WFRs

and in proposing a morphological model that is based on ‘words’ (not on

m

orphemes). The topics developed in Ar

o

n

o

ff’

swo

rk ar

e

at th

eb

a

s

i

so

f th

e

Lexicalist approach; although most morphologists never agreed completely with

A

ronoff’s proposals, this monograph can

s

t

ill

be co

n

s

i

de

r

ed

th

e

f

ou

n

d

ati

o

nal

wo

r

k

i

n lexicalism. We will briefl

y

describe some of its salient contributions, most of

w

hich have provoked a

g

reat deal of debate in morpholo

g

ical theor

y

.

4.2.1 The Word-based Hypothesis

A

ronoff ar

g

ues a

g

ainst morpheme-based theories of morpholo

gy

: morphemes

c

ann

o

t

se

r

ve

a

s

th

eb

a

s

i

so

f

wo

r

d

-f

o

rmati

o

n processes because the very notion of

m

or

p

heme is

p

roblematic

.

1

If words are to be analyzed exhaustively in morphemes,

o

ne is often left with isolated strings that cannot be interpreted as ‘meaningful

elements’.

C

onsider the so-called cranberry-morphs in

(

8a

)

:

A

ll these words refer to types of ‘

b

erries’, but, isolating the morpheme

b

err

y

in

(

8a

)

we are left with

c

r

an

d

huckle

,

which do not exist independently (cf. the

examples in (8b), where the analysis yields the English words

st

r

aw

an

d

black

). An

kk

even more extreme case is shown in

(

9

)

:

(

9

)

X

+

fer

+

: refer defer prefer transfer

r

X

+

mit

: remit demit … transmit

t

A

theory based on morphemes would have to recognize the stems

fer

and

r

mit

and

t

he prefixes

re-

,

de-

,

pre

-

,

t

r

ans

-

.

However

,

it is clear tha

t

i

n these examples “neither

t

he prefix nor the stem has any fixed meaning” (Aronoff 1976: 12). These units

w

ithout meanin

g

cannot be considered minimal lin

g

uistic si

g

ns (i.e. the

y

are not

Saussurean si

g

ns, arbitrar

y

constant unions of sound and meanin

g

). Aronoff’s

proposal is that morpholo

gy

must be explained on the basis of words, which are

i

ndeed true minimal signs:

(

10) All regular word-formation proc

e

sses

ar

ewo

r

d

-

b

a

sed.

A n

ew wo

r

d

i

s

formed by applying a regular rule

t

o a single already existing word.

B

oth the new word and the existin

g

o

ne are members of ma

j

or lexical

c

ate

g

ories (Aronoff 1976: 21).

1

Cf. Anderson

(

1992: 51 ff.

)

.

(8) a. cranberry huckleberry

b. strawberry blackberry

1

58

S

ER

G

I

OSC

ALI

S

E

&

EMILIAN

OGU

E

V

AR

A

The word-based hypothesis entails that English words such as cranberr

y

,

h

uckleberr

y

,

r

efer

,

prefer

,

t

r

ansmit

,

etc. are not formed by regular morphological

processes. They are listed in the Lexic

o

n as such. In Aronoff’s view

,

all word-

formation takes place in the Lexicon (whi

c

h lists only words and no morphemes in

t

he traditional sense

)

.

4.2.2

W

ord-Formation Rules

[

WFRs] specify a set of words on which [they] can operate. This set [...] we will term

t

h

e

base

of that rule. Ever

y

WFR specifies a unique phonolo

g

ical operation which is

performed on the base. Ever

y

WFR also specifies a s

y

ntactic label and

subcate

g

orization for the resultin

g

word, as w

e

l

l as a semantic readin

g

for it, which is a

function of the readin

g

of the base (Aronoff 1976: 22)

.

W

FRs, therefore, can have the following form:

These rules can be seen as an input/output mechanism, where the input is the

s

imple word and the output the complex one. Note that in the formulation

g

iven in

(11) suffixation is supposed to change the lexical category of the base (X

ĺ

Y

)

,

while

p

refixation is not su

pp

osed to do so

(

X

ĺ

X

)(

as for com

p

ounds, see section 4

below

)

.

W

FRs have a semantic counterpart: the meaning of the whole word is

‘composed’ by the partial meaning of the constituents and it is usually given as a

paraphrase:

In contrast to Halle

(

1973

)

, in Aronoff’s

m

odel words and affixes are represented

i

n different levels: words are in the lexicon while affixes are parts of rules, that is to

sa

y

, affixes are not lexical items. A WFR is a sort of ‘instruction’ to chan

g

e the

c

ate

g

or

y

of the base into another cate

g

or

y

(e.

g

. A

ĺ

N

)

and it is at the same time a

phonolo

g

ical and a semantic operation on the b

a

se (the former t

y

picall

y

addin

g

an

affix to the base, the latter changing its meaning).

12

This

g

eneral rule for compoundin

g

wa

s

actuall

y

proposed b

y

Allen (1978).

(

11

)

a. Suffixation: [WORD]

X

ĺ

[[WORD

]

X

+Su

f]

Y

e.g. happy

ĺ

ha

pp

iness

b. Prefixation:

[

WORD

]

X

ĺ

[

Pre+

[

WORD

]

X

]

X

e.g. happy

ĺ

unhappy

c. Compoundin

g

:[WORD]

X

,[

WORD

]

Y

ĺ

[[

WORD

]

X

+

[

WORD

]

Y

]

Z

12

e.

g

. apron, strin

g

ĺ

apron strin

g

(

12) happy+ness

ĺ

‘the state of being happy’

un+happy

ĺ

‘not happy’

apron, string

ĺ

‘string of the apron’

THE LEXICALIST APPROACH TO WOR

D

-

FORMATION

159

4.2.3 Productivity

I

n Aronoff’s view, only productive processes build the possible complex words

o

f a language: they may not still exist, but nevertheless conform to the

m

orphological rules of that language. Morphological processes can be classified

accordin

g

to their productivit

y

: while some

W

FRs are totall

y

productive, e.

g

. the

f

o

rmati

o

n

o

f a

dve

r

bs w

ith -l

y

(

[[ ]

A

+ -l

y

]

A

dv

,

which can be said to appl

y

to most

En

g

lish ad

j

ectives), others exhibit a lower productivit

y

(e.

g

. -

dom

, which onl

y

o

ccasionally forms new words

–

dolla

r

dom

,

gangsterdom

,

girldom

).

A

ronoff gives two different views on the issue of productivity: First, he proposes

(

and quickly dismisses) a method to calculate the productivity of a given WFR as

the ratio of possible input entries and attested output entries. Second, he points out

that productivity “goes hand in hand” with semantic compositionality: any complex

word whose properties are completely predictable from the interaction of the

Lexicon and the set of WFRs is a

p

ossible word that does not need to be listed

(

whether it has been attested or not

)

and the related WFR is a

p

roductive one.

I

nstead, if a word has some idios

y

ncratic feature (formal, semantic, s

y

ntactic), it

must be stipulated and not constructed by

rule. (For a detailed account of these and

y

m

any other issues related to morphological productivity, cf. Bauer 2001 and this

v

olume

)

.

4

.

2

.4 Restrictions on

W

FRs

One of the goals of syntactic theory is the definition of the class of possible

s

entences of a language. Similarly, one of the goals of morphological theory is the

definition of the class of possible words of a language. To achieve this goal, Aronoff

provides WFRs with a series of restrictions in order to determine correctl

y

(a) the

k

ind of information that is available to them, and (b) the kind of operations that the

y

c

an carry out (cf. Rainer, this volume, for a thorough and critical assessment of the

theoretical status of restrictions on WFRs

)

.

C

onsider the following schema:

(

13

)

Y

X

S

u

f

T

h

e

base

of a WFR (in this case a rule adding a suffix) is the set of all words that

c

an

subs

tit

u

t

e

th

e

¨

symbol dominated by X. X’s lexical category might not by itself

be sufficient to establish appropriately the base of a suffix. In order to exclude all the

words yielding ungrammatical X+Suf combinations from the base, restrictions must

be formulated re

g

ardin

g

different t

y

pes of information: there exist s

y

ntactic,

s

emantic, phonolo

g

ical and morpholo

g

ical restrictions on the base of a WFR.

160

S

ER

G

I

OSC

ALI

S

E

&

EMILIAN

OGU

E

V

ARA

Sy

ntactic restrictions

W

FRs have access to the syntactic properties of the base: generally, the base is a

m

ember of the major syntactic cate

g

ories N, A and V (e.g. the suffix -

able

atta

c

h

es

t

o V

,

not to N

,

A or P

,

c

f

.

readable

,*

d

og-abl

e

,*

nice

-

able

,*

ove

r-

able

)

; WFRs do

n

ot usually apply, for example, to articles, pronouns, etc. Furthermore, WFRs are

sensitive to the subcategorization frame of the base (-

able

attaches to

V

s which are

s

ubcate

g

orized as [+transitive] and not [–transitive], cf.

d

r

inkable

vs. *

dieable

).

S

emantic restrictions

Generally, we may say that derivational affixes ‘select’ the base to which they

attach also with respect to its meaning.

C

onsider

,

for instance

,

the Italian verb tenta

(

re

)

,

which may be used (at least)

with the following two meanings (a) ‘to attempt’, (b) ‘to tempt’. The WFR that adds

-

tivo

(yielding

tentativo

‘tentative’) selects meaning (a), while the WFR adding -

to

r

e

(

forming

tentato

r

e

‘tempter’) selects meaning (b):

(

14

)

tenta

(

re

)

tentativo tentatore

Meaning (a) + –

Meaning (b) – +

The semantics of the base is relevant in various wa

y

s to the functionin

g

of

m

orphological rules: they do not apply indistinctly to all the possible meanings of

t

he base but, instead, typically select one of them.

Phonological restrictions

W

FRs are sub

j

ect to phonological restri

c

tions when the ungrammaticality of

t

heir outputs depends exclusively on the phonological shape of the base. E.g. the

En

g

lish noun-formin

g

suffix -

al

attaches only to verbs with main stress on the last

l

s

y

llable:

(

15) a. tr

y

[

¥

t

ra

i

]

ĺ

tri

a

l

propose

[

p

r

7

¥

p

7

z

]

ĺ

p

ro

p

osal

arrive

[

¥

r

a

i

v

]

ĺ

arri

v

al

M

orphological restrictions

Some WFRs are sensitive to the morphol

o

gical makeup of the base, blocking the

d

erivation if the base has a particular internal structure.

b. deposit

[

F

K¥

R

b\KV

]

ĺ

*deposital

recover

[

T

K

¥

M¡X

(

T

)

]

ĺ

*r

ecove

ral

promise

[

¥R

TbOK

U

]

ĺ

*promisal

T

HE LEXICALIST APPROACH TO WORD

-

FORMATION

161

F

or instance

,

the Italia

n

adverb-forming suffix -

mente

is as productive as its

English counterpart -l

y

with simple and derived ad

j

ectival bases, but it does not

attach to compound bases, as can be seen from (16) and (17):

N

o

t

e

that -

mente

can be freely attached to the individual constituents of these

c

ompounded ad

j

ectives, whether they are simple (as in (16)) or derived (as in (17)).

The non-application of -

mente

to the examples in (16) and (17) depends on the

i

nt

e

rnal

s

tr

uc

t

u

r

eo

f th

eb

a

se.

4.2.5 Stratal

f

eatures

The Lexicon of a language may have variou

s

lexical strata

(

often due to contact

with other languages, borrowings, diachrony etc.). In order to express this

i

nformation, a system of stratal features has been developed (e.g. [

±

n

ative

]

[

±

l

atinate

]

,

[

±

g

reek], etc., cf. Saciuk 1969).

English is characterized by having two well-defined strata, a [+native] body of

G

ermanic origin and a [–native] stratum mainly of Romance origin. This kind of

d

istinction is relevant to the functioning

o

f many WFRs: different affixes “select”

bases from different

lexical strata.

t

C

onsider e.

g

. the suffix

-

it

y

,

which can attach to [+latinate] words – as in

(

19a

)–

but not to [–latinate] words

(

19b

)

:

(

19

)

a.

p

rofane

ĺ

profanity

v

i

v

i

d

ĺ

vividity

b. w

i

de

ĺ

*widity

strong

ĺ

*strongity

I

n this respect, -it

y

d

iff

e

r

s

fr

o

m th

esu

ffix -

ness

,

which does not discriminate

between [+latinate]

(

20a

)

and [–latinate]

(

20b

)

:

(

16

)

[dolce][amaro] + *mente ‘sweet-sour’

[

terzo

][

ultimo

]

+ *mente ‘lit. third last’

(

17

)

[storico][critico] +*mente ‘lit. historical-critical’

[

economico

][

sociale

]

+ *mente ‘lit. economic-social’

(18) dolcemente ‘lit. sweet+ly’

amaramente ‘lit. sour+ly’

?terzamente ‘thirdl

y

’

ultimamente ‘lastl

y

’

storicamente ‘historicall

y

’

criticamente ‘critically’

economicamente ‘economically’

socialmente ‘socially’

1

62

S

ER

G

I

OSC

ALI

S

E

&

EMILIAN

OGU

E

V

ARA

(

20

)

a. common+ness

strange+ness

b. happy+ness

w

hit

e

+n

ess

4

.2.6 Restrictions on the output of WFRs

T

he output of WFRs is also sub

j

ect to

r

estrictions. In Aronoff’s view

,

these are

either s

y

ntactic or semantic.

A

ll th

ewo

r

ds

that ar

ec

r

e

at

e

d b

y

WFRs must be members of a ma

j

or lexical

c

ategory (i.e. N, A, V or P). The catego

r

y of the output is specified by the WFR

i

tself. The output of a WFR is a labeled bracketing where the lexical categories of

the input and the output are explicitly si

g

naled, together with a boundary between

the constituents of the complex word. An output such as happiness has the following

r

epresentation:

(

21) [[happ

y

]

A

+ ness

]

N

T

he semantic restriction on the output requires that its meaning be

c

ompositionally derived from the meaning of the base. The meaning of the output is

r

epresented as compositional paraphrase containing a variable:

(

22

)

a. [[X]

V

+er

]

A

‘one who Xs habituall

y

, professionall

y

’

drinker ‘one who drinks habitually’

b. [un+ [X

]

A

]

A

‘n

o

t X’

unhappy ‘not happy’

T

he meaning of a complex word is al

w

ays compositional when it has been

c

reated b

y

a (s

y

nchronicall

y

) productive WFR. With time, a complex word ma

y

acquire unexpected or idios

y

ncratic meanin

g

s, i.e. meanin

g

s that cannot be derived

from its constituents, as, for example, in the alread

y

seen word

t

r

ansmission

.

4.2.7 Conditions

Alongside the system of restrictions, which define the possible words of a

l

anguage applying differentl

y

on each individual WFR, Aronoff put forward a set of

C

onditions devised to constrain morphological rules at a system-wide level.

Unitary Base Hypothesis (UBH)

An affix

c

ann

o

t atta

c

h t

o

any

lexical category, it selects words of one and only

o

ne cate

g

or

y

. So, when we find an affix that apparentl

y

selects more than one

c

ate

g

or

y

, we are forced to conclude that there are (at least) two homophonous

affixes. For exam

p

le, -

able

usuall

y

attaches to Vs (acce

p

tabl

e

)

, but it is also found

THE LEXICALIST APPROACH TO WOR

D

-

F

ORMATION

163

attached to Ns

(

cha

r

itable

). According to the UBH, therefore

,

there are two -

able

s

uffixes and this is supported by two facts. First, denominal -

able

ad

j

ectives may be

f

u

rth

e

r

de

ri

ved w

ith -

ness

bu

t n

o

t

w

ith -it

y

(

cha

r

itableness

,*

charitabilit

y

)

, while

deve

r

b

al -

able

ad

j

ectives do not show such preference (acceptabilit

y

,

acceptableness

)

. Second, the semantics of these formations is different

(

i.e. deverbal

-

able

ad

j

ectives can be paraphrased ‘capable of bein

g

X-

ed

’

,

while denominal -

able

adjectives mean ‘characterized by X’).

Since the UBH proved to be too strong, various refinements of Aronoff’s

o

riginal formulation have been proposed. For example, Scalise (1984) proposed that

t

he UBH might be held true only if, instead of plain categories, we take into

c

onsideration syntactic features such as [

±

N] and [

±

V]

; Scalise’s Modified UBH

wou

l

d

th

e

r

e

f

o

r

e

r

e

a

d

a

s

f

o

ll

ows:

an affix

se

l

ects as its base only items marked either

as

[

+N

]

or

[

+V

]

. Consider two Italian examples as the following:

Th

e

fir

s

t

su

ffix

c

an

be desc

ri

bed

a

s

ha

v

in

g

a [+V] base and the second one as

h

avin

g

a [+N] base. The h

y

pothesis reformulated in these terms seems able to

account for a

g

reat number of cases Actua

l

y

, adoptin

g

the MUBH, one can describe

c

orrectly the behaviour of all Italian

s

uffixes but one

(

cf. Scalise 1984: 143

)

,

although it is uncertain if this conditi

o

n holds equally well for every language

.

1

3

Binary Branching Hypothesis (BBH)

The Binary Branching Hypothesis states that morphological structures are

basically binary, regardless of their complexity:

(

24

)

re- think happy -ness high school

un- think -able industr

y

-al -ize -ation

1

3

See Iacobini & Scalise (to appear).

(

23

)

-

an

z

a

tolle

r

an

z

a

‘tolerance’

(V

ĺ

N)

fro

m

tolle

r

a

r

e

‘t

o

t

o

l

e

rat

e

’

lontanan

z

a

‘distance’

(A

ĺ

N)

fro

m

lontano

‘

d

i

s

tant’

-

esimo

incantesimo

‘enchantment’

(

N

ĺ

N

)

fro

m

incanto

‘spell’

umanesimo

‘humanism’

(

A

ĺ

N)

fro

m

umano

‘h

u

man’

1

64

S

ER

G

I

OSC

ALI

S

E

&

EMILIAN

OGU

E

V

AR

A

s

tudent film societ

y

committee scandal inquir

y

Thus, the trees in (25a) are possible morphological structures while those in

(

25b

)

are not:

(

25

)

a.

b.

* * *

This hypothesis corresponds to Aronoff’s idea that affixes attach to their base

o

ne at the time (‘one affix, one rule’) and, therefore, that there are no genuine

m

orpholo

g

ical rules that attach two (or more) affixes simultaneousl

y

. Some lin

g

uists

do not accept the validit

y

of BBH: Rainer (1989)

,

for example, ar

g

ues that there is a

c

lass of coordinate compounds (cf. an

g

lo-italo-american

)

where there is no reason

t

o postulate a binary-branching structure. In fact, the BBH has some limits of

a

pp

lication

:

1

4

it seems generally true for deri

v

ation and for subordinate com

p

ounds,

n

ot clearly so for multip

l

e coordinate com

p

ounds.

No Phrase Constraint

(

NPC

)

T

he No Phrase

C

onstraint

1

5

states that the base of a WFR is always a lexical

c

ate

g

or

y

(i.e. a word) and never a s

y

ntactic phrase. This restriction is decisive in

o

rder to separate morpholo

gy

and s

y

ntax. A

g

ain, there are counterexamples, but

the

y

are t

y

picall

y

found in compound

s

16

(

Aronoff’s concern, we must remember,

was entirely within derivation). As for derivation, some exceptions to the NPC can

be found (cf. the following cases from Allen 1978: 236):

(

26

)

[black and blue]ness

[

at home

]

ish

[open-air]y

14

In some theoretical approaches, this hypothesis is even dispensed with: cf. fo

r

example Štekauer (1998,

t

his volume

)

.

1

5

The label ‘No Phrase Constraint

’

was introduced by Botha (1981).

16

Cf. Lieber

(

1992

)

.

THE LEXICALIST APPROACH TO WOR

D

-

F

ORMATION

165

These examples can nevertheless be desc

r

ibed as involving lexicalized phrases.

If we systematically apply the same processes to similar constructions,

u

n

g

rammatical complex words are obtained:

(27) *[intelli

g

ent and attractive]ness

*[at school

]

ish

*[open wood]y

B

locking

Blocking expresses the general tendency of

t

h

e

L

e

xi

co

n t

o

a

vo

i

d

th

e

f

o

rmati

o

n

o

f

s

ynonyms. It may be stated as “the nonoccurrence of one form due to the simple

e

xistence of another”

(

Aronoff 1976: 43

)

.

There are two t

y

pes of blockin

g

: one that we can call s

y

nta

g

matic and one

paradi

g

matic (or, followin

g

Rainer 1988, token blockin

g

and t

y

pe blockin

g

,

r

espectively, cf. also Rainer this volume). The first type is illustrated by cases such

as the following:

The sim

p

le nouns glor

y

,

grac

e

an

d

fur

y

‘block’ the formation of the semantically

re

lat

ed

a

bs

tra

c

t n

ou

n

s*

g

loriosity, *graciosity, *furiosit

y

.

C

onsider now the following set of data:

Th

e

a

bs

tra

c

t n

o

minal in -it

y

can be formed because there is no blocking element

X. Blocking is strictly linked with productivity: the suffix -

ness

c

an in fa

c

t

be

a

dded

to

all th

e

X-

ous

adjectives in (28, 29) due to its high productivity.

In the second type of blocking, an affix will prevent the attachment of rival

affix

es

t

o

th

es

am

eb

a

se: so we do

n

o

t ha

ve *

occu

r

ation

o

r

*

occu

rr

ement

because of

t

h

ee

xi

s

t

e

n

ce o

f

occu

rr

ence

.

The two different t

y

pes of blockin

g

do not seem to operate across the board,

h

owever. The

y

can be said to express onl

y

a

g

eneral tendenc

y

of the lexicon and not

a stron

g

constraint on the functionin

g

of morpholo

g

ical rules.

17

1

7

Cf. Scalise et al.

(

1983

)

, Rainer

(

1988

)

(

28) X X+ous X+ity

glory glorious *gloriosity

grace gracious *graciosity

fury furious *furiosity

(

29) X X+ous X+ity

* curious curiosity

* precious preciosity

* specious speciosity

1

66

S

ER

G

I

OSC

ALI

S

E

&

EMILIAN

OGU

E

V

AR

A

4.2.8 Summar

y

on Word-Formation Rules

A

list of the

p

ro

p

erties of WFRs as it has been elaborated in the main core of the

l

exicalism

,

can be summarized as follows:

W

FRs

•

t

ake as their input only lexical items (they are thus ‘lexical’ rules)

•

h

a

ve

a

ccess

t

o

all th

e

inf

o

rmati

o

n a

s

sociated with a given lexical item

•

t

hey have, therefore, access to phonol

o

gical, morphological and categorial

p

ro

p

erties of words, but also to all kinds of semantic information

•

t

hey consist of a formal part (attachment of an affix) and of a semantic part

•

t

hey form new words but can be used also to analyze existing words

•

t

hey are structure-building rules (the structure they build is binary)

•

t

he

y

can operate on possible but not existent words

•

t

he

y

can be more or less productive

• appl

y

to one another’s outputs

W

FRs are different from other ru

l

es of grammar, because they

•

c

an change all the information associated with their base (they can thus change

l

exical categories, subcategorization features, a

r

g

ument structure, etc.)

• ar

e

l

oc

al

1

8

• are optional

W

FRs are subject to restrictions

•

o

n their base (cate

g

orial, phonolo

g

ical, morpholo

g

ical, stratal, semantic)

•

t

heir output belongs to a major lexical category and has compositional meaning

•

t

hey take as their base only words (that is, neither morphemes nor phrases,

although this point is not accepted by all lexicalist morphologists)

•

t

hey select a single category they can attach to (Unitary Base Hypothesis)

•

t

he

y

tendentiall

y

do not form s

y

non

y

ms of the base (blockin

g

)

5

.

SO

ME MA

JO

R I

SSU

E

S

The theoretical a

g

enda resultin

g

fro

m

H

alle

(

1973

)

and Aronoff

(

1976

)

was soon

ex

p

anded with new

p

ro

p

osals that had as a conse

q

uence the enhancement of

L

exicalism

(

which became a well-defined framework) besides inspiring a good deal

o

f research on both the theoretical and the em

p

irical as

p

ect. In this section we will

h

ave a quick look at some of the most interesting and thought-provoking issues

arising from research in the lexicalist tradition.

1

8

‘Local’ intended in the sense of Wasow

(

1977: 330

)

: “Whereas transformations are mappings between

entire phrase markers, lexical redundancy rules (i.e. WFRs) are mappings only between lexical

i

t

e

m

s.

”

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

Подождите немного. Документ загружается.