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

Подождите немного. Документ загружается.

280

LER

d) change within a derivational class,

e) other indigenous WF productivity

e start from the

eneral assumption that it is more difficult to inte

rate words

oming from a foreign language [marked alternative] than indigenous words

[

unmarked alternative]), 2) when these foreign words have unfitting properties, an

3) when these unfitting properties are ‘fitted’, i.e. accommodated to the properties of

a derivational class or of the respective language-specific system adequacy in

particular. Clearly a WFR must have optimal productivity in order to overcome the

difficulties of

(

)

and

(

)

and to enforce accommodation (3). Similarly abbreviations

are not formed via

rammatical WFRs but b

extra

rammatical means and are

therefore both mar

inal and marked within the lexical stock of a lan

t is ver

difficult to distin

uish the first two criteria in contemporar

lish,

because nowadays rather few foreign word

are loaned into English and English

FRs are rather liberal w

th their in

ut conditions.

umping these two criteria

together we can identify English right-hand-headed nominal compounding as fully

productive, because it may be freely app

ied to nouns of foreign origin, e.g. traffic

jam spiel, delicatessen sho

riterion c

)

can be illustrated

inal-n

ber-for

ng suffi

-t

ough

the domain of its application is severel

limited, i.e. to cardinal numbers, it ma

freel

attach to abstract numbers, as in

and

are

otential

rdinal numbers. In contrast, the homophonous noun-forming deadjectival suffix -

(

as in

) is completely unproductive, because it cannot form any new noun

from adjectives. Criterion d) applies mainly to diachronic substitution of one affix

by another one, where substitution is preceded

by a stage of competition between an

lder derivation with a less productive affix and a new variant which exhibits a more

productive suffix (cf. Bauer 2001: 177ff), as in apprais-al

vs.

apprais-emen

4.2 Dynamic vs. static morphology

Productivity has obtained a still more importa

t role in a recent develo

ment of

M, which distinguishes the two interacting components of

tatic morpholog

(

the

rganisation of stored morphological forms, including stored compounds and

derivations

)

and

ynamic morphology

(

based on morphological rules, with its core

n productive morphology). We assum

in performance competition between

dynamic and static morphology for all mo

phological constructions which can be

andled both b

rules and b

direct lexical access (cf. for performance the

orpholo

ical race model b

Baa

Schreuder 1991 and Frauenfelder &

Schreuder 1992

)

his means for word formation that

on the one hand, all acce

ted words are in

the domain of static morphology, because their word meaning is lexicalised. On the

ther hand, all

otential words

(

in the sense of Aronoff 1976

)

, at least those formed

by productive word-formation rules, are in the domain of dynamic morphology, but

for all accepted, i.e. lexicalised words, dynamic morphology

ust handles word-

formation meanin

ORMATION IN NAT

RAL MORPHOLOGY

281

4.3 Universal vs. t

polo

ical vs. s

stem-dependent naturalness

polo

ical adequac

(secti

n 3) ma

be understood as a filt

r an

rati

n universal naturalness/markedness

(

sec

ion 2), and language-specific syste

adequacy (4.1) as a filter and elaboration on typological adequacy. Each lower-level

filter can specify and even overturn prefe

ences of the preceding higher-order level

(

cf. Dressler et al. 1987, Bittner 1988, Wheeler 1993

)

s a consequence, lack of the typological and the language-specific system-

dependent filter leaves universal preferences intact. Therefore these play a highe

ole in the earliest phases of morpholo

acquisition b

small children (cf. Dressle

Karpf 1995) and in extra

rammatical or expressive morpholo

(Dressler 2000b,

see above section 2.1), such as in echo-word formation. S

mptomaticall

, small

hildren produce many more echo-words and similar extragrammatical

eduplications before they really start to acquire morphological grammar, i.e. before

system-independent preferences

al nat

raln

s are filtered by language-

specific system adequacy.

n such deductive and empirically accountable ways, the theory of NM

distinguishes different

types of naturalness and relates them between each other.

Wol

U. Dressler

Department o

Lin

uistics

University of Vienna

Berggasse

A-1090

ienna

ust

e-mail: wolfgang.dressler@univie.ac.a

REFERENCES

ronoff

Mark. 1976.

ord Formation in

enerative

rammar

Cambrid

e: MIT Press.

en, Rolf H. and Schreuder, Robert. 1991. “

ar and peace: Morphemes and full forms in a non-

nteractive activation parallel route model.” Brain and Lan

68, 2

auer

Laurie. 1983.

lish Word-Formation. Cambrid

e: Cambrid

e Universit

Press.

auer, Laurie. 1996. “No

honetic iconi

in evaluative morpholo

.” Studia Lin

uistica 50

189-204.

auer

Laurie. 2001.

orpholo

ical Productivit

Cambrid

e: Cambrid

e Universit

Press.

erko, Jean. 1958. The child’s learning of English morphology. Word 14. 150-177.

ittner, Andreas. 1988. “Is anything 'more natural?

onsiderations on establishing a hierarchy of

Naturalness Princi

les

(

)

.” Linguistische Studien

188

23-35.

loomer, Robert K. 1996. “Die pleonastischen Zusammensetzungen der deutsc

en Gegenwartssprache.”

merican Journal of Germanic Linguistics and Literatures

8, 69

ooij, Geert. 1992. “Compounding in Dutch.” Rivista di Linguistica

4, 3

7-5

ooi

, Geert; Lehmann, Christian; and Mu

dan, Joachim (eds.). 2000.

Morphologie/Morphology I

erlin: de Gru

ter

lark

Eve

;

Hecht

Barbara

;

and Mulford

Rand

C. 1986. “Coining complex compounds in English.”

Lin

uistics

24,

orbin

Danielle. 1987.

Morphologie dérivationnelle et

structuration du lexique

Tübingen: Niemeyer.

rocco Galèas, Grazia. 1990. “Conversion as a morphological metaphor.” In: J. Méndez-Dosuna and C.

Pensado

(

eds.

)

aturalists at Krems

alamanca: Ediciones de la

versidad de Salamanca

23-32.

DRE

LER

rocco Galèas, Grazia. 2003a. “The morpholo

ical

echnique of metaphoricit

in En

lish word-

formation.” In: E. Mela Athanaso

oulou

(

ed.

)

elected Papers from the 15

nte

national

ymposium on Theoretical and Applied Linguistics. Thessaloniki: Aristotle University, School of

English, 135-151.

rocco Galèas, Grazia. 2003b. “Compound adjectives in English: the type lion-hearted and good-

ed.

”

tudi Italiani di Linguistica Teorica e Applicata 32, 31-43.

rocco Galèas, Grazia and Dressler, Wolfgang U. 1992. “Trasparenza morfotattica e morfosemantica dei

omposti nominali più produttivi dell

’

taliano di oggi.” In: B. Moretti,

. Petrini and

. Bianconi

(

eds.

)

, Atti del 25. Congresso Società di Linguistica Italiana

Roma: Bulzoni

9-24.

e Knop, Sabine. 1987.

etaphorische Komposita in Zeitungsüberschriften

Tübingen: Niemeyer.

ressler, Wolf

U. 1982. “Zur semiotischen Be

ründun

einer natürlichen Wortbildun

slehre.”

urter Beiträ

e zur Sprachwissenscha

8. 72-87.

ressler, Wolf

U. 1985a. “T

polo

ical aspects of Natural Morpholo

.” Wiener lin

uistische

azette 36, 3-26

(

rinted in Acta Linguistica Hungarica 35, 1987, 51-70

)

ressler, Wolfgang U. 1985b.

orphonolog

Ann Ar

Kar

ma P

Dressler, Wolfgang U. 1985c. “On the predictiveness of Natural Morphology.”

ournal of Linguistics 21

321-337.

Dressler, Wolfgang U. 1988a. „Zur Bedeutung der Sp

achtypologie in der Natürlichen Morphologie.“ In:

Albrecht, J.

(

ed.

)

nergeia und Ergon

Tübingen: Narr, Vol.3, 199-208.

Dressler, Wolfgang U. 1988b. „Preferences vs. strict

niversals in morphology: word based rules.” In: M.

ammond and M. Noonan

(

eds.

)

, Theoretical morpholog

San Diego: Academic Press, 143-154.

ressler, Wolfgang U. 1990.

emiotische Parameter einer textli

guistischen Natürlichkeitstheori

Vienna: Verla

der Österreichisch

Akademie der

issenschaften.

ressler, Wolf

U. 1994. “LSP ‘von aussen’: Reflections on the 9th European LSP s

mposium.” In:

M. Brekke et al.

(

eds.

)

, Applications and Implications o

Current LSP Research. Ber

en:

forla

et, 950-969.

Dressler, Wolfgang U. 1995. “Wealth and poverty of functional analyses, with special reference to

functional deficiencies.” In: Sh. Miller and J. Mey (eds.), Form and Function in Language,

Proceedings of the First Rasmus Rask Colloquium

Odense: Odense Univ. Press

11-39.

ressler, Wolfgang U. 1999. “On a semiotic theory

f preferences in language.” In: M. Shapiro and M.

aley (eds.),

ssays in Semiotic Analysis, The Peirce Semiotic Papers

389-415.

ressler, Wolfgang U. 2000a. “Naturalness.” In: G

Booij, C. Lehmann and J. Mugdan (eds.),

orpholo

ie: ein internationales Handbuch zur Flexion und Wortbildun

I. Berlin: de Gru

ter, 288-

296.

ressler, Wolf

U. 2000b. “Extra

rammatical vs. ma

inal morpholo

.” In: U. Doleschal and A. M.

Thornton

(

eds.

)

xtragrammatical and Marginal Morpholog

. München: Lincom

1-10.

Dressler, Wolf

U. 2002a. “Nat

raln

ess

nali

” In

K. Dziubalska-Kolacz

k and J.

eckwerth

(

eds.

)

, Future Challen

or Natural Lin

uistics

Lin

83-101

Dressler, Wolfgang U. 2002b. “Naturalness and morphologi

al change.” In: B. D. Joseph and R. D. Janda

(

eds.

)

he Handbook of Historical Linguistics

Oxford: Blackwell

461-471.

ressler, Wolfgang U. and Karpf, Annemarie. 1995. “The theoretical relevance of

re- and

protomorphology in l

nguage acquisition.” In: G. Boo

j and J. van Marle (eds.),

earbook of

orphology 199

Dordrecht: Kluwer

99-122.

ressler, Wolfgang U. and Ladányi, Mária. 2000.

“

Productivity in Word Formation: a morphological

approach.” Acta Linguistica Hungarica

, 103

144

Dressler, Wolfgang U.; Mayerthaler, Willi; Panagl, Oswald; and Wurzel, Wolfgang U. 1987. Leitmotifs in

Natural Morpholog

. Amsterdam: John Benjamins.

Dressler, Wolf

U. and Merlini Barbaresi, Lavinia. 1994.

orphopra

matics

rlin

Gru

ter

Dressler, Wolf

U. & Merlini Barbaresi, Lavinia. 2001. “Morphopra

matics of diminutives and

mentatives.” In: I. Kenesei and R. M. Harnish (eds.),

erspectives on Semantics, Pra

matics, and

Discou

Amsterdam: John Benjamins, 43-58.

elber

Helmut and Budin

Gerhard. 1989.

erminologie in Theorie und Praxis

Tübingen: Narr.

rauenfelder

Uli H. and Schreuder

Robert. 1992. “Constraining psycholinguistic models of

orphological processing and representation: the rol

of productivity.” In: G. Booij and J. van Marle

(

eds.

)

earbook of Morpholog

991

Dordrecht: Kluwer

165-183.

all, Christopher. 1992.

Morphology and Mind

London: Routledge.

ORMATION IN NAT

RAL MORPHOLOGY

283

akobson

Roman. 1941. Kindersprache, Aphasie und allgemeine Lautgesetze. Uppsala: Almqvist &

iksell.

ensen

John T. 1990.

orpholo

: Word Structure in Generative Grammar

Amsterdam:

ohn

amins.

Kilani-Schoch

Marianne. 1988. Introduction à la morpholo

ie naturelle

Bern: Lan

, Leon. 1979. “The principle of transparenc

and semantic antinomies.” In: J. Pelc (ed.),

emiotics in

Poland

. Dordrecht: Reidel, 376-406.

ibben, Gary. 1998. “Semantic transparency in

he processing of compounds: consequences fo

epresentation, processing and impairment.” Brain and Language 61

30-44.

ieber, Rochelle. 1992. “Compounding in English.”

ivista di Linguistica

79-96.

Mayerthaler, Willi. 1981.

orphologische Natürlichkei

. Wiesbaden: Athenaion (English translation:

1988

orphological Naturalness

Ann Arbor: Karoma Press

)

Méndez Dosuna

Julián and Pensado

men. 1990. “How unnatural is Spanish V

ctor --> Vict-

t-or

nfixed diminutives in Spanish.” In: Méndez Dosuna, J. and Pensado, C. (eds.),

aturalists at Krems

Salamanca: Acta Salmanticensia, Estudios Filolo

icos 227, 89-106.

er, Ralf.1992. Compound Comprehension in Isolation and in Contex

Tübin

en: Nieme

er.

Olsen, Susan. 2001. “Copulative compounds: a closer look at the interface between syntax and

orpholo

.” In: G. Booi

and J. van Marle (eds.),

earbook o

Morpholo

2000

Kluwer

279-320.

Peirce, Charles S. 1965. Collected Papers. Cambridge: Harvard University Press.

Plag, Ingo. 2003. Word-Formation in English

Cambridge: Cambridge University Press.

Plag, Ingo. 2004. “Syntactic category information and the semantics of

derivational morphological rules.”

Folia Linguistica

38, 193

Rainer

Franz. 1993. Spanische Wortbildungslehre

Tübingen: Niemeyer.

Rainer, Franz. 2001. “Compositionality and paradigmatically determined allomorphy in Italian word-

formation.” In: C. Schaner-Wolles

Rennison, and F. Neubarth

(

eds.

)

, Naturally! T

rin

Rosenber

and Sellier, 383-392.

Ronneber

er-Sibold, Elke. 1980.

prachverwendun

, Sprachs

stem

,Ö

konomie und Wandel

Tübin

en:

Nieme

Scalise, Sergio. 1992. “Compounding in Italian.” Rivista di Linguistica

175-199.

Scherer

Bernd M. 1984. Prolegomena zuu einer ei

nheitlichen

eichentheo

Stuttgart: Stauffenburg

erlag

Seiler, Hansjakob. 1991. “Die Pri

zipien der deskriptiven und der

tikettierenden Benennung.” In: H.

Seiler

(

ed.

)

, Linguistic Workshop 3, 2-57.

Shaw

Howard J. 1979.

otivierte Kom

osita in der deutsche

und englischen Gegenwartssprache

Tübingen: Narr.

Siegel, Dorothy. 1977. “The adjacency constraint and the theory of morphology.”

NELS

8, 189-197.

kaliþ

Vladimír. 1979.

ypologische Studien

Braunschweig: Vieweg.

Stampe, David. 1973. “On Chapter Nine.” I

M. Kenstowicz and Ch. Kisseberth

(

eds.

)

ssues in

honological Theor

The Hague: Mouton, 44-52.

an Marle, Jaap. 1985.

n the Paradigmatic Dimension of Morphological Creativit

Vennemann, Theo. 1983. “Theories of lin

uistic prefer

nces as a basis for lin

uistic explanations.” F

olia

uistica historica

5-26.

heeler, Max W. 1993. “On the hierarch

of natura

ness principles in inflectional morpholo

.”

ournal

of Linguistics 29

95-111.

ijnen, Frank; Kempen, Masja; and Gillis, Steven. 2001. “Root infinitives in Dutch early child language:

an effect of in

ut.” Journal of Child Language 28

629-660.

illiams, Edwin. 1981. “Argume

t structure and morphology.” Linguistic Review 1

81-114.

underlich

Dieter. 1986. “Probleme der Wortstruktur.” Zeitschrift für Sprachwissenschaf

209-252.

urzel, Wolfgang U. 1984. Flexionsmorphologie und Natürlichkei

. Berlin: Akademie-Verlag.

urzel, Wolfgang U. 1988a. “The structura

ist heritage in Natural Morphology.” Linguistische Studien A

188, 99

116

urzel, Wolfgang U. 1988b. “Hermann Paul

–

nalogie und die Natürliche Morphologie.”

eitschrift für

ermanisti

7-5

urzel, Wolf

U. 1994. “Grammatisch initii

ter Wandel.” In: Jeßin

, B. (ed.),

prachd

namik. Au

em We

zu einer T

polo

ie sprachlichen Wandels

Bochum: Brockme

er, Vol.1, 7-114.

wanenbur

, Wiecher 1992a. “Compoundin

in French.” Rivista di Lin

uistica

221-240.

284

DRE

LER

wanenbur

, Wiecher 1992b. “La composition dans les lan

ues romanes et

ermaniques: essuie-

lace

ind-shield wi

er.”

TS Working Papers

OTS-WP-TL

92-113.

wicky, Arnold M. and Pullum, Geoffrey K. 1987. “P

ain morphology and expressive morphology.” In:

. Aske, N. Beery, L. Michaelis and H. Filip (eds.), Proceedings of the 1

Annual Meeting of the

erkeley Linguistic Societ

330-340.

285

tekauer, P. and R. Lieber (eds.), Handbook of Word-Formation, 285—313

2005 Sprin

er. Printed in the Netherlands

WORD-FORMATION IN OPTIMALITY THEORY

PETER A

KEMA AND AD NEELEMAN

1. INTR

1.1 Optimality Theory

grammar can be defined as the set of principles that distinguish the possible

orpheme combinations, word combinati

ns and sound combinations in a language

from the impossible ones. In traditional grammars a possible word, sentence o

yllable is one that satisfies all the principles pertaining to it. Data may be accounted

for b

a conspirac

of principles, but the principles themselves do not compete with

ne another. No

rinci

le is violated in order to avoid violatin

another principle. In

fact, no

rinci

les are violated at all in a

rammatical sentence;

lati

eve

n a

ingle grammatical principle i

exorably means ungrammaticality.

n recent years, theories of grammar ha

e come up in which this no longer holds

true, in particula

ptimality Theor

(henceforth OT) (see Prince and Smolensky

2004 [1993]). This theory emphasizes the role of competition in determining which

forms are grammatical and which are not. The crucial question is which of a numbe

f forms that compete for the realizati

n of a particular concept satisfies the

principles of

rammar better than the others

(

where ‘better’ is defined in a

recise

, to be discussed shortl

). This will be the

rammatical structure. This implies

that

rammatical structures can violate principles of

rammar – as lon

as there is no

ompetitor that does better. This also implies that different principles of gramma

an impose demands on structure that are i

direct conflict (meaning that in any

tructure at least one of them will be violated

)

et us sketch the outlines of an OT-style grammar in a bit more detail

uch a

grammar consists of two components. The first is a device, called

(

erator

)

that

rmin

ow e

be co

nto a structure. The demands that GEN

oses on structures cannot be violated.

(

Thus, there remains room for inviolable

rinci

les in OT

)

. Below, we will assume a minimal GEN com

onent fo

orphology, one in which an operation of merger is applied to morphemes, so that

ierarchically ordered structures for words are built. This parallels the building of

tructure in syntax, but the morphological GENerator is distinct from the syntactic

ne and builds structures specifically for the sub-word level. In other words, we

ssu

ewo

t th

esu

f operations in phrasal syntax such as

ead movement. For a defense of such a specific ‘word syntax’ component to build

For a detailed introduction see for example Kager 1999.

286

PETER A

KEMA

AD NEELEMAN

orphological structure, see Ackema &

eeleman (2004); we will have nothing

further to say on it here.

The second component of an OT-grammar is an

evaluation met

that

ooses

from the output of GEN the structure that best satisfies a set of universal

const

aints

These constraints are all violable. Their impact in a specific language follows from a

anguage-particula

constraint ranking

against which the various candidate

eev

ed.

Evaluation proceeds as follows. The struct

ures to be compared (which make up

he so-called candidate set) are first evaluated with respect to the hi

hest-ranked

traint

wo o

r m

es r

eceive an equal score, they are judged by

he next highest constraint. In case there is still more than one surviving candidate,

hey are judges on the third constraint, and so on. The candidate that finally survives

his procedure is optimal and there

y grammatical. The other ones are all

ngrammatical – they are blocked by the optimal one. As noted, a consequence of

ew o

traint int

n i

tha

no constraint is necessarily surface true. A

lowe

r rank

ed co

traint

be v

lat

in an optimal structure

n thi

cores better on a hi

her ranked constraint than its competitors. Even the hi

hest-

anked constraint can be violated, namel

in case there is no potential output that

does

lat

The question we will address in this chapter is whether the OT view on grammar,

with its emphasis on competition between forms expressing the same concept, can

be advantageously applied to problems of word formation

We will see that

indeed

there are a number of phenomena that seem to ask for an account in terms of

etition.

1.2 Competition in morphology

he idea that morphological forms can be

n competition, so that one form may

block another, is in fact crucially involved in one of the oldest regulatory principles

n linguistics, nowadays usually known as the

lsewhere Principl

ll-kn

English example is the regular past tense of the verb

i.e.

goed

Although

orphologically well-formed, this form seems to be blocked by the irregula

went

One pattern frequentl

found in cases of bl

ckin

is that the availabilit

of more

specific forms excludes the use of more

eneral ones. Thus, in the case

ust

entioned, the question is how to form the past tense o

that is

how to realize

he morpho-syntactic structure in (1a) (here and below we assume a realizational

odel of affixation; see Halle & Marantz 1993, Beard 1995; for the s

ecific

assumptions we make, see Ackema & Neeleman 2004). The English lexicon

ontains the following relevant morphemes:

-ed

(which is specified as past), and

went

(which is specified as the past of

)

. Since

went

spells out the most features in

a single morpheme, it is the most specific form, and the one favoured in the

ompetition (as indicated b

)

We include inflectional phenomena in this concept, because some of the relevant phenomena only occu

with inflectional morphology. We will indicate wh

re this is the case in the relevant

laces.

RMATI

N IN

PTIMALITY THE

(

)

a. [

]

(morphosyntax)

b. [/go/ /ed/] (morphophonology)

’

)

[/went/] (morphophonology)

The structure of inflectional paradigms can often be described in these terms.

The most

eneral form is usuall

called the elsewhere form, since it will be inserted

where no more specific form is available. The Elsewhere Principle, which re

ulates

ompetition in the wa

described above, w

introduced into

enerative

rammar b

nderson (1968) and Kiparsky (1973). The

eneral idea reportedly goes all the way

k t

Panini

s outlined in section 1.1., such a notion of competition forms the very basis of

but it can be found in more or less limited forms in other frameworks as well

suc

h a

sso

eve

Minimalism

(see Chomsky 1995) and Williams’ (2003)

Representation Theor

Below we will explore the type of morphological

phenomena that can be anal

zed in terms of competition. We will show that there

are three relevant t

pes of competition: (i) competition between different

orphemes (of which classical Elsewhere cases are an instance), (ii) competition

between different orderings of the same morphemes, and (iii) competition between

orphological and syntactic realization of the same concept. As a second aim, we

will consider in how far

T is a suitable framework to deal with these kinds of

orphological competition, or whether i

ses

iti

nal

iff

assumptions about grammar are required.

MPETITI

N BET

EEN DIFFERENT M

RPHEME

1 T

he basic case

ointed out in section 1, the classical case of com

etition between different

orphemes deals with the structure of inflectional systems, and can be described in

erms of the Elsewhere Principle. In order to illustrate the general reasoning in some

ore detail, we consider Dutch verbal ag

eement in the present tense. The relevant

paradigm is as follows

alt

rnati

wee

doub

in the stem in the singular and single

in the plural does not

epresent an

difference in the qualit

of the vowel, but is an idios

ncras

f Dutch ortho

raph

(lon

owels are spelled doubly in closed syllables).

(

2) ik loop wi

lop-en

I walk

we walk-

jij loop-t jullie lop-en

you walk-2.

you-pl walk-

hij loop-t zij lop-en

PETER A

KEMA

D NEELEMAN

The most economical way of describing the distribution of affixes is in terms of

nt f

es:

PAR

(

TICIPAN

)

(

RESSEE

)

and PL

(

RAL

)

. The

alizati

f infl

ccou

r if th

h l

ntain

following specified affixes; we use subscripts to distinguish between homophonous,

tin

t affix

es:

(

)

∅

↔

[

PAR

]

↔

[

PAR

]

↔

[

]

sew

-t

e assume that morphosyntactic structure is fully specified for person and

number features. The workings of the Elsewhere Principle can be illustrated in

arious ways, even for this simple inflectional system. Consider first the singular. In

he first person, the information in the morpho-syntactic structure is that the verb is

pecified as [PAR]. Two of the affixes in (3), namely

-en

, spell out features

hat are not present in morphosyntax in this case, and hence do not qualify as

andidates (thou

h see section 2.2). The choice, then, is between the zero endin

and

sew

ending. Since the former is more

richl

specified, it blocks the

att

n the second person singular, the question is how to spell out [

PAR

ADD

]

. In this

thr

ee c

es w

ith a f

re s

ecification that does not s

ell-out too

uch. Clearly, of these three, -

is the most richly specified, and hence favoured by

he Elsewhere Principle. The third person singular is characterized in morphosyntax

by the absence of phi-features. Hence, the only possible spell-out for it is

-t

In the plural, the morphosyntax will contain the same feature specifications fo

he various persons as in the sin

ular; in addition, a [PL

]

feature is present. As it

tands, the Elsewhere Principle cannot determine which of the affixes in (3) should

be used. Consider, for example, the first person plural, specified in the

orphosyntax as [

PAR

]

. There are three candidates com

atible with this

specification

∅

-t

Whereas least-s

ecified

-t

is blocked b

the other two

andidate realizations, the Elsewhere Princi

le cannot decide between

∅

t stands, because neither contains a superset

f th

es o

f th

euseof

a single form in the plural suggests that the features [

PAR

]

and

[

ADD

]

are neutralized

n this context. This phenomenon can be accounted for using a context-sensitive rule

f feature deletion that applies before spell-out (the point at which it is determined

which of the affixes is to be used

In the case at hand, the feature

[

PAR

]

is deleted in

he context of the feature

[

(

)

[

PAR

]

→

∅

/ __ [PL

]

The syncretism between the second person and thi

d person singular is accidental. Many dialects still

tin

t -

ending for the second person singular. The

ending derives from an older second

person plural ending, and is not an extension of the third person singular.

he walk

they walk-

ORMATION IN OPTIMALITY THEORY

289

The result for the first person plural is trivial, as its morphosyntactic

pecification will now be

ust [

]

. Hence, -

sused.

ue o

f th

thir

person, where the rule in (4) applies vacuously. After application of (4), the second

erson

lural will be s

ecified as

[

]

. This s

ecification excludes the use of

which s

ells out [PAR

]

, so that we are left with -

for this case as well.

(

The

pe of rule that deletes features prior to spell-out is referred to as impoverishment in

lit

rat

n Di

tri

Morphology; for relevant discussion, see Bonet 1991

and Harley and Noyer 1999).

The assumption that the second person is more richly specified than the first,

ather than the other way around, is supported by a curious instance of

mpoverishment: when the verb inverts with the subject, the zero affix is used for the

econd person singular, rather than -

, as shown in

(

)

(

This cannot be the effect

f a phonolo

ical rule of

deletion, as a com

arison with the third

erson verb in

(

)

, which retains its

-t

ending, makes clear).

This phenomenon can be anal

zed in t

f an

xt-

iti

overishment rule that deletes the feature

[

ADD

]

and a

lies under inversion

(

see

ckema & Neeleman 2004 for details

)

The discussion so far illustrates that elsewhere relations hold not only between a

ingle default form and the rest of the affix inventory, but in fact for any two affixes

whose features stand in a subset-superset relation. Yet, there is of course one

orpheme in many affix inventories that is the least specified, and hence the

ultimate elsewhere form. In the Dutch a

reement paradi

m, this is the completel

uns

ecified

-t

That thi

affix i

a tr

ue e

sew

form, rather than it bein

specified

for some third

erson feature, can be see

in so-called im

ersonal constructions. An

exam

le is the im

ersonal

assive:

In such structures there is no subject

for the verb to agree with, yet the

-t

endin

bli

atoril

shows up on the verb

The onl

to avoid the conclusion that -

does not spell out agreement features here, is to assume that

there is an empty third person singular expletive in

mpersonal constructions. It is unlikely that such

an expletive exists, given that it cannot satisfy the verb-second requirement that holds in Dutch main

uses: *

aat hier niet

elachen worden

‘

oes here not lau

hed be’. The postulation of an empt

expletive in fact comes down to making an untestable assumption.

(

5) a. Loop jij?

walk you

‘do you walk?’

b. Loopt Jan?

walks

ohn

‘Does John walk?’

(6) Hier

aat niet

elachen worden!

here goes not laughed be

‘There will be no laughing here!’