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

B

ASIC TERMINOLOGY

23

A

ronoff

,

Mark. 1976. Word Formation in

G

enerative

G

rammar. (Linguistic Inquiry Monographs, 1).

C

ambrid

g

e, Massachusetts: MIT Press.

B

audouin de Courtenay, Jan

.

1972.

A

Baudouin de Courtenay Antholog

y

,

ed. by Edward Stankiewicz.

B

loomington: Indiana University Press.

B

loch, Bernard. 1947. “En

g

lish verb inflection.” Lan

g

ua

ge

23, 399-418.

(

Re

p

rinted in Joos 1957

)

B

loomfield

,

Leonard. 1933. Lan

g

ua

ge

.

N

ew

Y

o

rk

:

H

o

lt

.

C

arstairs-McCarth

y

, Andrew. 1994. “Inflection cl

a

s

ses,

g

ender and the Principle of Contrast.” Lan

g

ua

g

e

7

0

,

737-88.

C

arstairs-McCarthy, Andrew. 2002a. An Introduction to English Morphology

.

Edinburgh: Edinburgh

U

niversity Press.

C

arstairs-McCarthy, Andrew. 2002b. “How stems and affixes interact: stem alternants as morphological

signata.” In: S. Bendjaballah; W. U. Dressler; O.E. Pfeiffer; and M.D. Voeikova (eds.), Morphology

2

000 (Proceedings of the 9th International Morphology Me

e

ting, Vienna, 2000

)

.

Amsterdam: John

B

en

j

amins, 49–57.

H

alle, Morris. 1973. “Prole

g

omena to a theor

y

of word formation.” Lin

g

uistic Inquir

y

4, 3

-

16

.

H

alle

,

Morris and Marantz

,

Alec. 1993. “Distributed Morpholo

gy

and the pieces of inflection.” In: K.

H

ale and S. J. Ke

y

ser

(

eds.

)

, The View

f

rom Buildin

g

20

.

Cambrid

g

e, Massachusetts: MIT Press, 111-

7

6.

Harris, Ro

y

. 1987.

R

eadin

g

Saussure

.

La Salle, Illinois: O

p

en Court.

Harris, Zellig S. 1942. “Morpheme alt

e

r

nants in linguistic analysis.” Language 18,169-80. [Re

p

rinted in

J

oos 1957

]

.

Hockett, Charles F. 1947. “Problems of morphemic analysis.” Languag

e

23, 321-43. [Re

p

rinted in Joos

1957

]

.

Hjelmslev, Louis. 1961. Prolegomena to a Theory of Language (revised English edition). Madison:

U

niversity of Wisconsin Press.

J

oos

,

Martin. 1957. Readings in Linguistics I: The Development of Descriptive Linguistics in America

1925

-

56

. Chica

g

o: Universit

y

of Chica

g

o Press.

L

ieber

,

Rochelle. 1992. Deconstructin

g

Morpholo

gy

: Word Formation in S

y

ntactic Theor

y

.

Chica

g

o:

U

niversit

y

of Chica

g

o Press.

Malinowski

,

Bronislaw. 1935. Coral Gardens and their Magi

c

(

Vol. 2

)

. London: Allen & Unwin.

McCarth

y

, John J. “A prosodic theor

y

of nonconcatenative morpholo

gy

.” Lin

g

uistic Inquir

y

12

,

373-418.

M

e

Đ

þ

uk, I

g

or. 1993-2000. Cours de morpholo

g

ie

g

énérale

(

5 volumes

)

. Montreal: Les Presses de

l

’

U

niversité de Montréal.

Mugdan, Joachim. 1986. „Was ist eigentlich ein Morphem?“ Zeitschrift für Phonetik, Sprachwissenschaf

t

und Kommunikationsforschung

39

,

29-43.

g

R

obins

,

R. H. 1990. A Short History of Linguistics (3rd ed.). London: Longmans.

Saussure

,

Ferdinand de. 1973. Cours de linguistique général

e

(ed. by Charles Bally, Albert Sechehaye

and Albert Riedlinger; critical edition by Tullio de Mauro). Paris: Payot.

Saussure

,

Ferdinand de. 1983.

Course in General Linguistics (translated by Roy Harris). London:

D

uckworth. (Republished 1986 by Open Court, La Salle, Illinois

)

Stump, Gregory T. 2001.

I

nflectional Morphology: A The

o

r

y of Paradigm Structure

.

Cambridge:

C

ambrid

g

e Universit

y

Press.

2

5

Štekauer, P. and R. Lieber (eds.), Handbook o

f

Word-Formation

,

25—4

7.

.

WORD-FORMATION AND PHONOLOGY

ELLEN M. KAI

SS

E

1. INTROD

U

CTION

I

f phonology deals in sounds – their p

r

o

duction, arrangement, alternations,

r

elative prominences, and so on – one might expect the ideal phonological process to

r

efer only to phonological entities: features, segments, tones, syllables, metrical

structure, and the like. One would not expect morphological information to play any

r

ole in determinin

g

whether or how the phonolo

gy

operated. Wh

y

should the

co

n

so

nant

c

l

us

t

e

r at th

ee

n

do

f th

eve

r

b

collá

p

se attract stress to the final s

y

llable

while the similar clusters at the end of the third

p

erson for

m

édit

+

s

o

r th

e

n

ou

n

l

áryn

x

do not? Why should the strings of syllables in

c

ò

nd

è

d

n

s

á

tion

,c

ò

mpens

á

tion,

and

P

è

n

nsylvánia, which are phonologically nearly identical as far as the

c

haracteristics that determine stress go,

r

esult in a secondary stress falling on the

second syllable of

c

ò

ndènsátion

but not on the second syllables of the latter two

w

ords? The phonologies of languages in gen

e

r

al and of English in particular, abound

i

n cases like these, where the morphological make-up of a word has a considerable

i

nfluence on its pronunciation.Word formation, in particular, has stron

g

effects on

En

g

lish phonolo

gy

: the presence of a secondar

y

stress on the second s

y

llable of

c

ò

ndènsátion

has ever

y

thin

g

to do with there bein

g

main stress on the correspondin

g

syllable of its base,

c

o

ndénse

. To take another example, the preantepenultimate

stress and string of three unstressed syllables in adverbs like

s

igníficantly,

húmorously, an

d

pérsonall

y

,

o

therwise very unusual in English, is obviously related

to

th

e

ir

de

ri

v

ati

o

n fr

o

m the innocuously antepenultimately stressed adjectives

s

igníficant, húmorous, an

d

pérsonal

.W

e will look in more details at cases like these

and others in this cha

p

ter.

R

easonin

g

alon

g

the same lines, if morpholo

gy

deals in meanin

g

ful units and

t

heir arran

g

ement, we could ima

g

ine that the ideal s

y

stem, from the morpholo

gy

’s

standpoint, would result in the same phonological event always being associated

with a given morpheme. There should be no allomorphy caused by the requirements

o

f the phonological system. Why should the plural morpheme in

cats

be [s

]

but [z

]

i

n

d

ogs? Why should the stresses in phótogràph fall on the first and third syllables of

t

he stem, but on the second syllable in photógraphy,

r

esulting in such wholesale

allophony in the vowel qualities and, for North American speakers,

i

n th

e/

t

/

a

swe

ll

?

Y

et a

g

ain, lan

g

ua

g

es are replete with phonolo

g

icall

y

mandated alternations of

m

orphemes. Indeed, in lan

g

ua

g

es with concatenative morpholo

gy

, much of the

phonolo

gy

seems to be tri

gg

ered b

y

the combination of sounds put to

g

ether b

y

m

or

p

heme combination.

Finally, if morphology were to have it

s

way, we would not expect the phonology

ever to

p

lace limits on what combinations

o

f meaningful units

t

he morphology could

26

ELLEN M

.

KAI

SS

E

put together. Yet in numerous well-docume

n

ted cases taken from diverse languages,

t

here are instances where morphological p

r

ocesses c

ann

o

t

eve

n

occu

r if a

ce

rtain

phonological output would result, or, to look at it another way, where the

m

orphology must make reference to phonological properties, properties frequently

not present in the underlying representation.

There is, therefore, an inherent tension between the goals of the morphology and

o

f the phonolo

gy

. And there is an interaction between them which must be modeled

i

n an

y

theor

y

, derivational or constraint-ba

s

e

d. In this chapter, we will look at some

o

f the ma

j

or processes of En

g

lish in which morpholo

g

ical effects can be seen in

phonological rules, and in which phonological effects can be seen in morphology.

C

learly we cannot hope to cover all such interactions, and the reader is encouraged

t

o consult the many detailed treatments of English phonology-morphology

r

elationships that have appeared in the literature. I have tried to refer to a cross-

section of them here, but my bibliography is only meant as a starting point and

l

eaves out more significant pieces of scholarship than

i

t i

s

a

b

l

e

t

o

m

e

nti

o

n

.

2. EFFECTS OF LEXICAL CATEGORY

,

MORPHOLOGICAL STRUCTURE

,

A

ND AFFIX TYPE ON PHONOLOGY

I

n English, the most striking and pervasive effects of morphology on the

application of phonological processes can be seen in the stress system. English

st

r

ess

rules

ar

ese

n

s

iti

ve

t

o

(a) the morphological category of the word being stressed – for example, verb

vs.

noun

;

(

b

)

whether or not the word has an affi

x

– for example, suffixed ad

j

ectives are

s

tressed like nouns while unsuffixed ones are stressed like verbs

;

(c) the type of affix involved – some affix

e

s

affect stress

,

others are ‘stress-neutral’

;

t

his explains for instance why the addition of the adverb-forming suffix -l

y

n

o

t

ed

above can result in stress four syllables from the end of the word;

(d) whether the word is related to an independently occurring word – the so-called

c

yclic effect, which in part explai

ns

th

ed

iff

e

r

e

n

ce

in

s

tr

ess be

t

wee

n

còndènsátion

an

d

P

ènnsylvánia.

2

.1 Effects of lexical category and of morphological complexity

The basic metrical system of English fa

v

ors moraic trochees

,

much like that of

m

any languages (Hayes 1995). One of the last two syllables of a word is stressed,

and alternating syllables before the stressed one also receive stress, resulting in a

phonologically pleasing alternating pattern of weak and strong syllables:

(

1

)

a.

p

úr

p

le

,

áq

ua

,

c

r

ímson

,

f

úchsia

,

ma

gé

nta

,

siénna

b.

ma

r

óon

,

ové

r

t

,

p

olít

e

,

ma

r

íne

,

r

obúst

c.

àmarýllis

W

OR

D

-

F

ORMATION AND PHONOLOGY

2

7

The ad

j

ectives above have stress on the penultimate syllable (1a) unless the final

syllable is extra-heavy, that is, contains two consonants or a long vowel (1b). When

t

he word gets long enough, as in (1c), a secondary stress appears two syllables bac

k

from the ri

g

hthand stress, as in the first

a

of

àmar

ý

llis

.

There aren’t ver

y

man

y

m

orpholo

g

icall

y

simple but lon

g

ad

j

ectives in En

g

lish to fill out the (1c) examples,

but there are dozens of other monomorphemic lon

g

words in En

g

lish that show us

this rhythmic repeating strong-weak

pattern, such as the place names:

k

M

ì

MM

ssiss

ípp

i

,

C

òlorád

o

,

S

hèndandóah

,

C

àrolína

.

V

erbs, like unsuffixed adjectives, are stressed on last syllable if it is heavy,

o

therwise on penult. The (a) forms in (2) below have long vowels or diphthongs in

t

he ultima, while the (b) forms have the two consonants needed to make a syllable

h

eavy after final consonant extrametricalit

y

is taken into account. The forms in

(

c

)

e

nd in a sin

g

le short vowel plus a sin

g

le consonant, so the

y

are stressed on the

p

enult.

(

2

)

a.

o

bé

y

,

atóne

,

a

rr

íve

b.

molést

,

us

ú

rp

,

tò

r

mént

,

1

collápse

c.

astónish

,

édit

,

devélop

V

erbs too show an alternatin

g

pattern when the

y

are lon

g

enou

g

h to require

a

dd

iti

o

nal

s

tr

esses:

(

3

)

démonst

r

àte

,

r

é

cogn

ì

z

e

,

mánifès

t

,

exp

é

rim

è

n

t

H

owever

,

observations made first in

C

homsk

y

and Halle (1968; henceforth

S

P

E

)

will show us how morpholo

gy

intervenes.

F

irst, the lexical category of the word matters in English

.

2

W

hile verbs and

u

nsuffixed adjectives show the quantity-sensitive trochaic pattern we have been

o

bserving,

S

P

E

argues at length that nouns in English do not show this

E

final/penultimate stress pattern, but rather have all the action shifted one syllable to

t

he left, so that stress appears on a heavy penult (forms in 4c) or, if the penult is

l

ight, on the antepenult (4a, b). The weight of the final syllable becomes largely

i

rrelevant, as the forms in

(

4b

)

show.

3

1

The secondary stress on this form is due to its initial syllable being heavy.

2

Some views of phonology are sufficiently restrictive as to refuse to categorize a process as phonological

i

f it must mention specific lexical categories

.

See Bermúdez-Otero and McMahon

(

in

p

ress

)

for

d

i

scuss

i

o

n

.

3

An added complication is that long vowels in the final syllable of a noun do attract stress (SPE: 77-79;

Ha

y

es 1982).

28

ELLEN M

.

KAISS

E

(

4

)

a. asp

á

ragus

,

jávelin

,

A

mé

r

ica

4

b.

exp

é

rimen

t

,

l

ábyrinth

,

t

é

mpes

t

,

há

r

vest

,

lárynx

c.

agénda

,

uténsil

,

ve

r

ánda

,

a

r

óma

,

appéndix

Morpholo

gy

also intervenes in another wa

y

: ad

j

ectives containin

g

a certain

g

roup of suffixes are stressed accordin

g

to

t

he noun

p

attern rather than

p

attern fo

r

v

erbs and underived adjectives. The adjectives in (1) are all underived – they contain

n

osu

ffix

.

S

PE

s

hows that derived adjectives in English do not show this

final/penultimate stress pattern, exhibiting instead the penult/antepenult behavior of

t

h

e

n

ou

n

s.

(

5

)

a.

ínnoc

+

ent

,

si

g

ní

f

ic+an

t

,

ómin

+

ous

,

ma

g

n

á

nim+ous

,

p

érson+al

,

muníci

p

+al

,

p

r

í

m+itive

b.

malígn+an

t

,

momént

+

ous

,

autúmn

+

al

,

àdjectív+al

,

ex

pé

ns+ive

N

otice in

p

articular that the suffixes -ent

/

-an

t

end in two consonants

,

but those

c

onsonants do not attract stress as they do in the simple adjectives

o

vért

,

robúst.

S

o

t

he mere presence or absence of a suffix may affect the phonology of stress.

S

P

E

(38ff) notes a similar affix-dependency for nouns. While, as we’ve seen, a heavy

penult normall

y

attracts stress in a noun, those nouns derived with the affix -

y

a

s

in

í

ndustr

y

, éner

gy

, módest

y

,

and

g

álax

y

,

s

kip the heav

y

penult and place stress on the

ante

p

enult.

5

2

.2 Cohering and non

-

cohering affixes

If the English stress system did not already seem intricate enough, it turns out

t

hat it involves one more ma

j

or complication. Not only lexical category and the

s

imple presence or absence of an affix affects En

g

lish stress. Like man

y

othe

r

l

an

g

ua

g

es, En

g

lish has two different kinds

o

f suffixes (and prefixes), often referred

to

a

s

cohering

and

g

n

on-coherin

g

.

6

The ad

j

ectival suffixes -

ous

an

d

-al

an

d

th

eo

th

e

r

s

4

While stress in En

g

lish is lar

g

el

y

rule

g

overned, there are cases where the stress rules strictl

y

limit but

do not uniquely predict the correct output. For instance, many vowel-final nouns in English receive

stress on the penult even when it is light (Nessly 1977). Thus, while A

mé

r

ica

an

d

P

ámela

ha

ve

th

e

g

enerally expected penultimate stress, vowel-final spaghétti

,

Colorád

o

,

an

d

M

ississ

í

ppi

s

h

ow

th

esub

-

r

egularity of penultimate stress.

5

Some treatments of En

g

lish stress treat noun-formin

g

-

y

and ad

j

ectival suffixes like -

ent

,

-

al

,

-ous

,

an

d

e

ven the multi-syllabic -

ative

an

d

-ator

y

as ‘extrametrical’, that is, that they are ignored in the syllable

c

ount. Extrametricality should not be confused

w

i

th being non stress-affec

t

ing or non-cohering (see

immediately following section, especially exam

ple (6)), as they do affect position of stress

m

i

n th

eb

a

se

t

o which they are added and h

a

v

e multiple other effects on the syllabification and segmental

phonolo

gy

of their bases.

6

A welter of overlappin

g

but not entirel

y

equivalent t

e

r

ms exists to refer to the bipartite division of

affix

es.

S

P

E

re

f

e

r

s

t

o

th

e

m a

s

stress-affecting

and

g

st

r

ess

-

neut

r

al

,

and uses the juncture symbols +

v

s.

#

i

n attaching them to bases. Whitney (1889), in his discussion of Sanskrit, uses the terms primar

y

an

d

s

econdar

y

.

Work after Siegel (1974) often uses the terms

class

1 an

d

class

2

,

while Lexical

Phonology and Morphology calls the

m

level

(or stratum)

l

1

an

d

level

(or stratum)

l

2

(Kiparsky 1982).

O

ne often finds reference to

stem

-

level

vs

word-level

affixes (based on the idea that cohering affixes

l

W

OR

D

-

F

ORMATION AND PHONOLOGY

29

i

llustrated in (6) are cohering – that is, they interact with the rest of the word in

d

etermining its phonology. Note that adding those suffixes can result in the main

stress falling in a different place than it

does

in th

eu

naffix

ed wo

r

d:

(

6

)

móment moméntous

áutumn autúmnal

ádjective àdjectíval

The presence of the suffix may also resu

l

t in the laxing of a vowel, according to

t

h

e

r

u

l

eo

f Trisyllabic Laxing, which requires vowels three syllables or more fro

m

t

he end of a word to be lax, so lon

g

as the followin

g

s

y

llable is unstressed:

(

7

)

o

#

men o

#



minous

mali

#

gn mali

#



li

gnant

i



ii

li

#

ne li

##



li

near

com

p

e

#

te compe

#



titive

p

e

#

nal pe

##



n

alty

A

nd a vowel-initial cohering suffix can provide the nucleus to which a stem-final

sonorant consonant will attach as onset.

W

hen the suffix is absent

,

that sonorant

m

ust become the nucleus of a syllable itself. (

S

P

E

:

85-86

).

(

8

)

d

isastr-ous/disaster

,

hindr-ance

/

hinder, cycl-ic/cycle, rhythm-ic/

/

r

hythm

To summarize, in the so-called cohering suffixes, the phonology is sensitive to

t

he presence of the suffix. However, English has many non-cohering or ‘neutral’

suffixes as well. These do not seem to have an

y

effect on the position or de

g

ree of

stress. The non-coherin

g

suffixes of En

g

lis

h

consist of virtuall

y

all inflectional

suffixes

,

7

plus a subset of the derivational ones. I list some of the more common

neutral derivational suffixes in (9), using

t

h

ec

rit

e

ri

o

n

o

f

w

h

e

th

e

r th

esu

ffix i

s

tak

e

n

i

nto account when assigning stress.

(

9

)

-

able

,

-er (agentive)

,

-

en

,

-ful

,

-hood

,

-

ish

,

-

ism

,

-

ist

,

-

i

z

e

,

-

less

,

-

like

,

-

ment

,

-

ness

,

-l

y

,

-

wise

,

-y (adjective-forming)

I

n (10) are some of the major stress-affecting suffixes.

see

m a

b

l

e

t

o

atta

c

h t

o bou

n

ds

t

e

m

s

a

swe

ll a

s

t

o

i

ndependent words while non-

c

ohering affixes attach

o

nl

y

to words). Coherin

g

affixes are said to be i

n

close

j

uncture, while non-coherin

g

ones are in open

j

uncture.

7

Kiparsk

y

(1982) makes the point that some inflectiona

l

suffixes have allomor

p

hs that are added at level

1 and have phonological (cohering)

e

ff

ec

t

so

n th

e

ir

b

a

ses:

keep/kep

t

,

hide

/

hidden

,

child

/

children.

30

ELLEN M

.

KAI

SS

E

(

10

)

-age, -al

,

-

an

,

-

ant

,

-ance, -ar

y

,

-

ate

,

-

ic

,

-

ion

,

-if

y

,

8

-it

y

,

-or

y

,

-

ous

,

-y

(noun-forming)

W

e saw in (2) that verbs are typically stressed on an extra-heavy ultima or on the

penult. However, inflectional suffixes do not affect the s

y

llable count nor cause

v

erbs to be stressed as if the

y

ended in two consonants. The same is true of a stress-

n

eutral suffix like verb-formin

g

-

i

z

e

:

(

11

)

astónishing

,

astónishes

,

édits

,

devélops

,

Áfricanize

,

partícularize

En

g

lish also has a

g

eneral rule that tells us which of a number of stresses in a

word will be the main stress – that is

,

which of the feet contains the head of the

m

etrical structure: it is the rightmost stress that has at least one syllable after it

.

9

In

o

ther words, the very last syllable of words with more than one stress is not usually

t

h

eo

n

e

that r

ece

i

ves

th

e

main

s

tr

ess.

1

0

Only a final foot that branches gets to contain

t

he head syllable of the word. So, in the following monomorphemic words with

alternating stresses, the last stress wins out

as main stress in (12a) because the last

t

s

tressed syllable has another syllable following it. In (12b), however, the second-to

l

ast-stress emer

g

es as the main stress and the final stressed s

y

llable receives onl

y

s

econdar

y

stress:

(

12

)

a.

Mì

ssiss

ípp

i

b.

P

ánamà

Thèodó

r

a

T

héodò

r

e

C

àrolína

C

árolìne

1

Tìcònderó

g

a ní

g

htin

g

àle

In the examples in (13), a stress-affectin

g

suffix counts for makin

g

a s

y

llable

n

on-final and thus eligible for the main stress.

(

13

)

díalèct

dìaléctal

ánecdòte

ànecdótal

ó

r

iènt

ò

r

ièntátion

P

á

ra

gù

a

y

P

à

ra

gú

a

y

an

máni

f

èst màni

f

èstátion

8

Raffelsiefen (2004) argues that -if

y

is not really stress-affecting. R

a

ther, it selects already-existing stems

with stress that may happen to fall in a different

place than in the underived adjective. Thus, for

t

i

nstance, she argues that

s

olídif

y

i

s based on the

(

stressed

)

stem of

s

olídit

y

rath

e

r than th

ewo

r

d

sólid

.

S

ee section 7 for further discussion.

9

This generalization was first noted, I believe, in

S

PE.

I

t has been restated and refined repeatedly in

subsequent

y

ears.

10

There are several suffixes

,

such as -

ése

,

which receive main stress in contradiction to this lar

g

e

generalization. Other sub-regularities as well as this one

,

such as the tendency of word-final verb

stems to be stressed

(

i

nj

é

c

t

,

conféss

)

are discussed at some length in Liberman and Prince (1977.)

11

The example is relevant, of course, only in the pronunciation where the final syllable is unreduced

[àjn].

W

ORD

-

FORMATION AND PHONOLOGY

31

However, the stress-neutral suffixes are not visible to this rule – they do not

c

ount as adding a final syllable and their presence therefore does not result in the

preceding syllable receiving main stress.

(

14

)

díalècthood

,

ánecdòte

-

like

,

mánifèstl

y

,

ó

ri

è

nting

Similarly, these neutral suffixes are not relevant in determining the position of

the rightmost stress (that is, the rightmost foot) – words containing them have stress

i

n the position it occupies in the word the

y

are derived from, even if that results in

unstressed closed s

y

llables, sequences of unstressed s

y

llables (lapses), or ri

g

htmost

s

tresses fallin

g

more than three s

y

llable

s

from the end of the word. Compare the

position of stress in the words in (15a), their unchanged derivatives with stress-

n

eutral affixes in

(

15b

)

, and their derivative

s

with stress-affecting affixes in (15c).

(

15

)

a.

r

ígid

b.

r

ígidness

c.

r

igídit

y

óri

g

in óri

g

inless orí

g

inal

móde

r

n

m

ódernism modérnit

y

hóspital

h

óspitalize hòspitálit

y

T

urnin

g

to phonolo

g

ical phenomena other than stress, we notice that the neutral

s

uffixes cannot rescue an otherwise uns

y

l

la

b

ifia

b

l

es

t

e

m-final

so

n

o

rant

co

n

so

nant

.

12

(

16

)

a.

hinde

r

b.

hindering

c.

hind

r

ance

rh

y

thm rh

y

thm-

y

rh

y

thmic

meter meterin

g

metrical

A

nd they do not provide the syllable count needed to trigger Trisyllabic Laxing.

(

17

)a

.me

#

ter

##

b.

me

#

tering

#

c.

me



t

r

ical

na

#

tion na

#

tionhood

##

na



tional

o

#

men o

#

menlike

#

o



minous

spi

#

ne spi

#

nelessness

##

N

eutral suffixes also do not trigger the rule of velar softening.

(

18

)

a. pirat

e

,

pirac+y

b.

meat

,

meat

#

y

W

e have now seen that En

g

lis

h

phonolo

gy

must have access to the

m

orpholo

g

ical structure of a word, its lexical cate

g

or

y

, and the distinction between

12

Indeed, Booi

j

(1977, 1995) ar

g

ues that, at least for Dutch

,

which shows a similar bifurcation in its

affixes, the effect that an affix has on syllabific

a

t

i

o

n i

s

th

e

main

c

rit

e

ri

o

n

o

n

w

hi

c

h t

odec

i

de w

hi

c

h

c

lass it falls into. Affixes that don’t interact in the basic syllabification of a word are considered to fall

o

utside the phonological word. Such affixes show a

m

ismatch between their morphological structure

(one morphological word) and their phonological struct

ure. For more discussion of ‘co-present’ and

tt

possibly conflicting morphologi

c

al and phonological structures, see Inkelas (1994).

3

2 ELLEN M

.

KAI

SS

E

o

f phonology, creating clashing stresses (

còn

dè

n

sá

tion

), strings of unstressed

syllables (

ó

r

iginless

)

s

,

heavy syllables with no stress on them (

rí

gid

í

ness,),

dd

no

n-

primary rightmost stresses followed by another syllable

(

día

lècthood

)

,

s

yllabic

s

onorants before vowels

(

hind

eri

d

ng)

ii

,

and other otherwise m

y

sterious complications

i

n the sound patterns of the lan

g

ua

g

e. We turn now to interactions of the opposite

k

ind, where morpholo

gy

seems to be required to take phonolo

g

ical factors into

a

ccou

nt

.

In

sec

ti

o

n 6

we w

ill t

o

r

e

t

u

rn

to d

i

scuss

an

o

th

er

k

ind of morphological

i

nfluence on phonology – the so-called

cyclic effect

seen in

t

còndènsátion

,

where

s

ome as

p

ects of the

p

ronunciation of a word are carried over into words formed

fr

o

m that

wo

r

d.

3. M

O

RPH

O

L

OG

Y LIMITED BY THE PH

O

N

O

L

OG

I

C

AL F

O

RM

O

F THE

B

A

S

E

O

F AFFIXATI

O

N

Th

e

lit

e

rat

u

r

eco

ntain

s

a n

u

m

be

r

o

f

c

a

s

es where phonology influences the ability

o

f morphology to act as it otherwise would if unfettered. Hargus (1993: 48-52)

c

ontains an excellent summary and evaluation of the most convincing cases of this

k

ind, though she does not talk about any English examples at length. In one common

t

y

pe, a suffix has different allomorphs dependin

g

on the len

g

th of the base, the

base’s stress pattern, or the se

g

ment in wh

i

ch the base ends. Thus in German

(

Hall

1990), one deverbal suffix has the allomorph -

e

r

ei

w

h

e

n a

dded

t

ove

r

bs

t

e

m

sw

ith

final stressed syllables, while -

ei

i

s added to other verb stems. Perha

p

s the most

o

ften-cited case of affixal sensitivity to

phonology is that of

the affixation of

f

-al

to

l

v

erbs to form nouns. Siegel (1974) discovered that this

-al

(not to be confused with

l

the ad

j

ective-forming -

al

found in

l

autumnal

)

can only be added to verbs with final

s

tr

ess.

Th

us we

fin

d

arríval, bestówal, reférral, avówal, renéwal, and so on

,

but not

*

édital

,

*énteral

,

o

r an

y

other non-finall

y

stressed verb plus -

al

.

1

3

A

n

o

th

er

t

horou

g

hl

y

documented case is that of -

i

z

e

(Marchand 1969, Or

g

un and Sprouse

1999, Raffelsiefen 2004), which can onl

y

be productivel

y

affixed to items with non-

final

s

tr

ess.

I

n both these cases

,

the interes

t

for the phonology-morphology interface

t

i

s clear. Stress is a derived, largely predictable phonological property, not present in

u

nderlying forms. If we did not have to deal with such cases, the simplest theory of

h

ow phonology and word formation interact would be for them to have a single

i

nterface. Word formation would assemble all the morphemes, after which

phonology would deal with the result.

1

3

Hargus (1993) argues that

-al

affixation is too lexically unpredictable to serve as a clear example of

l

m

orpholo

gy

dependent upon phonolo

gy

. For instance,

a

rr

ival

e

xi

s

t

sbu

t

*

derival

does not. While I

l

co

ntin

ue

t

o

think that final

s

tr

ess

i

s

a

n

ecessar

y

, if not sufficient, condition fo

r

-

al

affixation

,

the case

of

-

i

z

e

is more extensivel

y

documented. Sie

g

el also finds stress-dependent behavior for deverbal -

f

ul

(

f

or

g

ét

f

ul),

w

hich attaches to verbs with final stress

,

and

sy

llable countin

g

behavior for de-ad

j

ectival

-

en

.

t

wo kinds of affixes. In this way, morphology may interfere in the normal workings

W

OR

D

-

F

ORMATION AND PHONOLOGY

33

(

19

)

I

n such a model, which is basically that of

S

P

E

, there is no obvious way for a

m

orphological rule to have access to deri

v

ed phonological information such as the

position of stress. Such models may be called

non

-

inte

r

actionist

.

1

4

An

inte

r

actionist

model

,

on the other hand,

p

ro

p

oses that som

e

affixes can be added, followed b

y

some phonolo

g

ical operations, follow

e

d

b

y

additional affixation:

(

20

)

a

rr

ive

→

a

rr

íve

→

a

rr

íval

O

ne of the most widely adopted views of morphology-phonology interaction in

t

he 1980’s and earl

y

1990’s was Lexical Phonolo

gy

and Morpholo

gy

(

LPM

)

(Kiparsk

y

1982, Mohanan 1982, Kaisse and Shaw 1985, Har

g

us and Kaisse 1993

and man

y

others.) LPM adopts Sie

g

el’s interactionist view that morpholo

g

ical and

phonological operations can be interspersed, again with stress assignment potentially

occurring before some mor

phological affixation. We will discuss LPM and its

r

p

roblems in section 2.1.4.

I

n a non-derivational model like Optimality Theory, one can use constraint

r

anking to allow phonological considerations like those disfavoring clashing o

r

l

apsing stresses to outweigh the generall

y

productive ability of the morphology to

form new words. Raffelsiefen (2004) provides an interestin

g

wa

y

of lookin

g

at the

t

ension between morpholo

gy

and phonolo

gy

. She asks us to consider the lexicall

y

s

tr

essed su

ffix

es

-

ée

,

-

ése

,

an

d

-

ée

r

.

Each of these suffixes not onl

y

receives stress,

as we might expect of a suffix containing a tense vowel, but must be marked in the

le

xi

co

n t

o

r

ece

i

ve

main

s

tr

ess.

1

5

W

hen -

ée

,

-

ése

,

or

-

ée

r ar

e

a

dded

t

ob

a

ses

that

e

n

d

i

n unstressed syllables, no problem arises with respect to the resulting stress pattern.

Fo

rm

s

lik

e

ex

à

min

é

e

,

v

ì

sit

é

e

,

m

ò

untain

é

er

,

an

d

mèdicalése

do not have adjacent,

h

ence clashing stressed syllables. However, the phonology finds a different solution

f

o

r

e

a

c

h

o

f th

ese

thr

ee su

ffix

es w

h

e

n the morphology attempts to coin words on

b

a

ses w

ith final

s

tr

ess.

F

or

-

ése

,

the solution is a repair: stress is shifted off the final s

y

llable of the base

to a preceding syllable:

1

4

See Odden (1993) for a proposal about how one might replace the intera

c

ti

o

ni

s

t m

ode

l

scu

rr

e

nt in th

e

h

e

y

da

y

of Lexical Phonolo

gy

and Mo

r

pholo

gy

and return to a non-interactionist model that maintains

,

nonetheless, the segregation of

l

exical from postlexical phonology.

1

5

Recall that normally a final sylla

ble does not receive the main stre

a

s

s in English. Certain morphemes

and other strin

g

s, such as the

óon

o

f

ballóon

an

d

p

ò

nt

ó

on

mus

t

be

mark

ed

t

o

r

ece

i

ve

main

s

tr

ess

in

c

ontradiction to this

g

eneralization.

Morphology

Phonology

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

Подождите немного. Документ загружается.