The Phonology of Dutch

Illl 

PHONOLOGY OF THE WORLD'S LANGUAGES 

Series Editor: Jacques Durand, University of Salford 

The Phonology of Dutch

THE PHONOLOGY OF THE WORLD'S LANGUAGES 

The phonology of most languages has until now been available only in a fragmented 

way, through unpublished theses, or articles scattered in more or less accessible 

journals. Each volume in this series will offer an extensive treatment of the phonology 

of one language within a modern theoretical perspective, and will provide comprehensive 

references to recent and more classical studies of the language. The following will 

normally be included: an introduction situating the language geographically and 

typologically, an overview of the theoretical assumptions made by the author, a 

description of the segmental system and of the rules or parameters characterizing the 

language, an outline of syllable structure and domains above the syllable, a discussion 

of lexical and postlexical phonology, an account of stress and prominence, and, if space 

allows, some overview of the intonational structure of the language. 

While it is assumed that every volume will be cast in a modern non-linear framework, 

there will be scope for a diversity of approach which reflects variations between 

languages and in the methodologies and theoretical preoccupations of the individual 

authors. 

Published in the series: 

The Phonology of English 

Michael Hammond 

The Phonology of Armenian 

Bert Vaux 

The Phonology of German 

Richard Wiese 

The Phonology and Morphology of Kimatuumbi 

David Odden 

The Lexical Phonology of Slovak 

Jerzy Rubach

THE 

PHONOLOGY 

OF 

DUTCH 

Geert Booij 

OXTORD 

UNIVERSITY PRESS

OXPORD 

UNIVERSITY PRESS 

Great Clarendon Street, Oxford ox2 6op 

Oxford University Press is a department of the University of Oxford. 

It furthers the University's objective of excellence in research, scholarship, 

and education by publishing worldwide in 

Oxford New York 

Athens Auckland Bangkok Bogota Buenos Aires Calcutta 

Cape Town Chennai Dar es Salaam Delhi Florence Hong Kong Istanbul 

Karachi Kuala Lumpur Madrid Melbourne Mexico City Mumbat 

Nairobi Paris Sào Paulo Singapore Taipei Tokyo Toronto Warsaw 

with associated companies in Berlin Ibadan 

Oxford is a registered trade mark of Oxford University Press 

in the UK and in certain other countries 

Published in the United States 

by Oxford University Press Inc., New York 

© Geert Booij 1995 

The moral rights of the author have been asserted 

First published in paperback 1999 

All rights reserved. No part of this publication may be reproduced, 

stored in a retrieval system, or transmitted, in any form or by any means, 

without the prior permission in writing of Oxford University Press, 

or as expressly permitted by law, or under terms agreed with the appropriate 

reprographics rights organisation. Enquiries concerning reproduction 

outside the scope of the above should be sent to the Rights Department, 

Oxford University Press, at the address above 

You must not circulate this book in any other binding or cover 

and you must impose this same condition on any acquirer 

British Library Cataloguing in Publication Data 

Data available 

Library of Congress Cataloging in Publication Data 

The phonology of Dutch / Geert Booij. 

(The Phonology of the world's languages) 

Includes bibliographical references. 

I. Dutch language Phonology. I. Title. II. Series. 

PF13I.B67 1995 439.3'I15—dc20 94-34547 

ISBN 0 19 823869 X 

13579 108642 

Typeset in Times 

by J&L Composition Ltd, Filey, North Yorkshire 

Printed in Great Britain 

on acid-free paper by 

Biddies Ltd., Guildford and King's Lynn

PREFACE 

This book presents a survey of the phonological system of Dutch. Its primary 

aim is providing insight into that system as a whole, based on a considerable 

number of studies in the phonology of Dutch. Thus, this book can function as a 

reference work that provides access to more detailed studies of aspects of the 

phonology of Dutch that are available in the linguistic literature. 

The analyses proposed are couched in terms of present-day generative 

phonology, in accordance with the saying that 'description without theory is 

blind'. Although the book is not primarily meant as an original contribution to 

phonological theory, the analyses will appear to be relevant for a number of 

current issues in phonological theory. Moreover, a comprehensive analysis of 

the phonological system of one language can keep us from too hasty generalizations 

and theoretical conclusions based on the analysis of just a few 

phenomena. 

A substantial part of the first draft of this book was written while I was a 

visiting scholar in the Department of Linguistics of the University of 

Massachusetts in Amherst, from May to July 1992. I would like to express 

my sincere thanks to all the staff members of the department who made my 

stay there pleasant and fruitful. In particular, 1 thank my 'hostess', Lisa 

Selkirk, and the other two phonologists of the department, John Kingston 

and John McCarthy, for the inspiring and fruitful discussions I had with 

them and for their hospitality and helpfulness in practical matters. I also 

thank the Faculty of Letters of the Vrije Universiteit Amsterdam for making 

my stay in Amherst possible. 

A number of colleagues were so kind as to comment on a previous version 

of the manuscript: Renée van Bezooijen, Jacques Durand, Carlos Gussenhoven, 

Ben Hermans, Daan de Jong, René Kager, and Richard Wiese. Their questions 

and criticisms were very useful. 

This hook is dedicated to my wife, Herry, and our children, Su/.anne, 

Rebecca, and Indriaas. 

ACKNOWLEDGEMENTS 

My thanks are due to Cambridge University Press and to the authors for their 

kind permission to reproduce figures from Van den Berg, Gussenhoven, and 

Rietveld (1992).

CONTENTS 

List of Tables 

Abbreviations and Symbols 

x 

xi 

1. INTRODUCTION 1 

2. THE SOUNDS OF DUTCH: PHONETIC 

CHARACTERIZATION AND PHONOLOGICAL 

REPRESENTATION 4 

2.1. Introduction 4 

2.2. The vowel system 4 

2.2.1. Allophonic variation 6 

2.3. The consonant system 7 

2.4. Phonological features 9 

2.4.1. Laryngeal features 10 

2.4.2. Supralaryngeal features 10 

2.4.3. Underspecih'cation 12 

2.5. The representation of vowels and diphthongs 13 

2.5.1. Vowel length 13 

2.5.2. Vowel height 16 

2.5.3. Diphthongs 18 

2.5.4. The representation of the schwa 19 

2.6. Feature charts for vowels and consonants 20 

3. THE PROSODIC STRUCTURE OF WORDS 22 


3.2. The representation of syllable structure 23 

3.3. The syllable template of Dutch 24 

3.3.1. Appendices 26 

3.4. Syllabification 29 

3.4.1. The domains of syllabification 29 

3.4.2. Maximal Onset and Minimal Rhyme 30 

3.5. Co-occurrence constraints 33 

3.5.1. Rhyme constraints 33 

3.5.2. Onset constraints 35 

3.5.3. Coda constraints 40 

3.5.4. Pansyllabic constraints 42 

3.5.5. Sequential constraints 43 

3.6. The relation between morphological and prosodie structure 47

Vlll 

CONTENTS 

4. WORD PHONOLOGY 53 

4.1. Introduction: the organization of the grammar 53 

4.2. Phonological rules 57 

4.2.1. Devoicing and voice assimilation 58 

4.2.2. Nasal assimilation 64 

4.2.3. Hiatus rules 65 

4.2.4. Degemination 68 

4.3. Morpholexical rules 69 

4.3.1. Diminutive allomorphy 69 

4.3.2. -er- allomorphy 73 

4.4. Allomorphy in the non-native lexicon 75 

4.4.1. Affixal allomorphy 77 

4.4.2. Root alternations 78 

4.4.3. Alternations in velar nasal clusters 80 

4.4.4. Vowel lengthening 80 

4.4.5. Other cases of allomorphy 83 

4.5. Allomorphy in the native lexicon 86 

4.5.1. Vowel lengthening 87 

4.5.2. /da/-0 alternations and /o/-0 alternations 88 

4.5.3. /d/-deletion and /d/-weakening 90 

4.5.4. 0-/k/ alternations 92 

4.5.5. F-ricative devoicing 92 

4.6. Allophonic rules 93 

5. WORD STRESS 96 


5.2. Main stress patterns 99 

5.2.1. Secondary stress 105 

5.2.2. Cyclicity of stress assignment 109 

5.3. Word stress and native suffixation l K) 

5.3.1. Stress-neutral suffixes 111 

5.3.2. Stress-bearing suffixes 113 

5.3.3. Stress-shifting suffixes 114 

5.4. Compound stress 115 

5.5. Stress in prefixed words 119 

5.5.1. Non-native prefixes 119 

5.5.2. Nativized and native prefixes 120 

5.6. Bracketing paradoxe-s 123 

6. CONNECTED SPEECH I: WORD PHONOLOGY 125 


6.2. Schwa epenthesis 127 

6.3. Schwa deletion 128 

6.4. Vowel reduction 130 

6.5. Vowel shortening 136 

6.6. Intrusive stops 137

CONTENTS 

ix 

6.7. /i/-deletion 138 

6.8. /n/-deletion 139 

6.9. Conclusions 141 

7. CONNECTED SPEECH II: SENTENCE PHONOLOGY 143 


7.2. Phonology above the word level 145 

7.2.1. Voice Assimilation 146 

7.2.2. Nasal Assimilation 147 

7.2.3. Hiatus rules 150 

7.2.4. Degemination 151 

7.2.5. Palatalization 151 

7.2.6. /t/-deletion 152 

7.3. Sentence accent 154 

7.4. Rhythmic rules 161 

X. CONNECTED SPEECH III: CLITICIZATION 165 


8.2. Pronominal and adverbial clitics 166 

8.2.1. Prosodie integration 169 

8.2.2. Clitici/.ution and Final Devoicing 174 

8.3. ('litici/ation of other function words 176 

8.4. Contraction 178 

9. ORTHOGRAPHY 181 


9.2. The correspondence between sounds and letters 181 

9.3. Degrees ol ahstractness in spelling 185 

9.4. Spelling and higher-order linguistic structure 187 

References 189 

Index 203

LIST OF TABLES 

2.1. The Dutch vowels 4 

2.2. The consonants of Dutch 7 

2.3. Vowel height features 17 

2.4. Feature chart for Dutch vowels 20 

2.5. Feature chart for Dutch consonants 21 

3.1. Sonority Ranking Hierarchy 25 

3.2. The phonological forms of grammatical morphemes 48 

8.1. The Dutch personal pronouns 167 

8.2. Dutch singular possessive pronouns 169 

8.3. Dutch adverbial clitics 169 

9.1. The spelling of consonants 182 

9.2. The spelling of short vowels 183 

9.3. The spelling of long vowels 184 

9.4. The spelling of diphthongs IK5 

9.5. Survey of orthographical representation of P-rule effects 187

ABBREVIATIONS AND SYMBOLS 

A Adjective or argument part, 

adj. adjective pej. 

ant anterior pert', 

appr approximant pers. 

asp. aspirated pi. 

attr. attributive poss. 

C Clitic Group or Coda or PP 

consonant 

pres. 

cont continuant pron. 

cons consonant PVA 

Cor Coronal node 

CSR Compound Stress Rule R 

def. definitive rnd 

denom. de-nominal RVA 

dimin. diminutive SAAR 

dist distributed 

Dors Dorsal node sg. 

FM cxtrametrical son 

F foot or French SSC 

fem. feminine SSG 

G Germanic 

H high (tone) subj. 

HGI Homorganic Glide Insertion suft', 

inf. infinitive u 

IP Intonational Phrase U 

L low (tone) UAC 

Lab Labial node 

lat lateral V 

M modifier voc 

MP-ruIe morpholexical rule a 

MSC morpheme structure condition 

MSR Main Stress Rule 

masc. masculine c 

n. neuter q> 

N Noun or Nucleus w 

NP Noun Phrase 0 

NSR Noun Phrase Stress Rule / / 

O Onset I ] 

obj. object 

OCP Obligatory Contour Principle 

P predicate or Preposition ll'/i 

P-rule phonological rule L% 

particle 

pejorative 

perfect 

person or personal 

plural 

possessive 

Prepositional Phrase 

present 

pronoun 

Progressive Voice 

Assimilation 

Rhyme 

round 

Regressive Voice Assimilation 

Sentence Accent Assignment 

Rule 

singular 

sonorant 

Syllable Structure Condition 

Sonority Sequencing 

Generalization 

subject 

suffix 

unstressable 

Utterance 

Uniform Applicability 

Condition 

Verb or vowel 

vocoid 

variable ranging over + and - 

(showing agreement in these 

values) 

syllable 

Phonological Phrase 

Prosodie Word 

zero 

phonemic representation 

phonetic representation 

main stress 

secondary stress 

High boundary tone 

Low boundary tone

NTRODUCTION 

Dutch is the native language of most inhabitants of the Netherlands, which 

means that it has about 14 million native speakers in this country. Dutch 

belongs to the West Germanic branch of the Germanic languages. The West 

Germanic languages can be divided into two subgroups, Dutch and German 

versus Frisian and English. 

Dutch is indeed similar to German. As far as syntax is concerned, both have 

two basic word orders, SVO in main clauses, and SOV in embedded clauses; 

also, both languages exhibit the phenomenon of Verb Raising (raising of a verb 

or verbal projection from an embedded clause to the dominating clause). On 

ilk' oilier hand, Dutch is different from German in that it does not have such a 

rich inflectional morphology; there is no morphological expression of case 

except for pronouns that have spécifie object forms, and it has only two 

genders, neutral and non-neutral. 

The Dutch name for Dutch is Nederlands.* It is the official language of the 

Netherlands, but, in Friesland, Frisian is the second official language, which 

has certain rights in education and local government. Dutch is also one of the 

three official languages of Belgium, besides French and German. Dutch is 

spoken in the northern part of Belgium, in the provinces of West Vlaanderen 

and Oost Vlaanderen, Antwerpen, Limburg, and Brabant (that is bilingual) and 

it has about 6 million speakers there. The capital of Belgium, Brussels, is 

bilingual (Dutch and French). This means that there are about 20 million native 

speakers of Dutch in Europe. 

Standard Dutch is used both in the Netherlands and in Belgium in the media, 

in government, and in education. However, Belgian native speakers differ in the 

extent to which their Dutch differs from the standard Northern Dutch as spoken 

in the Netherlands. Some of them speak a variety that is almost identical to 

Northern Dutch, but most of them speak a kind of Dutch that is rather different 

from standard Northern Dutch in its syntax, morphology, phonology, and 

vocabulary. This variety is called Southern Dutch (Zuid-Nederlands)? 

1 

The word llnllanih is also used, because The Netherlands is also referred to as Holland, but 

strictly speaking the word llnlliiihl refers to two western pun inces ol' Holland, Noord-Holland and 

/.uid-Holland. which are economically and culturally dominant, and where one linds most speakers 

ol standard Dutch. 

' Southern Dutch is sometimes called Hemish (l'/

2 INTRODUCTION 

In one of the twelve provinces of the Netherlands, Friesland, about 400,000 

people (73 per cent of the population) speak Frisian, 1 but these Frisians also 

speak Dutch, i.e. they are bilingual. A substantial number of speakers of Dutch 

are bilingual in that they also speak a local dialect. In Belgium we lind 

Flemish, Brabantish, and Limburgian. There is also a great variety of dialects 

in the Netherlands, some of which are similar to High German (for example, 

Limburgian as spoken in the southern part of the province of Limburg), or to 

Low German (for example, the north-eastern dialects near the German border). 

For many speakers, their Dutch is coloured to some extent by the rural or 

urban dialect that they speak. These regional variants of Dutch are called 

regiolects, and they must be seen as intermediate language varieties, in 

between standard Dutch and specific dialects. Nowadays, younger people 

often appear to speak only such regiolects rather than specific dialects, in 

particular in the cities. 

Dutch is also spoken in former colonies of the Netherlands, Surinam and the 

Dutch Antilles, where it is the language of the government, and also used in 

education. However, there are only very few people in these countries for 

whom Dutch is their native language. As may be expected, there are specific 

Surinamese and Antillian varieties of Dutch. In Indonesia, another former 

colony of the Netherlands, Dutch only plays a role in that some law codes 

are still only available in Dutch. Remnants of spoken Dutch are found among 

Dutch immigrants in the USA (American Dutch), Canada, New Zealand, and 

Australia. 

Dutch also has a daughter language, Afrikaans, spoken in South Africa, both 

as a native and as a second language. It is also spoken in Botswana, Namibia, 

and Zimbabwe, and has about 5 million native speakers. Afrikaans derives 

from seventeenth-century urban dialects of Dutch, as spoken by the lust 

settlers in Cape Town, but was also influenced by Malay-Portuguese, and 

English. 4 

The phonology of Dutch presented in this book is the phonology of standard 

Northern Dutch. 

The first detailed twentieth-century study of the Dutch sound system is a 

traditional handbook on the phonetics of Dutch, Zwaardemaker and Hijkman 

(1928). In the 1930s, structuralists like N. van Wijk and A. W. de (iroot 

contributed to the study of the phonology of Dutch (for instance, Van Wijk 

1939). The only monograph-si/e structuralist phonology of Dutch is Cohen 1-1 

al. (1959). An often reprinted textbook is Van den Berg (195X). 

The study of the phonology of Dutch was boosted enormously through the 

rise of generative phonology, as will be clear from the references given in this 

book (cf. Booij 1990/j for a historical survey). Two introductions to the 

I I 'I'iorsma ( 19X5) lor a survey of the phonology and syntax ot l-'risian. 

1 

I .»ex unions ot i In- phonology of Afrikaans can he found in Wissin^! ( 14X2) ami ('omhrink anil 

IX' Sladler (19X7).

INTRODUCTION 3 

generative phonology of Dutch have appeared. Trommelen and Zonneveld 

(1979) and Booij (1981«). Zonneveld et al. (1980) is a collection of papers 

on Dutch phonology with the explicit aim of showing that Dutch is an 

interesting object language for theoretical controversies. Nevertheless, a 

generative phonology-oriented monograph on the phonological system of 

Dutch written in English was still lacking. s This book is meant to fill this gap. 

In the tirsl hall of this century Mime linguists were able to read Dutch. For instance, Edward 

Sapir reviewed publications by C. C. Uhlenbeck on Amerindian languages written in Dutch in the 

first volume of liitrrnaliunal Journal of American Liiif!iii\lic.'i (Sapir 1917). and Van Wijk, in his 

preface to Van Wijk ( 19.W), just presupposed that the fact that this book was written in Dutch did 

not impede its being read by his colleagues. Bloomlield wrote a short grammar of Dutch for the US 

army.

THE SOUNDS OF DUTCH: PHONETIC 

CHARACTERIZATION AND PHONOLOGICAL 

REPRESENTATION 

2.1. INTRODUCTION 

In this chapter I will first present a survey of the vowels and consonants of 

Dutch, and a phonetic characteri/.ation of each of them.' Subsequently, I will 

discuss some crucial insights of present-day generative phonology as to 

phonological features and multi-tiered phonological structure, and use these 

insights in proposing phonological representations for the sounds of Dutch, 

which will form the basis for the analysis of the phonological processes of 

Dutch in the following chapters. 

2.2. THE VOWEL SYSTEM 

Dutch has sixteen vowels, which can be classified as in Table 2.1. 

TAHIJ 2.1. '/'//c Dutch vowels 

Short vowels i, c, 3, Y, u 

Long vowels 

i, y, u, e, 0, o, a 

Schwa 3 

Diphthongs 

ci, œy, DU 

I-ig. 2.1 represents the positions of the non-diphthongal (= steady-state) 

vowels in the stylized vowel space. The vowels are usually distinguished 

from each other in terms of their place of articulation (as shown in Fig. 

2.1). roundedness, and length. Length, however, is not a purely phonetic 

property: the main reason for distinguishing seven long vowels is a phonological 

one: the long vowels behave as two 'units', whereas the short vowels 

behave as one unit (cf. Moulton 1962 and Section 2.5.1). In particular, whereas 

' A systematic comparison between the phonetics of Dutch and the phonetics of English is given 

in (iussenhoven and Broeders (1976). Gussenhoven (IW2c) is an illustration ot the International 

Phonetic Alphabet lor Dutch.

2.2. ÏHI-: vowi-:i. svs 11 M 

front central hack 

close 

half close 

half' open 

open \ a, u 

FK;. 2.1. The Dutch vowel space 

the non-close long vowels have an average duration of about 200 msec., the 

short vowels and the high long vowels /i,y,u/ have an average duration of about 

100 msec. (Nooteboom 1972: 45-7). On the other hand, before /r/ the high 

vowels do have the same (extra) length as the other long vowels. Thus, we 

seem to have a 'clash' here between the phonetic and the phonological 

classification of speech sounds. Below I will make a proposal as to how to 

account for this. 

The long vowels /e/ and /o/ are not only longer than their short phonological 

counterparts /e/ and /o/, but they also differ with respect to height: the long 

vowels are higher. Phonetically, the short counterpart of/e/ is III. The /Y/ is the 

short counterpart of 101. Finally, the /a/ is more central than the /u/. 

The schwa occupies the central place in the vowel space. It is similar to 

the /Y/ but pronounced more centrally and without rounding. The defining 

characteristics of the schwa are again primarily phonological ones: for 

instance, it can occur in word-final position, unlike the (other) short vowels, 

and it never bears stress (except in cases where function words such as Je /da/ 

'the' are emphatically stressed). 2 

The three diphthongs are characteri/.ed by the transition of a lower position 

in the vowel space to a higher position. That is, the two constituents only differ 

with respect to height. Their position in the vowel space is represented in Fig. 

2.2. 1 

In standard Dutch the long mid vowels /e, 0, o/ also receive a slightly 

diphthongal reali/.ation: at the end of their articulation there is a transition 

to the positions of /i/, /y/, /u/ respectively. 4 

In addition to the vowels discussed so far, Dutch also has marginal vowels, 

occurring in loan-words. The first set comprises phonetically long counterparts 

~ Cf. Stulterheiin ( 147S) lor .1 survey of the discussion between Dutch phonologiits concerning 

Ihe status of the schwa. 

1 

( 'I. 'l Marl ( l%y) and Koopmans-van Beinum ( 1964) for detailed phonetic information about 

these three diphthongs. 

4 In non-standard variants we do lind Ihe purely monophlhongal realisations of these vowels.

TKK SOUNDS OF DUTCH 

Fronl Central Back 

Close \ i y 

Hall close \ î Î 

Halt open 

Open 

f 

œ 

Fie;. 2.2. The Dutch diphthongs 

of native phonetically (but not always phonologically!) short vowels, as 

illustrated in (1). They only occur in stressed syllables: 

(1) [i:] analyse 'analysis' 

[y:] centrifwge 'spinner' 

[u:| rouge 'blusher' 

[c:| enquête 'inquiry' 

[ce:] oeuvre 'id.' 

[D:| zone 'id.' 

[u:] basketball 'id.' 

The second set of marginal vowels consists of nasal vowels that occur in some 

words borrowed from French: 

(2) |c) enfin 'anyway' 

|ti| parfum 'perfume' 

[5| chanson 'id.' 

[u] restaura/n 'id.' 

2.2.1. Allo/ihonic variation 

All long steady-state vowels are lengthened before /r/, if the vowel and the 

following consonant belong to the same (prosodie) word. In addition, the /r/ 

also has a centrali/.ing effect. In other words, long vowels tend to end in a 

schwa-like sound in this position. The lengthening effect is particularly strong 

for the high vowels: before the /r/ they have the same length as the non-high 

long vowels. 5 This centralizing effect opposes the diphthongal realization 

(raising in the last stage of articulation) of the long mid vowels, and hence 

the difference in 'colour' between the normal realization of the long mid 

vowels and that before /r/ is rather strong. This centralizing effect of the /r/ 

also explains why diphthongs cannot be followed by a tautosyllabic /r/: the 

realization of diphthongs requires raising in their second phase as opposed to 

s The only exceptions are the past-tense forms of a few verbs, in which the /i/ remains short, 

e.g., wierp [wirpl 'threw', stierf|stirf| 'died', zwierf |/.wirt'| 'wandered', /«•

2.3- TUI-: CONSONANT SYSTEM 7 

centrali/ation. After vowels followed by /j/ or /u/ we get the transitional 

vowels |i| and |u| respectively, which is sometimes reflected in Dutch spelling, 

as in Booij /boj/ 'id.' and nieuw /niu/ 'new'. 

The diphthongs /ci/ and /ivy/ cannot be followed by a /u/ in the same 

syllable, nor /DU/ by a tautosyllabic /j/, probably because the final stage of 

the diphthong requires an articulatory gesture with respect to backness which 

is the opposite of that of the next glide. 

Before the nasal consonants, the vowel /o/ is closer and more centralized 

than in other positions, i.e. it is pronounced as [o] in that position. We also find 

this realization of /.">/ in some other words, for example, op [up| 'on' versus 

mop [mop] 'joke'. There is, however, geographical and individual variation 

with respect to this phonetic variation. 6 

2.3. THE CONSONANT SYSTEM 

The consonants of Dutch are given in Table 2.2. 

TABU. 2.2. '/'/«• i-ntmniiiiiix nf Dutch 

Bilabial 

Lahio 

denial 

Alveolar 

Palatal 

Velar 

Glottal 

Plosives 

Fricatives 

Nasals 

Liquids 

Glides 

p, h 

ni 

f. v 

i' 

1, d 

s, / 

n 

l.r 

j 

k, (9) 

x, v 

D 

h 

The [y| is put in parentheses because it only occurs in non-native words such 

as youl 'id.' (football term), and as the contextual allophone of/k/ before a 

voiced plosive, as in zukiloek |/ugduk] 'handkerchief'. The alveolar consonants 

/s, /, t, n/ are palatali/ed before /j/, and then reali/ed as the postalveolur 

or prepalatal sounds [ƒ, 3, c, ji| respectively. The postalveolar fricatives also 

occur in loan-words, as in chique [Jik| ami jury l3y:ri|, but phonologically 

they can also be considered as combinations of/s, /./ and /j/, with the fricatives 

predictably being reali/ed as the postalveolar allophones. 7 

For many speakers of standard Dutch, in particular in the western part of 

Holland, the voiced-voicelessness distinction between/f/and/v/and/x/and/v/ 

is neutrali/eil at the beginning of a word, and sometimes also intervocalic-ally 

6 Cf. Schoulcn ( 198 I ) lor u discussion of Ihis phenomenon. 

l'ui 11 us & Mecs ( I'M I ) also consider /d.V and AJ7 as in /

8 THE SOUNDS OF Dl I ( M 

(word-finally, obstruents are always voiceless). In these cases, the voiceless 

variants are used.* Note, however, that even for these speakers there must be 

an underlying distinction between voiced and voiceless fricatives, at least 

morpheme-finally, because this distinction still plays a role in the selection 

of the proper past tense suffix of verbs: /do/ after voiced segments, /ID/ after 

voiceless obstruents, for example, legde (Icvda] 'laid' versus lachte [luxto] 

'laughed', and draafde |dravdo] 'raced' versus mafic |mufto| 'slept'.'' 

The /h/ is a voiced consonant, articulated with glottal friction, with the same 

shape of the vocal tract as that of the following vowel."' 

The articulation of the velar fricatives varies from postvelar to postpalatal. 

The latter variant is characteristic for speakers from the southern parts of the 

Netherlands and the Dutch speaking area of Belgium. The /x/ is often also 

palatali/.ed before /j/ as in wiegje |wiçjo| 'cradle'. Speakers of the Western 

variety of Dutch also reali/e it us a uvular fricative [x|. 

The /!/ is usually reali/ed as a rather clear |l| before vowels in word-initial 

position, with an alveolar contact, and as a dark (velari/.ed) [l| in syllable 

rhymes and intervocalically. Some speakers even reali/,e the /I/ as a vocoid in 

that position.' ' 

The /r/ may be reali/ed as an alveolar roll |r) (in particular in utteranceinitial 

position), as an alveolar flap |r|, as a uvular roll [R|, as a uvular fricative 

|XK or as a uvular approximant |K|. In postvocalic position /r/ may also be 

realized as a palatal approximant similar to [j|. This is a matter of individual 

and regional variation. 12 

The /u/ is a labiodental approximant. In coda position, it is reali/.ed as a 

bilabial vocoid, without contact between the two articulators, as in nieuw 

|niu) 'new', leeuw [leu| 'lion', and ruw |ryu] 'rough'. As we will see in 

Chapter 3, it forms indeed a natural class with the vocoid /']/ in that position. 

After back vowels before a next vowel we get insertion of a homorganic glide 

|Ul, as in gnoeen [vnuuon| 'gnus'. In other positions it is a labiodental 

approximant, for example, in water /uatar/ 'id.' and wreed /ured/ 'cruel'. In 

the south of the Netherlands and in Belgium, the bilabial apprimant [ß| is used 

instead of the labiodental approximant (cf. Zwaardemaker and Eijkman 192X: 

154-5, Gussenhoven and Broeders 1976: 54-5). 

The /j/ is a palatal vocoid, with the same phonetic properties as the [i] except 

I hi-- varialion is discussed in dusscnhoven and Urcmmer (19X3). It may he llial Ihc 

distinctive role of voice is taken over by the opposition lortis lenis (l)elnock I97X). 'I Indistinctive 

role of voice lor fricatives is also discussed in Slis and Van Hcugtcn ( 19X4). Historical 

aspects aie discussed in Dekeyser ( 197X). Information on the phonetic icali/alion of voiced slops is 

given in Slis and Cohen (1969). 

9 This argument was given in Kagcr ( 14X9: 221 ). 

" The Dutch /h/ differs honi the English /h/ in that il is voiced whereas the linglish /h/ is 

voiceless, cf. Anderson (19X6). Phonetically, however, it may have a partially voiceless reali/alion 

(Rietveld and l.oman 19X5). 

1 1 

('I. Van Reenen (19X6) toi some more information on the vocoid reali/alion ol Ihc /I/. 

12 

A survey of the dillerenl pronunciations of the /r/ is given in Gussenhoven and Hn 

(1976: 127) and in Collins and Mees (19X1: I6X 71).

2.4. PHONOLOGICAL FEATURES 9 

that it occurs in consonantal positions in the syllable; it has a shorter duration 

than the Iil. 

2.4. PHONOLOGICAL FEATURES 

A fundamental insight of present-day generative phonology is that the sounds of 

a language are not atomic, indivisible units, but should be decomposed into sets 

of phonological features (Chomsky and Halle 1968, Halle 1991). In Chomsky 

and Halle (1968) each segment is represented as a simple set of co-occurrent 

binary phonological features such as [ -voice], [+nusal|. However, in Clements 

( 19X51 and subsequent work, it has been argued that the set of phonological 

features should be internally structured. For instance, place features must be 

assumed to form a subunit of features since cross-linguistically they behave as 

a unit in assimilation processes such as the assimilation of nasal consonants to 

the following consonant with respect to place of articulation. 

In this book I assume the feature geometry for Dutch given in Fig. 2.3, based 

on proposals by Clements (1985), Sagey (1986), McCarthy (1988), Halle and 

Ladefoged (1988), and Halle (1991). 

Root (cons, son] 

Larynpeal |cont] |nasal) [lateral] |Place| 

|asp| (voice) 

Labial 

[round] |ant| (hack) |high| |niicl| 

I-ui. 2 A The feature geometry of Dutch 

The root node represents the phonological unity of the features that form 

together a phonological segment. The root node is annotated with the binary 

features [consonant) and [sonorant), the so-called major class features that 

divide the segments of a language into three classes: 

(3) [+cons, —son) obstruents 

|+cons, +son| sonorant consonants 

|-cons, +son| vowels

10 THE SOUNDS OF' DI K II 

Moreover, the feature [consonant] divides the segments into consonants versus 

vowels, and the feature [sonorant] divides them into obstruents and nonobstruents. 

The feature [consonant] specifies the difference in the degree of 

stricture of the vocal tract between vowels and consonants. Segments with the 

feature [+consonant| are consonants, segments with the feature [-consonant] 

are vowels. The latter have a free escape of the airstream. The feature 

[sonorant] also pertains lo the degree of stricture in the vocal tract, and 

distinguishes stops and fricatives from the other consonants and the 

vowels. 11 In addition to these major class features we also need the feature 

[approximant), as will be made clear in Section 3.3: obstruents and nasal 

consonants are [-approximant), the other segments are [+approximant|. 

2.4.1. Laryngealfeatures 

Laryngeal is a so-called class node, which unites features with respect to the 

ways in which the larynx is used in articulation. The basic laryngeal distinction 

for Dutch is that between voiced and voiceless segments. The feature [voice) is 

only distinctive with respect to obstruents: vowels and sonorant consonants arcalways 

voiced in Dutch. In this book I will take the conservative position that 

the terminal nodes in feature geometry are binary features, and therefore 

voiced segments are described as |+voice|, and voiceless segments are 

[ — voice). 14 

The feature [aspirated] can be used to characteri/e the /h/, a voiced glottal 

fricative. Aspirated sounds are produced with the vocal cords drawn apart, 

which gives a non-periodic (noise) component in the acoustic signal. The /h/ 

does not have supralaryngeal properties of its own, the shape of the vocal tract 

being determined by that of the following vowel. Therefore, the phonological 

representation of the /h/ consists of a root node that only dominates Laryngeal 

specified as [+aspirated|. All other segments are |-aspirated). 15 

2.4.2. Supralaryngeal features 

Supralaryngeal features are traditionally divided into manner features and 

place features. The manner features comprise (continuant], [nasal], and 

[lateral]. The feature [continuant] applies as follows: nasals, oral stops, and 

laterals are [—continuant], the other consonants and vowels are [-(-continuant]. 

'Sonoranl soumis aie produced with a vocal Iracl configuration sufficiently open that Ihe air 

pressure inside and outside Ihe mouth is approximately equal. Obstruent sounds are produced with 

a vocal tract constriction sufficient to increase the air pressure inside Ihe mouth significantly ovei 

that of the ambient air.' (Halle and Clements I U X.V ft.) 

' 4 Alternatively, one may assume that some features are privative or monovalent. 1-or instance. 

Mester and llo ( I9X U ) and Lombard! ( I'Ml ) proposed Ihal the feature |voice| is a privative feature. 

This makes Ihe prediction that phonological rules can only refer lo Ihe properly of voicedness, nol 

to voicelessness. Whether ilns prediction can he upheld, is a mailer for further rescnu h 

I Ins suggests that [aspirated! may he interpreted as a privalive feature.

2.4- PHONOLOGICAL FEATURES II 

In the production of continuants the vocal tract has such a form that the airflow 

through the midsagittal region of the oral tract is not interrupted. In the 

production of nasals, the nasal cavity is opened during the production. 16 

Although 'supralaryngeal' is a convenient term, this does not mean that it is 

a class node: sufficient evidence for the supralaryngeal features behaving as a 

unit is lacking (McCarthy 19X8). 

On the other hand, Place clearly is a class node because the cluster of place 

features may behave as a unit in phonological processes, tor instance in place 

assimilation for nasal consonants. 

In Dutch, three articulators play a role with respect to the place of articulation, 

the lips (labia), the tongue blade (corona), and the tongue body (dorsum). 

The relevant Place features are privative by nature. 

The lips play an essential role in the articulation of the labiodental consonants 

/f, v, u/. and the bilabial consonants /p, b, m/. The phonetic difference 

between labiodentals and bilabials does not play a phonological role, and this 

differente can therefore be spelled out by a phonetic detail rule. 

The class node Coronal refers to constrictions made with the corona, the 

tongue blade, as the active articulator. The alveolar consonants are furthermore 

|+anterior| whereas the postalveolar consonants are [— anterior]. Non-anterior 

sounds are produced with a constriction behind the alveolar ridge. There is no 

distinctive class of [-anterior] coronal consonants in Dutch, they are allophones 

of [+anterior| consonants before /j/, and the specifications for this 

feature are therefore predictable. Another binary feature that is dependent on 

the class node Coronal is [distributed]. The postalveolar coronal consonants of 

Dutch are predictably [+distributed| which means that they 'are produced with 

a constriction that extends for a considerable distance along the midsagittal 

axis of the oral tract; nondistributed sounds are produced with a constriction 

that extends for only a short distance in this direction' (Halle and Clements 

1983: 6). So [—distributed] correlates with the phonetic feature 'apical', and 

[+distributed| with the notion 'laminai'. Such specifications do not belong to 

the phonological representations of these sounds, but are to be predicted by 

redundancy rules. 

Palatal and velar consonants are produced by means of a constriction at the 

dorsum. The palatal glide /j/ is [ -back], 17 the velar consonants and the uvular 

"' As far as Dutch is concerned, it is possible ID consider the feature |nasal| as privative. This 

concclly predicts thai there are no rules of Dutch that crucially refer to the class of non-nasal 

segments. The same holds for the feature |lateral| that uniquely characteii/es the /!/, in ihe 

production of which the air stream does not pass through Ihe centre of the mouth hut at both sides 

of thé longue. However. Slenade ( 1987) argues in favour of a binary interpretation of (lateral). 

17 

In qualiKing the III as a dorsal glide we follow Chomsky and Halle (1968). Halle (1991) 

assumes that the glide |j| can be arliculaled by either Ihe ('oional or ihe Dorsal articulator. Both 

Keating (1991) and Lahm and l-vers ( 1991 ) argue that Ihe |j| should he represented as a coronal 

glide, which would also explain the allophonic effect of the /j/ on preceding coronal obstruents 

more straightforwardly. Since Ihe /j/ is identical to ihe vowel /i/ as to its phonetic properties, this 

would have Ihe consequence Ilial from vowels are also Coronal, as indeed proposed by l.ahiri and 

hivers (1991). I leave Ibis issue open heie.

12 im SOUNDS OF Dl U II 

[R] are [+back|. Both palatals and velars are |+high], the uvular [R| can be 

characteri/ed as [+back, —high). In the lexical forms, there is no need to 

specify Dorsal consonants for [back] since this is predictable: they are all 

[+back|. Also, we do not need a phonological feature for uvular |R| because it 

is an allophone of the alveolar [r|. 

The dorsum is the main articulator for vowels. The feature [back] distinguishes 

between front and back vowels, and the features [high] and [mid| 

account for vowel height. The issue of how to account for vowel height will be 

discussed in more detail in Section 2.5.2. 

Vowels also have to be qualified as either [+round| or [ -round). Since in the 

articulation of vowels the lips are also involved, these vowels will also be 

specified for the class node Labial, with the feature [round] dependent on this 

class node (Sagey 1986).'" 

2.4.3. Underspei•//) [-i-son] 

[+son] -» |-t-voice, +cont] 

[+nas| —> |+cons, +son, -appr, -cont] 

[-son] —> [+cons, -appr] 

|+asp| —» [+cons] 

[-Hat] —» [+cons, -(-son, + appr] 

Rules that predict major class features are crucial for the proper application of 

syllabification rules. Therefore, it is clear that such rules have to apply before 

the phonological derivation starts. I will assume that rules apply as soon as 

possible, unless stated otherwise. 

Another case of redundancy is that of the feature |+voice], which is 

predictable for sonorants. Yet, it must be available before the end of the 

phonological derivation, for instance because there is a postlexical rule in 

Dutch that Voices word-final fricatives before vocoids (cf. section 7.2.1). 

There is also a lexical rule that spreads the feature |+voice| of sonorants to 

the initial consonant of the past-tense suffix (cf. Section 4.2). In other words, 

'* For critical comments on Ihis po.silion, see Odden (1991).

2.5- THE REPRESENTATION OF VOWELS AND DIPHTHONC1S 13 

underspecification cannot always be interpreted as 'the perseverance of underspecified 

segments during the course of a derivation' (Archangeli 1984: 189), 

and as the filling in of the relevant specifications at the end of the phonological 

derivation.''' In this book, 1 will assume that predictable features are specified 

before the application of the phonological rules of Dutch, unless explicitly 

stated otherwise. 

Underspecification is also relevant in specifying the /v/. It suffices to specify 

it as a labiodental approximant at the underlying level. It is predictable that it 

is a vocoid after vowels, as in nieuw /niu/ 'new', and the feature [+vocoid] (cf. 

Section 3.3) will be filled in immediately, because certain syllable-structure 

constraints refer to [vocoid]. Moreover, as observed by Gussenhoven and 

Broeders (1976: 55), the /v/ remains vocoid after the addition of an inflectional 

schwa, as in nieuwe /niua/ 'new' as follows from filling in the value 

|+vocoid] as soon as possible, i.e., before the addition of the inflectional 

schwa. In all other positions, the /u/ will be specified as [ — vocoid]. 

Certain consonants bear specifications that make it possible to uniquely 

identify them by means of one or two features. For instance, [-(-lateral] 

uniquely identities the /I/, [+aspirated| the /h/, and each nasal consonant is 

uniquely identified by |+nasal| plus its Place feature. 

In Section 2.6 I will give a survey of the phonological specifications for 

Dutch consonants at the lexical level. 

2.5. THE REPRESENTATION OF VOWELS AND DIPHTHONGS 

In this section I will discuss the phonological representation of vowels and 

diphthongs on the basis of the feature system introduced in Section 2.4. 

2.5.1. Vowel length 

Although Dutch has a systematic opposition between short and long vowels 

(except for high vowels), the feature system introduced above does not contain 

a feature for length. The reason for this is that there are good arguments for 

representing length in structural terms rather than by means of a binary feature. 

It is generally accepted that Dutch short vowels count as one 'unit', and long 

vowels as two (Moulton 1962). In a syllable, a short vowel can be followed by 

at most two consonants, but after a long vowel only one consonant can occur. 

For instance, damp [dump) is a possible syllable, whereas daamp [damp] is 

not. To account for this difference, it has been proposed that long vowels are 

represented as sequences of two identical [—consonant] segments (Zonneveld 

and Trommelen 1980). 

|g Cl. Mester and 110(1989), Mohanan ( 1991). McCarthy and Taub (1992), and Hall (1993) for 

a critical evaluation of ihcoiies of underspecification.

14 THE SOUNDS OF DUTCH 

In non-linear phonology as developed in McCarthy (1981), Clements and 

Keyser (1983), and Levin (1985) it has been argued that phonological representations 

consist of at least two tiers, a skeletal tier and a melodic tier. The 

skeletal tier represents the abstract units to which the units at the melodic tier 

are linked, not necessarily in a one-to-one fashion. The phonological representations 

of the words kap |kup| 'cap' and kaap |kap| 'cape' will then be as 

follows (where the letters at the melodic tier stand for the feature bundle of ;i 

segment): 

(5) skeletal tier XXX X X X X 

I I I I V I 

melodic tier k a p k a p 

In other words, vowel length is represented as the linking of two elements on 

the skeletal or X-tier to one element on the melodic tier. A representation of 

long vowels as in (6): 

(6) X X 

I I 

a a 

is forbidden by the Obligatory Contour Principle (OCP) (McCarthy 1986), 

which says that identical adjacent elements on the melodic tier are prohibited. 

In other words, the OCP requires us to represent long vowels as doubly linked 

melodic elements. 

A desirable consequence of excluding representations like (6) is that at the 

melodic level long vowels are still represented as a unit. This is confirmed by 

the behaviour of long vowels with respect to certain phonological processes. 

Dutch vowels in unstressed syllables can be reduced to schwa. For instance, 

the word banaan /banan/ 'banana', with stress on the second syllable, can be 

pronounced as (bonanj. This reduction process always affects the whole long 

vowel, not only part of it. So we do have reduction of long |a| to schwa, but not 

to something like [uo|. The reduction process can now be straightforwardly 

characteri/.ed as delinking of the Place specification of the vowel as represented 

in (7), and subsequent filling in of the Place features of the schwa by a 

default rule: 

(7) Vowel Reduction [— cons] 

Place 

(in unstressed syllables) 

The hori/.ontal lines through the association line indicate delinking. That is, the 

Place features are removed from the vowel, which is now unspecified for 

Place. 

If we assumed long vowels to consist of two identical vowels, we would 

have to add a special proviso to the rule of Vowel Reduction that once one of

2.5. THE REPRESENTATION OF VOWELS AND DIPHTHONGS 15 

the vocalic segments has been reduced, the other one has to be reduced as well. 

Moreover, note that the reduction always results in one schwa, not in two. 

The representation of long vowels as one melodic unit linked to two X- 

positions thus nicely accounts for the dual nature of long vowels. It also 

enables us to give an adequate characterization of diphthongi/ation processes. 

It has long been noted that diphthongs normally arise out of long 

vowels (Hayes 1990). In the framework used here this is readily interprétable: 

in diphthongization the two Xs, which first shared all features, get 

different features for a certain dimension, often height, and therefore, the 

long vowel splits up into two melodic units, each linked to an X-position 

(Sluyters 1992). 

Another logical possibility is representing the length distinction by means of 

a binary feature [length). However, such an approach does not explain why 

long vowels behave like two units in co-occurrence restrictions, and why it is 

long vowels that diphthongi/.e. 

The lexical distinction between short and long vowels will therefore be 

expressed in the lexicon as in (8): 

(8) [u] X [u| X X 

I 

I 

a a 

The vowels /i, y, u/ have two X-positions, although they are phonetically 

short, except before /r/. Phonologically they behave like long vowels. First, 

like (other) long vowels, they allow for only one consonant to follow them in 

the same syllable: syllables such as the following are ill-formed: 

(9) */kimp/, */kymp/, */kump/, */kilm/. */kyIm/, */kulm/ 

Second, unlike short vowels, they can occur in word-final position. Third, 

when they occur in the word-final syllable, they select the same allomorph 

of the diminutive suffix as words with a long vowel in their word-final syllable 

(see Chapter 4):-° 

(10) traan 'tear'-traantje, schoen 'shoe'-sch[u]ntje versus pan 'pan'- 

p|u|nnetje 

raam 'window'-raampje, riem 'belt'-r[i]mpje versus bom 'bomb'- 

b|o]mmetje 

meer 'lake'—meertje, buur 'neighbour'-b[y:|rtje versus ster 'star'- 

it[e]rretje 

A question that remains to be answered in this connection is how we account 

for the difference in length between the 'normal' and the marginal high 

: " However, there arc a lew exceptional cases in which nouns with /i/ or /u/ in the word-final 

syllable select the allomorph that mutually occurs with short vowels, such as hlaciiiflji' 'hunch of 

Mowers' (cf. Van ll.icrmgcn 1 1 >5X, Heeroma 1959). Trommelen (1987) mentions some other 

marginal phenomena in which high vowels, in pailicular in non-native words, seem to behave 

as short vowels. Historically, the long and the short /i/ have merged in standard Dutch. In some 

dialects of Dutch, there is still a length difference for /i/ and /u/ with a distinctive function.


vowels. For instance, in the loan-word team 'team' |ti:m] the vowel is 

phonetically long, in contrast to the vowel of tien 'ten' [tin]. Following a 

proposal by Smith et al. (1989) and Hermans (1992), I will represent the 

difference between [i| and [i:] as a difference in linking: 

(11) [i] XX [i:] XX 

I \/ 

i 

i 

Both vowels will function as branching nuclei, but the second X of/i/ does not 

receive a phonetic interpretation, because it is not linked to the melodic tier. 

Hence, the /i/ surfaces as a short vowel, except before /r/. In non-high vowels, 

which are phonetically long, the segment will be linked to the second X by a 

rule (cf. Section 2.5.2). 

Interestingly, there is a length alternation in pairs of related words such as 

(12) analyse [ana'li:/o] 'analysis' 

analyseer |anali'ze:r] 'to analyse' 

centrifuge [sentri'fyizja] 'spinner' 

centrifugeer [sentrify'v.e:r] 'to spin-dry' 

A related alternation is to be observed for non-high marginal vowels, as in 

(13) gene ['zjeinaj 'shame' geneer [2ja'ne:r) 'to feel ashamed' 

zone ['zD:na] 'id.' zonéer [zo'ne:r] 'to divide into zones' 

In the latter case, the marginal vowels turn into their native counterparts in 

unstressed position. It is now possible to account for these alternations by 

means of a rule (14) that delinks the second X from the melodic tier in 

unstressed position. 

(14) X X 

\? 

(-cons) in unstressed syllables 

In the next section (2.5.2) it will be shown how delinking implies raising rather 

than shortening for non-high vowels. 

2.5.2. Vowel height 

Chomsky and Halle (1968) used the binary features (highJ and [low] for vowel 

height. Thus they can distinguish between three degrees of vowel height, since 

the combination (+high, +low| is logically impossible. Wang (1968) pointed 

out, however, that there are languages for which four degrees of vowel height 

have to be distinguished. 21 Dutch clearly requires four degrees of vowel height 

21 The same poinl is made by Kiparsky (1986), Lindau (1978), Clements (1989), and Hayes 

(1990).

2.5. THF. REPRESENTATION OF VOWELS AND DIPHTHONGS 

TABLE 2.3. Vowel height features 

+high, —mid 

+ high, +mid 

-high, +mid 

-high, -mid 

-back 

i- i:, y, y: 

i, e, Y, 0 

e, e:, œ, ce: 

+ back 

u, u: 

o 

D, o: 

u, a 

at the phonological level if we want to be able to give a proper characterization 

of the marginal vowels as well. This is possible by using the feature [mid] 

instead of the feature [low] (Table 2.3). 

The /œ/ only occurs as the first half of the diphthong /œy/. An allophonic 

rule will specify that /o/ is realized as [+high] before nasals. 

Certain redundancy rules hold for the system. For instance, low vowels are 

always [+back], and unrounded back vowels are always low. This can be 

expressed by the following rules: 

(15) a. [-cons, -high, -mid] -> [+back] 

b. [-cons, -round, +back| —» [-high, -mid] 

Clearly, the rule system should not contain both rules (15a) and (I5b) because 

they presuppose different lexical specifications: rule (15a) presumes that the 

height features are lexically specified, and [+back] predictable, whereas rule 

(15/>) presupposes the inverse. This is the case, of course, because feature 

specifications are mutually dependent. Below I will show that the choice 

between (15a) and (\5b) can be made on empirical grounds, and that (\5h) 

is to be preferred. By contraposition ('p -> q' is equivalent to 'not-q —» not-p') 

we can derive a second rule from rule 

(15) b' [-cons, +back] -» [+round] / 

I +high 

I +mid 

It is also predictable that |+high, -mid] vowels are always long, that is, have 

two Xs on the skeletal tier. 

Furthermore, except for loan-words, it is predictable that long mid vowels 

are [+high]: 

(16) [-cons, +mid] -> [+high] / X X 

This rule has a dual function: it predicts redundant features, but can also be 

applied in a feature-changing fashion in derived environments. Dutch has an 

unproductive rule of vowel lengthening in open syllables (cf. Chapter 4), as 

illustrated in (17):

l8 

THE SOUNDS OF DUTCH 

(17) g[D]d-g[o]den 'god'-'gods' 

sch[i)p-sch[e]pen 'ship'-'ships' 

w|c]g-w|e]gen 'road'-'roads' 

Note that /i/ and /e/ both change into /e/. This follows from rule (16) which 

will turn the lengthened low /e/ into the mid vowel /e/. This rule also accounts 

for the change of the marginal half-open vowels to half-close vowels in 

unstressed position. The delinking rule (14) creates a derived environment, 

in which rule (16) can apply, changing the melodic elements |e:| and |D:) into 

[e] and [o]. In order to account for the phonetic interpretation of non-high 

vowels with two Xs as long, I assume, with Hermans (1992), the linking rule 

(18) for non-high vowels that links the melodic element to the second, empty 

X. 22 

(18) X X 

/ 

I / 

i +mid i 

I-high) 

The regularity expressed in (\5b) also plays a role in phonological alternations, 

and therefore this rule must be preferred above rule (15a). Dutch has a 

rule of Learned Vowel Backing (cf. Chapter 4) that backs vowels in non-native 

morphemes followed by a non-native suffix: 

(19) fundament[e]l 'fundamental' fundament[a]lisme 'fundamentalism' 

milit[e:]r 'military' 

milit[a]risme 'militarism' 

direkt[0:]r 'director' 

direkt[o]raat 'directorate' 

Note now that |o] is indeed the back counterpart of the |0|, but that |e] and |c:] 

not only differ from [a] with respect to the feature [back], but also with respect 

to vowel height since the backed vowel is low. This is exactly predicted by rule 

(\5h), and therefore we must consider this rule as the correct generalization 

with respect to the relation between backness and roundness for low vowels. 

A chart of the underlying feature specifications for vowels is given in 

Section 2.6 (Table 2.4). 

2.5.3. Diphthongs 

The three Dutch diphthongs /ci/, /œy/ and /DU/ can be defined as sequences of 

non-identical vowels that form the (complex) nucleus of a syllable. The three 

Dutch diphthongs indeed behave distributionally as long vowels in terms of 

co-occurrence restrictions. They must be distinguished from vowel + glide 

combinations such as /a'}/ and /oj/ that occur in a few Dutch words such as mais 

/mqjs/ 'corn' and hoi /hoj/ 'hi'. Diphthongs can be followed by the same 

2 This linking rule also makes it possible to represent all native long vowels as having a second 

empty X-slot to which the vocalic segment on the melodic tier is not linked underlyingly (cf. 

Hermans 1992 for arguments in favour of this assumption).

2.5- THE REPRESENTATION OF VOWELS AND DIPHTHONGS 1C 

consonants as the long vowels (i.e., all consonants except /h/ and /rj/), whereas 

glides cannot be followed by other consonants (except alveolar obstruents). 

For instance, /tip/ is fine, but /ajp/ is impossible (Booij 1989a). 

The characteristic property of Dutch diphthongs is that the two halves only 

differ with respect to height. They are identical with respect to [back] and 

[round]. In other words, two short vowels can only form a complex nucleus in 

Dutch if the first one is a half-open vowel, and the second one a close vowel. I 

will therefore represent the three diphthongs as sequences of two short vowels, 

each linked to one X-position. Their diphthongal nature is then defined by 

template (20) for proper syllable nuclei (cf. Chapter 3). 

(20) N 

X 

—cons 

-high 

+ mid 

a F 

— cons 

+ high 

-mid 

aF 

In other words, a sequence of two short vowels forms a complex nucleus if the 

vowels are identical with respect to backness and roundness, and differ in 

height in the way indicated. 

2.5.4. The representation of the schwa 

The schwa, although phonetically short, behaves distributionally as a long 

vowel. Like long vowels, the schwa cannot be followed by a cluster of two 

consonants in the same syllable, and it can occur in word-final position, which 

is impossible for short vowels. Hence, we have to assign two skeletal positions 

to the schwa. 21 Just like the high long vowels, the schwa is phonetically short, 

and therefore the features of the schwa will only be linked to the first X. 

What is the underlying representation of the schwa in terms of features? As 

pointed out by Trommelen (1984: 77) the schwa does require an underlying 

representation since its occurrence is not completely predictable by rule on the 

basis of certain consonantal configurations. The following examples from 

Trommelen (1984: 77) illustrate this: 

(21) a. lamp 'id.' b. henn[a[p 'hemp' 

bank 'bench' monn[a]k 'monk' 

kans 'chance' Jann[a]s 'John' 

hemd 'shirt' lemm[3]t 'blade' 

" A similar proposal was made in Trommelen (1984). In Kager (1989) the hypothesis is 

pursued that the schwa is not linked to an X-slot at the underlying level.


Yet, the schwa can be considered as the unmarked, the default vowel of 

Dutch: it is the vowel that occupies the central position in the vowel space, it is 

the vowel to which full vowels can be reduced in unstressed position, and it is 

the vowel that is optionally inserted in non-homorganic consonant clusters. 

Phonetically, it is an unrounded central vowel, that is, neither front, back, high, 

nor low. This cannot be expressed by assuming a feature [+central| for the 

schwa, since this feature does not play any role in the vowel system of Dutch. 

In other words, the schwa may not have any phonological specification other 

than (—consonant). 

Note, however, that we have to be able to identify the schwa as such, since 

there are certain rules that crucially have to identify the schwa, for instance the 

rule of Prevocalic Schwa Deletion that only applies to schwa. Also, the schwa 

must be identifiable for stress rules since a syllable with a schwa never bears 

stress, and the template for well-formed prosodie words of Dutch also refers to 

the schwa. Therefore, I will represent the Dutch schwa as an empty vowel with 

the diacritic feature [unstressable], |uj for short, specified on the root. This will 

make it possible to identify the schwa for the proper application of phonological 

rules. Therefore, the lexical phonological representation of the schwa 

will be as follows: 

(22) X X 

I 

2.6. FEATURE CHARTS FOR VOWELS AND CONSONANTS 

The discussion of the phonological representation of Dutch vowels and consonants 

is summarized by the feature charts (Tables 2.4, 2.5). In these charts, 

the predictable features are circled. The uncircled feature specifications suffice 

to distinguish between the different consonants. In this chart, the glide /j/ is not 

mentioned because it has the same phonological properties as the /i/ except 

that it has only one position on the X-tier. Features that are either completely 

predictable or concern allophones, such as [anterior|, [distributed], [high], and 

[back] are omitted. 

TABLE 2.4. Feature chart for Dutch vowels 

i y u i e Y 0 o 3 e o e a a 

cons 

high + + + + e + © e ee 

mid + + + + -»- + + + 0 © 

back + 

rnd - + © -

- 

2.6. FEATURE CHARTS FOR VOWELS AND CONSONANTS 

TABLE 2.5. Feature chart for Dutch consonants 

21 

p b t 

d k f v s z x y m n r j l r u h 

cons 

® vu ® 

appr eeeeeeeeeeeeee® + 

com + + + + + + © 0 © © f f i 

B ffi ffi ffi 

nasal 90900900060+ + + 0© 

lat © © © © © © © © 0 0 Ö 0 0 9 + 9 

asp © © © © © © 0 9 9 9 9 0 © © © © 

Lab \ \ v \ v 

Cor v v v v v v v 

Dors v v v v 

© © 

© © 

© + 

v 

Note. A 'v' indicates that a privative feature is present.

THE PROSODIC STRUCTURE OF WORDS 


Words are not just sequences of sounds, like beads on a string; rather, they are 

organized into higher order phonological constituents such as the syllable and 

the prosodie word (also called the phonological word), two fundamental units 

of prosodie structure that will be discussed in this chapter. 

There are three basic roles for the syllable as a phonological unit: 

(a) The syllable is the most important domain of phonotactic restrictions, 

that is, a crucial domain for constraints on the co-occurrence of segments. For 

instance, we cannot determine whether the consonant sequence /pm/ is wellformed 

without taking the domain of the syllable into account: the sequence 

/pm/ is impossible if it belongs to one syllable (is tautosyllabic), whereas it is 

possible if there is a syllable boundary between the /p/ and the /m/ (i.e., /pm/ is 

heterosyllabic), as in the Dutch family name Ahma [upma| that consists of the 

syllables (up) a and (ma) 0 . 

(h) The syllable functions as a domain of phonological rules. The classical 

example from Dutch is the rule of final devoicing (traditionally called 

Auslautverhaerturitf) that devoices obstruents at the end of a syllable, which 

is usually formalized as follows: 

(I) [-son] -> [-voice] / —)„ 

The effect of this rule is demonstrated by the word pair hoed |hut] 'hat'- 

hoeden |hudon| 'hats'. In the singular form the morpheme-final /d/ occurs at 

the end of a syllable, since the singular form consists of the syllable (hud) CT , 

and hence it is devoiced. In the plural form, on the other hand, which consists 

of the syllables (hu) a (d3n) 0 , the /d/ is syllable-initial, and therefore it remains 

voiced. This example also illustrates that morphological structure and syllable 

structure do not coincide, because the morphological division of hoeden is 

hoed-en, -en being the plural suffix. This issue will be discussed in greater 

detail in Section 3.6. 

In early generative phonological publications on Dutch, final devoicing was 

sometimes formulated as a word-final rule, because we only find alternations 

between voiced and voiceless obstruents in word-final position. Yet, it can be 

shown that the rule also applies word-internally. First, it applies to acronyms 

such as ABVA (Algemene Bond van Ambtenaren 'General Union of Civil 

Servants'), which is syllabified as (ub) CT (va) a , and pronounced as [apfa],

3-2. THE REPRESENTATION OF SYLLABLE STRUCTURE 23 

with devoicing of the /b/. Second, foreign geographical names are also subjected 

to this rule, resulting in pronunciations such as |sitni] for Sidney. Thus, 

these facts support the claim that the correct generalization concerning the 

pronunciation of obstruents can only be made in terms of syllable structure. 

(r) The third role of the syllable is that it functions as the bearer of stress 

properties. Moreover, the location of stress in a word is dependent on the length 

('quantity') of the syllables of that word. This will be dealt with in Chapter 5. 

Syllables are concatenated into prosodie words, if possible. In simple cases, 

each word (in the morphological/syntactic sense) corresponds to one prosodie 

word. For instance, the word hoed-en 'hats' consists of two syllables that form 

one prosodie word. A compound like hoedenwinkel 'hat shop', on the other 

hand, consists of two prosodie words, which each form a domain of syllabification, 

hoeden and winkel, as will be explained in Section 3.6. 

3.2. THE REPRESENTATION OF SYLLABLE STRUCTURE 

Each syllable consists of an obligatory nucleus (either a vowel or—in some 

languages—a syllabic consonant), preceded by zero or more consonants (the 

onset), and followed by zero or more consonants (the coda). Nucleus and coda 

form the rhyme. In sum, the internal structure of the syllable is traditionally 

assumed to be as in (2).' 

(2) o 

Onset 

Rhyme 

Nucleus Coda 

For instance, the syllabic representation of a word like heen /ben/ 'leg' will be 

as m 

given in (3). 

A 

O R 

/ \ 

N C 

l\ I 

X X X X 

1 

This representation of syllable structure is only used in order to present the descriptive 

generali/ations concerning Dutch syllables as clearly as possible. No theoretical status is claimed 

for this particular model of syllable structure. Other models can be found in Clements and Kcyser 

( I'M Î) (Hat syllable with only a nucleus). Levin ( 1985) (syllable as X-bar projection of nucleus), 

and in Hyrnan (1985), Zee (1988), and Hayes (1989) (syllables consisting of morae).

24 THE PROSODIC STRUCTURE OF WORDS 

As (3) shows, syllable structure is not erected directly on the segments of the 

melodic tier, but on the X-tier. This is crucial for a proper account of Dutch 

syllable structure since long vowels, which correspond to two X-positions, are 

indeed to count as two positions in syllable structure. The relevant generalization 

is that Dutch rhymes contain at most three positions at the X-tier, that 

is, they contain either a short vowel and at most two consonants, or a long 

vowel or diphthong and at most one consonant. This restriction is illustrated in 

(4), with long vowels represented orthographically by a sequence of two 

identical letters: 

(4) aam, eim, amp, *aamp, *eimp, *almp, *ermp 

Since the schwa also has two X-positions (Section 2.5.4), it is correctly 

predicted to combine with at most one consonant in a rhyme. 

3.3. THE SYLLABLE TEMPLATE OF DUTCH 

The notion 'possible syllable of Dutch' will be defined by the following set of 

constraints: 2 

(a) The universal principle that is referred to as the Sonority Sequencing 

Generalization (Selkirk 1982). 1 will assume the following version, based on 

Zee (1988) and Clements (1990): 

(5) Sonority Sequencing Generalization (SSG) 

The sonority of consonants must decrease towards the edges of a syllable, 

where the sonority of consonants is defined by the following scale of 

decreasing sonority: 

Glide-Liquid-Nasal-Obstruent 

decreasing sonority 

This principle restricts the co-occurrence of segments in onsets and codas, and 

also explains the mirror image effects in these constraints, as illustrated in (6): 

(6) klem 'grip', *lkem melk 'milk', *mekl 

slop 'slum', *lsop pols 'wrist', *posl 

(h) The parameter of syllabicity. Universally, vowels can function as nuclei, 

but languages differ with respect to the syllabicity of consonants. Standard 

Dutch only has vowels as nuclei, whereas some non-standard dialects also 

allow for syllabic nasals. Note that this shows that the parameter is independent 

from the SSG because in these dialects liquids do not function as nuclei 

although they rank higher in the sonority hierarchy than nasals. 3 

Detailed studies of Dutch syllable structure can be found in Booij (1981«. ch. 6), Trommelen 

(1984). Van der Hulst (1984), and Kager (1989). 

' This point is also made by Clements 11990: 294), who refers to the findings of Alan Bell in this 

respect.

3-3- THE SYLLABLE TEMPLATE OF DUTCH 25 

(r) Language-specific restrictions on the number of X-positions in the 

different syllable constituents. As already stated above, the Dutch rhyme 

consists of at most three positions, and minimally two positions. The nucleus 

consists of at most two positions. 4 The Dutch onset contains at most three 

positions. 

(d) Language-specific constraints on the possible combinations of segments. 

For instance, Dutch does not allow for the onset /tl/ although this cluster does 

not violate the SSG. These constraints will be discussed in detail in Section 

3.5. 

The SSG refers crucially to classes of segments such as glides, liquids, etc. 

Therefore, our system of phonological features must be able to identify them 

as such. Moreover, as will be shown in Section 3.5, language-specific constraints 

also require the availability of features that can single out classes of 

segments such as glides and liquids. As proposed by Clements (1990), we can 

identify the relevant classes, and also derive the sonority ranking required, by 

making use of the major class features [vocoid], [approximant], and [sonorant|. 

The feature [vocoidl characterizes vowels and glides, and the feature 

[approximant] stands for 'an articulation in which one articulator is close to 

another, but without the vocal tract being narrowed to such an extent that a 

turbulent airstream is produced' (Ladefoged 1982: 10)."'' 

The Sonority Ranking Hierarchy can now be derived from the number of 

positive specifications for the major class features as shown in Table 3.1. 6 

Nuclei consist of either one or two segments, with at least one of them linked 

to a [—consonant] root. The nucleus (N) may contain two (short) vowels, that 

is, a complex nucleus, if they form one of the three Dutch diphthongs. The 

template for admissible complex nuclei has already been defined in Section 

2.5.3. 

TABI.K 3.1. Sonoritv Ranking Hierarchy 

Obstruent < Nasal < Liquid < Glide 

| voc | + 

|appr| + + 

|son| + + + 

Sonority Ranking 0 1 2 3 

In the case of nuclei containing only one short vowel, the nucleus must be 

followed by at least one consonant. In other words, a syllable cannot end in a 

short vowel. This is expressed in the Minimal Rhyme Constraint: 

1 

This np|H-:ns 10 he a universal tendency, cf. Kenslowicv and Rubach ( 1987), Booij ( 1989o) and 

the references given there. 

5 Cited after Clements (1990: 293). 

6 The variant I use here is that argued for in Zee ( I98X: 83).


(7) Minimal Rhyme Constraint 

Dutch rhymes consist of at least two X-positions. 

Hence, we find the types of Dutch rhyme (R) given in (8). 

"'K 

N 

X 

C 

X 

R 

N 

l\ 

X X 

R 

i\ 

l\ \ 

N c r 

l\ 

xxx : 

"I C 

l\ 

3-3- THE SYLLABLE TEMPLATE OF DUTCH 2J 

Evert /evar-t/ 'id.' (Christian name) 

kaars /kar-s/ 'candle' 

Evers /evar-s/ 'id.' (family name) 

laars /lar-z/ 'boot' 

h. Appendix of two segments 

koorts /kor-ts/ 'fever' 

Everts /evar-ts 'id.' (family name) 

ernst /ern-st/ 'seriousness' 

herfst /herf-st/ 'autumn' 

eerst /er-st/ 'first' 

aardigst /arday-st/ 'nicest' 

c. Appendix of three segments 

bedaardst /badar-dst/ 'calmest' 

vermoeidst /vermuj-dst/ 'most tired' 

promptst /promp-tst/ 'most prompt' 

It should be noted that an appendix with three coronal consonants, phonetically 

always the sequence [tst], is hard to pronounce; usually, the first |t] is not 

pronounced. 

We thus conclude that a well-formed prosodie word of Dutch consists of one 

or more syllables followed by an appendix of up to three coronal obstruents. y 

The concept of 'appendix' also explains why we can combine two or more 

obstruents with the same degree of sonority: the SSG does not pertain to 

appendices. 10 This also applies to words like gips 'plaster' and straks 

'soon'. Here, the final /s/ can be interpreted as an appendix even though the 

rhymes /ip/ and /ak/ do not have their maximal length. 

In Section 6.2 it will be argued that consonants are prosodified as appendix 

consonants if possible." 

A relevant observation in this connection is that Dutch vowelless suffixes 

always consist of one or more coronal obstruents, which makes it possible to 

always attach them to a stem without violating prosodie well-formedness 

conditions. 

Similar appendix phenomena seem to occur at the left edges of Dutch words: 

a Dutch onset consists of two positions, possibly preceded by /s/. The following 

word-initial clusters of three consonants occur: 

(11) spl- split 'id.' 

spr- spreeuw 'starling' 

9 II may be that the limit on the number of appendix consonants need not be slated as part of the 

phonology of Dutch since it may be assumed to follow from the fact that longer strings of coronal 

obstruents are simply unpronounceable. 

" This does not mean, however, that the selection of appendix consonants is completely 

arbitrary in terms of the SSG, since only obstruents can occur in that position. 

1 ' Hall ( 1992: 123), on the other hand, proposes for German that coronal obstruents are adjoined 

to the preceding syllable by a rule of Stray Consonant Adjunction, which implies that only those 

consonants are interpreted as appendices which cannot be lilted inlo the normal syllable template.


strskrskisxr- 

stroom 'stream' 

skriba 'scribe' 

sklerose 'sclerosis' 

schreeuw 'cry' 

Trommelen (1984) suggested that we should interpret the word-initial /s/ as an 

appendix. This would explain why only the /s/ occurs in clusters of three 

consonants. Moreover, the SSG violations in /sC-/ clusters would also be 

explained. However, as Van der Hulst (1984: 66) has pointed out, an appendix 

interpretation of the /s/ predicts that it freely co-occurs with all possible onsets, 

which is incorrect. For instance, although tw-, xl-,fl-, and fr- are correct onsets, 

the clusters stw-, sxl-, sfi-, and xfr- are impossible. 

Another problematical aspect of the appendix interpretation of the wordinitial 

/s/ is that it implies that sC- and sCC- clusters do not occur in wordinternal 

syllables, which is incorrect (Booij 1984), given syllabifications such 

as the following: 

(12) got-spe 'chutzpah' 

En-sche-de 'id.' (place-name) 

ka-ta-stro-fe 'catastrophe' 

ek-stra 'extra' 

ven-ster 'window' 

Therefore, I will assume the onset template (9) for Dutch. 

The claim that coronal appendices are only allowed at the right edges of 

words seems to be contradicted by the following words with a seemingly wordinternal 

appendix: 

(13) aartsbisschop 'archbishop' 

herf.v/kleuren 'autumn colours' 

koorttachtig 'feverish' 

However, the relevant notion 'word' to be used here is that of 'prosodie word', 

not 'grammatical word'. As will be shown below, the prefix acirts-, the 

compound constituent herfst, and the suffix -achtig form independent 

prosodie words. Therefore, a coronal appendix can appear in such cases. 

In sum, we have arrived at the templates (14) for Dutch syllables and 

prosodie words. 

The role of the prosodie word as a domain of phonotactic constraints is not 

restricted to the distribution of appendices. Other conditions on prosodie words 

will be discussed in Section 3.6, where it is shown that a prosodie word cannot 

begin with a schwa, and that it has to contain at least one syllable with a full 

vowel. 

Furthermore, Dutch does not allow for geminate consonants within prosodie 

words, unlike languages such as Finnish, Icelandic, and Italian. In other words, 

although, for example, the syllables /buk/ and /kur/ are both well-formed, they 

cannot be combined into the prosodie word /bukkur/. If a sequence of identical

(14) a. Syllable: 

3.4. SYLLABIFICATION 

a 

x\ 

O R 

/l\ /l\ 

(X) (X) (X) X X (X) 

29 

b. Prosodie word co 

\ 

a" Appendix 

-son]" 

CorJ n > l, m < 3 

consonants arises within a prosodie word due to some morphological operation, 

as in grootte /yrot+to/ 'size', an obligatory rule of Degemination applies, 

resulting in this case in the phonetic form [yrota]. Since there is a phonological 

rule (P-rule) expressing this constraint, the prohibition on geminates need not 

be part of the prosodie word template. This shows that the phonotactic 

constraints on the possible words of a language are not expressed by just 

one type of constraint: these constraints are defined by an array of different 

types of rules: syllable-structure conditions, prosodie word conditions, sequential 

constraints (Section 3.5.5), and phonological rules (P-rules, Section 4.2). 12 

Together, they define the notion 'possible prosodie word of language L'. 

3.4. SYLLABIFICATION 

The syllabification of Dutch words is predictable on the basis of the syllable 

template given above (14a). Before syllabifying a word, we will have to 

determine the domains of syllabification. 

3.4.1. The domains of syllabification 

In Section 3.3.1 I pointed out that we have to distinguish between the grammatical 

word and the prosodie word, and that the appendix appears on the right 

periphery of prosodie words. Crucially, a prosodie word does not always 

correspond to a grammatical word. This also explains why in a compound 

such as handappel '(lit.) hand apple', 'eating apple' with the morphological 

structure \[hand\^\appel\^\ N , the final /d/ of hand is not syllabified with the/a/ 

2 The point thai phonetic constraints arc the effects of both siatic constraints and phonological 

rules is made in Postal (iy6S). Morpholexical rules, i.e. phonological rules conditioned by 

morphological or diacritic properties, do not form part of the definition of 'possible prosodie 

word of a language'.


of appel in conformity with the universal principle that a consonant is syllabified 

with a following adjacent vowel: hand and appel form two different 

domains of prosodification, and hence each corresponds with a prosodie word 

because the syllables in a domain of syllabification form one prosodie word. 11 

The following mapping rules define the relation between morphological 

structure and syllabification domains for Dutch (Booij 1977, 1981«, 1985«): 

(15) a. In compounds, each constituent that can form a grammatical word of 

its own is an independent domain of syllabification. (For instance, in 

handappel the two constituents hand and appel are syllabified separately, 

and hence the /d/ of hand devoices: (huntupol).) 

h. Prefixes form independent domains of syllabification, as is clear from 

the syllabification of prefixed words: there is always a syllable 

boundary coinciding with the prefix boundary (for instance ontaard 

'degenerate' with the syllabification pattern (Dnt) 0 (ard) n ). 14 

c. Most suffixes do not form an independent domain of syllabification, 

but form a prosodie word with the preceding material, as illustrated in 

Section 3.1. There are, however, a number of suffixes with at least one 

full vowel, such as -achtig '-ish', -baar '-able', -dom 'the set of', -held 

'-ness', -ling '-ling', -loos '-less', -schap '-ship', and -zaam 'inclined 

to', that do form independent domains of syllabification, and thus 

form prosodie words of their own. They are all consonant-initial 

except for -achti^. 

The fact that word-internal compound boundaries and prefix boundaries 

always form edges of a syllabification domain seems to be a universal 

tendency (Rubach and Booij 19907?). It can be expressed as follows: 

'Lexical syllabification does not apply across |' 

(where '|' is a left morphological bracket). 

3.4.2. Maximal Onset and Minimal Rhyme 

Syllabification can be seen as the matching of the syllable template of a 

language with the segmental string of a word in such a way that a maximal 

number of segments is syllabified into a minimal number of syllables (Itô 

1989). Furthermore, syllabification is subject to the universal principle that a 

consonant before a syllabic segment always forms a syllable with that syllabic 

segment, the universal CV-rule. 

Syllable template matching can in principle take place from left to right or 

from right to left. If it takes place from right to left, the result is that, in a 

11 As pointed oui in Booij ( I''XK/)). we eannol assume thai the prosodie word ilsell is the domain 

ol syllabification since this would lead to a paradox: the prosodie word presupposes the presence ol 

syllables. 

4 This does not necessarily imply that prefixes arc always prosodie words of their own. as will 

be shown in Section 3.6.

3-4- SYLLABIFICATION 3! 

sequence of consonants between two syllabic segments, as many consonants as 

possible are assigned to the onset. In other words, right-to-left syllabification 

gives us the Maximal Onset effect. Dutch, like many other Indo-European 

languages, appears to require maximal onsets. 15 

If we conceive of the prosodie word template as comprising the syllable 

template, as proposed in (\4b), the prosodie word template also has to be 

matched from right to left. This correctly implies that word-final coronal 

obstruents will be matched with the Appendix position, even if they could 

also have filled a rhyme position, as in gans 'goose' (cf. Section 6.2). 

A Dutch rhyme consists of at least two X-positions, the Minimal Rhyme 

Constraint. This correctly predicts that word-final syllables cannot end in a 

short vowel. (Exceptions are some exclamations such as he [he], joh [ja], bah 

[buI, and goh [va], and French loan-words such as cachet [kaj"jt| 'id.' and 

bidet [bide] 'id.'.) On the other hand, word-internal rhymes ending in a short 

vowel seem to be admissible: 

(16) adder [udar] 'snake', letter [letar] 'id.', opper [spar] 'upper' 

Nevertheless, I will assume, with Van der Hulst (1984) and Kager (1989), that 

the Minimal Rhyme Constraint applies word-internally. This explains why we 

do not find words with a word-internal short vowel followed immediately by 

another vowel, for instance /hiat/ (compare /hiat/ 'hiatus'): after a short vowel 

a consonant is required in order to get a minimal rhyme. 

The question now is how the universal CV-rule and the Minimal Rhyme 

Constraint can be reconciled in the syllabification of a word like adder [udor| 

'snake'. The Maximal Onset effect of right-to-left syllabification will of course 

be overruled by the absolute Minimal Rhyme Constraint" 1 as the following 

examples show (the hyphen indicates the relevant syllabification): 

(17) as-ter 'id.' 

as-pect 'id.' 

vis-[k]ose 'viscose' 

Pas-[x|a 'Pesach' 

a|k|-ne 'acne' 

Note that although st-, sp-, sk-, sx-, and kn- are possible onsets of Dutch, these 

clusters are nevertheless heterosyllabic in these words, in order to comply with 

the Minimal Rhyme requirement. 17 

n Instead of assuming iliivrlionality of syllahilicalion, the Maximal Onset effect may also be 

derived from the requirement that syllable contacts are optimal (cf. Clements 1990) which implies 

that, lor instance, hi is better than h-l since in the latter case we get a had syllable contact: 

obstruent before sonorant, whereas in an optimal syllable contact the more sonorous consonant 

comes lirsl. 

6 The priority of language specilic minimal rhyme constraints in matching syllabic templates 

with strings is also argued for Archangeli (1991). 

1 

Intuitions as to the syllahilicalion of intervocalic /si/ appear to vary, as shown in Rietveld 

(1983). It may be that at (he surface level the /s/ in. e.g. aster is also ambisyllabic. Evidence other 

than intuitions on syllabification seems to be lacking here.

32 THE PROSODIC STRUCTURE OK WORDS 

The only way in which a word like adder can be syllabified without 

violating the universal CV-rule and the Dutch Minimal Rhyme Constraint is 

by assigning the intervocalic /d/ to both syllables. In other words, the /d/ is 

ambisyllabic (see (18)). 18 

(18) o 

rX X X X X 

The ambisyllabicity of the /d/ will block its devoicing in syllable-final 

position. To see this, rule (I) should be reformulated as delinking rule (19). 

(19) Coda 

I 

X 

I 

[-son] 

[+voicej 

After delinking, the consonant no longer bears the feature l+voice], and hence 

it will be realized as a voiceless consonant. 

Note that the /d/ of adder does not satisfy the structural description of this 

rule, because the /d/ is also linked to the onset of a second a. The required 

blocking of the application of the devoicing rule can be achieved by the socalled 

Uniform Applicability Condition (UAC) from Schein and Steriade 

(1986: 727): 

(20) Uniform Applicability Condition (UAC) 

Given a node n, a set S consisting of all nodes linked to n on some tier T, 

and a rule R that alters the content of n: a condition in the structural 

description of R on any member of S is a condition on every member of S. 

The relevant node is [— sonorant) which does not lose (+voice) when it is 

ambisyllabic because in that case one of the nodes to which the X-slot is 

linked, is not a coda, as required by the rule. 19 

This description of final devoicing, however, does not account for the fact 

that final devoicing also applies to voiced obstruents in Appendix positions, 

such as the /z/ of laars /laarz/ 'boot' that is pronounced as [s]. Lombard! 

'* When the internal structure of the syllable is irrelevant, it will not be represented. 

|g Note that in this case the Linking Constraint proposed by Hayes (1986): 'Association linos in 

structural descriptions are interpreted as exhaustive', also makes the correct prediction, as is 

pointed out by Van der Hulst (1985). However, as we will see in Chapter 4. the Linking Constraint 

also incorrectly blocks the application of Final Devoicing to voiced obstruent clusters—which 

share the feature |+voice]—in coda position.

3-5- CO-OCCURRENCE CONSTRAINTS 33 

(1991) therefore proposed to assign a different interpretation to final de voicing: 

the (in her theory, privative) feature [voice] is only licensed in Dutch 

onsets. Therefore, it must (by convention) be delinked from any other position. 

This results in final devoicing of obstruents in both Coda and Appendix 

positions. The condition that [voice] is only licensed in onsets should be 

taken to apply at a particular level of phonological derivation: not at the 

underlying level, but at the word level (cf. Chapter 4). In this analysis, we 

do not have to invoke the UAC to explain why the /d/ of adder is not devoiced: 

its feature [voice] is licensed by its being linked to the onset of the second 

syllable. 

The nucleus of Dutch syllables contains either one or two X-positions, and 

the first has to be linked to [-consonant]. Thus we get the possible nuclei in 

(21). 

(21) N N N 

I l\ l\ 

X X X X X 

[—cons] [-cons] -cons' 

-high 

+mid 

ccF 

'—cons' 

+high 

-mid 

aF 

short vowels long vowels diphthongs 

Given this set of possible nuclei, the glide [j] which is a short /i/ in underlying 

forms will always be assigned to an onset or a coda position, as required, except 

when it is preceded by the Id in which case it will be interpreted as the second 

part of a diphthong. For instance, in the word aai /ai/ 'to caress' it is impossible 

to interpret the vocoid sequence as a complex nucleus. Hence, the only possible 

solution is to assign /i/ to the coda. Similarly, the postvocalic vocoid Ivl will 

automatically be interpreted as a glide when the syllable template is matched 

with the phonological string of a word, as in nieuw Imvl 'new'. 

3.5. CO-OCCURRENCE CONSTRAINTS 

3.5.1. Rhyme constraints 

As stated above, Dutch rhymes consist of at most three positions. Trommelen 

(1984: 121) observed that word-internal three-positional rhymes tend to end in 

an obstruent rather than a sonorant consonant. 20 This restriction accounts for 

20 Kager and Zonneveld (1986) proposed the even stronger constrain! that word-internal rhymes 

of simplex words consist of at most two positions. As Kager (1989: 205-7) points out, there is a 

substantial number of exceptions/counterexamples to this constraint.


the opposition between the unacceptable words in (22a) and the existing words 

in (22b,c): 

(22) a. *teem-po b. tem-po 'id.' c. bauk-siet 'bauxite' 

*oom-nibus om-nibus 'id.' buus-te 'bust' 

*terrn-po tem-po 'id.' kloos-ter 'cloister' 

*marn-darijn man-darijn 'mandarin' hyp-nose 'hypnosis' 

However, as pointed out by Booij (1984) and Kager (1989), this rhyme 

constraint has quite a number of exceptions, some of which are listed here: 

(23) aal-moes 'alms', pien-ter 'smart', Maar-ten 'id.' (proper name), aar-de 

'earth', vaan-del 'banner' 

Moreover, the constraint does not hold for morphologically complex words: 

(24) Plural nouns with the suffix -en Past-tense forms of verbs 

maan-den 'months' 

talm-de 'hesitated' 

beem-den 'meadows' 

kerm-de 'moaned' 

baar-den 'beards' 

naal-den 'needles' 

Conceptually, it is not attractive to propose syllable templates that only hold 

for simplex words. It is a basic insight that words are structured in two ways, in 

terms of a morphological and a prosodie hierarchy, which need not be 

isomorphic. Therefore, I interpret Trommelen's observation as a (prosodie) 

Morpheme Structure Condition (cf. Section 3.5.5): morpheme-internal 

syllables with three-positional rhymes tend to end in an obstruent consonant. 

Within rhymes, diphthongs cannot be followed by a glide or by /r/: 

sequences such as /civ/, /ceyu/, /ouj/, and /eir/ are ill-formed. 21 /eir/, /ceyr/ 

and /our/ are only pronounceable by inserting a schwa before the /r/. This 

constraint may be explained in terms of co-articulation: the phonetic realization 

of a diphthong requires a movement of the tongue in the direction of the 

position of the high vowels, whereas the /r/ has a centralizing effect, and 

requires a movement of the tongue into the direction of the centre of the 

vowel space (cf. Koopmans-van Beinum 1969). 

This constraint also shows the importance of the syllable and its constituents 

as domains of phonotactic restrictions since the Dutch diphthongs do occur 

before a heterosyllabic /r/, as in Beira [bcira] 'id.' (place-name), neuron 

[nœyron] 'id.', and Laura [loura] 'id.'. 

!l 

An exception is the loan-word reveil /raveij/ 'revival'. 

In Dutch orthography we find spellings like ouw and auw in words such as jouw 'your' and rauw 

'raw', suggesting that in these words a /u/ follows the back diphthong. However, these letter 

sequences should be taken as standing for just the back diphthong. This is also proven by the quite 

frequent spelling mistake of usingyou. which is the correct spelling for the personal pronoun, 2 sg. 

'you', as the spelling for the possessive pronoun 2 sg. 'your', which is phonetically undistinguishable 

from the 2 sg personal pronoun Similar pairs of homophones which are distinguished in the 

spelling are hou 'hold' (I sg. pres.) vs. Iwuw 'hew' (l sg. pres.), hui 'cold' vs. kauw 'chew' (1 sg. 

pres ), and nou 'now' vs. nauw 'narrow'.

(25) Diphthong-r Constraint 

*([-cons][-cons] +cons )RHYME 

+appr 

+cont 


The next constraint to be discussed is that short vowels cannot be followed 

by /v/ or /z/, except in the loan-words razzia [razija] 'raid', puzzel [pvzal] 

'puzzle', and mazzel (muzal] 'good luck'. This restriction is illustrated in (26): 

(26) a. kaas /kaz/ 'cheese', kazen /kazan/ 'cheeses', Pasen /pasan/ 'Easter', 

passen /pasan/ 'to fit'; but */kaz/, */pazan/ 

h. graaf /yraf/ 'graph', grafen /yrafan/ 'graphs', graaf/yrav/ 'earl', 

graven /yravan/ 'earls', maf /maf/ 'mad', maffe /mafa/ 'mad', 

(inflected form); but: */muv/, */mava/ 

We therefore have to assume the constraint (27): 

(27) /V/-/Z/ Constraint *R 

N 

— son 

+cont 

+voice 

(Lab) 

I Cor I 

The /V/-/Z/ Constraint explains why native speakers of Dutch are inclined to 

pronounce puzzel as [pyzal|, with a long vowel before the /z/. It can be 

formulated as a rhyme constraint because it concerns short vowels, which 

always require a following tautosyllabic consonant. In other words, it will 

alsoexclude the sequence short vowel +/v/or/z/+ vowel in polysyllabic words. 

Another rhyme constraint is that the /rj/ cannot be preceded by long vowels 

or diphthongs: -eng, *-eeng, *-eing. It will be discussed in Section 3.5.5 

because it can also be interpreted as a sequential constraint. 

3.5.2. Onset Constraints 

All Dutch consonants except the velar nasal can appear in syllable-initial 

position. 23 This is expressed by constraint (28). 

u In Kager (1989: 196) it is indeed analysed as a rhyme constraint. 

M Kager (1989) suggested that this follows from assuming the underlying representation /Ng/ 

for velar nasals, in combinalion with the SSG. However, both Booij (1980) and Trommelen (1983) 

have argued that this analysis is untenable.


(28) Velar Nasal Constraint *O 

I 

X 

I 

l+nas 1 

L Dors] 

Interestingly, the velar nasal does occur in word-internal onsets after a short 

vowel as in engel |erpl| 'angel'. Here, the velar nasal is ambisyllabic: 

(29) o .o 

X X X X X 

l 

Therefore, the Uniform Applicability Condition will prohibit the Velar Nasal 

Constraint from applying. 24 After long vowels, a word-internal /g/ would be 

syllabified as an onset because of the constraint on velar nasals mentioned in 

the preceding section, but this will be excluded by the onset constraint (28). 

Hence, we do not find words like zaanger /zarjar/. 

Let us now turn to CC-constraints. To begin with, the /h/ does not combine 

with other consonants in onsets: 

(30) /h/-onset Constraint 

A branching onset may not dominate |+aspirated]. 

A more general condition is that an onset never contains two sonorant 

consonants: combinations of nasals with liquids or glides, or liquids with 

glides are excluded: 

(31) Branching-onset Constraint 

A branching onset must dominate [-sonorant]. 

Obstruents combine with glides, liquids, and nasals. A survey of possible 

clusters is given in (32): 

(32) pj- riempje 'little belt' 

tj- traan/y'e 'little tear' 

kj- koninfye 'little king' 

dj- aJ/eu 'farewell' (loan-word) 

tw- twee 'two' 

dw- (/waas 'stupid' 

sw- .yvvingen 'to swing' (loan-word) 

/.w- zweet 'sweat' 

fj- fjord 'id.' (loan-word) 

24 Alternatively, one might say in the framework of Goldsmith ( I WO) and Lombardi (1991 ) thai 

velar nasals are only licensed in codas. The velar nasal in the onsel of the second syllable of engel 

is then 'parasilically licensed', through its link with the coda of the previous syllable.

sj- 

Z J~ 


.v/'aal 'shawl' (loan-word) 

J UT y ''d-' (loan-word) 

I will exclude /dj-/ and //j-/ as possible onsets because they only occur in loanwords, 

and because word-internally they do not form onsets in non-borrowed 

words such as diminutives. For instance, the word handje /handja/ 'hand', 

(dimin.) is syllabified as (hund) CT (J3) 0 as shown by the fact that syllable-final 

devoicing applies: |hantja|. Therefore, the onset constraints (33) will be 

assumed. 

(33) a. *bw-, *pw-, *vw-, *fw- *O 

/\ 

Lab 

Lab 

b. *xj, *vj-, *xw-, *yw- *O 

X 

p-cont] 

L Dors) 

X 

+appr 

(Lab 

I Dors I 

c. *bj-, *dj-, *vj-, *zj- *O 

X 

X 

[+voice] [+voc~| 

|_ DorsJ 

Constraint (33a) also excludes labial obstruents before /m/, a correct 

prediction. 

The first three clusters of (32) also illustrate that the set of possible wordinitial 

clusters need not be identical to the set of syllable-initial clusters (cf. 

Booij 1983): pj-, tj-, and kj- do not occur as word-initial clusters except in 

loan-words. In other words, the prosodie word is also a domain of phonotactic 

restrictions, as already proven by the appendix phenomena. 

Obstruent plus liquids combine rather easily, as the following survey shows: 

(34) Obstruents + liquids 

pi- plant 'id.' 

bl- bloem 'flower' 

pr- pruim 'plum' 

br- bruin 'brown' 

tr- frein 'train'

J 


drflvlfrvrsixlxrylyr- 

droog 'dry' 

y/es 'bottle' 

r/as 'flax' 

/raai 'beautiful' 

vriend 'friend' 

\/ank 'slim' 

chloor 'chlorine' 

chroom 'chrome' 

£/as 'glass' 

#ras 'grass' 

Thus, the following clusters are out: //-, dl-, zl-, zr-, and .vr-. This is formulated 

in the constraints (35). 

(35) Obstruent + Liquid Constraints 

a. *tl-, *dl-, *sr-, *zr- *O 

[acontl facontj 

L Cor J L CorJ 

b. *zl- *O 

X 

X 

[+voicel 

L Cor J 

A remarkable onset type is the combination /ur-/ that occurs in a few words 

such as wreed 'cruel' and wrijven 'to rub'. This sequence is an exception to the 

SSG since the /u/ is a non-vocoid approximant here, with the same degree of 

sonority as the /r/. This explains that many speakers of Dutch actually 

pronounce the w as [v] rather than as [u], i.e., as a fricative obstruent, thus 

avoiding violation of the SSG. 

Finally, we have to consider obstruent + nasal clusters. Generally, they do 

not combine, with the exception of the clusters listed here: 25 

(36) Obstruent + Nasal 

sm- .wiaak 'taste' 

sn- snee 'cut' 

kn- knie 'knee' 

Other combinations occur, but only in a few exceptional words (partially loanwords), 

namely pn-, as in pneumatisch 'pneumatic', fn- as in fnuiken 'to 

'"' There is a universal tendency in languages lhal onset clusters of obstruents plus nasals are 

very marked, whereas obstruents combine easily with glides and liquids (cf. BrosHow (IWI) ).


cripple', and #«- as in gnuiven 'to gloat'. The ill-formed clusters and the 

rare onsets pn-,fn-, and gn- are excluded by constraint (37). 

(37) Obstruent + Nasal Constraint *O 

/\ 

X, X 

|+nas| 

Condition: 

Xj / |+cont, -voice, Cor] or [-cont, Dors) 

Combinations of stops and fricatives are excluded by the SSG that ranks 

them as of equal sonority. Such combinations do occur, however, in some 

loan-words like psalm 'id.', tsaar 'tsar', and Xantippe 'id.' (the letter X stands 

for [ksj. 26 Moreover, the /s/ occupies a special position in that it combines with 

both stops and fricatives: 

(38) sp- spin 'spider' 

st- steen 'stone' 

sk- igelet 'skeleton' 

sf- sfeer 'sphere' 

sx- .vc/iool 'id.' 

We might try to to explain this exceptional behaviour by assuming that /s/ is a 

word-initial appendix. However, it was argued in Section 3.3.1 that this 

solution cannot be upheld. So the generalization seems to be the following: 27 

(39) In sC(C)-clusters, the sequence sC may violate the SSG. 

As noted in Section 3.3.1, there is only a very restricted set of CCC clusters: 2 * 

(40) spr- spreeuw 'starling' 

str- stroom 'stream' 

skr- skriba. 'scribe' 

spl- .v/7/inter 'id.' 

ski- .vA'/erose 'sclerosis' 

sxr- schreeuw 'cry' 

Thus, the following descriptive generalization can be made: 

(41) In CCC onset clusters the final CC is either a possible Stop + Liquid 

cluster, or a velar fricative + /r/ 

This will exclude sequences such as /stl-/ because /tl-/ is ill-formed, but also 

sequences like /sfl-/ and /sfr-/. 

6 This implies lhat other languages may make a difference in sonority between stops and 

fricatives. 

7 The special nature of the combination /s/ plus stop in lhat it behaves as a unit, a sort of 

complex segment, appears to be a cross-linguistic universal, as has frequently been noted in the 

literature (cf. Ewen 19X2. Selkirk 19X2, Stcriade 1982, Broselow 1991). 

In loan-words we also lind /stj-/ and /skw-/ as in sieward 'id.' and squaiinin 'id.' respectively.

40 THF PROSODIC STRUCTURE OK WORDS 

3.5.3. Coda constraints 

The only consonant that never occurs in codas is the /h/, as expressed in (42). 

(42) /h/-coda Constraint *C 

I 

X 

I 

l+asp] 

This can be related to the fact that the /h/ needs to be followed by a vowel in order 

to receive a spécification for its Place features: it is parasitic on a next vowel. 

Since glides only occur after long vowels, they do not combine with other 

consonants in codas because otherwise the rhyme would contain more than 

three positions, and thus they can only be followed by coronal obstruents, the 

appendix consonants. 

Of the logically possible liquid + nasal clusters, only /-1m, -rm, -rn/ occur: 

(43) -1m helm 'helmet' 

-rm arm 'id.' 

-rn kar« 'to churn' 

That is, we have to exclude /-In/ and liquid + velar nasal clusters (before the 

velar nasal no coda consonant whatsoever can appear). This is expressed in (44). 

(44) Liquid + Nasal Constraints *C *C 

[latj r+nasl l+nas 1 

L CorJ 

L DorsJ 

The occurrence of clusters of liquids and obstruents is almost unrestricted, 

as the following survey shows: 

(45) -lp help 'id.' 

-Ib 

-If e///elf/ 'id.' 

-Iv elf /elv/ 'eleven' 

-Ik melk 'milk' 

-ly a/j; 'alga' 

-Ix 

-rp har/7 'id.' 

-rb 

-rf amo// 'amorphous' 

-rv kor//korv/ 'basket' 

-rk ker£ /kerk/ 'church' 

-rx monarch 'id.' 

-ry er# 'very'

3-5- CO-OCCURRENCE CONSTRAINTS 4! 

There are no examples of/-Ib/ and /-rb/: liquids apparently do not cluster with 

a /b/. The non-occurrence of /-lx/ at the underlying level is very probably an 

accidental gap since phonetically [lx] does occur. Of course, /-Ib/ and /-rb/ 

would never surface as such in codas because of final devoicing. 

Liquids can also be followed by a coronal obstruent, but since these 

obstruents can also be interpreted as appendices, these clusters cannot be 

used as evidence for coda constraints: held /held/ 'hero', milt /milt/ 'spleen', 

hard /hard/ 'id.', hart /hart/ 'heart', hals /hal/7 'neck', als /als/ 'if', hars /hars/ 

'resin', vers /verz/ 'verse'. 

Nasal consonants cluster with stops, but not with fricatives; the nasal is 

always homorganic with the following plosive. Only in heterosyllabic clusters 

do we find clusters of nasals with fricatives (e.g. kamfer /kumfar/ 'camphor' 

and angina /urjvina/ 'id.') and with heterorganic stops (e.g. imker /imkar/ 'bee 

keeper'), which once more stresses the relevance of the syllable as a domain of 

phonotactics. Like the liquids, nasals do not combine with /b/. Thus we have 

the constraints (46) and (47). 

(46) Nasal + Fricative Constraint *C 

|+nas] 

|+cont] 

(47) /b/-coda Constraint *C 

X 

"-cont 

+voice 

. Lab_ 

The SSG predicts that clusters of fricatives and stops do not occur, except in 

the Appendix. However, we do find /-sp, -st, -sk/: 

(48) -sp we.vp 'wasp', ge.sy? 'buckle', rasp 'grater' 

-st a.v/-ma 'asthma', i.v/-mus 'isthmus' 

-sk grote.si 'grotesque', obeliil 'id.' 

We have to allow for /-st/ as a possible coda because we cannot interpret the 

word-internal cluster /-st/ as in astma as an appendix. The clusters /-sp/ and 

/-sk/ also behave as real coda clusters, not as appendices, in that they cannot be 

preceded by a long vowel or diphthong. 29 Note also that they cannot be 

interpreted as complex segments (two melodic elements linked to one X) 

because distributionally they count as two X-positions. So we have to allow 

g 

The only exceptions are a tew geographical names such as Weesp, and the word hruusk 

'brusque', a loan from French.


for the /s/ to violate the SSG both in onset and coda; in the coda the violations 

are more restricted, however, since only stops combine with the /»/. 

3.5.4. Pansyllahic constraints 

Pansyllabic constraints are constraints governing sequences that do not fall 

within onset, coda, or rhyme constituents (Cairns 1988: 229), and which tend 

to limit recurrences of the same or similar segments within a syllable. 

As Janson (1986: 192) observed, in many languages the /j/ does not occur 

before /i/; this is also the case for Dutch as expressed by (49). 

(49) /ji/-Constraint *O 

( . . . 1+vocJi l+vocjj . . . ) 0 

Note that the sequence /vu/ is permitted in Dutch, because in prevocalic 

position the /u/ is a non-vocoid. 

Another constraint on CV-sequences is that /h/ does not occur before schwa. 

This may have to do with the fact that /h/ lacks a supralaryngeal specification, 

and the schwa, being the default vowel, also lacks a supralaryngeal specification. 

In other words, a syllable /ha/ would have no specification of Place 

whatsoever. That is, the following constraint: 

(50) /h9/-Constraint 

*( . . . [+asp] [-cons] u . . . ) 0 

can be related to the following general constraint:" 

(51) Place Constraint 

A a must dominate Place. 

This constraint correctly predicts that there are no syllables that consist of a 

schwa only. However, the /ha/-Constraint is more specific since, for instance, 

/hat/ is impossible even though /at/ is a possible syllable. 

Another type of pansyllabic constraint in Dutch concerns the co-occurrence 

of two liquids in a syllable: the sequence /IVI/ is strange with long vowels, and 

the sequence /rVr/ is out when the vowel is short: 12 

In Van der Hulst ( 1984) some of the constraints on consonant clusters in ousels and codas are 

formulated in terms of minimal sonority distance requirements on adjacent consonants, /ec ( 19X8) 

argues that co-ocurrence constraints such as those formulated here should he preferred to the 

minimal sonority distance approach. 

1 

This was suggested to me hy Carlos (iussenhoven. 

' 2 One might wonder whether the constraint holds across syllable boundaries It that were true, 

Ihe constraint would be a morpheme structure constraint rather than a syllable structure constraint. 

However, we do find words like Lola 'id.' (Christian name) and rara 'guess what?' which suggest 

that it is a syllable structure constraint alter all.

3.5. CO-OCCURRENCE CONSTRAINTS 43 

(52) 1V1 a. lol /bl/ 'fun', lel /lel/ 'lobe', lal /lal/ 'to jabber', lul /Ivl/ 'penis' 

*lool /lol/, *leel /lel/, *laal /lal/, *luul /lyl/, *loel /lul/, *liel /Ml/ 

A), raar /rar/ 'strange', roer /rur/ 'rudder', Ruurd /ryrt/ 'id.' (proper 

name) 

*ror /ror/, *rer /rcr/, *rar /rar/, *rur /m/ 33 

A similar constraint is found for glides: we never find two /j/s or two /u/s in 

one syllable: *joej, *jooj, *jaaj, *wiew, *wuuw, with as exception weeuw /ueu/ 

'widow'. Since glides only occur after long vowels, the length contrast which 

played a role with respect to /r/ and /I/ is not relevant here. 

The constraint can be expressed as in (53). 

(53) Double Approximant Constraint *o 

Conditions: 

If X, is |+lat|, N is branching 

If Xi is |+cont, Cor], N is non-branching 

O R 

/\ 

N C 

Xi X, 

l+appr] [+appr] 

3.5.5. Sequential constraints 

As has been amply demonstrated above, many phonotactic restrictions should 

or can be seen as restrictions on possible syllables. In standard linear generative 

phonology, in which the notion 'syllable' did not play a role, phonotactic 

restrictions were expressed as morpheme structure conditions (MSCs), 

whereas Hooper (1972) argued that all MSCs could be replaced with syllable 

structure conditions (SSCs). Hooper's position is clearly too strong since 

languages also exhibit restrictions on combinations of segments within morphemes 

that are valid whether these segments belong to the same syllable or 

not (McCarthy 1986, Davis 1991). 

A possible example of an MSC is the restriction that velar nasals in Dutch 

can only be followed by the schwa, as observed by Kager and Zonneveld 

(1986): sequences like /urjal/ are ill-formed in contrast to /arjsl/. This is 

expressed by the following negative constraint: 

(54) * [+nas| |-cons] 

I 

Dors 

I 

Place 

" There happened to he a Dutch TV programme with the name Kur. hut this word is indeed 

experienced as an intentional coining.


However, such a cross-syllabic constraint can also be seen as a constraint on 

the possible prosodie words of Dutch. 

Dutch prosodie words are also subject to the constraint that the velar nasal 

does not occur after a long vowel; this constraint is expressed in (55). 

(55) Velar Nasal Constraint *N 

X X X 

f+nas 1 

L Dors) 

The glides /j/ and /u/ do not combine with all vowels: the /j/ only occurs 

after long back vowels, and the /u/ only after long front vowels: 

(56) a. roei /ruj/ 'row', mooi /moj/ 'beautiful', saai /saj/ 'dull' 

h. nieuw /niu/ 'new', uw /yu/ 'your', leeuw /leu/ 'lion' 

The constraints (57) exclude the non-occurring combinations: 

(57) Glide constraints 

a. *N b. *N 

I 

A 

\/ I 

X X X X X 

l+vocj [+back] Lab I 

[-back] Dors) 

Constraint (57a) excludes glides after short vowels. The only exceptions are 

loan-words such as mais /mqjs/ 'maize', boiler /bojlar/ 'id.', and de tail /detaj/ 

'id.', and the word hoi /hoj/ 'hello'. Constraint (51b) expresses that /j/ occurs 

after back vowels, and /v/ after front vowels. An additional constraint is that 

after /0/ we do not find glides at all. 

Glide constraint (57a) also shows that Dutch diphthongs should be analysed 

as complex nuclei rather than as sequences of a vowel and a consonantal glide, 

since otherwise we cannot explain the difference in status between, for instance, 

/ei/ and /oi/. It also explains why after real diphthongs (i.e., complex nuclei) all 

kinds of consonants can appear in the coda. For instance, we find eik, eip, and 

cil, whereas *aik , *aip, and *aim are impossible (cf. Booij 1989a). These 

sequential constraints are not confined to the syllable, but they do not necessarily 

have the morpheme as their domain: it can also be the prosodie word. 

There are also constraints on clusters of consonants independent from the 

tautosyllabic or heterosyllabic status of these clusters. One generalization, 

proposed for 1 English by Yip (1991) also appears to hold for Dutch: 

(58) Consonant Cluster Condition 

In consonant clusters, consonants may have at most one other articulator 

feature than Coronal.


For instance, Dutch allows for clusters of a non-coronal obstruent plus a 

coronal one: 

(59) pt heliko/?rer 'helicopter' 

ft refter 'dining-room' 

kt akte 'act' 

xt echter 'however' 

However, combinations of a dorsal consonant with a labial consonant are 

either excluded or extremely rare: 

(60) Excluded clusters: km, mk, xm, mx, pk, kp, fk, xf, fx, kf, xp, px 34 

Moreover, in clusters with one coronal obstruent, the coronal is preferably the 

second consonant: /pt/ rather than /tp/, and /kt/ rather than /tk/. This seems to 

be a universal tendency (Clements 1990: 313). Again, the constraint can also 

be seen as a constraint on prosodie words since such clusters do not arise 

through morpheme concatenation: the only cohering suffix that begins with a 

non-coronal consonant is the allomorph -pje of the diminutive suffix, but this 

morpheme only appears after a stem ending in /m/. 

The examples of obstruent clusters given here all concern voiceless obstruents. 

This is no coincidence: as observed by Zonneveld (1983) one hardly finds 

any cluster of voiced obstruents within morphemes. Those that do occur are 

mainly found in loan-words, such as labda (lubda) 'lambda' and budget 

Ibvdzjetl 'budget'. This suggests a negative MSC that forbids double-linked 

[+voice| in the lexical representations of morphemes: 

(61) * X X 

[+voice] 

This constraint will only apply to obstruents if other segments have no 

specification for [voice] at the lexical level. In this case, its status as MSC 

is certain since we do find clusters of voiced obstruents within prosodie words, 

for instance in the past-tense forms of verbs with stems ending in a voiced 

obstruent, for example, tob-de [tDbda] 'toiled' (sg.). 

Consonant clusters are also subject to another tendency, the Syllable Contact 

Law, which says that in an optimal heterosyllabic consonant cluster the 

first consonant is more sonorous than the second one. In other words, liquids 

and nasals should precede obstruents in optimal syllable contacts. For instance, 

/mp/ is preferred to /pm/, and /nt/ to /tn/ (Clements 1990). It is indeed a correct 

generalization about Dutch that the first C in a heterosyllabic CC is a liquid or 

nasal, and the second one an obstruent. It is not an absolute restriction on 

pronounceability, though. Clusters with two obstruents occur, subject to 

14 Exceptions are loan-words like pigmee Ipixme] 'pygmee', Afghaan (afxan] 'Afghan', 

drachme (druxrna] 'drachma', the word imker [imkar] 'bee keeper', and the brand name Agfa 

[uxfa|. Nasals before another obstruent do not occur in the list of excluded clusters because they 

agree in place of articulation with the following consonant.

46 TUK PROSODIC STRUCTURE OF WORDS 

certain restrictions as given above, and even clusters in which an obstruent 

precedes a sonorant consonant, but these latter clusters are indeed rare. Such 

words are often loan-words, such as acne /ukne/ 'id.' and drachme /druxma/ 

'drachma', and we also find them in family names like Ahma and Postma 

which historically derive from morphologically complex words. 

The Syllable Contact Law is particularly strong before schwas. For instance, 

a form like ordner /ordnar/ 'file' is very marked compared with ander /andDr/ 

'other', and the effect is that a consonant cluster before a schwa often has the 

form of a well-formed syllable coda. 35 

Another constraint on the segmental composition of Dutch morphemes (a 

constraint observed for English by Davis (1991)) is that in the sequence sCVC 

identical non-coronal obstruents are avoided. It does not play a role whether 

the two identical Cs belong to the same syllable or not. The following 

examples illustrate this constraint: 

(62) CVC versus sCVC 

poep 'shit' *spoep 

toet 'face' stoet 'procession' 

koek 'cake' *skoek 

gigant 'giant' *sgigant 

The constraint may be of a more general nature since the same restriction holds 

for other well-formed onset clusters CQ followed by a vowel + Q (where 

subscript , indicates identity). 

(63) mam 'mum' *smam 

non 'nun' *snon, *knon 

lol 'fun' *slol, *plol, *klol, *blol 

raar 'strange' *praar, *kraar, *braar 

Such configurations are also strange when they are heterosyllabic: *smama, 

*knona, *krara, etc. Therefore, the following constraint holds: 

(64) Identical Consonant Constraint 

* CC; -V-Cj 

Morpheme structure conditions may also have prosodie dimensions. For 

instance, above we saw that in lexical morphemes three-positional rhymes 

tend to end in an obstruent consonant. Another tendency is that short vowels do 

not appear so easily before obstruent + liquid clusters, as observed by Van der 

Hulst (1984). For instance we do not have /zebra/ along with zebra /zebra/ 

'id.'. There are, however, exceptions to this constraint, as in the words Accru 

/ukra/ 'id.', agglomeratie /uvlomorati/ 'agglomerate', hopla /hopla/ 'upsydaisy', 

Biafra /biafra/ 'id.', ACLO /aklo/ (acronym), Oslo /Dslo/ 'id.', Islam 

" In Kager and Zonneveld (19S6), this observation about the nature of consonant clusters 

before schwa is adduced as evidence lor the hypothesis lhat word final schwas are extra 

metrical. See Kager (1989: 212-14) for criticism of this hypothesis

3-6. RELATION BETWEEN MORPHOLOGICAL AND PROSODIC STRUCTURE 47 

/islam/ 'id.', and so on. So, clearly, it is not an absolute pronounceability 

constraint on sequences of segments. 

3.6. THE RELATION BETWEEN MORPHOLOGICAL AND PROSODIC 

STRUCTURE 

Lexical morphemes of Dutch appear to be subject to the following constraints: 

they have at least one full vowel; 

they do not begin with a schwa. 

This can be interpreted as follows: a lexical morpheme (i.e. a morpheme 

belonging to one of the lexical categories Noun, Verb, Adjective, Adverb, or 

Preposition 36 ) must be able to form a well-formed prosodie word of its own, 

where a prosodie word is defined as a sequence of one or more well-formed 

syllables, at least one of which contains a full vowel. It is necessary to have at 

least one full vowel in a prosodie word because otherwise there is no syllable 

available for the location of main stress—schwa cannot bear stress—in words 

of major lexical categories (N, V, A). That is, a lexical morpheme minimally 

consists of a foot. 

As pointed out in Section 3.5.4, a syllable that consists of a schwa only is illformed 

because a syllable must be specified for Place. However, a prosodie 

word cannot begin with a schwa, even if it is followed by a tautosyllabic 

consonant, that is, even though the syllable with the schwa does dominate 

Place. Observe that function words with the structure /a/ + consonant do exist, 

for example, een /an/ 'a'. This implies that such function words do not form 

prosodie words of their own. 

Thus, the segmental string corresponding to a lexical morpheme will be 

dominated by a prosodie word node. The same applies to those suffixes that 

form an independent domain of syllabification such as -achtig '-like': since 

this string contains a full vowel and does not begin with a schwa, it will be 

dominated by a prosodie word node. On the other hand, function words such as 

determiners and pronouns can begin with a schwa, and/or have schwa as their 

only vowel, in particular in their weak or clitic forms. This also applies to 

weak (clitic) forms of some adverbs, and to prefixes and suffixes, all nonlexical 

morphemes. A survey of the phonological forms of these grammatical 

morphemes is given in Table 3.2. 

In so far as they do not contain a full vowel, these morphemes will not be 

dominated by a prosodie word node after syllabification has applied. In the 

case of suffixes, which normally do not form a syllabification domain of their 

"' Prepositions are ambiguous: Ihey tonn a class of words thai cannot be extended by morphological 

means like the other categories, but they do project into phrases, i.e. PPs. Note that Dutch 

Ps do form prosodie words (except the preposition te /la/ 'at'): they always contain at least one full 

vowel, and also have appendices, as in naaxl /nast/ 'besides'.


TABLE 3.2. The phonological forms of grammatical morphemes 

Determiners 

de 'the' /da/ (-1- def. art., 

masc,. fern., sg.; def. 

art., pi.) 

het /at/ 'the' (strong form 

[het]) (def. art., n., sg.) 

een /an/ 'a' (indef. art., 

sg.) 

Pronouns 

ik /ak/ 'I' (strong form IM) 

me /ma/ 'me' (strong form 

mij /mei/) 

m'n /man/ 'my' (strong 

form mijn /mein/) 

je /ja/ 'you' (subj.) (strong 

form jij /jei/) 

je /ja/ 'you' (obj.) (strong 

form jou /j DU/) 

je /ja/ 'your' (strong form 

jouw /jou/) 

hem /am/ 'him' (strong 

form /hem/) 

haar /ar/ 'her' (strong form 

/har/) 

het /at/ 'it' (strong form 

/het/) 

d'r/dar/ 'her' (poss. pron.) 

(strong form haar /har/) 

er /ar/ 'it' (before 

postpositions) (strong form 

/er/) 

we /ua/ 'we' (strong form 

wij /uei/) 

ze /za/ 'they' (strong form 

zij /zei/) 

ze /za/ 'them' (strong form 

hen /hen/) 

z'n/zan/ 'his' (strong form 

zy'/i /zein/) 

Adverbs 

er /ar/ 'there' (strong form 

/Er/) 

daar /dar/ 'there' (strong 

form /dar/) 

eens /as/ 'once' (strong 

form /ens/) 

Preposition 

te /ta/ 'at' 

Prefixes 

be- /ba/, ge- /ya/, ver- 

/var/, er- /ar/ 37 

Suffixes 

-e /a/ (f. suff.) or 

substantivizing suff. or 

inflection of attr. 

adjectives) 

-el /al/ (de-nominal 

verbalizing suff.) 

-(e)lijk /(a)lak/ '-able' 

-en /an/ (plural suff. (verbs 

and nouns)) 

-end /and/ (près. part. 

suff.) 

-(e)nis /anls/ '-ness' 

-er /ar/ '-er' 

-erd /ard/ (de-adjectival 

nominal suff. for 

personal names, pej.) 

-erig /aray/ '-ish' 

-erij /arei/ '-ing' 

-ig /ay/ '-ish' 

-sel /sal/ (de-verbal obj. 

names) 

-ster /star/ (f. suff.) 

-te /ta/ '-ness' (past t.) 

-tje /tja/ (dimin. (with 

allomorphs -/ja/, -/atja/, 

-/pja/, -/kja/)) 

Circumfixes 

ge-en /ya-an/ (past part. 

circumfix for strong 

verbs) 

ge-t/d /ya-t/d/ (past part. 

for weak verbs) 

ge-te /ya-ta/ 'de-nominal 

collective' 

own, this does not mean that the suffixal string is not dominated by a prosodie 

word node, because they fuse with the preceding segmental material. (Note 

that for suffixes that contain only schwas it is predictable that they do not form 

prosodie words of their own!) In the case of prefixes and (the weak forms of) 

determiners, pronouns, and adverbs (all independent domains of syllabification), 

their segmental material will be dominated by syllable nodes, but not by 

prosodie word nodes. As we will see, they therefore have to be prosodically 

integrated at the post-syntactic level of prosodie structure. This of course also 

" Ver- and er- also have phonetic forms with the full vowel [e], at least for some speakers. The 

prefix ont- does not have a variant with schwa, presumably because after a schwa a sequence of (wo 

consonants is not possible, and because vowel reduction does not occur in vowel-initial syllables 

(except /c/ before /r/).


applies to clitic forms like k T and t 'it' which consist of a consonant only, 

and therefore do not even get a syllable node. 

Even if a morpheme forms an independent domain of syllabification this 

does not guarantee that it will also be a prosodie word. For instance, the 

prefixes and clitic words discussed here are independent domains of syllabification, 

but not prosodie words. 

Note that this analysis also explains why the /n/ of the strong form of the 

adverb eens /ens/ 'once' has to be dropped in its weak form [as): once it has 

only a schwa, this word cannot be a prosodie word any more, and the sequence 

/ns/ is no longer licensed because there is no Appendix position that can 

accommodate a second consonant in the syllable after the schwa which itself 

occupies two positions. 

It is a basic insight of prosodie phonology that the prosodie structure of a 

word is not necessarily isomorphic to its morphological structure. For instance, 

the word hoeden 'hats' has the following two non-isomorphic structures as 

shown in (65). 

(65) Word 

Morpheme Morpheme 

3 n 

As pointed out above, most Dutch suffixes prosodify with the stem to which 

they are attached, except for a special class of suffixes which form prosodie 

words of their own. The first type of suffix may be called 'cohering suffix', the 

second type 'non-cohering suffix' (cf. Booij 1982a, 1983). Each constituent of 

a compound forms a prosodie word of its own, as was shown above for the 

compound handappel 'eating apple', where syllable-final devoicing applied to 

the word-internal /d/. This assumption is also supported by the way in which 

Prevocalic Schwa Deletion applies. This phonological rule of Dutch deletes 

schwas before vowels, thus avoiding hiatus (cf. Section 4.2.3). It applies 

within the domain of a word, but not across a prefix boundary, across the 

internal boundaries of compounds, or before the vowel-initial suffix -aclitii^. 

This restriction on the domain of application of schwa deletion follows 

straightforwardly by the assumption that this rule has the prosodie word as 

its domain: 

(66) a. Deletion in: 

kaden /kada + an/ 'quays' (kadanj 

/.ydig //eido + ay/ 'silky' l/eidsx] 

Romein /roma + ein/ 'Roman'/ [romein]


b. No deletion in: 

zijdeachtig /zeida + uxtay/ 'silky' [zcidaaxtax] 

zijdeinkomsten /zeida + igkomstan/ 'silk revenues' [zddairjknmstan) 

beantwoorden /ba + antuordan/ 'to reply' [bauntuoirdan] 

A complex word may lose its semantic transparency, however, and thus it can 

lose its internal morphological structure. An example of this is provided by 

aardappel 'potato'. Since a potato is no longer considered to be a kind of 

apple, the word aardappel appears to be prosodified as one word: 

(67) Stage I: morphological structure 38 [ |aard| N (appel] N | N 

prosodie structure ( (aard) c )J (ap) a (pel) 0 ) œ 

phonetic form |a:rtupall 

Stage II: morphological structure (aardappel) 

prosodie structure ( (aar) 0 (dap) a (pel) CT ) M 

phonetic form [airdupol] 

The same development took place in the compound tandarts |tundurts| 

'dentist' (lit. 'tooth-doctor', with the original structure land-arts), and in the 

word reuzachtix |r0/.uxtox) 'fantastic' (lit. 'giant-like') with the suffix -achtln. 

In the case of reuzachtig, the literal interpretation 'giant-like' also exists, and 

then we get the phonetic form [r0saxtax], as expected. In the prefixed verb 

herinneren 'to remember' (lit. to 're-internalize') with the prefix her- 're-', the 

effect of losing its morphological transparency is not to be seen in the voiced 

realization of word-internal obstruents, but in the fact that the /r/ of the prefix 

always forms the onset of the next syllable: (her) a (rin) CT (ner) 0 , not 

(her) CT (in) CT (ner) c . The prefix re- may also lose its transparency, as in reageren 

'to react' which requires obligatory insertion of a glide between the two 

adjacent vowels, that suggests that it has become one prosodie word. 

That affixes may function as independent prosodie words correctly predicts 

that they are subject to the rules for word stress, which have the prosodie word 

as their domain. For instance, in the prefix over- /ovar/ 'id.' and the suffix 

-aclitig /uxtav/ '-like', the first syllable receives stress because the second one 

contains a schwa, and is therefore unstressable. completely in accordance with 

the Dutch word-stress rule. 19 

The non-isomorphy of morphological and prosodie structure is also demonstrated 

by the phenomenon of conjunction reduction ('gapping') in complex 

words (Booij 1985a): a part of a word can be omitted under identity with a part 

of another word, for instance (the italic part is deleted): 

* The /a/ of aar Je /ardD/ 'earth' is deleted although it does not stand before a vowel in the same 

prosodie word. This schwa deletion is due to a lexically governed rule of schwa deletion that also 

deletes schwas before consonants, as in f»nli' + si Imp -» hnndxi hn/i 'message'. 

19 

There is a cross-linguistic tendency for prefixes to behave as independent prosodie words, or 

as independent domains ol syllabification; cf. Booij and Kubach (1984) on English, Polish, and 

Dutch, and Cohn (1989) on Indonesian.


(68) | [land||/w«H>] ] en | [tuin||bouw] j 'agri- and horticulture' 

( [rood\achtix\ of'[ (groenJachtig] 'reddish or greenish' 

The crucial condition is that the gapped part forms one or more prosodie 

words, and that it is adjacent to the conjunction. 40 It is not enough that there 

is morphological identity. For instance, the suffix -ig, a cohering suffix with 

the same meaning as -achtig cannot be gapped because it does not form a 

prosodie word of its own: 

(69) *[ [rod] -| of [ Igroenjigl 'reddish or greenish' 

On the other hand, the cohering suffix -er and the cohering suffix -s can be 

gapped together with the part of the segmental string to which they prosodically 

cohere. That is, houwers can be gapped: although it is not a morphological 

constituent, it is a prosodie word: 

(70) morphology: [ | [ [land][/wMw| |c/ |.v| en [ [ [ [tuin](bouw] Jerjs] 

prosody: (Idnâ^houwers)^ (en),,, (tuin^bouwers)^ 

Thus, land- en tuinbouwers 'agriculturers and horticulturers' is a well-formed 

expression. 

Gapping of the right of two identical constituents is also possible. In that 

case, the right prosodie word must be adjacent to the conjunction, as in 

herenjassen en -schoenen 'men's coats and -shoes'. Rightward gapping also 

proves that prefixes do not form independent prosodie words, although they 

form independent domains of syllabification: gapping of the second of two 

identical prefixes is impossible: 

(71) *befietsen of—lopen 'to cycle on or to walk on' 

The two constituents of a Dutch compound often have either an /s/ or an /s/ 

in between them, the so-called linking phonemes. The question when they 

appear is a morphological issue (cf. Van den Toorn 1982). What is relevant 

here is that, although the linking phonemes do not form part of one of the 

constituents from the morphological point of view, they fuse prosodically with 

the preceding prosodie word, as can be seen from the way in which they 

syllabify and from the fact that underlyingly voiced obstruents remain voiced 

before the linking schwa. For instance, hondebrood 'dog's food' has the 

following morphological and prosodie structures: 

(72) morphological structure: [ |hDnd] N 3[brod| N ] N 

prosodie structure: ( (h3n) 0 (d9) 0 )a,( (brod) a ) w 

phonetic form: 

[hDndabrot] 

'" A more correct slatemem of this latter condition is that the gapped constituent should occur at 

the periphery of a phonological phrase, since gapping also applies before a preposition as in Hij 

verwisselde de dagblad-voor de wetkblac(journalùtiek '(lit.) He exchanged the daily journalism for 

the weekly journalism' where voor is a preposition that begins a new phonological phrase 

according to the rules proposed in Nespor and Vogel (1986). Cf. Booij (


We now also make the right prediction with respect to the behaviour of such 

compounds under conjunction reduction: since honde is the first prosodie 

word, that will be the remnant after gapping: 

(73) honde- en kattebrood 'dog's food and cat's food' 

In sum, we have seen that the mapping of morphological structure on to 

prosodie structure is systematic, but not isomorphic. This insight will play an 

important role in the analysis of phonological processes in the following 

chapters.

WORD PHONOLOGY 

4.1. INTRODUCTION: THE ORGANIZATION OF THE GRAMMAR 

A traditional distinction in the description of the phonological processes is that 

between word phonology and sentence phonology (cf. Booij 1981/>). Rules of 

word phonology apply within words, whereas rules of sentence phonology, 

which are characteristic of connected speech, (also) apply in larger, phrasal 

LEXICON 

PHONOLOGY 

List of underived words 

Rules 

of 

Morphological rules 

word 

phonology 

Words 

Syntax 

Rules of connected speech 

Phonetic form of sentences 

FIG. 4.1. The organization of the grammar

54 WORD PHONOLOGY 

domains. Clear cases of rules of word phonology are the rules for word stress 

(Chapter 5), and the so-called morpholexical rules,' rules that are conditioned 

by specific morphemes or only apply to restricted sets of lexical items. For 

instance, the rule of Learned Vowel Backing that was introduced in Section 

2.5.2 is typically a morpholexical rule because it only backs mid vowels in 

[-native] words. The question now is, where in the grammar the rules of 

phonology must be assumed to apply, and how they interact with other 

components of the grammar. 

Suppose that we assume the following simple principle as to when 

phonological rules apply: 

(1) Apply a phonological rule whenever its structural description is met. 

This implies that a phonological rule will apply as soon as the relevant 

environment is available. Since words are available in the lexicon, words 

that meet the structural description of a rule will, in principle, undergo that 

rule in the lexicon. Rules of connected speech that apply within phrases, on the 

other hand, can only apply after the creation of phrases in syntax. Thus, the 

organization of the grammar emerges as in Fig. 4.1. 

The boxes around the two sets of phonological rules do not indicate 

completely separate subcomponents of the phonological component (some 

rules apply both in the lexicon and postlexically), but are meant to indicate 

that only a subset of phonological rules is able to interact with morphological 

operations, or can apply before syntax. To make this more concrete, let us look 

at the derivation of the word elitarisme 'elitism': 

(2) 1st cycle: [elit3] N 

2nd cycle: 

morphology 

[ [elit3] N e:r] A 

phonology: (Prevocalic Schwa Deletion) 0 

3rd cycle: 

morphology 

[ [ [elit] N e:r] A izm3] N 

phonology: Learned Backing 

a 

Lexical phonetic form: 

[elitanzma] 

This cyclic derivation of the word elitarisme follows from the assumption 

that morphological and phonological rules apply in tandem, the core of the 

theory of Lexical Phonology (Kiparsky 1982, Booij 198lè, Booij and 

Rubach 1987, 1991). By having phonological and morphological rules interspersed, 

it is correctly predicted that morphological rules may apply to 

derived phonological representations (Anderson 1979, Booij 1988b, Hargus 

1988: 59). Since the rules of Prevocalic Schwa Deletion and Learned Backing 

are applicable at the lexical level, they will apply in conformity with 

principle (1). 

However, in Chapter 3 it was argued that the rule of Prevocalic Schwa 

' This term is taken from Anderson (1974).

4-1. INTRODUCTION 55 

Deletion applies within the domain of the prosodie word. This implies that 

information about the prosodie structure of words must also be generated at 

the lexical level. This also follows from principle (1), since prosodification 

rules are phonological rules. Moreover, it makes the correct prediction that 

morphological rules may have to refer to prosodie properties of their input 

words (cf. Booij 1988/?, 1992r, Booij and Lieber 1993). On the other hand, 

the syllabification of the segmental string of a word may change due to 

morphological and phonological processes. For instance, the final schwa of 

elite /elito/ is deleted before the next vowel-initial suffix. Since the schwa 

heads a syllable, this syllable node gets lost. Also, after the addition of -ixnie 

the coda of the last syllable, /r/, of elitair will become the onset of the first 

syllable of -isme. Therefore, I will assume the following (re)syllabification 

procedure (steps (a) and (b) precede (c), but are only applicable after a 

morphological or phonological rule has applied) (cf. Itô 1989, Rubach and 

Booij I990a): 

(3) Continuous syllabification 

a. Coda Erasure in the word-final syllable after attachment of a cohering 

suffix; 2 

h. erasure of the o-node and the prosodie structure dominated by it if the 

head of o (the nucleus) is deleted; 

c. matching of syllable and prosodie word templates. 

This procedure will again be illustrated by the derivation of the word 

elitarisme as given in (4). 

Let us now look at the rule of syllable-final devoicing of obstruents. If it 

applied as early as possible, we would get wrong phonetic forms such as 

Ihutan) for hoeden: on the first cycle the /d/ of the morpheme /hud/ is 

syllable-final, and hence it would devoice, thus resulting in the wrong phonetic 

form. In other words, this rule should apply after morphology: the plural 

suffix -en will induce resyllabification of the /d/ as onset of the next syllable, 

and hence it remains voiced, as required (Booij 1981«: 119). Such rules are 

called word-level rules since they apply after morphology, but still at the 

lexical level. That is, we have to allow for rules of word phonology to apply 

;it a later level, the word level. This means that there are three levels at which 

rules can apply (Kiparsky 1985, Booij and Rubach 1987). The first level will 

be called the cyclic level because at that level morphology and phonology 

apply cyclically, as shown. 

"' Instead of Coda leisure. Levin ( 1985) anil Hall ( 1992) assume thai only Ihe last consonant of 

a ccxla can ho rcs>llahilieil. through the structure-changing application of Ihe universal CV-rule. 

However, there arc case«, in which more lhan one consonant has to he rosyllabilieil. as in angstig 

(ui|),,(sl,ixi„ 'learliil' derived Irom uiifixl (urjst) 0 'leur'. Also, in liandjc (hun) 0 (IJ3) 0 'little hand' 

the AI/ is lust sylhbiliod as coda of the lirst syllable because Dutch docs not allow for dj- onsets 

This causes final devoicing. and subsequent resyllabification because tj- is a possible onset of 

Dutch. Note that icsyllahiliealion takes place here even though the onset is already filled by the 

consonant /j/. ('I. also Ruhach and Booij (1990«).


(4) 1st cycle: 

Syllabification 

0) 

o a a 

e li t9 

2nd cycle: 

-a/r-affixation 

+ prosodification 

Prevocalic Schwa Deletion 

Resyllabification 

(deletion of third a) I 

3rd cycle: 

-isme-affixation 

+ prosodification 

(including Coda Erasure) 

Learned Vowel Backing 

(0 

a a a a 

e li t3 e:r 

co 

/K 

a a a 

e li te:r 

o a o 

te: riz ma 

Lexical phonetic form 

[elitanzma] 

(5) Levels of application 

Cyclic level 

Word level 

Postlexical level 

By locating syllable-final devoicing at the word level, we also predict that 

postlexical resyllabification of consonants across word boundaries will not 

affect the voice properties of word-final obstruents. For instance, in the phrase 

een hond aan de lijn 'a dog on the leash' the sequence of words hond aan may 

be syllabified as (hDn) 0 (tan) 0 , with the word-final obstruent of hond /hond/ 

forming an onset. However, the underlying /d/ is devoiced, so the phonetic 

form [hondan] is impossible. 

For other phonological rules of Dutch that apply obligatorily within the 

domain of the prosodie word there is no crucial evidence that they are cyclic 

rules or word level rules. This applies to the phonological (P-rules) discussed 

in Section 4.2. If nothing is said, they will apply as early as possible, in 

conformity with principle (1), hence at the cyclic level. In Chapter 5 it is 

shown that the Main Stress Rule for the location of main stress on words is

4-2. PHONOLOGICAL RULES 57 

indeed crucially cyclic. Consequently, syllabification must also be cyclic, since 

stress is a property of syllables. 

The cyclic application of rules is subject to the Strict Cycle Condition 

(Kiparsky 1985): they only apply in derived environments, unless they add 

information (i.e. are not structure-changing); in the latter case they may also 

apply in underived environments. This makes the right predictions for the 

phonological rules discussed in Section 4.2. since they either apply in derived 

environments only (for instance, Prevocalic Schwa Deletion), or they add 

information, and hence apply in underived environments as well, as required 

(e.g. Nasal Assimilation). Final Devoicing, as we have seen, is a word-level 

rule. This follows from the fact that it is only conditioned by syllable structure, 

and syllable structure does not count as derived environment (Kiparsky 1985). 

Hence, since it is a structure-changing rule, it will be blocked from application 

at the cyclic level, as required, and apply as soon as possible after the cyclic 

level, i.e., at the word level. 

Not all phonological rules that apply within words are obligatory rules. 

There are also rules that optionally apply, such as Schwa Epenthesis, a rule 

that breaks up coda consonant clusters (as in harp /harp/ 'id.' pronounced as 

[hurop)). Such rules are often stylistically governed in that they are characteristic 

of less monitored forms of speaking (casual speech). They will be dealt 

with in Chapter 6, and will be argued to belong to the postlexical level. 

In this chapter I will discuss phonological processes that apply obligatorily 

within words, with the exception of word stress which is dealt with in Chapter 

5. Chapter 6 analyses optional rules that apply within words, and Chapter 7 

deals with rules of connected speech above the word, which are of course 

located at the postlexical level. 

4.2. PHONOLOGICAL RULES 

Phonological alternations differ as to the nature of their conditioning environments. 

Some alternations, such as that between voiced and voiceless obstruents 

in morpheme-final position are conditioned by phonological conditions only, 

in this example the position of the obstruent in the syllable. Such rules are 

called 'automatic phonological rules', phonological rules, or P-rules. Other 

alternations are also conditioned by non-phonological factors. For instance, the 

alternations in Dutch diminutive nouns are conditioned crucially by the presence 

of the diminutive suffix, and these alternations do not occur in other 

types of complex word. Rules which only apply to words with specific 

morphemes or to specific sets of items that are not characterizable in purely 

phonological terms, are called morpholexical rules or MP-ruIes (Anderson 

1974). This section will deal with P-rules, Sections 4.3 to 4.5 with MP-rules. 

P-rules that apply obligatorily within the domain of the prosodie word form

58 WORD PHONOUXiY 

part of the lexical phonotactics of a language, i.e., they define constraints on 

the notion 'possible prosodie word of language L'. In this respect they differ 

from MP-rules, which do not have such a phonotactic role. For instance, there 

is an MP-rule that inserts a schwa before the diminutive suffix after a short 

vowel followed by a sonorant consonant, as in kannetje /kan+tja/ 'small jug' 

with the phonetic form | kiin.it ja|. Nevertheless, this does not mean that /n/ and 

A/ cannot be adjacent in this phonological configuration, as is proven by the 

existence of another diminutive, kantje /kunt+tja/ 'piece of lace' with the 

phonetic form |kuntja|. This shows that the MP-rule of Schwa-insertion 

does not have a phonotactic role, and why it is important to distinguish these 

two kinds of rule. Related to this, MP-rules can have lexical exceptions, as 

illustrated below, which is understandable because they do not express absolute 

constraints on possible sequences of sounds within a prosodie word, 

whereas P-rules do not have exceptions at the lexical level. 

4.2.1. Devoicinfi and voice asaimilaiion 

Dutch has the following two P-rules of voice assimilation, traditionally called 

Progressive Assimilation and Regressive Assimilation respectively. 3 

The rule of Progressive Assimilation says that a fricative is devoiced after a 

voiceless obstruent: 

(6) Progressive Assimilation 

opvallend 'remarkable' /pv/ [pf| 

stoepzout 'pavement salt /p/./ [ps| 

o/>#raving 'excavation' /pv/ |px| 

zoiUvat 'salt tub' 

/tv/ |tf| 

zou/ruur 'hydrochloric acid' A// |ts| 

straa^'oot 'gutter' /ty/ [tx| 

da^venster 'dormer' /kv/ [kf | 

dakgoot 'gutter' 

/ky/ [kx| 

afval 'trash' 

/fv/ |f:| 

a/zuigen 'to extract' 

/fz/ |fs| 

a/âng 'failure' /fy/ |fx| 

a.vvat 'ashbin' /sv/ |sf | 

wa.vrak 'laundry bag' 

/s/,/ [s:| 

wasgoed 'laundry' 

/sy/ |sx| 

per/ivogel 'unlucky person' /xv/ [xf] 

lar/izak 'laughing machine' /x// |xs| 

lardas'laughing gas' /xy/ |x:| 

In three examples we see a geminate fricative in the phonetic form. These 

geminates are usually shortened through Degemination (Section 4.2.4). 

1 Cf. Tops (1974), Van der Hulst (1980), Booij (1981«), and /oiiiievekl ( 19X3).


All these examples are complex words, consisting of at least two prosodie 

words, that is, we may call them prosodie compounds. Note that obstruent 

clusters of this type are very rare within morphemes: they do not form optimal 

syllable contacts (cf. Section 3.5.5). Nevertheless, there is evidence that this 

devoicing rule also applies within prosodie words, as in advies |utfis] 'advice' 

and in acronyms like ABVA /ubva/ [upfa] and AKZO /ukzo/ [uksoj. Phonological 

rules like Progressive Assimilation, which apply obligatorily within 

prosodie words and compounds, become optional in larger domains: for 

instance, in phonological phrases. 

The rule of Progressive Assimilation is fed by Final Devoicing since 

obstruents that are voiceless due to the latter rule also devoice when they 

follow underlyingly voiced fricatives: 

(7) handzaam 'handy' /d/./ [ts] 

hebzucht 'avarice' /bz/ [ps] 

graa/zucht 'digging urge' /vz/ |fs| 

The second rule of voice assimilation is that of Regressive Assimilation. It 

says that voiceless obstruents become voiced before a following voiced stop 

(i.e. /b/ or /d/): 

(8) Regressive Assimilation 

klapband 'flat tyre' 

/pb/ [b:] 

opdruk 'imprint' 

/pd/ [bd| 

eefbaar 'edible' /tb/ |db| 

potdicht 'tight' 

/td/ [d:] 

kooMoek 'cookery book' /kb/ |gb| 

/.akdoek 'handkerchief' /kd/ [gd| 

a/fcellen 'to ring off' /fb/ |vb| 

stq/doek 'duster' 

/fd/ [vd] 

ka.vboek 'cash book' /sb/ [zb] 

misdaad 'crime' 

/sd/ [zd] 

lar/;/)ui 'fit of laughter' /xb/ [yb] 

\achduif 'laugher' 

/xd/ [yd] 

In this case, the effects of Final Devoicing are undone by voice assimilation. 

For instance, in niesbui 'fit of sneezing', the s represents an underlying /z/ 

(witness niezen 'to sneeze'), which can be devoiced in coda position, but is 

then revoiced before the next /b/. As was the case with Progressive Assimilation, 

the rule applies both within and across prosodie word boundaries. The 

effects within prosodie words may be seen in words like asbest [azbest] 

'asbestos' and Lesbos [lezbDs] 'id.'. 

The two rules of voice assimilation can be formulated as in (9) and (10). 

Since PVA is a more specific rule than RVA, PVA will apply first in 

accordance with the Elsewhere Principle (Kiparsky 1982) which says that 

when two rules compete, the more specific one gets priority, and blocks 

application of the other one. Once PVA has applied, for instance to zoutvat

6o 


(9) Progressive Voice Assimilation (PVA) 

[—son] [— son] 

[+cont] 

[ — voice] [+voice] 

(10) Regressive Voice Assimilation (RVA) 

[-son] [-son] 

[-voice] 

[+voice] 

'salt tub', thereby devoicing the /v/, the application of RVA is no longer 

possible anyway, since the second obstruent no longer bears the feature 

[+voice]. 

No ordering of PVA with respect to Final Devoicing is necessary for words 

like hebzucht 'avarice', with the underlying cluster of obstruents /bz/: the 

application of Final Devoicing will feed PVA: first the /b/ turns into [p], 

and this [p] changes the following /z/ into [s]. 

When at the word level the rule of Final Devoicing applies to a word like 

handboei /hundbuj/ 'handcuff', with underlyingly voiced obstruents, the /d/ 

will be devoiced, and then revoiced by RVA. Then, Final Devoicing cannot 

reapply, because RVA has created a structure in which the feature [+voice] is 

doubly linked, to both obstruents, and therefore the Uniform Applicability 

Condition (cf. Section 3.4.2) will block another application of Final Devoicing, 

which has the form given in (II). 4 

(11) Final Devoicing 

C 

I 

X 

I 

[-son] 

[+voice] 

4 Obstruents in Appendix position also devoice. We might therefore assume lhat, at the word 

level, appendix consonants are Chomsky-adjoined to the Coda. Another solution is proposed by 

Lombardi (1991 ): the feature [+voice] is only licensed in Onset position, and therefore delinked by 

convention in all other positions, i.e. Coda and Appendix. This analysis makes Chomsky-adjunction 

of appendices to codas superfluous. Gussmann (1992: 42). in his analysis of Polish voice 

assimilation, assumes the same restriction as to the licensing of |+voice| for Polish.


The rules of voice assimilation do not apply only within words, but also 

across (grammatical) word boundaries, that is, in phrases. That is, the rules 

apply postsyntactically because their domain of application, the Intonational 

Phrase, will only be available postsyntactically. However, in larger domains 

the application of such rules is not completely obligatory and can be suppressed. 

5 I already mention them here, in the chapter on word phonology, 

because they apply obligatorily within the smaller domain of the word. 

Another complication as to the phrasal application of these rules is that the 

demonstratives die 'that' and Jat 'that' can be optionally subject to Progressive 

Voice Assimilation. Thus, we find the following phonetic realizations for 

dat and die in the sentence Is dat juist, op die manier? 'Is that OK, in that 

manner?' (Van Haeringen 1955): 

(12) [izdat] or [istat], [Dbdi] or [opti] 

In Section 8.3. it is shown that these facts follow from the optional encliticization 

of these function words to a preceding word. 

Verbal inflectional suffixes exhibit a different behaviour with respect to 

voice assimilation, which requires detailed discussion. 

The Dutch past-tense suffix for regular ('weak') verbs is /ta/ or /da/: /ta/ 

after stem-final underlyingly voiceless obstruents, and /da/ elsewhere, i.e., 

after voiced obstruents, sonorant consonants and vowels: 

(13) a. stem 

a. klap /klup/ 

zet /zet/ 

haak /hak/ 

maf /maf/ 

vis /vis/ 

lach /lux/ 

b. krab /krub/ 

red /red/ 

kloof /klov/ 

raas /raz/ 

1 leg /lev/ 

c. roem /rum/ 

zoen /zun/ 

meng /men/ 

roer /rur/ 

rol /rol/ 

aai /aj/ 

ski /ski/ 

'to applaud' 

'to put' 

'to crochet' 

'to sleep' 

'to fish' 

'to laugh' 

'to scratch' 

'to save' 

'to split' 

'to rage' 

'to lay' 

'to praise' 

'to kiss' 

'to mix' 

'to stir' 

'to roll' 

'to caress' 

'to ski' 

klapte 

zette 

haakte 

mafte 

viste 

lachte 

krabde 

redde 

kloofde 

raasde 

legde 

roemde 

zoende 

mengde 

roerde 

rolde 

aaide 

skiede 

past tense 

[klupta] 

[zeta] 

[hakto] 

[mafta] 

[vista] 

(luxta) 

[krubda] 

[reda] 

(klovdaj 

[razda] 

[Icyda] 

[rumda] 

[zunda] 

[merjda] 

|ru:rda] 

[rr>ldD] 

|ajda| 

[skida] 

According lo Cammenga and Van Reenen (1980) RVA of /s/ is also optional in compounds 

like misdaad 'crime' and niesbui 'fit of sneezing' which consist of two prosodie words, and in 

which therefore the two obstruents are separated by prosodie word boundaries.


The past-tense forms given here are the singular forms. The plural forms are 

created by adding the verbal plural suffix -en /on/. The final schwa of the pasttense 

suffix deletes before the plural suffix due to the rule of Prevocalic Schwa 

Deletion, and thus the surface forms of the past tense plural forms consist of 

the verbal stem plus /dan/ or /tan/ (e.g klapten, krabden, roemden). 

The first question to be answered is: why do not we just say that there are 

two competing past-tense suffixes, which appear in complementary environments?'' 

The reason is that in such an analysis we would not account for the 

fact that the two suffixes are almost identical, and that, moreover, the variation 

is not purely arbitrary from a phonological point of view: the variant with the 

voiceless consonant shows up after voiceless sounds. It would be easy to 

describe a situation in which -te appeared after voiced segments, and -de after 

voiceless segments. Therefore, a phonological analysis is called for. I will 

assume an underlying form /Do/, where the initial coronal stop is unspecified 

for [voice], and a P-rule that spreads the Laryngeal node from the preceding 

segment to the underspecified coronal stop (the '•' indicates the root): 

(14) Laryngeal Spreading 

• | son] 

Laryngeal 

Since the laryngeal node dominates either [+voice] or [-voice], the past 

tense suffix is either [da] or [ta]. 7 

The question now arises why syllable-final voiced obstruents in past-tense 

forms such as krabde /krub-do/ |krubdD| 'scratched' are not subject to syllable-final 

devoicing. The relevant part of the representation of krabde will then 

be as in (15) (for ease of exposition I refer directly to the feature [voice] 

instead of the class node Laryngeal): 

(15) C O 

I I 

X X X X X X X 

I I I I I I 

k r u b d o 

[+voice| 

6 Cf. Van Marie (19X5) for an analysis of competing affixes in Dutch, ('arslairs ll l )XX) has 

shown Ihat even when the choice between competing suffixes is ilelerniined by phonological 

properties of ihe stem, this does nol necessarily imply lhal the suffixes derive from a common 

underlying form. 

7 Zonneveld ( 19X3) proposed to account for Ihe progressive assimilation of Ihe /d/ of Ihe past 

tense suffix by positing an underlying denial fricalne lor lhe/d/ which is then subject lo Ihe iule ol 

Progressive Voice Assimilation. Such an abstract solution is made superfluous here.


Representation (15) will not be subject to rule (II) due to the Uniform 

Applicability Condition (UAC). Alternatively, as pointed out above, we may 

assume, with Lombardi (1991), that there is no rule of Syllable-final Devoicing. 

but that the feature [+voice] is only licensed in onsets, and that, through 

the double linking, this feature is also licensed for the /b/. In other words, this 

would be a case of 'parasitic licensing'. 1 * 

The role of the UAC also appears to be crucial when we take another /t/-/d/ 

alternation into account. The past participles of regular verbs are formed by 

prefixing t>e-, unless the verb begins with a [+native] prefix that does not bear 

stress, and by suffixing /t/ to the stem if the stem ends in a voiceless obstruent, 

and /d/ elsewhere. In other words, we have the same regularity here as with 

respect to past-tense suffixes, and the same rule of Laryngeal Spreading 

applies: 

(16) Verb Past participle 

klap 'to applaud' ge-klap-t 

krab 'to scratch' ge-krab-d 

roem 'to praise' ge-roem-d 

ski 'to ski' 

ge-skie-d 

The /d/ only shows up if the participle is inflected, that is, followed by a schwa, 

he-cause then it occurs in an onset. If it remains in word-final position, it will 

devoice due to Syllable-final Devoicing. In the case of gekrabd, all final 

obstruents will be voiceless, so /bd/ is realized as [pt]. 

The representation of gekrabd will look as follows after the application of 

Laryngeal Spreading: 

(17) X X X X X X X X 

I I I I I I I 

y 3 k r u b d 

[+voice| 

The UAC correctly predicts that in this configuration, Syllable-final Devoicing 

applies because both obstruents are dominated by the Coda node (the second 

one, the appendix consonant, after being Chomsky-adjoined to the coda), and 

thus we get the correct phonetic form |yokrupt|." 

The rule of Laryngeal Spreading may also be used to underspecify wordinternal 

voiced stops in words like Bagdad [buydut| 'Baghdad' and Egbert 

[cybortl 'id.' (male name). Laryngeal Spreading will then create obstruent 

x 

C'I. Goldsmith (1WM: VIS ) lor I ho notions 'piosodic licensing' and 'parasitic licensing'. 

Goldsmith also observes that coda positions license fewer lealuies than onset positions, '(ienu 

nates, however, are associated with both a coda position and an onset position, so. while lhe> gel 

their licensing from their onset position, they then give rise to the presence ol segmentai material in 

the coda that would otherwise not have a chance ol appearing there' (IWO: 336). 

'' Note, however, thai the Linking Constraint (Hayes I'JXo) would block application of Syllablefinal 

Devoicing, which is incorrect, since the phonetic form |v,3krupl| must he derived.

64 WORD PHONOI.OCY 

clusters doubly linked to [+voice], which are thus exempted from Final 

Devoicing, as is required. More generally, Laryngeal Spreading may be used 

to express the fact that in underived sequences obstruent clusters always agree 

in voice. For instance, the voicelessness of the A/ in akte 'act' need not be 

specified in its underlying form. 

4.2.2. Nasal assimilation 

In many languages nasal consonants are often homorganic with a following 

obstruent. This is also the case in Dutch. In particular, tautosyllabic nasalobstruent 

clusters are always homorganic. The only systematic exceptions are 

nasal consonants followed by a coronal consonant, the appendix consonants. 

Such clusters occur usually in morphologically complex forms. 

(18) |m] dawp 'damp', rawp 'disaster', gember 'ginger' 

|rrj| kawfer 'camphor' 

[n| tand 'tooth', kam 'side', ruwd 'cow' 

[n] orawje 'orange', Spawje 'Spain', bonje 'fight' 

[n] dawk 'thank', bank 'couch', /.ink 'sink', a|gy|list 'anglicist', ta[rjg|o 

'id.' 

Consequently, in this configuration nasals need not be specified for Place, and 

the Place features will be spread from the following consonant as shown in (19). 

(19) Nasal Assimilation 

• l+cons) 

AX 

[+nas] 

Place 

Within morphemes, heterosyllabic nasal-obstruent clusters are usually also 

homorganic, as illustrated in (18). In a few words this is not the case, for 

instance in imker 'bee keeper'. Therefore, the nasal of this word has to be 

specified as Labial underlyingly. Likewise, in zing+t 'sings', the velar nasal 

will already be specified as such in the morpheme zing //'in/- 

In compounds and phrases it is the coronal nasal /n/ only that assimilates to a 

following consonant. Examples are the following, with the prepositions aan 

/an/ 'to' and in /in/ 'id.' and the negative prefix on- /on/: 

(20) in Parijs 'in Paris' [imparcis] 

aanbod 'offer' 

lambot] 

onfatsoenlijk 'indecent' [orçfutsunbk] 

onvast 'unstable' 

|nrnvust| 

onwaar 'untrue' 

|:>rni>a:rl 

onmogelijk 'impossible' 

in Madrid 'id.' 

(Dmoyalak] 

|imadnt|

onjoviaal 'unjovial' 

in juli 'in July' 

onkies 'indecent' 

in chaos 'in chaos' 

ongewoon 'uncommon' 

4-2. PHONOLOGICAL RULES 

[onjovijal] 

[injylil 

[inxans] 

[onyauon] 

Non-coronal nasals do not assimilate, as illustrated by the following examples: 

(21) wangzak 'cheek-pouch' *[uunzak| 

damkampioen 'draught champion' *[dankumpijun] 

This restriction to /n/ can be accounted for if we assume that the morphemefinal 

coronal nasal is unspecified as to Place. It will then receive its place 

specification from Nasal Assimilation if it occurs before a consonant-initial 

word in the domain in which Nasal Assimilation applies. 10 If not, a default rule 

will provide the Place specification [Coronal | at the end of the phonological 

derivation, thus expressing that [Coronal] is the unmarked value for Place with 

respect to nasals." However, in Section 4.3.2 it will be shown that certain 

rules of Lexical Phonology have to refer to the feature [Coronal] of nasals. In 

Chapter 6 it will be argued that there are additional reasons for specifying 

coronal nasals as such before the end of the phonological derivation. 12 This 

implies that rule (19) cannot be taken to cover all cases of nasal assimilation: a 

separate feature-changing postlexical rule for the assimilation of the coronal 

nasal is required. 

4.2.3. Hiatus rules 

Two adjacent vowels within a Dutch prosodie word are very rare. In other 

words, Dutch does not favour word-internal onsetless syllables. The first vowel 

will always be long because syllables cannot end in a short vowel (Section 

3.3). If the first vowel is /a/, a glottal stop will be inserted at the phonetic level 

if the next vowel belongs to the syllable with main stress: 13 

(22) paella/paelja/'id.' [pa?elja| 

aorta /anrta/ 'id.' 

[a?orta| 

Kaiinda /kuunda/ 'id.' |ka?unda[ 

versus 

chaos /xaos/ 'id.' 

[xaos] 

farao /farao/ 'Pharaoh' [farao] 

" The domain specifications of rules will be discussed in Chapter 7. 

" Cf. Paradis and Prunel (1984, 1941) loi cross-linguistic evidence for this claim. 

12 This conclusion is confirmed by the findings of McCarthy and Taub (1992) and Hall (199.!) 

with respect to Coronal underspccilication for English and German respectively. 

' ' Jongenburger and Van Heuven ( 1991 ) showed that a glottal Mop is always inserted before a 

vowel-inilial word after a pause. According lo Gussenhoven (personal communication) wordinternal 

glottal-slop insertion only occurs in loot-initial position, i.e. not in words like


If the first vowel is a schwa, the schwa will be deleted. In all other cases except 

after/a/a homorganic glide will be inserted between the two vowels. I will first 

present the data concerning Homorganic Glide Insertion (HGI). The generalization 

is that the inserted glide has the same properties with respect to 

backness and roundness as the preceding vowel (Gussenhoven 1980: 177). 

Moreover, the glides are predictably high: 14 

(23) After rounded front vowels 

duo /dyo/ 'id.' 

flüor /flyor/ 'id' 

fonduen /fondyon/ 'fondue' (verb) 

üien /ceyan/ 'onions' 

réuen /r03n/ 'male dogs' 

Edui-.rd /edyurd/ 'id.' 

januari /janyari/ 'January' 

intui'tie /intyiti/ 'intuition' 

ruine /ryina/ 'ruin' 

After unrounded front vowels 

dieet /diet/ 'diet' 

bioscoop /bioskop/ 'cinema' 

Indriaas /indrias/ 'Andrew' 

Gea /yea/ 'id.' (f. name) 

Geo /yeo/ 'id.' (m. name) 

zee+en /zean/ 'seas' 

ree+en /rean/ 'deer' 

vijand /veiund/ 'enemy' 

After back vowels 

Ruanda /ruanda/ 'Rwanda' 

Boaz /boas/ 'id.' 

hou+en /hDuan/ 'hold' (Verb) 

[dyqo] 

Irlyipr] 

(fondyqon] 

lœyqan] 

[r0ipn| 

[edyqurt] 

[janyijaril 

[intyqitsi] 

(ryqina] 

[dijet] 

[bijnskop] 

|mdrijas| 

lyejal 

lyejo] 

[zejan| 

[rejon| 

|vfijunt| 

[ruuanda] 

[bouus] 

[houuDn] 

Note that the front rounded glide |q| does not occur as an underlying segment 

in Dutch. This explains why native speakers of Dutch are inclined to interpret 

this glide as either |j] or [u|, depending on the stress pattern, or give variant 

answers to the question 'which glide occurs in a word like dw>T. In particular, 

after a front round vowel with main stress the glide is often perceived as [j|. 

Given this array of facts, it seems appropriate to split the formalisation of 

HGI into two parts: insertion of an X-position, and subsequent spreading of the 

features of the preceding vowel as given in (24). 

The glide with the feature specification [+backj will surface as |u|, the glide 

4 The lirst generative analysis cil' 11(11 is presented in /onneveld (1978: 64-73). As Gussenhoven 

(1980) has pointed out, there are three ralher than two transitional glides. My analysis is based 

on Gussenhoven (1980).


(24) Homorganic Glide Insertion 

a. Insert X in the context p-voc 1 [+voc] 

l+highj 

b. Spreading X X 

I"'* 

[+voc] 

with the feature specification [ — back, —round] as [j], and the glide with 

[-back, +round] as [q]. 15 If the first vowel is /e/ or /o/, the feature [+midj 

which is spread to the glide position, will change into [—mid] by convention 

since glides are predictably high, that is, [+high, -mid] (cf. Booij 1989a). 

Note, moreover, that the inserted vocoid will be predictably interpreted as a 

consonant since it occurs after long vowels. Finally, note that the back glide is 

labiodental rather than bilabial, whereas the preceding back vowels are 

bilabial. This is predictable since the Dutch rounded back glide is always 

labiodental. 

The rule of HGI does not imply that [j] is never to be found after back 

vowels since, as shown in Section 3.5.5, /j/ does occur there at the underlying 

level, as in loeien /luj+3n/ 'to moo', with the first-person singular form loei/luj/. 

There are a few pairs of words such as koe 'cow' [ku] —koeien 'cows' [kujan] 

where we have to assume two lexical allomorphs, in this case /ku/ and /kuj/. ' 6 

The domain of obligatory HGI is clearly the prosodie word since there is no 

obligatory insertion of a glide at the boundary between the constituents of a 

compound or other complex words consisting of more than one prosodie word. 

Usually, a glottal stop is inserted at the beginning of a vowel-initial prosodie 

word. Yet, it is possible to insert glides in such environments in casual speech. 

As will be discussed in Section 7.2, many phonological rules exhibit the 

property that they are obligatory within the prosodie word, and optional in 

larger domains: 

(25) koeachtig 'cow-like' [ku?axtax] or [kuuaxOx] 

zeearend 'sea eagle' [zeParant] or [zejarant] 

The second hiatus rule is that of Prevocalic Schwa Deletion which also 

applies obligatorily within a prosodie word: 

(26) Romein /roma+ein/ 'Roman' [romein] 

elitair /elit3+c:r/ 'elitist' [elite:r] 

codeer /koda+er/ 'encode' [kode:r] 

kaden /kada+an/ 'quays' [kadon] 

15 As shown by Van Heuven and Hoos (1991), inserted glides have a distinctly shorter duration 

than underlying glides. The former are transitional glides only. This is reflected by the representations 

for the inserted glides proposed here, namely as Xs that are linked to the Place features of the 

preceding vowel. 

16 Some southern dialects of Dutch only have the form /kuj/, i.e. the singular form of 'cow' is 

koei.


Since the domain of application is the prosodie word, the prevocalic schwas in 

the following words will normally be pronounced: 

(27) zijd[3]-achtig 'silk-like' 

b|o]-antwoorden 'to reply' 

mod[3]-opleiding 'fashion academy' 

In casual speech the schwa may also be deleted across prosodie word boundaries, 

but hardly ever if the schwa forms part of a prefix. This will be discussed 

in Section 7.2.3. 

The rule of Prevocalic Schwa Deletion will therefore read as in (28). 

(28) Prevocalic Schwa Deletion 

Delete X X before |-cons] 

I 

[-consj u 

Glide Insertion will not apply after a schwa, as required, because Homorganic 

Glide Insertion only applies after vowels specified as [+high], a feature 

that the schwa does not bear. 

4.2.4. Degemination 

Dutch does not allow for geminate consonants within prosodie words. Consequently, 

degemination is obligatory within prosodie words as soon as a cluster 

of two identical consonants arises. In larger domains such as compounds and 

phrases the rule is optional. The following examples, of course all complex 

words, illustrate this process: 

(29) eet /et+t/ 'to eat' (3 sg. pres.) (et| 

voedt /vud+t/ 'to feed' (3 sg. pres.) |vul| 17 

zette /zet+ta/ 'to put' (past) 

[zeto| 

voedde /vud+da/ 'to feed' (past) 

[vuda] 

gezet /ye+zet+t/ 'to put' (past part.) 

[yazet] 

gevoed /ya+vud+d/ 'to feed' (past part.) (yavutj 

kiest /kis+st/ 'delicate' (superl.) 

|kist| 

kies /kis+s/ '(something) delicate' 

|kis| 

grootte /yrot-Ha/ 'size' 

[yrota] 

fietsster /fits+star/ 'cyclist' (fem.) 

[fitstor] 

onmiddellijk /on+midal+Iak/ 'immediately' [omidalak] 

The suffix -sel appears systematically to exclude base words ending in /s/ or/z/, 

thus obviating the need for degemination in -se/-words (Booij 1977: 123). 

Since the rule is optional across prosodie word boundaries, it is possible to 

differentiate phonetically between complex words with and without a geminate, 

as in the following examples: 

17 The /d/ first becomes A/ through the Final Devoicing rule (11) given in Section 4.2.1

(30) ver-assen 'to cremate' [verusanj 

ver-rassen 'to surprise' [vcr:uson| 

4-3- MORPHOLEXICAL RULES 69 

The rule of Degemination (31) deletes one of two adjacent identical consonants. 

(31) Degemination 

Xj Xi -» X: 

[+cons] [+cons] [+consj 

Domain: 

Obligatory in prosodie words, optional in larger domains 

4.3. MORPHOLEXICAL RULES 

4.3.1. Diminutive ullomorphy 

The Dutch diminutive suffix has five allomorphs: -tje, -je, -pje, -kje, and -etje. 

This suffix appears mostly after nouns, but also after some underived adjectives 

(e.g. blondje 'blond girl' < blond 'id.'), some verbs (e.g. speeltje 'toy' < 

speel 'to play'), and a few prepositions (for example, uitje 'outing' < uit 'out'), 

and creates neuter nouns. The regularities as to where the different allomorphs 

appear can be summarized as follows (Cohen 1958, Ewen 1978, Gussenhoven 

1978, Booij 1981«, Trommelen 1984): 

(«) -je U 9 ! appears after stem-final obstruents; 

(b) -etje [3tj3) appears after sonorant consonants if preceded by a short 

vowel with primary or secondary stress; 

(r) -pje appears after /m/ except in the cases sub (b); 

(d) -kje appears after /rj/ except in the cases sub (b); 

(e) -tje appears elsewhere. 

The following diminutives illustrate the use of the different allomorphs (the 

accent ' above a vowel letter indicates main stress, and the accent ' indicates 

secondary stress): 

(32) a. allomorph -je 

lip 'id.' 

lipje [lipjs] 

hand 'id.' 

handje [hantja] 

hek 'gate' 

hekje [hekjo] 

lief 'sweet' liefje |litja| . 

klas 'class' 

klasje [klusjol 

lach 'laugh' 

lachje [luxjaj 

b. allomorph -etje 

n'ng 'ring' 

sen'ng 'lilac' 

leerling 'pupil' 

ringetje 

seringetje [sanrjotjo] 

leerlingetje [leirlirjatja]

70 WORD PHONO1.0CV 

wandeling 'walk' 

bóemerang 'boomerang' 

bom 'bomb' 

bon 'ticket' 

horizon 'id.' 

bal 'ball' 

tor 'beetle' 

hotel 'id.' 

kanon 'gun' 

c. aUomorph -pje 

riem 'belt' 

bodem 'bottom' 

helm 'helmet' 

album 'id.' 

valium 'id.' 

wi'gwam 'id.' 

ti. dllomorph -kje 

koning 'king' 

paling 'eel' 

sdrong 'id.' 

kâmpong 'village' 

e. aUomorph -tje 

ree 'deer' 

stro 'straw' 

traan 'tear' 

wiel 'wheel' 

haar 'hair' 

doctor 'id.' 

ponton 'pontoon' 

professor 'id.' 

wandelingetje [uundalirjatja] 

boemerangetje [bumarurjatjo| 

bommetje [bDmatja] 

bonnetje [bonatja] 

horizonnetje [hoiri/.anatja] 

balletje [bulatja] 

torretje [taratjaj 

hotelletje [hotelatja] 

kanonnetje [kanonatja] 

riempje |rimpja| 

bodempje [bodampja] 

helmpje [helmpja] 

albumpje [ulbvmpjo] 

valiumpje |valijampjo| lx 

wigwampje (uixuumpja] 

koninkje [komrjkja| 

palinkje Ipalirjkjo) 

saronkje |sarr>rjkja| 

kamponkje |kumpDn.kja| 

reetje |retja| 

strootje [strotjo] 

traantje (trantja) 

wieltje |uiltjo| 

haartje [hairtja) 

doctortje [doktartja] 

pontontje [pnntantjal 

professortje Iproftsortja] 

We would like to express the phonological similarity of the five diminutive 

suffixes by deriving them from one underlying form, /tja/, through a number of 

MP-rules. Note also that the choice of a specific allomorph is not completely 

arbitrary. For instance, the p-initial variant appears aller a stem-final /m/, and 

the ^-initial variant after the velar nasal, suggesting that some kind of Place 

assimilation is involved. The deletion of A/ ties in with the tendency in Dutch to 

simplify clusters of obstruents through /t/-deletion. Note, however, that it is 

normally the nasal consonant that assimilates to a following obstruent, whereas 

here it is the obstruent that assimilates to a preceding nasal. This emphasi/.es the 

'" Since the letter u in valium can be interpreted as /v/ as well, an alternative diminutive form 

also occurs: voliiunmetje. In Ihis lallcr form Ihe final syllabic of the base word receives secondary 

slress, hence (he allomorph -elje.

4-3- MORPHOLEXICAL RIJLES -J ] 

differences between the P-rule of Nasal Assimilation and the assimilatory MPrules 

involved here. The MP-rules can now be formulated as in (33)-(35). |l) 

(33) Insert /a/ in the following context: 

R 

N 

I 

X 

C 

I 

X 

I 

|-cons) [+son| f^sonj ja 

[Corj 

Condition: 

R is the rhyme of a stressed syllable 

(34) • (-son] ja 

l+nas) 

Place 

Place 

(35) Delete [-son, Cor] in the context |-son| —ja 

In the case of a word like senn^cije, underlyingly /sanrj+tjo/, rule (33) will 

have to take precedence over rule (34), and therefore, rule (33) has to be 

ordered before rule (34). 20 

As may be expected in cases of allomorphy governed by MP-rules, there are 

exceptional diminutive forms. First, there is a number of words that in addition 

to the regular form also have a variant with -etjc: 

(36) 

wiel 'wheel' 

bloem 'flower' 

Jan 'John' 

brug 'bridge' 

weg 'road' 

heg 'hedge' 

Regular form 

wieltje 

bloempje 

Jannetje 'Jane' 

brugje 

wegje 

hegje 

Variant 

wieletje 

bloemetje 

Jantje 'Johnny' 

bruggetje 

weggetje 

heggetje 

l will relcr HI specific segments by iiu'ans of the corresponding phonemic symbol when no 

phonological gcneiali/alion is involved. We might a Ko represent the first segment of the diminutive 

suflix without Ihe specification Coronal. In that case, the rules that derive -pje and -kje do not 

have 10 delink the feature (Coronal], and can he restricted to a spreading operation. The feature 

[Coronal] would then be tilled in by a défailli rule lhal says that (Coronal) is the default value for 

consonants. 1 he proposal can be found in Lahiri and Evers (1991: 97). However, as pointed oui in 

the next section. Ihe feature (Coronal| has to he available at Ihe lexical level. 

Trommelen ( 1W4) is an attempt to deine Ihe different allomorphs without sliess conditions (cf. 

Boin| i l l )K-l) lm a review!. However, from a theoretical point of view Ihe fact that allomorphy is 

dcpi ndent on stress is nol embarrassing at all. since the stress patterns of Ihe base words have 

aheady been derived when Ihe rules of diminutive allomorphy apply. 

This is not piedicted by the hlscwhere Principle since the set ol' representations to which (33) 

applies is not a subset of lhal for (34); rather. Ihe sels are overlapping.


In the case of bloemetje there is also semantic differentiation: bloemetje can 

also mean 'bunch of flowers', unlike bloempje. Trommelen (1984: 48) 

observes that some native speakers of Dutch are also inclined to insert a 

schwa after stems ending in short vowel + sonorant consonant, although the 

vowel does not bear main stress, for example, wigwammetjt, kampongetje. 

Probably, loan-words such as wigwam 'id', and kampong 'village' (Indonesian 

loan) can be interpreted as (prosodie) compounds, which induces secondary 

stress on the second syllable. 

A second class of exceptions has to do with stem allo'morphy: some native 

nouns have two allomorphs, one with a short vowel and one with a long vowel. 

The allomorph with the long vowel occurs in some plural forms, diminutives, 

and verbs, but there is no systematicity here: 

(37) Singular Plural Diminutive Derived word 

schjijp 'ship' 

w[e]g 'road' 

p[u]d 'path' 

st[a]d 'city' 

b[a]d 'bath' 

gl[a]s 'glass' 

sch[e]pen 

w|e]gen 

p|a)den 

st[e|den 

b|ajden 

gl[a]zen 

sch|e|pje 

w|t-]ggetje 

p[a|dje 

st[u]dje 

b[u|dje 

gl|a]sje 

gl[u|sje 

sch[i)pper 'skipper' 

st[e]delijk 'urban' 

b[a|den 'to bathe' 

gl|a|zig 'glassy' 

In the case of glasje/glaasje we find semantic differentiation: a glaaxje is a 

drinking glass or the slide used under a microscope, whereas glasje may refer 

to a spectacles lens. 

Another type of stem allomorphy found with diminutives is the use of a stem 

without the normally occurring final stem-final schwa: 

(38) karbonade 'carbonade' karbonaadje (kurbonatjo] 

parachute 'id.' 

parachuutje [parasjytjs] 

machine 'engine' 

machientje [masjintjn] 

Schwa deletion before -tje is an irregular phenomenon, and speakers might 

also choose not to delete the schwa. Thus, the form parachute! je also occurs. 

In most relevant words, this schwa deletion does not occur. 

The type of allomorphy that we find for the diminutive suffix also occurs in 

de-adjectival adverbs derived with the suffix -tjes, which can be analysed as 

the morpheme sequence /tja+s/. We get exactly the same allomorphy as for 

(39) zacht 'soft' 

stil 'quiet' 

stiekem 'stealthy' 

gewoon 'ordinary' 

zachtjes 'softly' [zuxjos] 21 

stilletjes 'quietly |stilotjos| 

stiekempjes 'stealthily' |stikompjos| 

gewoontjes 'ordinarily lyauontjas] 

The /I/ is deleted due lo the /(/-deletion rule I» he discussed in Section 7.2.6.

4-3- MORPHOLEXICAL RULES 73 

Diminutive forms are also subject to P-rules. When the stem ends in a 

voiced obstruent, these voiced obstruents stand in syllable-final position after 

the application of the relevant MP-rule, for example, /hand+tja/ becomes 

/handja/ with the syllabification (hand) a (ja) CT (/dj/ is not a possible onset). 

Syllable-final Devoicing will then devoice the stem-final /d/, and thus we get 

(huntja] with, after resyllabirication, the syllabification pattern (han) 0 (tja) 0 . 

Another P-rule, /t/-deletion in obstruent clusters (Section 7.2.6) will delete 

the stem-final obstruent as in the diminutive form of the noun kast 'cupboard' 

/kust+ja/, that is realized as [kusja]. 

4.3.2. -fr-allomorphy 

The suffix -er is used in three different functions: it forms comparative forms 

of adjectives, it forms subject names from verbs (cf. Booij 1986), and it forms 

de-nominal nouns with a variety of interpretations (cf. Booij 1988a). All three 

suffixes have the allomorph |dar| after a base ending in /r/, but a competing 

rule is that -aar occurs instead of -er in the case of the nominal (thus de-verbal 

and de-nominal) use of -er, after a base ending in schwa + coronal sonorant 

consonant (Smith 1976): 

(40) Adjective 

rood /rod/ 'red' 

goochem lyoxam] 'smart' 

heikel /hcikal/ 'risky' 

bitter /bitar/ 'id.' 

zuur //yr/ 'sour' 

(41) Verb 

eet /et/ 'to eat' 

/eur /z0r/ 'to nag' 

bezem /bezarn/ 'to sweep' 

bibber /bibar/ 'to tremble' 

kibbel /kibal/ 'to quarrel' 

oefen /ufan/ 'to train' 

(42) Noun 

Amsterdam /amstardum/ 'id.' 

wetenschap /uetansxup/ 'science' 

Bijlmermeer /bcilmarmer/ 'id.' 

Diemen /diman/ 'id.' 

Assen /uson/ 'id.' 

Lochern /bxam/ 'id.' 

Comparative 

roder [rodar] 'redder' 

goochemer [yoxamar] 'smarter' 

heikeler (hdkolor] 'riskier' 

bitterder [bitardarj 'more bitter' 

zuurder [zyirdar] 'sourer' 

De-verbal ninin 

eter [etarj 'eater' 

zeurder [z0:rdar] 'nagger' 

bezemer [bezamar] 'sweeper' 

bibberaar [bibara:r] 'trembler' 

kibbelaar [kibalairj 'quarreler' 

oefenaar [ufana:r| 'trainer' 

De-nominal noun 

Amsterdammer 'inhabitant of A.' 

wetenschapper 'scientist' 

Bijlmermeerder 'inhabitant of B.' 

Diemenaar 'inhabitant of D.' 

Assenaar 'inhabitant of A.' 

Lochemer 'inhabitant of L.' 

These data show how Dutch avoids surfacing of the sequence [rVr], a cluster 

that is avoided cross-linguistically (cf. Dressler 1977, Shannon 1991). The

74 WORD PHONOl.OtiV 

-erl-aar alternation only occurs in nouns, 22 whereas the -erl-cicr alternation has 

a general character. 

The -erl-aar alternation can be dealt with in two ways. One analysis would 

simply state that these are two competing suffixes: -aar occurs in a specific 

phonological environment, -er is the general suffix, and occurs elsewhere (cf. 

Van Marie 1985 for comparable morphological analyses). However, that 

analysis does not account for the strong phonological similarity between -er 

and -aar (historically both derive from the Latin suffix -arius). Moreover, there 

are actually two nominal suffixes -er with different meanings, a de-verbal and 

a de-nominal one. So we would also have to assume two -aar suffixes. 

Therefore, I propose an MP-rule that changes -er into -aar within nouns: 

rule (43) 

(43) -erl-aar rule 

Insert [-high, -mid) in the context: 

[-cons) u +cons |-cons] u r 

+son 

Cor 

I 

Since the rule inserts Place features, it will only apply to empty vowels, i.e., 

schwas. Thus, the second schwa will receive the Place specification of the /a/, 

which is the only long low vowel. Note that the rule mentions the lexical 

category Noun which excludes adjectives from its domain, as required. That is, 

it is clearly an MP-rule. 23 

The rule for -der can be seen as a P-rule since its structural description can 

be completely couched in phonological terms. It will insert /d/ in the sequence 

/r—ar/. It will also apply before the de-verbal suffix -erij /orti/, and insert a /d/ 

in, for example, klierderij 'nagging' from underlying /klir+arci/. Similarly, it 

applies before the suffix -erii> as in liter 'with literary pretensions'. 24 

(44) /d/-insertion" 

Insert /d/ in the context: 

r — [— consj u r 

: The morpholexical nature of the -«•;•/ am alternation is also evident from the fact that -um 

occurs in some words that do not completely satisfy Ihe relevant phonological requirements, lor 

instance ilic/iiini 'servant' and Icnmi 'teacher', where the coronal consonant is preceded by a full 

vowel instead of the required schwa. Irregularly.

4-4- ALLOMORPHY IN THE NON-NATIVE LEXICON 75 

The MP-rule competes with the rule of /dAinsertion in nouns like schipperaar 

with the underlying form [ [sxip3r]v3r]N- The MP-rule has to have priority 

although this does not follow from the Elsewhere Principle since the class of 

words to which the MP-rule can apply is not completely a subset of the class of 

input words for the rule of/d/-insertion: the sets partially overlap. The correct 

ordering follows from the assumption that in the unmarked case MP-rulcs 

precede P-rules (Anderson 1974). 

4.4. ALLOMORPHY IN THE NON-NATIVE LEXICON 

In Section 2.5.2, I introduced the rule of Learned Vowel Backing which says 

that in non-native suffixes mid vowels become [+back] before an adjacent nonnative 

suffix. Crucially, this rule only applies in the non-native part of the 

lexicon. Before discussing a number of specific alternations in the non-native 

part of Dutch, I will discuss this notion 'non-native lexicon' in some more 

detail. 

The lexicon of a language consists of at least two parts: a list of existing 

words, and a set of word-formation rules that serve to expand this list. In the 

case of Dutch, it appears that the morphological rules clearly distinguish 

between native and non-native input words. For instance, of the two synonymous 

suffixes -iteit '-ity' and -held '-ness', the first one only attaches to nonnative 

input words, 26 whereas the native (Germanic) suffix -held attaches to 

both native and non-native input words of the required syntactic category: 

(45) absurd 'id.' absurditeit/absurdheid 'absurdity' 

gewoon 'ordinary' *gewoniteit/gewoonheid 'ordinariness' 

This asymmetry between native and non-native word formation is pervasive 

throughout Dutch word formation (Booij 1977: 131-9), and we may formulate 

the relevant morphological principle as follows: 

(46) Non-native suffixes only attach to non-native bases 

The class of non-native suffixes involved comprises at least the following 

ones: 27 

(47) Suffix Base word Derived word 

-aal muziek 'music' muzikaal 'musical' (adj.) 

-aan parochie 'parish' parochiaan 'parishioner' 

-aat doctor 'id.' doctoraat 'doctorate' 

-abel accept-eer 'to accept' acceptabel 'acceptable' 

-age percent 'per cent' percentage 'id.' 

-air elite 'élite' elitair 'élitist' 

-andus doctor 'id.' 

doctorandus 'MA' 

6 Exceptions to this generalization are the words stommiteit 'stupid act' andßauwiteit 'silliness'. 

21 Non-native prefixes are discussed in Chapter 5.

-ast 

-eel 

-eer 

-ces 

-esk 

-ein 

-eur 

-eus 

-iaan 

-ica 

-icus 

-ide 

-ieel 

-ief 

-iek 

-1er 

-ier 

-iet 

-ieus 

-ine 

-isch 

-iseer 

-isme 

-ist 

WORD 

gymnasium 'id.' 

fundament 'foundation' 

profeet 'prophet' 

Nepal 'id.' 

ballade 'ballad' 

Rome 'id.' 

ambassade 'embassy' 

rancune 'rancour' 

presbyter 'id.' 

elektron 'electron' 

elektron 'electron' 

chloor 'chlorine' 

exponent 'id.' 

agressie 'aggression' 

motor 'engine' 

juweel 'jewel' 

Israel 'id.' 

metropool 'metropolis' 

mode 'fashion' 

nectar 'id.' 

algebra 'id.' 

banaal 'trivial' 

absurd 'id.' 

propaganda 'id.' 

PHONOLOGY 

gymnasiast (lit.('gymnasium student' 

fundamenteel 'fundamental' 

profeteer 'to prophesy' 

Nepalees 'Nepalese' 

balladesk 'ballad-like' 

Romein 'Roman' 

ambassadeur 'ambassador' 

rancuneus 'vindictive' 

presbyteriaan 'Presbyterian' 

elektronica 'electronics' 

elektronicus 'electronic engineer' 

chloride 'id.' 

exponentieel 'exponential' 

agressief 'aggressive' 

motoriek 'locomotion' 

juwelier 'jeweller' 

Israelier 'Israeli' 

metropoliet 'metropolitan' 

modieus 'fashionable' 

nectarine 'id.' 

algebraisch 'algebraic' 

banaliseer 'to trivialize' 

absurdisme (lit.) 'absurdism' 

propagandist 'id.' 

It is not always easy to determine whether an underived word is still felt as 

non-native, but morphological behaviour is a reliable indication. 

So we might conceive of the non-native lexicon as a list of non-native words 

and a set of non-native word-formation rules. This subcomponent of the 

lexicon may then also comprise the MP-rules that account for the different 

types of allomorphy in affixes and roots. However, the notion 'non-native 

lexicon' cannot be seen as a completely distinct component of the grammar, 

because there are cases in which native affixation may take place in between 

cases of non-native affixation. For instance, in the word ongrammaticaal 

'ungrammatical' the native prefix on- 'un-' has been added to the non-native 

complex adjective f>rammatikaal 'grammatical', itself derived by suffixation 

of -aal '-al' to grammatika 'grammar'. Yet, we can add the non-native suffix 

-iteit '-ity' to the adjective ongrammaticaal, because the prefix on- is not the 

head of the word. Hence, the feature |— native] percolates from the head 

grammaticaal to the whole word ongrammaticaal (cf. Williams 1981, Lieber 

1989) which therefore allows for another cycle of non-native suffixation. 

As we will see in Chapter 5, the distinction between native and non-native 

lexicon will also play a role in the analysis of word stress. 

In the following subsections 1 will discuss a number of non-native MP-rules.

4-4- ALLOMORPHY IN THE NON-NATIVE LEXICON 

4.4.1. Affixal allomorphy 

77 

A number of non-native affixes exhibit allomorphy. They will be discussed in 

this subsection. 

One type of allomorphy is mentioned in Section 2.2.5, the backing of long 

mid vowels. The following examples illustrate the working of this rule: 

(48) Base word 

-eel —> -aal 

fundament+eel 

eventu+eel 

sentiment+eel 

ration+eel 

mor+eel 

-air -> -aar 

milit+air 

vulg+air 

popul+air 

-eur -> -oor 

direct+eur 

superi+eur 

inspect+eur 

inferi+eur 

-eus —> -oos 

nerv+eus 

religi+eus 

curi+eus 

mon.stru+eus 

Derived words 

fundament+al+isme, fundament+al+ist 

eventu+al+iteit 

sentiment+al+isme 

ration+al+isme, ration+al+iteit 

mor+al+isme, mor+al+ist 

milit+ar+isme, milit+ar+ist 

vulg+ar+isme, vulg+ar+iseer 

popul+ar+iteit, popul+ar+iseer 

direct+or+aat 

superi+or+iteit 

inspect+or+aat 

inferi+or+iteit 

nerv+os+iteit 

religi+os+iteit 

curi+os+iteit 

monstru+os+iteit 

Rule (49) expresses this generalization concerning non-native suffixes with 

mid front vowels: 

(49) Learned Vowel Backing T-consl -» [+back] / — [+cons] ] s [ . . . ] s 

L+midJ 

Condition: 

S(uffix) = [-native] 

It is not sufficient here to assume that the rule is a cyclic rule, and hence 

applies in derived environments only. For instance, in tonelist /tonel+ist/ 

'playwright' the /e/ is not backed since it is not part of a non-native suffix. 

The subsequent lowering of the [+back] counterpart of /e/ and A::/ to /a/ 

follows from the redundancy rule that says that unrounded back vowels are 

low (Chapter 2, rule (\5h)). The 'learned' nature of this backing rule is also 

evident from the fact that for most native speakers the noun derived from 

nerveus 'nervous' is nervcusiteit 'nervousness', without Learned Vowel Backing, 

rather than nervositeit.


A second type of allomorphy is characteristic for the non-native suffix -iek 

-likl that shows up as [is] before non-native suffixes that begin with an /i/ or /i/: 

(50) kathol-iek 'catholic' /katol-ik/, kathol-ic-iteit [kutolisiteit], kathol-icisme 

[kutolisismo] 

sympath-iek /sim-pat-ik/ 'sympathetic', sympath-is-eer |simpatise:r] 

excentr-iek /eks-sentr-ik/ 'eccentric', excentr-ic-iteit [eksentrisitcit] 

There are also exceptions to the rule. For instance, the final /k/ of antiek 

'antique' does not show up as [s] before the suffix -iteit, but either as [k] or 

as [kw|: antiquiteiten 'antiquities' |untik(w)itdton). 

The non-native suffix -ear that expresses agents or instruments exhibits two 

allomorphs before the feminizing suffix -e: -eus /0s/ and -rie /ris/ which is 

found in particular after stems ending in a coronal stop: 

(51) mont-eur 'engineer' mont-eus-e 

control-eur 'checker' control-eus-e 

ambassad-eur 'ambassador' ambassad-ric-e 

conduct-eur 'conductor' conduct-ric-e 

There are also some non-native prefixes with a specific allomorphy pattern. 

The negative prefix in-, a borrowing from Latin, still exhibits the allomorphy 

that it had in Latin: 

(52) im- [im| impopulair 'unpopular' 

in- [inj intolerant 'id.' 

in- [irjl inconsistent 'id.' 

il- [il] illegaal 'illegal' 

ir- [ir] irrationeel 'irrational' 

Note in particular that, whereas Dutch does have Nasal Assimilation, it does 

not have complete assimilation of nasals before liquids, as is the case here. The 

same allomorphy is shown by the prefix con- /kon/, also a Latin borrowing: 

(53) com-[kom] commemoreren'to commemorate' 

con- [kon] consistent 'id.' 

con- [korj] conclaaf 'conclave' 

col- [kol] collaboratie 'collaboration' (pej.) 

cor- [kor] correlatie 'correlation' 

The prefixes a- /a/ 'id.' and de- /de/ 'id.' have specific allomorphs before 

vowel-initial stems: an- /an/ and des- /des/ respectively, as in an-organisch 

'inorganic' and des-interesse 'disinterest'. 

4.4.2. Root alternations 

Some non-native roots that end in A/ alternate with either /s/ (after a consonant), 

or /ts/ (after a vowel): 28 

2 * Some speakers have /Is/ aller a sonorant consonant, as in tolerantie which will then be 

pronounced as |tolr>runtsi|, hut only in very careful speech, because in more casual speech A/s tend 

to delete in CtC clusters.

4-4- ALLOMORPHY IN THE NON-NATIVE LEXICON 79 

(54) akt+ie 'action' [uksi] 

adopt+ie 'adoption' [adnpsij 

tolerant 'tolerant' [tolaruntj 

president 'id.' [president] 

Kant |kunt) 'id.' 

convert+eer 'to convert' 

[knnvcrteir] 

(55) rat+io 'id.' [ratsijo] 

relat+ie 'relation' (relatsi] 

akt+ief 'active' [aktif] 

adopt+eer 'to adopt' [adopte:r], 

adopt+ief 'adoptive' [adnptif] 

tolerantie 'tolerancy' [tolarunsi] 

president+ieel 'presidential' 

Ipresidcnsjel) 

Kantiaan 'Kantian' [kunsijan] 

convers+ie 'conversion' (konversi] 

rat+ificeer 'to ratify' [ratifiseir] 

relat+ief 'relative' (relatif) 

relat+eer 'to relate' [relate:r] 

milit+air 'militairy' [milite:r] 

milit+ie 'militia' [militsi] 

polit+ie 'police' [politsi] polit+iek 'politics' [politik] 

polit+ioneel 'police-' [politsijonel] 

stat+ion 'id.' [statsijon] stat+isch 'static' [statis] 

The generalization is that underlying ft/ changes into /ts/ or /s/ before the 

morpheme /i/, or before morphemes that begin with an /i/ which is followed by 

a vowel in the same morpheme, as in -io, -iaan, -ion. The relevant rules will 

therefore refer to the notion 'morpheme /i/ or morpheme /i/ followed by a 

vowel (M = Morpheme): 29 

(56) a. t -> ts / (-cons) — [i X . . . | M X * |+cons] 

b. t -> s / [+cons] — [i X ... IM X * |+cons| 

The rules will not apply to simplex words like spaghetti 'id.' and Haiti 'id.' 

because they do not contain a morpheme /i/. Moreover, there are also exceptions 

to the rule even where a nominalizing morpheme -le seems to be involved as in 

sympathie 'sympathy' [simpati |. In the case of Kantiaan there is variation: it may 

also be pronounced as [kuntijan]. Many speakers of Dutch realize intervocalic 

As/ as [s], for instance in politie which will then be pronounced as [polisi], 

One may doubt whether such alternations should really be accounted for by 

rule. There is no transparent phonological generalization given words such as 

sympathie. Moreover, native speakers might not even establish relations 

between allomorphs like those of /polit/ in polit+ie 'police' versus polit+iek 

'politics', given the lack of semantic relationship between these words. So we 

might assume that both allomorphs are listed in the lexicon as parts of the 

words in which they occur, and that, instead of productive rules, we have 'via 

rules' (Vennemann 1972) that express the systematic* in the distribution of 

allomorphs. A via rule does not derive one form from another, but only states 

" v In a complete formali/uliun ol' these rules, one cannot uniquely refer Io A/ as the locus of the 

rules if Ivoicel is a private feature: the rule will jusi refer to coronal slops. Note, however, that /d/s 

ilo not alternate, compare komt'ilir 'comedy', (unii'iluinl 'comedy player'. This seems to be a 

problem for the privative interpretation of | voice]. However, the rule has exceptions anyway (e.g. 

.\\iii/Kiiliii' /simpati/ with the phonetic form |simpati|).


that two forms are formally related. For instance, rule (56a) would be 

reformulated as follows: 

(57) [. . . |-cons) ts] M in the context - - [i X . . . ] M alternates with 

[ . . . [-cons] t ] M elsewhere (X * [+cons| ) 

4.4.3. Alternations in velar nasal clusters 

As pointed out by Trommelen (1984: 165-6) velar nasal clusters exhibit three 

different types of alternation. 

First, words ending in |rj] may have an allomorph ending in |qy] before nonnative 

suffixes: 

(58) difto|rj] 'diphthong' difto[rjy]eer 'to diphthongi/.e' 

A[n]elsaksisch 'Anglo-Saxon' a[rjy]list 'Anglicist'" 

Secondly, [rjy] alternates with [rjk] before a coronal obstruent: 

(59) fu|rjy|eren 'to function' fu(rjk|tie 'function' 

larylrjyjaal 'laryngeal' lary[rjk|s 'larynx' 

A third alternation is that between |rjy] and |kj, as in 

(60) fi[rjy|eren 'to make up' fi|k]tie 'fiction' 

restri[rjy]eren 'to restrict' restri(k|tie 'restriction' 

As will be clear from the unsystematic nature of these alternations, the relevant 

allomorphs will have to be listed as parts of the complex words in which they 

occur. 

4.4.4. Vowel lengthening 

Non-native words ending in a syllable with a VC rhyme that does not bear 

main stress exhibit vowel lengthening: the vowel of the last syllable is 

lengthened before vowel-initial suffixes: 

(61) doct[o]r 'id.' doct[o]r-aat 'doctorate' 

doct[o|r-aal 'MA degree' 

doct|o|r-andus 'MA' 

This lengthening is systematic, as is illustrated here for a number of suffixes: 

(62) Base word Derived word 

profess|D]r 'professor' profess[o]r-aal 'professorial' 

profess|o|r-aat 'professorship' 

profess[o|r-abel 'fit for professorship' 

cons|v]l 'id.' 

cons|y]l-air 'consular' 

Nep|a|l 'id.' 

Nep[a]l-ees 'Nepalese' 

Äron|D]ff 'id.' 

Aron[o|v-iaan 'Aronovian' 

" The pronunciation without the velar fricative also occurs.

elektr[D]n 'id.' 

alcoh[rt]l 'id.' 

4-4- ALLOMORPHY IN THE NON-NATIVE LEXICON 8l 

mot[o]r 'engine' 

kan|ojn 'canon' 

nect[u|r 'id.' 

Isru|L-]l 'id.' 

Jak[D)b 'Jacob' 

Jez|v]s 'Jesus' 

No|a)ch 'Noah' 

alfab[e)t 'alphabet' 

dem[D]n 'id.' 

sat(u]n 'id.' 

elektr[o]n-ica 'electronics' 

elektr[o]n-icus 'electronic engineer' 

alcoh(o]l-icus 'alcoholic' 

alcoh|o)list 'alcoholic' 

mot[o]r-iek 'way of moving' 

kan[o]n-iek 'canonical' 

nect[a]r-ine 'id.' 

Isra[e)l-ier 'Israeli' 

Jak[o|b-iet 'Jacobite' 

Jez[y|-iet 'Jesuit' 

No[a|ch-iet 'Noachite' 

alfab[e)t-isch 'alphabetic' 

dem|o]n-isch 'demonic' 

sat[a]n-isch 'satanic' 

The lengthening is triggered by [-native] suffixes, but the base should also be 

[-native], as is proven by the following observation. We can add the 

[-native] suffix -iaan to personal names such as the name of the Dutch 

linguist Uhlenbeck /ylanbck/. We then get Uhlenbeckiaan 'follower of 

Uhlenbeck', but the /c/ in the final syllable is not lengthened, although it 

does not bear main stress: [ylanbekijan]. The rule should therefore be 

formulated as (63). 31 

(63) Vowel Lengthening (non-native morphemes only) 32 

Insert X in the context X; X X 

I I I 

[-cons] [+cons| [-cons] 

Condition: 

X; is not in a syllable with main stress" 

The condition that the base-final syllable does not bear main stress explains 

why we do not get lengthening in the following cases, where the base word has 

main stress on the final syllable: 

(64) Base word Derived word 

kan[D|n 'gun' 

kan[D]nn-ier 'gun man' 

kol[n]s 'giant' 

kol[o]ss-aal 'gigantic' 

" A seemingly exceptional case of lengthening is ufftoilixch |ufxodis| 'idolatrous' derived from 

«/.i;


mod|c|I 'model' 

kart|D]n 'cardboard' 

tir[u]n 'tyrant' 

tromplejt 'trumpet' 

klarin[L'|t 'clarinet' 

mod[i:lll-eer 'to model' 

kart|r>|nn-age 'cardboard manufacture' 

tir|u]nn-iseer 'to tyrannize' 

tromp|t|tt-ist 'trumpeter' 

klarin|t;)tt-ist 'clarinettist' 

The stress condition on vowel lengthening shows that the base words must 

already have their stress patterns assigned before vowel lengthening applies, 

because after affixation the main stress will shift rightward, thus erasing the 

stress differences between the base words. For instance, in alcohólicus 

'alcoholic', main stress is on the syllable (ho) 0 , and this would incorrectly 

block lengthening if lengthening applied at the end of the derivation. Inversely, 

in, for example, trompettist the penultimate syllable is unstressed. Yet, the 

vowel of that syllable should not lengthen since it did have main stress on the 

previous cycle. In sum, the facts discussed here clearly require a cyclic 

application of rules: on the first cycle, stress is assigned, on the second cycle 

Vowel Lengthening applies, and then again the stress rule. 

In Odden (1990) it is suggested that we may avoid cyclic stress assignment 

by making the relevant rules sensitive to lexical stresses. Indeed, words like 

trompet have exceptional stress in that normally words ending in a VC syllable 

do not have final stress. However, lengthening does not occur if the base word 

is monosyllabic, that is, in words for which the location of the main stress is 

completely predictable, as in tonnage 'number of tons' derived from ton 'id.', 

or in blokkeer 'to block' derived from blok 'block'. Another interesting case is 

librettist [librctistl 'id.' derived from libretto /lihrtto/ 'id.'. The base word has 

a regular, penultimate main stress. After regular deletion of the final vowel of 

the base word, it is the /e/ which is potentially subject to lengthening, but this 

is blocked by the presence of the regular main stress on the relevant syllable, as 

assigned on the first cycle. Thus it is correctly predicted that the /e/ does not 

lengthen, although no lexical stress is involved here. 

Non-native words in -or and -on are special in that vowel lengthening also 

occurs before the plural suffix -en /an/ (note that normally the plural suffix 

after syllables without main stress is /s/): 

(65) doct|D]r 'id.' doct[o|r-en 'doctors' 

profess|o|r 'id.' profess|o]r-en 'professors' 

juni[D)r 'id.' juni[o]r-en 'juniors' 

seni|D]r 'id.' seni[o|r-en 'seniors' 

mot[D]r 'engine' mot[o]r-en 'engines' 

dem|D]n 'id.' dem|o)n-en 'demons' 

elektr|D|n 'id.' elektr[ojn-en 'electrons' 

We may also select the other plural suftix of Dutch, -.v. In this case, the /.->/ is 

not lengthened. Also, the main stress of the word, which is on the syllable 

before -or, will then remain on that syllable, whereas it is shifted to the syllable 

with [o] in case the vowel is lengthened as shown in (66).

(66) doctors 

professors 

Juniors 

seniors 

motors 

démons 

eléktrons 

4.4. ALLOMORPHY IN THE NON-NATIVE LEXICON 

doctoren 

professoren 

junioren 

senioren 

motoren 

demonen 

elektronen 

The implication of this phenomenon is that these plural forms have to be 

exceptionally fed back into the non-native component of the lexicon which 

will then make them undergo vowel lengthening and the Main Stress Rule (cf. 

Section 5.2). The Main Stress Rule (MSR) will then assign main stress to the 

syllable with the lengthened vowel (67). 

(67) 1st cycle: motor 

MSR 0 

2nd cycle 

motor + an 

Lengthening 

o 

MSR 

ó 

Phonetic form [motóran] 

4.4.5. Other cases of allomorphy 

Allomorphy in the non-native lexicon is not restricted to the alternations 

discussed in the preceding sections. There are more examples, which often 

have an unsystematic character: we simply have to learn that a specific 

allomorph of a word has to be used when it enters derivation. The following 

list of examples is by no means an exhaustive list of all the possibilities. We 

clearly deal here with a 'pan-European' kind of lexicon of roots, since similar 

alternations are found across the learned vocabularies of European languages: 

(68) Base word 

a. orkest 'orchestra' 

b. gymnasium 'id.' 

c. cursus 'course' 

d. minist|3]r 'id.' 

apost[o|l 'apostle' 

c. g[c:]ne 'shame' 

ƒ. dimensie 'dimension' 

functie 'function' 

g. Plato 'id.' 

h. Portugal 'id.' 

i. Jezus 'Jesus' 

j. Paulus 'Paul' 

k. trauma 'id.' 

drama 'id.' 

Derived word 

orkestr-eer 'to orchestrate' 

gymnasi-ast '(lit.) gymnasium pupil' 

curs-ist 'student' 

minist[e]r-ieel 'ministerial' 

apost|o]l-isch 'apostolic' 

g[o]n-ant 'shaming' 

dimension-eel 'dimensional' 

function-eer 'to function' 

platon-ist 'Platonist' 

Portug-ees 'Portuguese' 

Jezu-iet 'Jesuit' 

Paulin-isch 'Pauline' 

traumat-isch 'traumatic' 

dramat-isch 'dramatic'


horizon 'id.' 

horizont-aal 'horizontal' 

/. chaos 'id.' chaot-isch 'chaotic' 

eros 'Eros' 

erot-isch 'erotic' 

m. orgel 'organ' organ-ist 'id.' 

n. commiss-ie 'committee' committ-eer 'to commit' 

In some cases the allomorphy is more systematic. There are roots ending in /tr/ 

or /yl/ such as arbiter /urbitr/ 'id.', center /scntr/ 'centre', regel /reyl/ 'rule', 

and filter /filtr/ 'id.' which occur without a schwa before non-native suffixes, 

but with a schwa in formations with a native suffix. The root ends in a cluster 

that cannot be syllabified because the final /r/ is more sonorous than the 

preceding obstruent, and thus the SSG would be violated if they formed a 

coda. In the non-native lexicon syllabification is made possible through 

attachment of a vowel-initial suffix. If no affixation takes place, the unsyllabifiable 

/r/ is rescued by inserting a schwa before the /r/ before the word 

enters the native morphology: 

(69) arbitreer, arbitrage versus arbiter, arbiteren 

centreer, centraal versus center, centeren 

reglement versus regel, regelen, regeling 

filtreer, filtraat versus filter, filteren, filtering 

The rule that inserts the schwa must therefore apply after non-native morphological 

processes, but before the application of the native morphology: 

(70) Insert /a/ before an extrasyllabic liquid. 

Another regularity is that the stem-final vowel of words may disappear 

before a vowel-initial non-native suffix if that stem-final vowel is not 

stressed. However, this vowel deletion is not systematic: 

(71) Amerika 'America' Amerik-aan 'American' 

Canada 'id.' 

Canad-ees 'Canadian' 

propaganda 'id.' propagand-ist 'id.' 

piano 'id.' 

pian-ist' 'id.' 

cello 'id.' 

cell-ist 'id.' 

solo 'id.' 

sol-ist 'soloist' 

inflatie 'inflation' inflat-oir 'inflatory' 

Exceptions are words like egoist 'id.' and maoist 'Maoist'. The final vowel of 

the base word does not disappear if it bears the main stress of the word, as in 

hobo 'oboe'-hoboist 'oboist'. 

In the following word pairs there is an alternation between /e/ or /e/ and 

schwa. The underlying /e/ and /e/ obligatorily change into a schwa in 

unstressed position: 

(72) juweel /jyuel/ 'jewel' juwel-ier [jyualiir) 'jeweller' 

profeet /protêt/ 'prophet' pröfet-éer [profateir) 'to prophesize' 

arrest /urest/ 'id.' 

àrrest-éer [urasteir] 'to arrest'

4-4- ALLOMORPHY IN THE NON-NATIVE LEXICON 

This alternation is related to the phenomenon of vowel reduction (Section 6.4): 

vowels, and in particular the /e/ and the /e/, optionally reduce to schwa in 

unstressed position. The difference is that in the words discussed here the 

reduction is obligatory. The rule can be stated as in (73). 

(73) Vowel Reduction [—cons] 

+mid 

— round 

Condition: 

In unstressed syllables only 

The rule is lexically governed. For instance, in percentage /persentaZja/ 'id.', 

derived from percent /pcrscnt/ 'per cent' the vowel of the unstressed second 

syllable can still be pronounced as a full |ej. Optional vowel reduction is 

accounted for by a separate rule (cf. Section 6.4). There is also variation 

between speakers here: some are rather categorical in their reduction of these 

vowels, and thus they may have more words marked in their lexicon as subject 

to this rule. 

Note, that, as in the case of vowel deletion, information about the stress 

pattern of the words involved may be crucial, either that of the base word, as in 

the case of vowel deletion, or that of the derived word, as in the case of vowel 

reduction. 

It is not always clear whether the allomorphy occurs in the stem or in the 

suffix. In the case of dimensioneel one might either say that -eel selects a 

specific allomorph dimension, or that dimensie selects a specific allomorph of 

the suffix -eel, viz. -oneel. If the allomorphy shows up in related words with 

different suffixes, there is evidence that it is a case of root allomorphy. For 

instance, the allomorph dramat of drama 'id.' occurs in dramatisch 'dramatic', 

dramaturg 'dramatist' and dramatiek 'dramatics'. A case of allomorphy of the 

suffix can be seen in the word pairs perceptie 'perception'—perceptueel 

'perceptual' and contract 'id.'—contractueel 'contractual', and in exponent 

'id.'—exponentieel 'exponential' an allomorph with /i/ shows up. If we 

consider the /u/ and the /i/ part of the suffixes, we can predict the deletion 

of the stem-final /i/ of 'perceptie, and also the loss of the final schwa of principe 

'principle' in the adjective principieel 'principled'. 

Clearly, we have reached here the borders of phonology, and such observations 

could be relegated to morphology as well. That applies even more so in 

pairs of related words like the following: 

(74) plagiaat 'plagiary' plagieer 'plagiarize' 

kandidaat 'candidate' kandideer 'to nominate as candidate' 

where a whole morpheme -aat disappears before the next suffix. Such rules 

have been called truncation rules by Aronoff (1976).


In sum, the non-native roots and affixes that are used in Dutch appear to 

have their own systematics and irregularities that do not extend to the native 

lexicon. The relevant rules will therefore be assigned to a specific non-native 

subpart of the lexical phonology of Dutch. Moreover, words will sometimes 

have to be listed with their specific allomorphs in the lexicon. 

4.5. ALLOMORPHY IN THE NATIVE LEXICON 

Above, I discussed some examples of regular allomorphy in specific native 

suffixes. Some suffixed words exhibit allomorphy in that there is an alternation 

between an allomorph with initial schwa, spelled e and one without it. This 

applies to the suffixes -ling, -loos, -lijk, and -nix: 

(75) twee-ling 'twins', naar-ling 'pain in the neck', verschopp-e-ling 'outcast', 

hov-e-ling 'courtier', lamm-e-ling 'dead loss', vertrouw-e-ling 

'confidant' 

baan-loos 'jobless', bodem-loos 'bottomless', werk-loos 'unemployed' 

moed-e-loos 'dispirited', werk-e-loos 'idle' 

aanzien-lijk 'considerable', aandoen-lijk 'moving', klaag-lijk 'plaintive' 

graf-e-lijk 'of a court', werk-e-lijk 'real', gemoed-e-lijk 'agreeable', 

vrouw-e-lijk 'female' 

vuil-nis 'trash', treur-nis 'sorrow' 

begraf-e-nis 'burial', laf-e-nis 'refreshment', vermoei-e-nis 'weariness' 

The schwa-initial allomorph seems to be the preferred one after an obstruent, 

but there are no absolute rules here, as the examples lammeling and werkloos 

illustrate. 14 One may raise the question whether the schwa belongs to the 

suffix, or to the stem. As a matter of fact, there is evidence that the schwa 

should indeed be considered here an extension of the stem: -ling and -loos 

belong to the set of suffixes that form a prosodie word of their own, and the 

schwa clearly does not belong to that prosodie word, but to the preceding 

one. (Prosodie words cannot even be schwa-initial!) For instance, the 

prosodie structure of moedeloos '(lit.) without courage, dispirited' is 

( (mu) CT (d3) 0 )o)( (losJo)^. So these schwas might be seen as kinds of linking 

phonemes between stem and suffix (cf. Section 3.6 on linking phonemes). 

Native roots also exhibit a number of types of allomorphy, with the common 

property that the relevant alternations are always lexically governed. For 

instance, a completely irregular pattern of vowel alternations is found for 

i4 

The phonological status of this schwa was discussed in Kooij ( 1977). Interestingly, Shannon 

( 1991 ) argued that the insertion of a schwa alter obstruents has to do with the fact that a cluster of 

an obstruent followed by a sonorant consonant does not form an optimal syllable contact. In 

optimal syllable contacts, the sonorant consonant precedes the obstruent (cf. Section 3.3.5). This 

also explains why it is sonorant-inilial suffixes that exhibit this allomorphy.

4-5- ALLOMORPHY IN THE NATIVE LEXICON 87 

nouns derived from strong or ablauting verbs." Sometimes, the vowel of the 

de-verbal noun corresponds with one of the vowels of the verbal paradigm, but 

even that is not always the case, as the examples given here show: 

(76) Verb Verbal form Noun 

komen 'to come' k[D|m (pres. sg.) k[3]mst 'coming' 

snijden 'to cut' sn[eld (past sg.) sn[e]de 'cut' 

sluiten 'to close' 

slot 'lock' 

ruiken 'to smell' 

reuk 'smell' 

zingen 'to sing' 

gez[a]ng 'song' 

geven 'to give' g[a[f (past sg.) g|a]ve 'gift' 

g(a|ven (past pi.) g[i]ft 'gift' 

Therefore, it makes no sense to analyse such vowel alternations in terms of 

rules. 

Another kind of irregular vowel alternation is found for a few words, where 

/y/ alternates with /DU/, as in nu-nou 'now', duw-douw 'push', and stuw-stouw 

'to stow'. They differ in that the form with the diphthong is more informal, 

and may have different connotations and different conditions of use, as is 

particularly clear for nu versus nou. 

There are also lexically governed alternations where some regularity is 

involved. They are discussed in the next subsections. 

4.5.1. Vowel lengthening 

In a number of words a short vowel alternates with a long vowel, as in the 

following pairs of singular and plural nouns: 

(77) Singular Plural 

schip 'ship' [sxip] schepen (sxepan) 

smid 'smith' [smit] smeden [smedan] 

spel 'game' [spel] spelen |spelan| 

weg 'road' |uex| wegen [ueyon] 

god 'god' [y:>t| goden [yodon] 

hol 'hole' [hol] 

holen [holan] 

lot 'id.' [lat] 

loten [lotan] 

hertog 'duke' [hertox] hertogen [hcrtoyan] 

dal 'valley' [dul) dalen [dalan] 

bad 'bath' [bat] 

baden [badan] 

stad 'city' |stut| 

steden [stedan] 

The last example is exceptional in that |u| alternates with [e] rather than with 

|;i|. Note that (e| is the long counterpart of both [i] and [e], as discussed in 

Section 2.5.2. 

" The vowel alternations in the verbal paradigms of these strong verbs are a purely morphological 

matter, and will therefore not be discussed in this book.


Historically, this alternation is due to a process of open-syllable lengthening. 

Synchronicaliy, however, it is a purely lexically governed phenomenon. It also 

occurs in other types of complex word, as in sch\e.\peling 'crew member', 

sm[t\derij 'smithy', l[o]terij 'lottery', hert[o]gelijk 'ducal' and st[e\delijk 

'urban'. But even for those morphemes that exhibit this alternation we also 

rind the allomorph with the short vowel in open syllable, as in schipper 

[sxipar] 'skipper', spelletje [spelatja] 'game', goddelijk (y^dalak] 'divine' 

and badderen [budaran] 'to take a bath' (informal). Inversely, we find the 

long vowel in all forms of the following verbs, even when the relevant syllable 

is closed, as is the case with the first-person singular forms: 

(78) inscheep 'to embark' /insxep/ 

smeed 'to forge' /smed/ 

speel 'to play' 

/spel/ 

verafgood 'to idolize' /vcrufyod/ 

loot 'to cast lots' /lot/ 

daal 'to go down' /dal/ 

baad 'to take a bath' /bad/ 

The long vowel is also found in a closed syllable in a compound like 

scheepsheschuit '(lit.) ship's biscuit'."' 

4.5.2. /da/-0 alternations and /a/-0 alternation* 

In medieval Dutch a phonological process took place in which the sequence 

/da/ was deleted both word-finally and in intervocalic position. It was a 

process subject to lexical diffusion, that is, it affected a number of words, 

one by one, and then it stopped. Consequently, there are a number of words in 

Dutch with two forms, one with, and one without /do/. As may be expected, the 

two allomorphs often got different meanings, or at least a stylistic differentiation 

(the Je-less allomorph is more informal, or the allomorph with de 

archaic): 

(79) Word-finally 

snede 'cut' 

snee 'cut, slice (of bread)' 

armoede 'poverty' armoe 

lade 'drawer' 

la 

weide 'meadow' wei 

koude 'cold' 

kou 

heide 'heath' 

hei 

moede 'tired' 

moe 

mede 'with' 

mee 

/.eide 'said' 

zei 

'' These alternations are discussed in Zonneveld (1978) where a so-called abstract analysis of 

them is given, and in Booij (19810: 69-72) who argues against the abstract analysis.

4-5- ALLOMORPHY IN THE NATIVE LEXICON 89 

IntervocalicaUy 

broeder 'brother' broer 

moeder 'mother' moer 'female animal, female screw' 

ijdel 'vain' 

ijl 'thin' 

buidel 'pouch' 

buil 'lump' 

voeder 'animal food' voer 

neder 'down' 

neer 

More examples are given in Zonneveld (1978: 73-86). An example of a 

resulting meaning difference is the case of broeder, a word that can also be 

used for someone in a religious order, for a male nurse, and as a form of 

address in church meetings, versus broer that can only be used in its literal 

meaning 'brother'. One even gets subtle meaning differences in verbs derived 

from these nouns. For instance, the verb voeren 'to feed' can be used for 

referring to the feeding of both animals and children, whereas voederen can 

only refer to the feeding of animals. In complex words, usually only one of the 

allomorphs is correct: we do not find heiveld, only heideveld 'heath', nor 

broertwist besides the correct hrocdertwist 'row between brothers'. 

The phonological generalization that can be made is that de deletes after 

long vowels and diphthongs, in word-final position, or before /y, k, m, 1, r/ 

(Zonneveld 1978: 73). However, it will be clear that this generalization can 

only be seen as a via rule. Moreover, the process only affected native words: 

words of Romance origin do not alternate. 

Another phonological process that was active in medieval Dutch is wordfinal 

schwa deletion after a voiced coronal obstruent, which resulted in word 

pairs like the following: 

(80) einde 'end' eind 

stonde 'hour' stond 

aarde 'earth' aard 

gaarde 'garden' gaard 

keuze 'choice' keus 

wijze 'manner' wijs 

leuze 'slogan' leus 

Since this proces was also subject to lexical diffusion, that is, affected words on 

a word-by-word basis, there are regions of Dutch where the schwa-deletion rule 

affected a word before df-deletion, thus bleeding the latter rule. For instance, in 

North Holland, weide became weid, a form to which ^-deletion does not apply, 

whereas in standard Dutch the form wei occurs (cf. Verhoeven 1974). 

Schwa deletion also occurs before the adjectival suffix -.y, and in compounds 

at the end of the first prosodie word (Van Marie 1982). The absence of the 

schwa is lexicalized in the case of compounds: 

(81) aarde 'earth' aards 'earthly' 

Drente'id.' 

Drents'of Drente' 

(geographical name)


Suriname 'Surinam' 

aard(e) 'earth' 

eind(e) 'end' 

schande 'shame' 

Surinaams 'Surinamese' 

aard-appel 'potato' (*aarde-appel) 

aard-beving 'earthquake' (*aarde-beving) 

aard-bei 'strawberry' (*aarde-bei) 

eind-verslag 'final report' (*einde-verslag) 

schand-knaap 'catamite' (*schande-knaap) 

schand-paal 'pillory' (*schande-paal) 

schand-vlek 'blot' (*schande-vlek) 

4.5.3. Idl-deletion and Idl-weakening 

In a number of words, the underlying /d/ optionally alternates with [jj or [u], 

before a schwa, which is usually either an inflectional suffix or the first 

segment of a derivational suffix. 37 The form with the glide has a more 

informal character: 

(82) a. breed /bred/ 'wide' bred+e |breja| 'wide' (attr.) 

rood /rod/ 'red' rod+e |roj3] 'red' (attr.) 

goed /yud/ 'good' goed+e [yuja] 'good' (attr.) 

goed+ig [yujax] 'complacent' 

b. goud /youd/ 'gold' goud+en [youuan] 'golden' 

oud /Dud/ 'old' oud+e [Duua] 'old' (attr.) 

These alternations are lexically governed. For instance, it is impossible to 

pronounce hoeden 'hats' /hud+an/ as [hujan], and in standard Dutch the 

phonetic realization [Duuars] for ouders 'parents' /nud+ar+s/ is also impossible, 

even though the morpheme oud 'old' does exhibit this alternation in 

the inflected form oude. The same applies to ouderling 'presbyter', 

whereas, on the other hand, the form with |u] is the only possible form in 

ouwehoeren 'to nag' (from ouwe hoer '(lit.) old whore, nagger', where, 

again, ouwe is the only possible form: oude hoer is only possible in the 

literal interpretation 'old whore'). These facts show that, although certain 

phonological generalizations can be made, yet all the allomorphs have to be 

listed in the lexicon, and complex words will be entered in the lexicon with 

the one or two allomorphs that they allow for. This is reflected by the fact 

that in Dutch orthography the occurrence of the glide in cases like ouwe is 

represented in the spelling. 

In the case of morphemes with a front vowel or a diphthong the glide that is 

created is homorganic with the preceding vowel. Morphemes with back vowels 

are different in that they have /j/ rather than the homorganic glide /u/. I will 

therefore assume that the /d/ alternates with /j/ after steady-state vowels, and 

17 An exception is heneden 'down' wilh the allomorph |bonepn| for which there is no evidence 

for internal morphological structure. 

K 

/onneveld ( 1978). Some dialects of Dutch, particularly in the south-western part of Holland, 

allow for considerably more cases of /d/-weakening, cf. Zonneveld (WK).

4-5- ALI.OMORPHY IN THE NATIVE LEXICON 91 

that the /d/ alternates with 0 after diphthongs. 39 In the latter case, the gap will 

he predictably filled with a homorganic glide. This is stated in (83). 

(83) a. d -> j / X X 

I 

[—cons] [—cons] 

h. d -» 0 / X X 

I 

I 

[-cons] [-cons] [-cons] 

There is a second process of /d/-deletion: a number of verbs ending in 

diphthong + /d/ have an allomorph without the /d/. In this case, there is no 

conditioning schwa environment, and, again, the /d/-deletion is lexically 

governed: 

(84) glijd 'glide' /gleid/ glij 

rijd 'ride' /rcid/ rij 

snijd 'cut' /sncid/ snij 

houd 'keep' /hsud/ hou 

lijden 'suffer' /leid/ lij 

Phonologically similar verbs like belijden 'to confess' and mijden 'to avoid' do 

not exhibit this alternation, thus showing the lexically governed nature of 

this alternation. The verbs involved are all high frequency strong verbs that 

form their past tense through vowel alternation (and sometimes additional 

modifications). The rule can be stated as (85). 

(85) d -» 0 / X X —]v 

I 

I 

[-cons] [-cons] 

Consequently, we also get two past-tense forms for these verbs, except for 

houden which has the extra irregular past tense form hield: 

(86) Present Past singular Past plural 

glijd/glij gleed/glee gleden/gleeën [glejan] 

rijd/rij reed/ree reden/reeën [rejan] 

snijd/snij sneed/snee sneden/sneeën [snejan] 

The [j] in the past tense plural forms results from Homorganic Glide Insertion. 

Note that, although we also have a special rule for d-j alternations, as pointed 

out above, this rule appears not to apply to past-tense forms such as vermeden 

'avoided' (pi.) and leden 'suffered' (pi.). In other words, it is not the d/j-rule 

that should be held responsible for the occurrence of glides in past-tense forms. 

By listing both the allomorph with /d/, and that without /d/ for the relevant 

w Zonneveld (1978: 32) presents a slightly different analysis in that he assumes thai /d/ also 

deletes aller front vowels, as in reden/reeën 'ride' (past pi.). In my analysis a form like ree is 

derived from the J-lcss verbal allomorph rij 'ride' (pres.). The latter form is derived by a separate 

rule for verbs ending in diphthong + /d/.


verbs, we predict that they will also have two past-tense allomorphs, and that 

we only find /j/ instead of /d/ in past-tense forms if the correlating presenttense 

stem allows for /d/-deletion. This is a correct prediction. For instance, 

since belijden 'to confess' has no /d/-deletion, the past tense plural form 

beleden cannot be realized as [balejan], and this is a systematic pattern for 

these verbs. Similarly, although we have rijer 'rider' derived from the allomorph 

rij 'to ride', we do not have belijer 'confessor' alongside with belijder 

because the verb belijden does not have the allomorph belij. 4>> 

4.5.4. 0-1 kl alternations 

In a number of words the morpheme-final /rj/ alternates with the cluster /rjk/ 

(Booij 1980, Trommelen 1984): 

(87) oorspro|rj] 'origin' oorspro|rj|k+elijk 'original' 

toega[rj| 'access' toega[rj]k+elijk 'accessible' 

aanvalrj] 'start' aanva[rj)k+elijk 'initial' 

jo[rj] 'young' jo[rj]kvrouw 'damsel' 

spri[rj] 'to jump' spri[rj|khaan 'grasshopper' 

la[rj] 'long' 

la(rj]kmoedig 'long-suffering' 

The historical background of this alternation is that the morpheme-final nasal 

derives from the sequence /rjg/. The /g/ devoiced in certain positions, among 

others at the end of a syllable (the rule of Syllable-final Devoicing), as 

illustrated by jonkvrouw with the structure ( |jong| A [vrouw| N ]N '(lit.) young 

woman'. 41 Later, the velar plosive /g/ developed into a velar fricative, and was 

dropped after nasals because Dutch does not allow for codas with nasal + 

fricative. Such clusters only occur as heterosyllabic clusters as in fungeren 

/fvrjyeran/ ' to function'. Synchronically, we have to list words with the 

allomorph ending in |k| in the lexicon, because we also find cases where the 

|k] does not show up in exactly the same environment: 

(88) jong [jDrjJ 'young' jongvee Lpf) ve ] 'young cattle' 

bang (barjl 'afraid' bangelijk [barplak] 'fearful' 

spring [sprirjl 'to jump' springnet |spnrjnet] '(lit.) jumping net' 

In other words, the alternation is not productive any more. 

4.5.5. Fricative devoicing 

As pointed out in Booij (1977: 87) the final fricatives of native morphemes 

devoice before the schwa-initial suffixes -elijk and -enis. Since these fricatives 

" The past-tense forms of zullen 'will' are irregular in that the sg. form is always without /d/ 

(zou), whereas the plural has both forms, zouden and zouwen |z3uuan]. Compare also kon 'could' 

(sg.) with konden 'could' (pi.), with the same type of irregularity. 

1 

The devoicing of obstruents before -elijk will be discussed in the next subsection.

4-6. ALLOPHONIC RULES 93 

will occur in onset position this devoicing cannot be related to the automatic P- 

rule of Syllable-final Devoicing. Moreover, the devoicing is very restricted: it 

applies to labial and coronal fricatives before -elijk, and to labial fricatives 

before -enis: 

(89) a. graaf/grav/'earl' 

lijf /leiv/ 'body' 

vrees /vrez/ 'fear' 

begraaf/bayrav/ 'to bury' 

droef /druv/ 'sad' 

laaf /lav/ 'feed' 

compare: 

b. heb /heb/ 'to have' 

land /lund/ 'land' 

besnijd /basneid/ 'to circumcise' 

verrijs /vcrreiz/ 'to rise' 

gra(f lelijk 'of a court' 

lij[f lelijk 'bodily' 

vre[s|elijk 'horrible' 

begra[f Jenis 'burial' 

droeff Jenis 'sadness' 

la|f jenis 'refreshment' 

onhe|b]elijk 'rude' 

lan|d)elijk 'rural' 

besnij[d]enis 'circumcision' 

verrij[z]enis 'resurrection' 

The generalizations involved are expressed by the MP-rules (90a, b). 

(90) a. p-cons 

+cont 

Lab 

b. —cons 

+cont 

Cor 

[+voice| / — alok, anis 

l+voice] / — alak 

4.6. ALLOPHONIC RULES 

Allophonic rules differ from the automatic P-rules discussed above in that they 

do not involve alternations between segments that occur in lexical, that is, 

underlying, forms. They have the same property of being determined by 

phonological conditions only. 

An allophonic rule that was already mentioned in Section 2.2.1 is the rule 

that lengthens long vowels before /r/. It is a rule that applies obligatorily, 

within the domain of the prosodie word. The effect of the rule is shown below 

(cf. Streekstra and De Graaf 1979): 

(91) koraal 'choral' [ko:ral| 

Erik 'id.' 

[eirik] 

oregano 'id.' |o:reyano| 

pirouet 'pirouette' |pi:ruuet| 

uur 'hour' 

[y:r| 

boer 'farmer' [bu:r] boer+en (pi.) [bu:ran] 

boor 'drill' [bo:r] bor+en (pi.) |bo:ran]

94 WORD PHONOLOüY 

kaars 'candle' [ka:rs] kaars+en (pi.) [kaïrsan] 

toorn 'rage' [to:rn] toorn+en 'to rage' |to:rnan| 

koers 'direction' |ku:rs| koers+en 'to direct' |ku:rson| 

Lengthening does not apply in the past-tense forms of some strong verbs that 

have exceptional rhymes in which a long vowel is followed by two (nonappendix) 

consonants in the coda, as in zwierf [zwirf] 'wandered', wierp [uirpl 

'threw', bedierf [badirf] 'spoiled', verwierf [veruirf] 'acquired', and stierf 

|stirf| 'died'. These facts are accounted for if we assume that the underlying 

vowel of these forms is /i/ rather than /i/, which is then turned into (i| after 

lengthening of long vowels before /r/. This solution is possible because we do 

not find combinations of/i/ and /rp/ or /if/ at the underlying level. Moreover, it 

explains why at the surface these syllables can violate the Maximal Rhyme 

Constraint. Lengthening in these exceptional morphemes is also impossible in 

plural forms, where the /r/ is syllable-final: 

(92) zwierven |/.i>irvon| 

wierpen [uirpsn] 

bedierven [bodirvon] 

verwierven |vcruirvon| 

stierven [stirvan] 

In compounds in which the second constituent begins with an /r/, this /r/ 

cannot have a lengthening effect because it does not belong to the same 

prosodie word. This explains the contrast in the realization of vowels in pairs 

of words like the following: 

(93) keur 'to select' [k0:rj keur-ing 'selection' |k0:nrj| 

versus 

keu-ring 'pig's 'ring' morphological structure [keu| N |ring| N ] N 

phonetic form [k0nrj| 

Lengthening also applies to the vowel of the non-native suffix -cur, as in 

direct+eur 'director' (di:rtkt0:r|. However, in the suffix sequence -eux-c 

derived from -eur-e the vowel is not lengthened. This confirms that lengthening 

is a word-level P-rule: if it were a cyclic rule we would get a lengthened 

/0:/ in the sequence -euse. In sum, the lengthening rule must be considered a 

word-level rule that applies in the domain of the prosodie word. 42 Its post- 

2 This rule is a lexical rule since il is an obligatory rule of word phonology. Thus, it forms a 

problem for the hypothesis thai lexical rules are structure preserving (Kiparsky 19X5). i.e., (hat they 

do not create new types of segment, but only segments which occur at Ihe underlying level. A 

similar problem was noted for German by Hall (19X4): (he allophomc rule lhal accounts foi the 

occurrence of the palatal allophone |c| and the velar allophone |x| of the palalal/alvcolar fricative 

after front and back vowels respectively, is also a lexical rule. The same point (die existence of 

non-structure-preserving lexical rules) is made by Harris (19X7. 19X9). As proposed by Hooi j and 

Rubach ( 19X7) and supported by Iveison and Salmons ( 1992) who |x>ml out that the German rule is 

a word-level rule as well, the hypothesis of structure prcscivingness should be restricted to the 

cyclic rules, i.e., word-level rules need not he strucluiv-preserving. Thus, the rule of vowel 

lengthening discussed here does not form a problem loi the theory since it is a word-level rule.

4-6. ALLOPHONIC RULES 95 

cyclic (word-level) nature follows from the fact that the sequence long vowel + 

/r/ never forms a derived environment. 

A second allophonic rule is the rule that palatalizes the coronal obstruents 

and nasals /s, z, t, n/ before /j/. This kind of palatalization is the shift of the 

primary place of articulation, not the addition of a second place of articulation: 

(94) katje 'kitten' [kufja] 

tasje 'little bag' Itujp] 

jury 'id.' 

l3Jy:ri| 

sjaal 'shawl' [Jjal] 

atjar 'id.' 

[utjar] 

pasja 'pasha' [pujja] 

In the case of handje 'hand' (dim.) we even get double palatalization: (hunt jo| 

(or [hanca]). This example shows that palatalization also affects the /n/. Note 

that while the /j/ is considered to be [Dorsal), the palatalized coronals are 

usually considered to be [Coronal, —anterior], that is, postalveolars. 

Palatalization should be expressed as a spreading rule. Sagey (1986: 108-9), 

who discusses the same rule for English, interprets this process as a rule that 

spreads [—back) to the Coronal node, with the concomitant effect that the 

Coronal node is specified as [ -anterior] (underlyingly, there is no specification 

for [anterior] in Dutch). The rule is given in (95). 

(95) Palatalization O 

I 

|-appr| l+voc] 

I 

I 

Cor Dors 

x J 

I-back| 

The rule applies obligatorily within prosodie words, and optionally across 

prosodie word boundaries. 41 

An alternative and simpler analysis of palatali/ation is presented in Lahiri and Hvcrs ( 1991) 

who argue lhal the l]l and the front vowels have the Place node [Coronal, -anterior], and that 

palatalization is spreading of Ihis Place node. In Ihis way, palatali/.ation can be expressed more 

dmvtly as sprcadinjj ul the Ailic.ulator node. The same point is made in Clements and Hume 

( 1 {) 'H). Jacobs and Van de Weijer (1992) proposed that the /j/ has a Place node lhal dominates both 

[Coronal, —anterior] and [Dorsal, —back], and that Ihis Place node is spread to the preceding 

coional consonant. A similar position is advocated by Pulleyblank (1989). 

Lahiri and Evers (1991: 99) also point out lhat there is variation between speakers as to the 

phonetic rcali/ation of the A j/ sequence in diminutive suffixes. Whereas in my dialect it is indeed a 

l>alaiali/cd /I/: |t'|, some speakers reali/e this sequence moie like |ts'|.

WORD STRESS 


Dutch is a language with word stress. That is, one of the syllables of a word is 

perceived as the prominent one, the main stress of the word. In words with 

more than one syllable, other syllables may also have stress, of a lower degree, 

so-called secondary stress. 

The phoneticians Cohen, 't Hart, and Collier have shown that main stress is 

realized as a pitch movement on the relevant syllable, if the word in question 

bears a sentence accent.' In other words, a syllable with main stress is a 

potential locus of sentence accent. A pitch rise in the most prominent syllable 

of a word will give a sentence accent to that word if the pitch rise starts before 

the vocalic part of the syllable. Pitch lowering assigns accent if the lowering 

starts after the beginning of the vocalic part of the stressed syllable. In other 

words, it is the timing of the pitch movement that is crucial. Thus we get 

intonational patterns like the following, the so-called hat pattern which is 

mein zon u i l an l e u z i n 

l 

Time (sec.) 

1.5 

FIG. 5.1. Stylized pitch contour of the sentence Mijn :i>on wil een leeuw zien 'My son 

wants to see a lion' 

Source: Based on Nooteboom and Cohen (1976: 151) 

1 

Cf. Cohen and 't Hart (1967), 't Hart and Cohen (1973), 'I Hart and Collier ( l975),.Collier and 

'I Hart (19X1), Nooteboom and Cohen (1976), 't Hart, Collier, and Cohen (1990).

5-1. INTRODUCTION 97 

frequent in Dutch (Nooteboom and Cohen 1984: 156-7), with two accent 

assigning pitch movements (Fig. 5.1). 

The accents are realized as pitch accents that consist of movements with 

respect to the low and the high declination line for F 0 , the fundamental 

frequency. 

This does not mean that stress is to be equated with pitch movement. Stress 

is a more abstract property, which also manifests itself in vowel duration and 

(lack of) vowel reduction, and these phenomena also play a role in the 

perception of syllables as being stressed (Beekman 1986). Note also that the 

location of stress plays a role in some morphological rules and lexical 

phonological rules, that is, at a level where pitch movements are not available 

yet. 

Moreover, a pitch accent can also be placed on a syllable without lexical 

stress, for instance for metalinguistic reasons, as in Heb je versie of verbum 

gezegd! 'Did you say versie ('version') or verbum ('verb')?' 2 (The words 

versie and verbum have lexical stress on the first syllable.) 

In recent years some phonologists have argued, also with respect to Dutch, 

that sentence stresses should not be seen as pitch movements imposed on the 

basic declination line, but as tone elements (High or Low) associated with the 

syllables that bear sentence stress, with predictable tone contours between 

the sentence stresses. 1 This approach maintains the insight that sentence stress 

is related to pitch movement. 

Secondary stress is determined by a rhythmic principle, that is, it creates an 

alternation of stressed and unstressed syllables. For instance, in the word 

automaat 'automaton', main or primary stress falls on the final syllable, and 

secondary stress on the first syllable, whereas in dominee 'parson' we find the 

reversed stress pattern. 

Both primary and secondary stress have a potential effect on duration. For 

instance, in a pair like bacterie ' bacterium '-difterie 'diphtheria' the identical 

second syllables (te) 0 differ in duration because this syllable only bears 

primary stress in the first word. And when automaat /nutomat/ is pronounced 

in isolation, the syllable (mat) c has the longest duration, the unstressed syllable 

(to) a is the shortest, and the duration of the secondary stressed syllable is in 

between that of the other two (Nooteboom 1972, Slootweg 1988) which shows 

that secondary stress also has effect on duration. 4 The imposed stress pattern 

also has an effect on the possibility of vowel reduction: only unstressed vowels 

can be reduced to schwa. That is, only the /o/ of (to) 0 can be reduced to a 

' Example from Sluijter (1992). 

' Cf. Gussenhoven ( 1984, 1988. 1991 ), Ladd (1992) and the references given there, and Van den 

Berg, Gussenhoven, and Rietveld (1992). Critical comments on this view are given in Collier 

(1989). 

4 The duration of a syllable also depends on its position in the word. As Slootweg (1988: 14) 

points out, '|T|he duration of the unstressed linal syllable is nearly as long as the duration of a 

stressed one in the same position.' In other words, the final syllable af automaat would have a long 

duration anyway, even if it were unstressed.

9ö 

WORD STRESS 

schwa, resulting in the phonetic form [automat]. Secondary stress also plays a 

role in determining the place to which a stress is moved in case of stress shift, 

as will he illustrated below. 5 

Following Liberman and Prince (1977), Prince (1983), Selkirk (1984/7), and 

Halle and Vergnaud (1987) I will represent the stress pattern of a word by 

means of a grid. 6 Each syllable has an entry on line 0 of the grid, indicated by 

'-', the place holder of the syllable on the grid. Each syllable with stress will 

get an asterisk (*) on line 1, the foot level, where metrical feet will be created. 

The optimal Dutch foot consists of a stressed syllable, followed by an 

unstressed one. On the next level, line 2, the word level, the syllable with 

main stress will have an entry. This also applies to monosyllabic words. In 

other words, the degree of stress is indicated by the number of asterisks that a 

syllable has on the grid. So the stress representation for the words automaat 

and dominee will be as follows: 

( 1) au to maat do mi nee 

line 0 

(* ) (*) (* ) (*) line 1 

* * line 2 

The parentheses indicate that a stressed syllable (a 'peak') forms a constituent 

(a foot) with a following unstressed syllable (a 'trough') if possible. For 

convenience's sake, 1 will usually omit line 0 in the stress representations, 

and also the parentheses that indicate constituency, unless it is crucial for the 

analysis. 

In compounds, each prosodie word has a stress grid. In Dutch compounds, 

the primarily stressed syllable of the first constituent becomes the main 

stressed syllable of the whole compound. So we add an asterisk for that 

syllable on the next level, line 3, as illustrated for the compound dorpsdomincc 

'village parson, rector': 

(2) dorps do mi nee 

* line 1 

line 2 

* line 3 

s 'I'lic relation between accent and duration in Dutch is also dealt with in Felting ( 1991 ) and 

Sluijter (1992). 

'' Kager (19X9) also uses the grid for the representation of the stress patterns of Dutch words. 

whereas Trommelen and /.onneveld (19X

5-2. MAIN STRESS PATTERNS 99 

This gives rise to a so-called stress clash, a clash between two adjacent 

stresses. A situation of stress clash is defined as a situation in which two 

asterisks are adjacent at two consecutive levels (Liberman and Prince 1977). It 

is indicated here by the boxed part of the representation. The only way to 

remedy this is to shift the line-2 asterisk of the syllable (do) a to the last 

syllable. We cannot shift the asterisk to the syllable (mi) a because a wellformedness 

condition on grids is that a syllable that has an asterisk at line n, 

also has to have an asterisk at line n — 1 (Prince 1983). Also, we cannot move 

the line-3 asterisk to the syllable (do) 0 because in that case the stress clash still 

exists. So, the stress clash is resolved by the operation Move *, with the 

following effect (called Trochaic Reversal): 

I nti'/iaic Reversal 

(3) dorps do mi nee 

* * * line 1 

* * line 2 

* line 3 

In Section 5.2. I will discuss the stress patterns of underived words and nonnative 

complex words. The subsequent sections deal with the stress patterns of 

the different categories of morphologically complex words. 

5.2. MAIN STRESS PATTERNS 

There are two general principles for the distribution of stresses (primary or 

secondary) in all Dutch words: 

(4) Schwa Restriction 

A syllable headed by a schwa never receives stress. 

(5) Optimal Grid Principle 

A prosodie word has an alternating stress pattern. 

Principle (4) restricts principle (5). For instance, in the word sappelen 

/supDlan/ 'to toil', a stressed syllable is followed by two unstressed syllables: 

the final syllable cannot get the stress that would give the word an alternating 

pattern, because it contains a schwa. In formal terms, the Schwa Restriction 

means that only a syllable with a nucleus that has a Place specification can 

receive an entry on line 1. 

Let us now look in more detail at the location of primary stress in underived 

words. From the historical point of view, Dutch stress is a mixture of three 

patterns (Van Marie 1980): a Germanic pattern, with stress on the initial 

stressable syllable, a French pattern with stress on the final stressable syllable, 

and a Latin pattern with stress on the penultimate syllable, or on the 

antepenultimate if the penultimate is light (i.e. the rhyme of the penultimate 

consists of a short vowel):

IOO 

WORD STRESS 

(6) German: pijler /peilar/ 'pillar', bevel /bavel/ 'command' 

French: kanon /kanon/ 'gun', parade /parada/ 'id.' 

Latin: kanon /kanon/ 'canon', valium /valivm/ 'id.' 

This does not mean that stress is completely unpredictable from the synchronie 

point of view. Minimal pairs are scarce: kanon 'gun' v. kanon 'canon', and 

servies [servis] 'dinner service' v. Servisch |servis] 'Serbian'. The position of 

stress often correlates with segmental structure, as illustrated by the etymologically 

related pair of words pilaar /pilar/ 'p'\\\&r'-pijler /pcilar/ 'pillar'. In 

pilaar we have the French stress pattern, whereas in pijler we see the Germanic 

stress pattern: in Germanic, stress shifted to the first syllable. This led to 

reduction of the second syllable in that its vowel changed into a schwa. Moreover, 

stressed /i/s diphthongized, changing /i/ into /ei/. Put more generally: since 

the words in the different layers of the vocabulary exhibit partially different 

patterns of segmental composition, it is possible to formulate generalizations 

concerning the locus of main stress on the basis of segmental composition. 

However, many cases of unpredictable stress will remain. For instance, words 

ending in -ie have main stress on the penultimate syllable, on the antepenultimate 

syllable, or on the final syllable (words of French origin); compare: 

(7) a. olie /oli/ 'oil' 

kanarie /kanari/ 'canary' 

Hérry /hcri/ 'id.' (female name) 

bikini /bikini/ 'id.' 

b. alibi /alibi/ 'id.' 

Rfmini /rimini/ 'id.' 

c. melodi'e /melodi/ 'melody' 

chemie /xemi/ 'chemistry' 

energie /enerzji/ 'energy' 

Sofie /sofi/ 'Sophie' 

The regularities that are discussed here, appear to hold for both underived 

words and complex words derived by non-native morphology, but there are 

more exceptions in the subset of native underived words. The Main Stress Rule 

proposed here is a rule that looks at the last three syllables of a word, that is, it 

applies from right to left. The basic generalizations can be found in Van der 

Hulst (1984), Trommelen and Zonneveld (l989«), and Kager (1989), the latest 

in-depth studies of Dutch word stress. 7 They can be summarized as follows: 

(8) Words ending in a vowel: penultimate stress: olie 'oil', kanarie 'canary', 

etc., with two classes of exceptions: 

(a) some word have antepenultimate stress, for example, words in -ia. 

-io: aria 'id.', ratio 'id.', etc. 

7 Kager (1989) provides an excellent survey of the descriptive gcncrali/ations concerning the 

stress patterns of underived and non-native words ('level-1 stress'), with a wealth of examples, and 

also gives a critical summary of the theoretical literature on Dutch word stress before 19X9. 1 will 

not give the (long) list of references here since they can be found in Kager's book.

5.2. MAIN STRESS PATTERNS 101 

(b) some words have final stress: melodfe 'melody', trofee 'trophy', 

kadó 'gift', etc. 

(9) Words ending in long vowel + consonant (WC), short vowel + CC 

(VCC), or a diphthong (+ optional C): final stress: kanaal 'channel', 

perkament 'parchment', kopij 'manuscript', paradijs 'paradise'. Exceptionally, 

main stress may be located on a penultimate or antepenultimate 

syllable, subject to the following restriction: primary stress cannot be 

on the antepenultimate syllable if the penultimate is closed or contains 

a diphthong (Kager 1989: 227): lichaam 'body', hospitaal 'hospital', 

andijvie 'endive', fndex 'id.', olifant 'elephant', appendix 'id.', juffrouw 

'miss', sélderij 'celery'. 

(10) Words ending in short vowel + consonant

102 WORD STRESS 

In order to derive the other patterns, we have to make use of the notion 

'extrametricality': a word-final constituent, in this case the syllable, can be 

made invisible to the MSR by making it extrametrical (Hayes 1982). Extrametricality 

can either be imposed by rule, or be marked in the lexical representation 

of a word. For words ending in -VC we impose extrametricality by 

rule: 

(13) Extrametricality Rule 

Make word-final syllables with a VC-rhyme extrametrical 

This rule applies to words like bivak /bivak/ 'bivouac' and Michael /mixael/ 

'id.'. So we create a monosyllabic foot on the first syllable of bivak, ignoring 

the second syllable, and a binary foot on the part (mixa) of Michael. Of course, 

words in -VC that have final stress, must be marked as exceptions to this rule 

of extrametricality. Rule (13) expresses that a majority of words in -VC do not 

have final stress, but there is a substantial number of words ending in -VC with 

final stress, for example, karton 'cardboard' and kanon 'gun'. 

Extrametricality may also have to be marked on specific lexical items. At 

the underlying level, we do not have syllables, so we cannot mark syllables as 

extrametrical in underlying representations. Instead, we mark the relevant 

vowel, the head of the syllable, as extrametrical, and this property can then 

be percolated from the head to the syllable node, once the syllable nodes have 

been introduced by the syllabification algorithm. 

In' a word like olifant /olifant/ 'elephant', the /u/ will be marked as EM 

(extrametrical). Consequently, foot construction will apply to the part /oli/, and 

thus locate stress on the first syllable. Subsequently, a second stress will be 

introduced by the general rule that creates rhythmic alternations of stressed and 

unstressed syllables, and thus creates optimal grids, the Optimal Grid Rule. 

(14) /o 1 ! a n t/ 

l 

[EM] 

Syllabification 

a 

(a) 

l i f u n t 

Grid construction 

MSR 

Optimal Grid 

Note that the extrametricality feature is only relevant for the location of main 

stress, since the last syllable of, for example, olifant does get secondary stress.


Certain regularities as to extrametricality can be found. For instance, words 

ending in VC.ç, such as climax /klimuks/ 'id.' and larynx /lanrjks/ 'id.' usually 

have prefinal stress, and thus their final syllables may be made extrametrical by 

rule. 

How do we account for cases of antepenultimate stress like calcium 'id.' and 

aria 'id.'? In these words, the last syllable should also be marked as extrametrical. 

There are certain generalizations involved here that have to do with 

the prefinal and final syllables since specific combinations lead to final extrametricality, 

in particular with penultimate syllables containing /i/. The historical 

background is that these are words from Latin or formed after a Latin 

model, and the corresponding /i/ in Latin was short, thus giving rise to light 

syllables, to be skipped in stress assignment. In addition, we find Germanic 

geographical names with initial stress and specific endings such as -uwe: 

(15) -ium /ivm/: gymnasium 'id.', geranium 'id.', critérium 'criterion' 

-i(C)o /iCo/: ratio 'id.', légio 'many', Pinócchio 'id.', libido 'id.', 

risico 'risk' 

-ina /ina/: pagina 'page', vagina 'id.' 

-ika /ika/: Amerika 'America', Afrika 'Africa', lógika 'logic' 

-ier /ior/: agrariër 'farmer', Australie 'Australia', Proletarier 

'proletarian' 

-icus /ikvs/: medicus 'doctor', lógicus 'logician', chemicus 

'chemist' 

-ia /ia/: 

paria 'pariah', aria 'id.', malaria 'id.' 

-uwe /yu9/: Bétuwe 'id.' (geographical name), Véluwe 'id.' 

(geographical name) 9 

Such generalizations concerning certain /(/-combinations (but not all, compare 

•\nlffde /svlrida/ 'sulphide' and marine /marina/ 'navy') and -uwe can be 

expressed by rules that predict the feature [extrametrical]. 

Words ending in -or are special in that the final syllable does not receive 

main stress, but the penultimate, which can be an open syllable: 

(16) motor 'engine' 

condensator 'id.' 

alligator 'id.' 

So here, the 'light' penultimate syllable is not skipped. The correct stress 

pattern can be derived by making the final /r/ in word-final -or extrametrical 

by rule. This more specific rule will take precedence over the general rule that 

makes final VC-syllables extrametrical. The MSR will then assign main stress 

to the penultimate syllable. 

A second type of lexical exception mechanism is to be introduced for words 

9 These words are exceptions to the generalisation in Kager and Zonneveld (1986) that it is the 

syllabic before the schwa that always receives main stress, i.e. pre-schwa syllables cannot be 

skipped in main stress assignment.


ending in vowels that have final stress, words like melodie and trofee. These 

words will have to be marked with an exception feature |+F) (for l+French)), 

and we will have to add to the MSR that a trough cannot dominate the feature 

[+F]. Thus, a monosyllabic foot will be created at the end of such words. 

Finally, there remain a number of words, in particular, geographical names, 

that have initial stress, which cannot be assigned by making use of extrametricality 

of the last syllable because they consist of four syllables, or three 

syllables of which the second is heavy. The only way to derive the correct 

stress pattern is to assume a minor stress rule for words marked as [+G] (for 

I+Germanic]) that assigns stress to the first stressable syllable. Examples are 

geographical names like: 

(17) Wägeningen /uayaniipn/, Éverdingen /evordirjan/, Ämerongen 

/amorDrjan/, Leeuwarden /leuurdan/ 'id.' 

Note that in Leeuwarden the penultimate syllable is closed, and hence it cannot 

be skipped by the MSR. Again, we might invoke redundancy rules to predict 

[+G|, for instance on the basis of the property 'geographical name'. The minor 

stress rule that is triggered by (+G] takes precedence over the Main Stress 

Rule, as predicted by the Elsewhere Principle. 

The (+GI rule also has to apply to a few quadrisyllable words of Latin origin 

with inital stress such as infinitief 'infinitive' and accusatief 'accusative' which 

emphasizes that features such as [+G] have no diachronic status. 10 

The way in which the MSR is formulated predicts how words that are 

incorrectly stressed behave. For instance, words like notulen /notylon/ 'minutes', 

emeritus /emeritvs/ 'id.', and normaliter /normaliter/ 'normally' require 

extrametricality to get the correct antepenultimate stress, and people indeed 

tend to give these words penultimate stress, as would be the case if there were 

no extrametricality (in that case the last vowel of normaliter is pronounced as 

schwa). Similarly, the word hysterie /histeri/ 'hysteria' which is marked as 

[+F|, is often pronounced with prefinal stress. Also, many speakers of Dutch 

pronounce Leeuwarden with prefinal instead of initial stress." 

The distinction between native and non-native words is not always easy to 

make. As long as a word does not form input for non-native word formation, 

we have no evidence. Non-native suffixes, on the other hand, form a clearly 

distinguishable class on the basis of morphology, and this correlates with their 

phonological behaviour. When they end in VVC or VCC, they always bear 

main stress, whereas non-native lexical morphemes ending in VVC or VCC 

can be extrametrical, as illustrated by index and larynx. The suffixes -ica and 

•ikux (as in elektron-ika 'electronics', elektron-ikus 'electric engineer') have 

'" However, some speakers of Dutch pronounce these words with final siress. 

" Cf. Neijl and Van Heuven (1992) for a discussion of the types of exceptional word stress in 

Dutch, and the available exception mechanisms. 

12 Compare indif+eer and laryng+aal which prove the non native nature of their hase words. 

because -eer and -aal are [-native).


an extrametrical final syllable, and thus get main stress on the presuffixal 

syllable. 13 

5.2.1. Secondary stress 

As pointed out above, words may also have secondary stresses, in conformity 

with principle (5), as illustrated below: 14 

(18) au to maat do mi nee 

* * * * 

* * 

fo no lo gie in fi ni tief ka ta lo gus 

* * * * * * 

* * * 

ste ro i de 

Wa ga nin gan 

* * * * 

* * 

en cy clo pe die 

en cy clo pe disch 

* * * 

* * 

en cy clo pe do loog en do cri no lo gie 

* * * * * * 

* * 

l assume that the relevant rule, the Optimal Grid Rule, applies after the main 

stress of a word has been determined. The Optimal Grid Rule adds asterisks on 

line 1. Subsequently, binary, left-dominant feet are created when possible. In 

some of these examples, it is impossible to create a really perfect grid of 

rhythmically alternating syllables, for instance in fonologie. In this word, the 

third syllable cannot receive secondary stress because it would clash with the 

next main stress. The crucial observation is that it is the first rather than 

the second syllable that receives secondary stress. In other words, Dutch 

prefers stresses at (prosodie) word boundaries. This principle, described in 

Booij (198la: 166-9), is sometimes referred to as the Hammock Principle 

(Van Zonneveld 1985). The mirror image case is infinitief with secondary 

stress on the final rather than the prefinal syllable. Note also that in words 

In female names such a.s Jti/iiina anil Hcntlcnka main stress is on the penultimate syllable. 

Such names can be analysed as sullixed with the feminine sut'lix -a, attached lo the stems of 

Latinized forms of masculine names such as .liiiinn us and Hcndenk-us. So they have the normal 

Mress pattern of suffixed non-native words, and this explains the difference in stress with words 

like 1'1,'klnni iku and i'li'kin'in-ifus, where different suflixes are involved. 

The respective meanings of these words are 'automaton', 'parson', 'phonology', 'infinitive', 

'catalogue', 'steroid'. 'Wagcningen'. 'encyclopaedia', 'encyclopaedic', 'encyclopaedist', 

'endocrinology'.

IO6 

WORD STRESS 

with initial stress like Wageninxeti, a secondary stress appears on the third 

syllable. When two secondary stresses are assigned there may be some variation 

as to the location of the word-internal secondary stress: in èncyclopèdolóof> 

it is on the fourth syllable, in endocrinologie on the third (Hoeksema and 

Van Zonneveld 1984). The difference may have to do with the interpretation 

of endo- as a non-native prefix. 

These observations lead to the conclusion that we should not derive 

secondary stresses by iterative application of the Main Stress Rule, which 

creates sequences of feet, and thus an alternating pattern, because the Main 

Stress Rule would derive secondary stress on the second syllable, instead of on 

the first syllable ol'fonologie. Secondly, whereas the MSR is a directional rule, 

applying from right to left, the rule for secondary stresses must be able to apply 

from both sides of the word, depending on the location of the main stress, as 

shown by the pair infinitlef-fOnologie, and the pair horizon 'id.'—klarinet 

'clarinet'. 

An additional but logically independent issue is whether the assignment of 

secondary stresses in Dutch is sensitive to the segmental composition of the 

syllable ('syllable weight'). The analysis given here assumes that this is not the 

case for Dutch. The opposite position is taken by Kager ( 1989) who argues that 

there is certain evidence that closed syllables and syllables with a diphthong 

have inherent stress. The evidence concerns certain stress patterns, vowelreduction 

possibilities (to be discussed in detail in Section 6.4), and Trochaic 

Reversal. 

Kager (1989: 290) observed that it is possible to have secondary stress on 

the second syllables of the words melancholiek 'melancholic' and gerontologie 

'gerontology', alongside the possibility of initial secondary stress. This would 

follow from stress being assigned by weight. However, in many words we get 

word-initial secondary stress even when the second syllable is closed, as in 

identiteit 'identity', tuberculose 'tuberculosis' and conservatief 'conservative'. 

So it seems that non-initial secondary stress is not a systematic phenomenon. 

Note, moreover, that non-initial secondary stress also occurs in some words 

although the second syllable is light, as in piraten'/ 'pirate actions' and 

grammàtikâal 'grammatical'. 

Vowels in open syllables indeed reduce easier than vowels in closed 

syllables, and diphthongs never reduce to schwa. Nevertheless, vowels in 

closed, that is, 'heavy' syllables can reduce. For instance, in anekdote 

/antkdoto/ 'anecdote' the second syllable is closed. Yet, reduction of the 

vowel is possible, which forces Kager to add a rule that reduces the stress 

of closed syllables in certain configurations. In the approach taken here, the 

second syllable does not take stress, and hence the vowel can reduce. The fact 

that diphthongs do not reduce may be related to a more general tendency, 

namely that diphthongs do not undergo rules which apply to stable long 

vowels. For instance, they also do not undergo the rule of word-initial vowel 

shortening (Section 6.5) that applies in stressed syllables.


Trochaic Reversal is the phenomenon discussed above in (3) with respect to 

the stress pattern of dorpsdomincc. It seems that this kind of reversal sometimes 

leads to a shift of stress to the final syllable of a bisyllabic word forming 

the right constituent of compounds, as in Bosatlas Md.' (Bos is the author of 

this atlas), jeugdherherg 'youth hostel', and hloedlichaam 'blood corpuscle'. 

The rhythmic reversal is illustrated for Bosatlas, for analyses with and without 

inherent weight: 

(19) With inherent weight Without inherent weight 

bos at las bos at las 

* (*) _> * 

* * 

Kager's argument in favour of the weight analysis is that stress shift does not 

occur on words ending in a vowel. For instance, in postgiro 'giro', stress 

cannot be shifted to the last syllable. Gussenhoven (1984: 305), however, 

provides the example tandpasta 'toothpaste' where Trochaic Reversal does 

occur, and hence the second vowel can reduce: |to.mp3sta|. Also, Visch (1989: 

129) gives a number of examples in which the secondary stress cannot shift 

although the relevant syllable to which the stress would move, is heavy: 

âartsvijand 'arch-enemy', hûlprôbot '(lit.) help robot', and borstsleraad 

'breast ornament'. In short, Trochaic Reversal phenomena do not lead to 

clear conclusions. We should also keep in mind that in words such as dominee 

'parson', the last, open syllable gets secondary stress anyway, also in Kager's 

analysis, 15 although it is a 'light' syllable. 

Moreover, it appears that the inverse of Trochaic Reversal, Iambic Reversal, 

which takes place in phrases, is sometimes possible with stress shift to open 

syllables, as the following examples (from Kager 1989: 279) illustrate: 16 

(20) vocaal 'vocal' vocale steun 'vocal support' 

neutraal 'neutral' neutrale opstelling 'neutral position' 

primair 'primary' primaire kenmerken 'primary features' 

An argument in favour of the analysis presented here is that the rule for 

assignment of secondary stress applies throughout the lexicon, i.e., also to 

complex words with native, stress-neutral suffixes, that do not trigger application 

of the MSR. For instance, the word wandeling /wundalirj/ 'walk', with the 

stress-neutral suffix -ing /irj/ bears secondary stress on the final syllable, which 

is confirmed by the fact that it selects the diminutive allomorph -etje that only 

occurs after stressed syllables. On the other hand, bisyllabic words ending in 

-ing like paling 'eel' get the allomorph -kje. Note that this also speaks against 

assigning inherent stress to closed syllables, since then the final syllable of 

" Kager derives this secondary stress by assigning lexical stress to Ihe vowel of the last syllable 

l

I08 

WORD STRESS 

paling would bear stress, and we can no longer predict the choice of the correct 

diminutive allomorph. 

Another relevant native suffix is -nis: in vuilnis 'garbage', the second 

syllable is stressless, and the underlying /!/ can reduce to schwa. In droefenis 

'sadness', on the other hand, the last syllable receives rhythmical stress, and 

thus it is correctly predicted that here the /i/ does not reduce. Again, this shows 

that the Optimal Grid Rule applies to the native complex words as well. 

In sum, there are two choices to be made for each language with respect to 

secondary stress: 

(a) is it weight-sensitive or only determined by rhythmic alternation? 

(b) does it apply in the same direction as the MSR or from the other 

direction? 

These choices appear to be independent. According to Halle and Kenstowic/. 

(1991), English is weight-sensitive, but secondary stress does not apply in 

the same direction as the MSR, and Spanish is just the other way round: 

secondary stress is not weight-sensitive, but applies in the same direction as 

the MSR, i.e., from right to left. According to Kager (1989) Dutch secondary 

stress is weight-sensitive, and is assigned in the same direction as the MSR, 

whereas here, the opposite choice is made for both parameters, which makes 

Dutch in this respect equal to, for instance. Polish (cf. Rubach and Booij 

1985). 

In conclusion, the following phonological rule of Dutch accounts for 

secondary stress: 17 

(21) Optimal Grid Rule 

Assign line-1 *s to syllables within a prosodie word under the following 

conditions: 

(a) no adjacent *s; 

(b) if possible, insert * on syllables at word edges (the Hammock 

Principle). 

In the initially stressed Wapeningen a secondary stress on the final syllable is 

7 The analysis in which main stress is assigned lirst. and then secondary stress by a separate 

rule, was proposed in Booij (1981«: ch. 9), and is also defended in Van /.onneveld (1985) and Van 

dei Hulst ( 1984). As argued in Rubach and Booij (1985), secondary stress in Polish is assigned 

according to the same principles of 'main stress first' (in the penultimate), and then secondarystress 

assignment from both word edges. Vogel and Scalise ( 19X2) defend this approach tor Italian, 

and Halle and Kenstowicz ( 1991 ) argue in favour of such a rule of secondary-stress assignment for 

English, in the case of English only from the left edge of the word, i.e. on the part of the word 

before the main stress. In their opinion the fact that the secondary-stress rule starts from the word 

edge follows from the 'cross-over constraint' that says that 'melritication' (the construction of 

metrical feet) can only affect elements that have not yet been parsed. Roca (1986, forthcoming) 

also argues that secondary stress in Spanish is determined by rhythmic principles rather than by an 

iterative Main Stress Rule. Other evidence is summari/ed in Van de Vijver (1993). 

Strictly speaking, secondary-stress assignment is not one rule in the case of Dutch because it 

applies from both word edges. One may conceive of it as imposing a certain template on line 1 of 

the metrical grids of words.


impossible because it contains a schwa. Hence, the secondary stress is located 

in the prefinal syllable. 

The principle that adjacent stresses are forbidden only holds within prosodie 

words. For instance, as shown above, the compound dorpsdominee does allow 

tor adjacent stresses, even after the stress clash has been removed. 

5.2.2. Cyclic i ty of stress assignment 

If the MSR applies when possible, the rule will apply cyclically without any 

further stipulation, given the organization of the grammar as outlined in 

Section 4.1. The question is whether this is a desirable result. In Chomsky 

and Halle (1968), the rule for English word stress is assigned cyclically. On 

each subsequent cycle, main stress is reassigned, and the primary stress of the 

previous cycle is lowered by one degree by convention. If we applied this 

theory to Dutch, the following stress patterns for words such as profeteer 'to 

prophesize' and kanaliseer 'to canalize' would result: 

(22) 1st cycle: 

MSR 

2nd cycle: 

MSR 

Stress Lowering 

profeet 

1 

+eer 

1 

2 

kanaal 

1 

+iseer 

1 

2 

So, we derive the wrong stress patterns profeteer and kanaliseer instead of the 

correct patterns with secondary stress on the word-initial syllables. The 

correctness of the given patterns can be deduced from the vowel-reduction 

possibilities: in both words, the vowel of the second syllable can reduce: 

[prof9te:r|, [kanalize:r]. In the case of profeteer we might explain this by a 

rule of stress shift that shifts the secondary stress from its prestress position, in 

order to get a more optimal rhythmic pattern. In the case of kanaliseer, this 

explanation does not go through since the cyclic derivation does derive an 

alternating pattern, as shown above.' x In other words, derived words, at least 

those derived by means of non-native suffixes, behave as underived words with 

respect to stress. Two different conclusions may be drawn from this observation: 

either main stress is assigned cyclically, but a Stress Erasure Convention 

erases it when a new cycle is created through the addition of an affix, or the 

MSR is a rule that applies after the non-native morphology. The first proposal, 

cyclic assignment plus a Stress Erasure Convention, is that of Halle and 

Vergnaud (1987: 83). This convention states that existing stresses within the 

8 There are a few words, though, with non-initial stress thai correlates with the position of the 

main stress in the base word: piraat—piraterij, grammatika—grananatikaal. In grammankuul it is 

the vowel of the first rather than that of the second syllable that can be reduced to schwa. They 

might be seen as lexically governed exceptions lo the Stress Erasure Convention. Comparable 

lexical exceptions appear lo exist for English. For instance, in originality the second syllable bears 

Mress, instead of the first one, a reflection of the main stress on the second syllable of its base word 

original (Halle and Kenstowicz 1991: 491).

I 10 

WORD STRESS 

domain of the prosodie word to which the MSR reapplies are erased. I will 

assume it a* well, since there is another argument for cyclicity of stress 

assignment: morphological and phonological rules may require information 

about the stress pattern of a word. For instance, the adjectival suffix -ief only 

attaches to nouns ending in unstressed -ie (in other cases the suffix -ixch is 

used, cf. Booij and Rubach (1987)): 

(23) a. agressie 'aggression' agressief 'aggressive' 

indicatie 'indication' indicatief 'indicative' 

restrictie 'restriction' restrictief 'restrictive' 

b. hysten'e 'hysteria' hysterisch 'hysterical' 

demokrati'e 'democracy' demokratisch 'democratic' 

algebra 'id.' 

algebraisch 'algebraic' 

Note that the words in unstressed -ie are the ones with the regular stress 

pattern, so we cannot refer to some diacritic feature of words like agressie** 

In Chapter 4 we also encountered a case where the correct application of a 

rule of vowel lengthening is dependent on information about the stress pattern 

on the previous cycle: the rule of Vowel Lengthening in the non-native lexicon 

(Section 4.4.4). 20 

In sum, like English (Halle and Vergnaud 1987: 83), Dutch has a class of 

cyclic suffixes that do not themselves form domains of stress assignment; they 

therefore trigger reapplication of the MSR and erase the existing stress pattern 

of the base. Below, we will see that the class of cyclic suffixes is somewhat 

larger than the class of non-native suffixes. 

5.3. WORD STRESS AND NATIVE SUFFIXATION 

Above we saw that the MSR holds for (native and non-native) underived 

words, and for non-native complex words. This layer of the lexicon is often 

referred to as 'stratum 1 ' or 'level 1 ' . The only clear level-1 rule that we have 

encountered so far is the MSR. The rules discussed in Chapter 4 were either 

19 An alternative analysis is proposed in Odden IIWO). Oclden assumes that in words like 

/HVC'/"/»K/

5-3- WORD STRESS AND NATIVE SUFFIXATION III 

P-rules, not restricted to a particular layer of the lexicon, or MP-rules, which 

were either characteristic for non-native or for native words. 

The native suffixes of Dutch can be divided into three classes: 

(a) stress-neutral suffixes 

(b) stress-bearing suffixes 

(c) stress-shifting suffixes. 

They will be discussed in the next subsections. 

5.3.1. Stress-neutral suffixes 

Most native suffixes of Dutch are stress-neutral, that is, they do not influence 

the position of the main stress of their base word. There are two categories of 

stress-neutral suffixes to be distinguished since some of them form prosodie 

words of their own, the so-called non-cohering suffixes: 

(24) a. Cohering suffixes 

-e /a/ (several functions) 

-el /al/ (denominal verbalizing suff.) 

-en /an/ (pi. suff.) 

-er /ar/ (several functions, plus allomorph /ar/) 

-erd /ord/ (creates de-adjectival pej. names) 

-erig /aray/ '-ish' 

-ing /irj/ '-ing' 

-nis /nis/ '-ness' 

-s /s/ (substantivizing suff., pi. suff., gen. suff.) 

-sel /sal/ (creates de-verbal obj. names) 

-st /st/ (super., de-verbal nominalizing suff.) 

-ster /star/ (feminizing suff.) 

-t /t/ (de-verbal nominalizing suff.) 

-te /ta/ '-ness' (past tense) 

-tje /tja/ (dimin., + 4 allomorphs) 

b. Non-cohering suffixes 

-achtig /oxtay/ '-like' 

-baar /bar/ '-able' 

-dom /dom/ '-dom' 

-held /held/ '-ness' 

-ling /lirj/ '-ling' 

-loos /los/ '-less' 

-schap /sxap/ '-ship' 

It is almost completely predictable whether a native suffix will be non-cohering 

or not. The rule is: a native suffix that could form a prosodie word of its own, 

1 

The non-cohering suffix -zaam is dealt with in Section 5.3.3, where -biiar also recurs.

I 12 

WORD STRESS 

does it, except for -ing and -nis. 22 By definition, non-cohering suffixes are not 

cyclic suffixes, and hence they do not affect the main stress of their input 

words. In order to account for the stress-neutrality of the stress-neutral cohering 

suffixes, we have to mark them as non-cyclic suffixes. In most cases, 

this marking can be predicted from the fact that these suffixes are either 

nominalizing or inflectional [+native] suffixes. 

On the other hand, the Optimal Grid Rule is a purely phonological rule, and 

applies if possible. Thus, in wandeling /wundolirj/ 'walk' derived from the 

verb wandel 'to walk', the MSR assigns main stress to the initial syllable. After 

the affixation of -ing, a non-cyclic suffix, the MSR does not reapply, but the 

Optimal Grid Rule is applicable since there is now a syllable with a full vowel, 

introduced by the suffix -ing, and it will assign secondary stress to the last 

syllable. 

The Optimal Grid Rule is, when nothing is said, a cyclic rule (although its 

effects will be erased by the Stress Erasure Convention). This is in harmony 

with the fact that in the allomorphy of the diminutive suffix the location of 

secondary stress may play a role, as is the case for wandeling 'walk' with the 

diminutive wandelingetje. If the Optimal Grid Rule were a word-level rule, 

and thus applied after the MP-rules triggered by the native morphology, the 

relevant information that the last syllable of wandeling has secondary stress, 

would not be available. 

The effect of adding stress-neutral suffixes with only schwas is that we may 

get complex words with a sequence of unstressed syllables, as illustrated by 

gemakkelijker, the comparative form of gemakkelijk 'easy', itself derived from 

gemak /y3mak/ 'easiness' by means of the suffix -elijk /alak/: 

(25) Y3 ma ks Is kar 

* line 1 

* line 2 

It should also be noted that the stress-neutrality of these suffixes does not 

follow from their containing schwas. For instance, in the adjective judaxxerig 

[jydusaraxl 'Judas-like', the stress is not on the last full vowel, as would be 

expected if we determined the stress pattern anew, but on the first one, 

corresponding with the main stress on the first syllable of Judas 'id.'. 

The non-cohering native suffixes have an internal stress pattern. For instance, 

in -achtig the first syllable bears stress. The other non-cohering suffixes are 

monosyllabic. Like monosyllabic words of lexical categories, they will automatically 

receive main stress, i.e., *s at level 1 and level 2. Of course, the fact 

that monosyllabic prosodie words have main stress is not specific for level 1, 

but is valid across the lexicon. Similarly, since across the lexicon a schwa 

cannot receive stress, the initial stress of -achtig /axtoy/ is predictable. Words 

like rood-achtig 'reddish' and eet-baar 'edible' are compounds of two 

22 Some speakers always rcali/.e -nis as |nas|. Thus, lor those speakers, -nis is no exception, hut 

is predictably a cohering suffix.

5-3- WORD STRESS AND NATIVE SUFFIXATION 113 

prosodie words, and thus they will be subject to the stress rule for compounds 

to be discussed in Section 5.4 which states that the first prosodie word gets an 

asterisk on line 3: 

(26) (rood),,, 

* 

* 

* 

(ach tig) u 

* 

(baar) œ 

* line 1 

line 2 

line 3 

5..?.2. Stress-bearing suffixes 

There are four native suffixes that bear the main stress of the words that they 

form: 

(27) -erij /arci/ schiet-erij 'shooting', et-erij 'eating' 

-es /es/ voogd-es 'fern, guardian' (fem.) 

-in /in/ hel-din 'heroine', leeuw-in 'lioness' 

-ij /ei/ voogd-ij 'guardianship, abd-ij 'abbey' 

These words are native suffixes as proven by the fact that they attach typically 

to native base words. Therefore, these native suffixes have to be marked as 

cyclic suffixes. Moreover, the final consonants of the suffixes -es and -in have 

to be marked as exceptions to the rule of final VC-extrametricality discussed 

above, because they bear the main stress of the word, or they are to be marked 

as I+F]. 

It is worth discussing how the grammar developed here will derive the stress 

pattern of maatschappij 'society' which is derived by suffixation of -ij to the 

base word maatschap '(lit) mateship' that in its turn contains the non-cohering 

suftix -schap. In maatschap the first syllable bears main stress, and the second 

syllable bears secondary stress because it is a prosodie word in its own right. In 

other words, stress is as predicted by the Compound Stress Rule to be 

discussed below. In maatschappij the cohering suffix -ij is added to the 

prosodie word -schap. The cohering nature of this suffix is also clear from 

its resyllabification effect: the final /p/ of -schap forms a syllable with -ij. So 

we get the following stress patterns: 

(28) (maat) (0 (schap),,, (maat) w (schap pij) (0 

* * * * line 1 

* * * line 2 

* line 3 

Since -ij is a cyclic suffix, it will delete previously assigned stresses. The 

second syllable of maatschappij is indeed stressless and the vowel can be 

reduced to schwa: [matsxapei]. The removal of the original stress pattern of 

maatschap follows from the Stress Erasure Convention. Finally, note that the 

original prosodie structure does not seem to play a role in the determination of 

the main stress. That is, we have to assume that after stress erasure, the MSR

I 14 

WORD STRESS 

which normally applies in the domain of the prosodie word, can extend its 

domain, and also apply across a prosodie word boundary. 

5.3.3. Stress-shifting suffixes 

A third category of native suffixes requires the main stress of the word to be 

located on the last stressable syllable before the suftix. They are all adjectival 

suffixes (cf. Schultink 19X0): 

(29) -baar /bar/ overdraagbaar 'transferable' 

-end /and/ omarmend 'embracing' 

-ig /ay/ 

driehoekig 'triangular' 

-isch /is/ afgodisch 'idolatrie' 

-(e)lijk /alok/ hartstochtelijk 'passionate', aanzienlijk 'considerable' 

-s /s/ goedlachs 'merry' 

-zaam /zam/ mededeel/aam 'communicative' 

A remarkable detail is that the suffix -haar only attracts main stress if the base 

word is a so-called separable complex verb. 21 For instance, overdraagbaar is 

derived from the separable complex verb overdragen 'to transfer'. On the other 

hand, when we add -baar to the complex verb beïnvloeden 'to influence' which 

is not a separable complex verb, the main stress remains where it was, on the 

syllable in. Furthermore, if -loos is preceded by a schwa, it may also attract 

stress, as in hartstochtelooa 'passionless'. Finally, stress shift is also observed 

for the sequence of suffixes -loos-heid, where main stress is located on -loos, 

as in werkeloosheid 'joblessness'. 24 The relevant generalization is: 

(30) Assign main stress to the last stressable syllable before the suffix in 

adjectives ending in -haar, -end. -/'#, -isch, -(e>lijk, -s, and -zaam. 

Clearly, these suffixes also have to be qualified as cyclic suffixes, in that they 

erase the existing stress pattern, and trigger reassignment of Main Stress. The 

suffixes -end, -ix, and -(e)lijk are then completely regular: they cannot bear 

main stress themselves since they do not contain full vowels. The other three 

suffixes are irregular in that they require main stress on the syllable before the 

suffix. 

Just as in the case of maatschappij, the effect of this special rule is that the 

second prosodie word of the adjective bears main stress: 

(31) Morphological structure 11 [mat| N sxup| N al3k] A 

Prosodie structure (mat)^ (sxupglak)^ 

Stress grid line 1 * * 

line 2 * 

" In main clauses the verbal pan of separable complex verbs stands in second position, whereas 

the particle is clause-final; cf. Booij (IWO«) for a morphological analysis of these verbs. 

The proper analysis of these complicated facts is discussed in Neijt and /onneveld (14X1 ). 

KOOIJ and Van der Niel (1985), and De Haas (1985, 1991).

5-4- COMPOUND STRESS 115 

To summarize, there is a number of adjectival native suffixes that have to be 

qualified as cyclic suffixes. 

5.4. COMPOUND STRESS 

The basic stress pattern for compounds is that the first of the two constituents is 

the most prominent one (the strong-weak pattern). The most productive 

category of compounds is that of nominal compounds. Verbal compounds 

are rare (but do exist), and adjectival compounds also form a productive 

category, but their constituents are usually non-compounds (i.e. adjectival 

compounding does not allow for recursivity). 25 The following rule accounts 

for compound stress: 

(32) Compound Stress Rule (CSR) 

Add an * to the leftmost prosodie word on the next line. 

The following derivations illustrate the (cyclic) application of this rule to the 

compound onderzoekfonds 'research fund': 

(33) Klonderl [zoek]] [fonds]] 

1st cycle: MSR * * * line 1 

* * * line 2 

2nd cycle: CSR * line 3 

3rd cycle: CSR * * line 3 

* line 4 

By convention, the application of the CSR at line 4 also introduces an * for 

fonds at line 3 because before each application of the CSR the two constituents 

involved must be represented at the same number of lines. This correctly 

expresses that in the word under discussion the constituent fonds is more 

prominent than the constituent zoek. Recall that the stress grids are assigned 

to prosodie constituents. Therefore, the CSR also applies to complex words 

that are not compounds from the morphological point of view, but nevertheless 

consist of two prosodie words, for instance the complex adjectives kleurloos 

'colourless' and draagbaar 'portable', and the complex nouns schoonheid 

'beauty' and leerling 'pupil'. 

Let us now look at the derivation of the stress grid of kankeronderzoek 

'cancer research' as given in (34). 

The then-resulting grid may seem to exhibit a configuration of stress clash. 

Stress clash is defined as the configuration in which two *s are adjacent at line 

" Cf. Langeweg (1988), Visen (1989), and Booij (1992ft) for a survey of Dutch compounding, 

and the different stress patterns.

no 

(34) ([kanker] 

1st cycle: MSR * 

2nd cycle: CSR 

3rd cycle: CSR 

WORD STRESS 

( (onder] [zoek]]] 

* 

line l 

line 2 

line 3 

line 3 

line 4 

n and line n - l (Liberman and Prince 1977, Prince 1983). The clash will be 

resolved by Trochaic Reversal, that is, movement of the line-3 * of the syllable 

on to the syllable zoek. In other words, the pattern 1-2-3 is changed into 1-3- 

2. Similar examples are taalwetenschap '(lit.) language science', staatseigendom 

'state property', and hoofdschakelaar 'main switch' (Visch 1989: 131). 

The interpretation of stress shift as 'Move *' correctly predicts that if the 

second constituent of a compound is bisyllabic, and thus forms one bisyllabic 

foot, the stress shift is impossible, because it would result in an ill-formed grid, 

with no grid entry for the final syllable at line I: 26 

(35) post gi ro 

* 

line 1 

* © -> line 2 

* line 3 

Therefore, stress shift is impossible in the following cases (data partially from 

Visch 1989: 128): 

(36) póstgiro 

blóedplasma 

grâansîlo 

jâarprèmie 

kóorkanon 

n'dderhàrnas 

*póstgirö 

*bloedplasmà 

*graansilô 

*jâarpremie 

*kóorkanön 

*n'dderharnàs 

'giro' 

'blood plasma' 

'grain elevator' 

'annual premium' 

'choir canon' 

'knight's cuirass' 

Exceptions to this restriction are words like Bosatlas and tandpasta mentioned 

above, and a few compounds (Visch 1989: 128) such as bloedlichaam 'blood 

cell' and booreiland 'oil rig': they allow for Trochaic Reversal to apply. 

Bosatlas and tandpasta may be seen as lexicalizations that form only one 

prosodie word (the Optimal Grid Principle will then assign secondary stress to 

the last syllable), whereas words like lichaam 'body' and eiland 'Lsland' can be 

interpreted as formal compounds, thus allowing for stress shift: 

ltl 

In Visch (1989: 126) Trochaic Revcisal is analysed in the iree-cum-grid framework of Hayes 

(19X4) This moans that the grid functions primarily as an interpretational device. When a stress 

clash arises, restructuring and relabelling of the metrical tree takes place. Trochaic Reversal is 

forinali/ed as the adjunction of a constituent that does not contain the Designated Terminal 

Element of the constituent in which the siress clash lakes place to the preceding constituent, 

which becomes strong, since an adjoined constituent is always weak.

5-4- COMPOUND STRESS iiy 

(37) boor ei land 

* * * line 1 

* * * line 2 

* ©-» - line 3 

* line 4 

The analysis given here also correctly predicts that Trochaic Reversal does not 

always apply to compounds with a second constituent ending in the suffix -ing: 

(38) prijsstijging -» *prijsstijgîng 'price increase' 

weersverwachting —> * weersverwachting 'weather forecast' 

bóswandeling —» bóswandeling 'forest walk' 

wéersveràndering —> wéersverandering 'weather change' 

In the last two examples, the final syllable of the second constituent receives 

stress from the Optimal Grid Rule, and hence it is a proper location for 'Move 

k ' to move an asterisk to: 

(39) prijs 

* 

stij 

* 

ging versus bos 

* 

wan 

* 

de ling 

* 

* * * (*) —> 

* * 

There are certain exceptions to the CSR, established compounds with main 

stress on the second constituent instead of on the first: 

(40) [staats][bosbeheer] 'forestry commission' 

[kinder|(bijslag] 27 'child allowance' 

[stad|[hiiis] 'town hall' 

[boeren][zoon| 'farmer's son' 

An even more exceptional stress pattern that is sometimes heard is arbeidsvoorwaardenbeleid 

'terms of employment policy' with main stress on voor, 

the morphologically most deeply embedded constituent. This type of stress 

pattern is the regular one for cases where a phrase functions as the first part of 

a compound. I will assume that within the phrase, stress is assigned by a noun 

phrase stress rule (NSR) which assigns stress to the head of a nominal phrase 

within compounds, as illustrated by the derivation (41). 

On line 3, an additional * will be added to huis by convention because at that 

level both constituents should be represented. 

Verbal compounding is unproductive in Dutch. A verbal compound like 

stofzuigen '(lit.) to dust-suck', i.e., 'to vacuum clean' has probably arisen 

through reinterpretation: the regular nominal compound ( [stof] N [zuiger]n]n 

'vacuum cleaner' with the de-verbal head zuiger can have been reinterpreted as 

the de-verbal -er-noun of the verbal stem stofzuig. 2 * Prosodically, they fall 

27 Some speakers of Dutch have regular main stress on this compound: kinderbijslag. 

2H Cf. Booij (1989/j) for a more elaborate analysis of these phenomena.

I l8 

WORD STRESS 

(41) [|[oude) A [mannenj N | NP |huis| N | N 

1st cycle: MSR * * * line 1 

* * * line 2 

2nd cycle: NSR 

3rd cycle: CSR 

* line 3 

* * line 3 

* line 4 

under the domain of the CSR that correctly predicts main stress on the first 

constituent. 

Adjectival compounds exhibit two different stress patterns. Most of them 

have main stress on the second constituent (Visen 1989: 105, Trommelen and 

Zonneveld 1989«: 246, Backhuys 1989): 

(42) hulpbehoevend '(lit.) help-needing' 

gastvrij 'hospitable' 

zelfrijzend 'self-raising' 

kiesgerechtigd 'enfranchised' 

doofstom 'deaf-mute' 

However, in attributive position there is stress shift (Iambic Reversal) to the first 

constituent, as in een gastvrije ontvangst 'a warm welcome' (cf. Section 7.4). A 

mimimal pair illustrating the difference with nominal compounds is tâalpolitiek 

tnoun) 'language policy' versus tâalpolitiek (adj.) 'language-political'. 

A second class of complex adjectives has main stress on the first prosodie 

words, that is, the strong-weak pattern (Backhuys 1989): 

(43) zéeziek 'sea-sick' 

schi'etgraag 'trigger-happy' 

/.fnvol 'meaningful' 

lévensmoe 'world-weary' 

kleurrijk 'colourful' 

Formally, the right constituents are adjectival heads, but semantically they are 

comparable to adjectival suffixes, since their lexical meaning has faded. Indeed, 

they have the same stress pattern as complex adjectives such as kleurloos 

'colourless' and vruchtbaar 'fruitful'. In other words, these adjectives fall 

under the general CSR. 

The following type of adjective has been claimed to form a special category: 

(44) |[ree] N |bruin| A ] A '(lit.) deer brown', i.e., 'fawn coloured' 

[[peper] N [duur] A ] A '(lit.) pepper-expensive', i.e., 'very expensive' 

[|reuze] N [leuk] A ] A '(lit.) giant funny', i.e., 'very funny' 

|(knetter) v [gek] A ] A '(lit.) crackling mad', i.e., 'completely mad' 

Ilgeel] A [groen] A ] A 'yellow-green' 

[aarts[lelijk| A ] A 'very ugly'

5-5- STRESS IN PREFIXED WORDS 119 

Van Heuven (19S6: 84; 1987) showed for bisyllahic adjectival compounds that 

although the focused constituent is indeed accented, as in Zij is een beeldschoon 

meisje 'She is a very beautiful girl', the intensity and duration of the two 

constituents are more alike than in the case of simplex bisyllabic adjectives 

such as concreet 'concrete'. That is, they seem to be equally stressed. Neijt 

(1990), in a reaction, argued that these adjectives fall under the weak-strong 

patterns, and that durational rules may account for the similarity in duration of 

the two syllables. 29 Spoken in isolation, a word like beeldschoon does have 

accent on the final syllable in the form of a down-stepped tone. 

Compound prepositions also fall under the weak-strong pattern, when they 

are used intransitively. When they form part of a prepositional phrase, the 

main stress shifts to the first constituent: 

(45) bovem'n bovenin de kàmer 'at the top of the room' 

onderaan ónderaan de bèrg 'at the foot of the mountain' 

bovenop bovenop het dàk 'on top of the roof' 

achteraan achteraan de rij 'at the back of the row' 

In sum, adjectival and prepositional compounds are exceptions to the CSR 

in that they have the reverse stress pattern weak-strong. However, this only 

applies to adjectives of which the head still has a distinct lexical meaning. 

5.5. STRESS IN PREFIXED WORDS 

5.5./. Non-native prefixes 

Dutch has a number of non-native prefixes which attach to non-native base 

words or to roots that do not always occur as words by themselves: 

(46) a- apathisch 'apathetic' 

ab- absorbeer 'to absorb' 

ad- adhesie 'adhesion' 

ante- antedateer 'to predate' 

apo- apotheose 'apotheosis' 

con- concentreer 'to concentrate' 

de- decomponeer 'to decompose' 

in- inclusief 'inclusive' 

inter- intervenieer 'to intervene' 

per- perforeer 'to perforate' 

post- postdateer 'to postdate' 

pre- prelude 'id.' 

re- representeer 'to represent' 

trans transponeer 'to transpose' 

29 Cf. Sluijter ( 1492) for further comments on this issue.


As we have seen in Section 3.6, prefixes are independent domains of syllabification. 

However, non-native prefixes may lose their transparency, and thus 

become one prosodie word with the stem. This is illustrated by the verb 

reageer 'to react' in which Homorganic Glide Insertion applies obligatorily. 

It may also be illustrated by the possible syllabification pattern 

(tran) 0 (spo) a (neer) a for the verb transponeer. In terms of secondary-stress 

patterns they often seem to form one domain with their verbal stem, because 

they get secondary stress (they will never bear main stress given the fact that 

the MSR basically assigns stress to one of the word-final syllables): 

(47) dateer antedateer 

componeer dècomponéer 

presenteer representeer 

poneer transponeer 

For instance, in representeer the first syllable bears secondary stress. Hence, 

reduction of the vowel of the first syllable is impossible, unlike that of re- in 

reductie 'reduction': 

(48) representeer [représente:^, [reprgsenteir], (reprosanteir], 

*|r9pres9nte:r| 

reductie |redvksi|, [radvksil 

In other cases, for instance the verb reanimeer /re-animer/ 'to reanimate', the 

prefix may still be recogni/.ed because of the transparent meaning of re- in this 

case: there exists an independent word animeer 'to animate'. In any case, nonnative 

prefixes do not get primary stress, but at most secondary stress, either by 

virtue of being a prosodie word of their own, or because they are in the domain 

of the Optimal Grid Rule. The rule that states that prefixes in complex verbs do 

not bear main stress will be given below. Prefixed words of other lexical 

categories such as prelude will be considered as simplex words, just like 

prefixed verbs that have lost their transparency, and hence they are stressed 

by the MSR. 

5.5.2. Nativized and native prefixes 

Some originally non-native prefixes have become native in that they also 

attach to native base words, unlike the non-native prefixes discussed above, 

and sometimes they even occur as independent words, for instance anti-, co-, 

contra-, des-, ex-, inter-, meta-, non-, para-, pro-, semi-, sub-, super-, ultra-. 

That is, they behave like compounds, and also get the stress pattern of 

compounds that is derived by the CSR. In many examples given here the 

'" Another example is the phonetic realization of abortus 'abortion'. Usually, the prefix ab- will 

no longer be recognized here. Those who still recognize the prefix will pronounce this word as 

(up) 0 (3r) 0 (tYS) 0 , but most speakers pronounce it as (

5-5- STRESS IN PREFIXED WORDS 121 

right constituent is a native word. The right constituent (the head) is a noun, 

and the syntactic category of the whole word is therefore predictable: 

(49) änti-houding 'adverse attitude' 

co-piloot 'co-pilot' 

contra-gewicht 'counterweight' 

des-interesse 'disinterest' 

éx-gelovige 'ex-believer' 

mter-faculteit 'inter-faculty' 

méta-taal 'metalanguage' 

non-issue 'id.' 

semi-arts 'assistant doctor' 

siib-groep 'sub group' 

super-mark! 'supermarket' 

All these prefixes are independent prosodie words, and thus such prefixed 

words are subject to the CSR. 

Adjectives with such prefixes pattern differently in that, like, for example, 

adjectives prefixed with on- (see below), they have main stress on the adjectival 

head when they occur in predicative position (Langeweg 1986): 

(50) inter-continentaal 'intercontinental' 

para-medisch 'paramedical' 

post-koloniäal 'postcolonial' 

semi-direct 'id.' 

ultra-modern 'id.' 

Within the set of native prefixes we have to distinguish two classes. Some of 

them get main stress. They should be considered independent prosodie words 

since they contain full vowels, have an internal stress pattern when they are 

polysyllabic, form an independent domain of syllabification, and one of them, 

aarts- 'arch-' even has an appendix of two coronal obstruents, which is only 

possible if it is a prosodie word. These prefixes attach to nouns, and the 

syntactic category of the complex word is always that of the base word: 

aarts-/arts/ aartsbisschop'archbishop' 

aartsvader 'patriarch' 

oer- /ur/ oermens 'primitive man' 

oerwoud 'primeval forest' 

on- /on/ 

onmens 'brute' 

onzin 'nonsense' 

her- /her/ hérbouw 'rebuilding' 

hérfinanciering 'refinancing' 

onder- /ondar/ onderverhuur 'to sublet' 

ónderbouw 'to underpin' 

over /ovar/ overgang 'transition' 

óverbesteding 'excess spending'


Since these prefixes are defined as independent domains of syllabification, and 

since they also fulfil the requirements for prosodie word status, words with 

these prefixes will be prosodie compounds, and thus subject to the CSR that 

correctly assigns primary stress to the first prosodie word of complex nouns, in 

these cases the prefix. So no special rule is required for the derivation of the 

stress patterns of words with these prefixes. 

The second category of native prefixes is formed by those that are always 

stressless; they are all verbalizing prefixes: 

(52) be- /ba/ bedi'jk 'to dike' 

er- /er/ ervaar 'to experience' 

ge- /ya/ geloof 'to believe' 

ont- /ont/ ontmoet 'to meet' 

ver- /ver/ vertrouw 'to trust' 

As pointed out above, although they are independent domains of syllabification, 

those with schwa cannot form a prosodie word of their own. 

There is a second category of unstressed verb-creating native prefixes, 

which also occur as prepositions or particles. They all fulfil the requirements 

for being a prosodie word, and for the bisyllabic ones, the location of the main 

stress is completely predictable because the second syllable contains a schwa. 

Since they have specific meanings when used as prefixes, 1 consider them ;is 

prefixes, not as the first parts of compounds. Thus, we can correlate their 

préfixai nature with the fact that they do not receive the main stress of the 

word, as we would expect if they were compounds: 

(53) aan- /an/ aanvaard 'to accept' 

achter- /uxtar/ achterhaal 'to recover' 

door- /dor/ 

doorloop 'to pass' 

her- /her/ 

herbouw 'to rebuild' 

mis- /mis/ 

misdraag 'to misconduct' 

om- /om/ 

omklem 'to grasp' 

onder- /nndor/ ondersteun 'to support' 

over- /ovar/ 

overstroom 'to overflow' 

vol- Aol/ 

voldoen 'to suffice' 

voor- /vor/ 

voorkóm 'to prevent' 

weer- /ver/ 

weerspiegel 'to reflect' 

So the rule that accounts for the stress pattern of these prefixed words is the 

following: 

(54) In prefixed verbs, assign * on the next line to the rightmost prosodie 

word. 

This rule will account for verbs with either native or non-native prefixes. It 

competes with the CSR in that it also applies to sequences of prosodie words. 

Since it is the more specific one, referring to the category 'prefixed verb'

5-6. BRACKETING PARADOXES 123 

whereas the CSR does not mention morphological information, it will get 

priority by the Elsewhere Principle." 

Finally, there are some prefixes with variable stress (Schultink 1964, De 

Vries 1975). The prefixes her-, over-, and onder- when used with verbal bases, 

follow the rule just given, unless the first syllable of the base verb does not 

bear main stress. In that case, the prefixes receive main stress: 

(55) 

heronderover- 

Basc verb 

formuleer 

financier 

belicht 

Derived verb 

herformuleer 'to reformulate' 

ónderfinancier 'to underfinance' 

overbelicht 'to overexpose' 

According to the literature on Dutch stress, the negative prefix -on behaves 

in a marked fashion when attached to adjectival bases: either the prefix or the 

adjective bears main stress, depending on the syntactic position (Van den Berg 

1970, Schultink 1979): in predicative position, main stress is either on the base 

or on the prefix. There appears to be variation here, at least in transparent 

adjectives: the main stress is on the adjective, but in order to emphasize the 

negative meaning, on- can also get main stress. 32 The prefix usually bears main 

stress in attributive position: 

(56) Jan is onaardig/onaardig 

'John is unkind' 

een onaardig mens 

'an unkind person' 

This also applies to adjectives with the prefixes aarts- and oer-, as in aartslelijk 

'very ugly' and oergezellix 'very cosy'. The stress behaviour in attributive 

position can be interpreted as a case of Iambic Reversal, i.e., a case where 

stress clash is resolved by shifting stress to the left (cf. Chapter 7; at line 4, the 

rule for phrasal stress applies): 

(57) (on) (aardig) 

line I * * 

(mens) 

* 

line 2 * * * 

line 3 CSR


1991) which involve stress. The classic example is the English word ungrammatically. 

I will discuss here its Dutch equivalent OHgrammaticaliteit. The 

problem arises because a distinction between level-1 phonology and level-2 

phonology is assumed which correlates with a difference between level-1 

morphology (= non-native morphology) ordered before level-2 morphology 

(native). On this view, non-native morphology is stress-shifting because the 

MSR is restricted to level 1, and this explains that native morphology is stressneutral. 

The paradox then is that from the morphological point of view -iteit is 

added to the adjective ongrammaticaal, whereas from the phonological point 

of view it seems that first -iteit is added to grammaticaal, thus causing shift of 

the main stress to the final syllable of -iteit, and then, at level 2, on- is added, 

which then has no effect on the location of main stress on the part grammaticaliteit. 

Thus, both of the following two structures seem to be necessary: 

(58) Morphological structure: ] [ongrammaticaalIA JAiteitjN 

Phonological structure: [on| [grammaticaal) A iteit] N ] N 

In the analysis presented here and the preceding chapter, I do not make use 

of level ordering. The word imgrammaticaliteit clearly shows that non-native 

affixation (with -iteit) may follow native affixation (with on-), so the ordering 

of all non-native affixation before all native affixation is empirically inadequate. 

13 Since in the adjective ongrammaticaal the base adjective grammaticaal 

is the head of the word, and since it bears the feature [—native], 

the whole adjective ongrammaticaal is [-native] by percolation (cf. Williams 

1981), and thus attachment of the [—native] suffix -iteit is allowed. Note that 

we need a morphological principle that restricts [-native] suffixation to 

[-native] bases anyway, since we also have to exclude [-native] suffixation 

to [+native] roots: a word such as *roditeit '(lit.) reddity i.e., 'redness' derived 

from the [+native] adjective rood is impossible. 

Instead, the suffix -iteit, being a [-native] suffix, is predictably cyclic. 

Hence it will erase the existing stress pattern of ongrammaticaal (the Stress 

Erasure Convention), and induce reapplication of the MSR to the newly 

formed string ongrammatikaliteit, thus assigning main stress to the last syllable. 

Subsequently, the Optimal Grid Rule will assign secondary stress, which 

results in the stress pattern ongrammatikaliteit? 4 

As noted above, there are also [+native] suffixes that are nevertheless cyclic. 

This forms another problem for the theory of level ordering because they 

would be level 1 in terms of stress, but level 2 as far as morphology is 

concerned. 

" See Booij (I982a) for more empirical evidence from Dutch. 

14 Arguments for the same analysis of the English equivalent ungrammatically are given in 

Halle and Vergnaud (1987: 80-5). However, they also argue that all phonology follows all 

morphology, a claim I do nol support.

CONNECTED SPEECH I: WORD PHONOLOGY 


In the preceding chapters I have presented a survey of rules concerning the 

distribution of sounds within prosodie words. These rules of word phonology 

have an obligatory character (although some also apply optionally in larger 

domains). In connected speech, words are subject to rules which are often 

optional, in that they are dependent on style of speech and speech rate. These 

rules serve to reduce the articulatory efforts of the speaker, and to get a 

smooth transition from one segment to the next. Therefore, most of these 

processes are either reduction processes or assimilation processes. For 

instance, the realization of a vowel as a schwa requires less articulatory effort 

than the realization of a full vowel (Koopmans-van Beinum 1980). In addition, 

we also find cases where insertion or apparent insertion of a segment serves 

ease of articulation. 

The use of these processes is often characteristic of less monitored, casual 

speech. It is characteristic of casual speech that ease of production gets priority 

over ease of perception, because speakers can only afford to give priority to 

ease of production in more informal situations. 

It is not the case that all rules diminish articulatory efforts in the same 

direction. For instance, schwa epenthesis has the effect of breaking up nonhomorganic 

consonant clusters (but increases the number of syllables to be 

pronounced), but there is also a process of schwa deletion that decreases the 

number of syllables, but creates consonant clusters. 

Although the rules to be discussed here are usually optional, it may be that 

use of one makes use of another one obligatory, as will be shown below. 

A related category of rules is formed by the rules that typically depend on 

speech rate. Examples are the rules of Schwa Deletion and /i/-deletion. The 

basic effects of such rules are that certain segments are either not pronounced, 

or 'masked' because they show temporal overlap due to decreased magnitude 

in space and time of the segments involved (Browman and Goldstein 1990). 

They may be called fast-speech rules, but without further empirical research it 

is impossible to decide whether there are rules which only depend on speech 

rate, and not on degree of monitoring as well. 

Although casual speech may often be fast speech, the two notions are not 

identical (Zwicky 1972). One can speak casually at low speed, or one may not 

use the rules typical of casual speech, and yet speak fast (Nolan 1992: 264).

126 CONNECTED SPEECH I: WORD PHONOLOGY 

However, since casual speech and fast speech often go together, the actual use 

of a casual-speech rule may also exhibit the effects of higher speech rate, and it 

is not easy to distinguish between the two factors. 

In addition to their being possibly optional, connected speech rules are 

also different from the obligatory rules of word phonology in that they may 

create types of syllable that are not allowed for by the lexical syllabification 

algorithm. This applies for instance to the rule of schwa deletion, and it is 

another reason for considering them as postlexical rules rather than as lexical 

rules, since the latter may not violate the lexical syllable-structure 

constraints.' 

As pointed out above, the actual application of rules of connected speech is 

typically dependent on the style of speech (i.e., degree of monitoring) and/or 

speech rate. In addition, a number of other factors play a role: 

lexical variation: words with a higher frequency, that is, with a higher 

activation level in the lexical memory, are more often subject to such 

rules than words with a lower frequency; 

dialectal variation: rules may be applied differently depending on the 

geographical or social background of the speaker; 

the linguistic properties and environments of the segments involved; for 

instance, mid vowels reduce much more easily than high vowels, and 

more easily in open syllables than in closed syllables. 

In other words, these rules are typically variable rules in the sense of Labov 

(1972). 

Although the connected speech processes of Dutch are described here in 

categorical terms, they may be of a more gradual nature than these descriptions 

suggest. Phonologists base their inventories of connected speech processes 

on perception, which is inherently categorical, but such inventories are 

not more than a first necessary step in the unravelling of connected speech 

processes (Kohler 1990, 1991). For instance, the phonologist of Dutch 

perceives a vowel in unstressed position as either the full vowel or the 

schwa, but it may be that the actual vowel has been slightly reduced only, 

thus leading to its still being perceived as a full vowel, although some 

reduction has applied. 

In this chapter I deal with connected speech phenomena that apply within 

words. The domains of these rules are prosodie constituents such as the coda, 

the syllable, or the prosodie word, hence they only apply within words. The 

next chapter covers reduction and assimilation processes that apply across 

(prosodie) word boundaries, and Chapter 8 deals with the phonological behaviour 

of clitics, another typical connected speech phenomenon. 

1 

In other words, these rules arc not always structure-preserving, which might be a reason lor 

considering them as postlexical rules; cf. Kiparsky (1985).

6.2. SCHWA EPENTHESIS 127 

6.2. SCHWA EPENTHESIS 

In non-homorganic consonant clusters in coda position, a schwa may be 

inserted: 

(1) kalm 'quiet' [kalam] 

arm 'id.' 

[aram] 

help 'id.' 

Ihelapj 

harp 'id.' 

[hurapl 

herfst 'autumn' [herafstj 

elf 'eleven' [dof | 

melk 'milk' [melakj 

werk 'work' [ucrak] 

alg 'alga' [alsx| 

erg 'very' [erox] 

urn 'id.' 

[vran] 

hoorn 'horn' [horan] 

Schwa insertion is impossible if the second of the two consonants involved is 

an appendix consonant, that is, /s/ or /t/, or if the cluster is a nasal followed by 

a homorganic consonant: 

(2) a. hart 'heart' *[huratl, hars 'resin' *[hur3s], markt 'market' [maraktj, 

but *[murk3t], halt 'stop' *[hulot], band 'tape' *|bun3t), hals 

'neck' *|hul3s|, kans 'chance' *|kunas) 

h. damp 'id.' *[dam3p], bank 'id.' *[burj9k| 

These facts suggest the following rule of Schwa Epenthesis: 2 

(3) Schwa Epenthesis 

Insert a in Coda 

X —X 

This rule will not apply to clusters with coronal obstruents because these are 

syllabified as appendices. This supports the 'active' interpretation of the 

Appendix position: even in a word like hart 'heart', the /t/ must be considered 

as an appendix although this is not required by the length of the rhyme. 

Homorganic clusters of a nasal plus obstruents form partial geminates: they 

share tne features on the Place tier. Indeed they show the property typical of 

geminates, integrity: the two halves cannot be separated by a vowel. Note, 

however, that insertion of a schwa is not blocked here by the prohibition on 

crossing association lines since the schwa has no Place tier at the underlying 

level. So the following configuration could arise: 3 

2 This process is discussed in BOOIJ (l9Xli;: 156), Berendsen and Zonneveld (1984), and De 

Haas (19Xh). 

1 The consequence ol treuling the epenthetic vowel as un underspecilied vowel, i.e. the 

impossibility ol using the Prohibition on Crossing ol Association Lines as a blocking mechanism 

in this case, was pointed out by l rvin ( 19X5: X7-X). The analysis in De Haas (19X6) makes use of 

this condition, and therefore Levin's objection applies to that analysis.


(4) [+cons] [—cons) u [+cons] 

Place 

Let us therefore assume the following condition on linking: 'Consonants that 

share their melodic specifications partially or wholly must be adjacent on the 

X-tier'. 4 This rules out the above configuration, and thus schwa-epenthesis is 

blocked. 

In words like kern /kern/ 'core' and urn /vrn/ 'id.', the two coda consonants 

do not share their Place features, unlike nasal-stop clusters, and hence schwa 

insertion does take place in these cases. 

The fact that the rule also applies to words like toorn 'anger' and hoorn 

'horn' proves that the final /n/s in these extra-long syllables should not be seen 

as (exceptional) appendix consonants, but as the last consonant of exceptionally 

long codas. Underlyingly, there is no schwa in these words, and we have 

minimal pairs such as toorn /torn/ 'anger'—toren /toran/ 'tower', hoorn /horn/ 

'horn'—horen /horan/ 'to hear'. 

Finally, note that schwas cannot be inserted in word-final obstruent clusters 

such as /sp/ and /sk/, as in wesp 'wasp' and in the suffix -esk. This implies that 

we require the first consonant to be sonorant, or that we treat the special 

clusters /sp/ and /sk/ as a kind of complex segment (cf. Ewen 1982), linked 

to one X on the X-tier. 

From a functional point of view it is quite natural that only heterorganic 

consonant clusters are broken up since they require more articulatory effort 

than homorganic clusters. 

6.3. SCHWA DELETION 

When a Dutch word has two consecutive syllables headed by schwa, the first of 

these schwas may be deleted, provided that the resulting onset consonant 

cluster be an obstruent + liquid cluster, universally the most favoured type 

4 This suggestion was made by John McCarthy (personal coinnuiniculion). Note that this 

condition is restricted here to consonants which indeed usually exhibit local types of assimilation, 

whereas vowels can influence each other across consonants (e.g. Umlaut, vowel harmony), as 

observed in Clements (1985). There are a few exceptions though, like the Sanskrit Nali rule thai 

makes a nasal retroflex if a retroflex consonant precedes it. There may be other segments between 

the consonants involved, hut no coronal consonants (Schein and Sleriade l9Xn). Theiefore, 

Clements and Hume (1993) proposed a different analysis in which non-adjacent consonants can 

share single features, but non-adjacent consonants are blocked from sharing class nodes since 

vowels have the same class nodes. 

5 There are southern and western dialects of Dutch in which schwas in this configuration appear 

to be underlying, since they also have the schwa in heterosyllabic clusters as in \\crkcn 'to work' 

] (cf. Berendscn and Zonneveld 1984).

6.3. SCHWA DELETION 12Q 

of onset cluster. In the examples below the schwa that deletes is represented by 

e. 

(5) soepele 'smooth' [suplo| 

koperen 'copper' 

[kopran] 

knabbelen 'to munch' [knublon] 

bibberen 'to tremble' (bibron] 

snuffelen 'to search' [snvflsn) 

offeren 'to sacrifice' [nfran] 

hevelen 'to syphon' 

(hevlan) 

kietelen 'to tickle' 

|kitlon] 

wandelen 'to walk' 

Iwundlan] 

wisselen 'to change' [wislan] 

mazelen 'pimps' 

Ima/.lon] 

gemakkelijk 'easy' 

lyamuklok] 

rochelen 'to hawk up' [roxlan] 

kegelen 'to play skittles' [keylan] 

What is remarkable here is that /tl/, /dl/, and //I/ form onsets. Universally, 

they are permitted, but in Dutch they do not occur at the lexical level, as 

shown in Section 3.5.2. In the case of /dl/ and //I/, the intuitive syllabification 

of these clusters as onsets is also proven by the fact that the /d/ and the 

/z/ do not devoice. The rule involved here should therefore not only express 

the deletion of the schwa and the configuration in which this is possible, but 

also that the pre-schwa consonant forms an onset with the following 

consonant. Note, moreover, that the rule does not apply to other configurations, 

even if they resulted in syllabifiable (but not always tautosyllabic) 

clusters: 

(6) tekenen /tekonon/ 'to draw' *|teknon| 

be/.emen /bezaman/ 'to sweep' *|bezm3n|/*[besmon| 

redenen /redonan/ 'reasons' *[rednon|/*[retnon] 

rammelen /rumolan/ 'to rattle' *[rumlon| 

Therefore, this process has to be interpreted as a rule that forms obstruentliquid 

clusters across the schwa, whereby the schwa deletes: 

(7) Schwa Deletion 

Delete o in the context [-son]- ~-voc~| 3 

L+apprj 

This rule deletes the schwa of the first relevant syllable. Hence, that syllable 

node is no longer headed by a |-consonant] segment, and will disappear by 

convention. Thus, resyllabification of the original onset obstruent to the next 

onset is possible. This relinking is in accordance with the universal conditions 

mi onsets (the SSG), but not with the language-specific constraints of Dutch that 

forbid /tl-, dl-, /.I-/. So we might assume that language-specific constraints on 

syllable structure are turned off at the postlexical level.


Note that a schwa does not delete if the next syllable is headed by a full 

vowel: 

(8) geraamte /yaramo/ 'skeleton' *[v.rarnt3] 

beloven /balovon/ 'to promise' *[blovon| 

beraad/barad/'meeting' 

*[brut| 

The rule decreases the articulatory effort required for the pronunciation of 

the relevant words in that there is one syllable less to produce. Admittedly, 

consonant clusters are created but this kind of consonant cluster, obstruent + 

liquid, is the universally favoured kind of CC-onset, and may therefore be 

assumed to be relatively simply to produce. 

6.4. VOWEL REDUCTION 

Vowel reduction is the phenomenon that an underlyingly full vowel is optionally 

realized as schwa in unstressed syllables. 7 The following examples 

illustrate this process: 

(9) banaan /banan/ 'banana' [banan] 

lokaal /lokal/ 'class-room' [lokal| 

metaal /metal/ 'metal' [motalj 

muzi'ek /myzik/ 'music' [ma/ik| 

minuut /minyt/ 'minute' [manyt] 

Vowel reduction is subject to a number of conditions. It should first be noted 

that speakers vary somewhat as to the vowel reductions they allow for. 

Furthermore, there is a strong frequency effect in that words with high 

frequency reduce much more easily. High vowels do not reduce easily: for 

instance, reduction of the first vowels of Liane 'id.' /liana/ and Suzanne 'id.' 

/syzana/ is almost impossible: *|lajana|, *|sazuna). Yet, the unstressed 

vowels in minuut and muziek reduce reasonably easily, because they are 

high frequency words that are often used in casual speech. 8 

Reduction is even easier when the vowel is in interstress position. So the 

combined effects of frequency and position makes the reduction of the high 

vowels in the following examples quite natural: 

(10) dominée /domine/ 'parson' |domone| 

asp/'rine /uspirina/ 'aspirin' |usparina) 

lucifer /lysifcr/ 'match' |Iysafer) 

6 Certain lexicalized cases are exceptions, such as ^cloven (ylovan) to believe, xi-n-fiinm-i-nl 

lyrel.irmeirt] 'reformed, Protestant', terug (lrYx|, and tcn-< tit |trcxt| 'rightly'. Moreover, some 

native speakers also delete schwas in syllables which are followed by syllables with primary or 

secondary stress in other words, as in apparaat /uparat/ (upral) 'apparatus' (Schwa Deletion is 

applicable after vowel reduction of the unstressed /a/), kaïn-laan /kupnlan/ |kuplan| 'chaplain' and 

konpelinf! /knpDlirj/ (k.iplirjl 'coupling'. High frequency of words is an additional factor here. 

S Cf. Martin (1968), Booij (1977, 198la, I982/?), and Kager (I9X'J|. 

* In Fidelholt/ (1975) the importance of frequency and the reluctance of high vowels to reduce 

is pointed out with respect to English.

6.4- VOWEL REDUCTION 131 

Important structural conditions are that vowels do not reduce when they are 

syllable-initial, or when they begin with /h/. This correlates with the observations 

made in Section 3.5.4 and Section 3.6 that a prosodie word cannot begin 

with schwa, and that the sequence /ha/ is also impossible: 

(11) anàal 'anal' *[anal] 

elite 'élite' *[alita] 

erotisch 'erotic' *[arotisj 

maoist 'Maoist' *|maaist| 

heraut 'herald' 

heroisch 'heroic' 

humaan 'human' 

*[h3rDut] 

*[h3roi>is) 

*[haman] 

Furthermore, reduction in open syllables is preferred (with certain classes of 

exceptions to be discussed below), except when the syllable-final consonant is 

ambisyllabic. In the latter case, reduction is possible: 9 

(12) syllabe/silaba/'syllable' [salabo] 

misschien /misxin/ 'perhaps' [masxin] 

supporter /svpDrtar/ 'id.' 

[saportar] 

dessert /desert/ 'id.' 

[dasert] 

rapport /report/ 'report' 

[rapr>rt| 

Reduction does not apply in word-final syllables (again, there are exceptional 

classes). Diphthongs do not reduce: 

(13) pleidooi/pleidoj/'plea' *[pladoj] 

seizoen /seizun/ 'season' *[sazun] 

Pauh'en /poulin/ 'Pauline' *[palin| 

The following rule expresses the phonological conditions mentioned so far: 

(14) Vowel Reduction 

[+cons) [-cons] X ) 0 Y) tó 

Place 

Place 

Conditions: 

X = 0or [+cons], Y * 0, o is unstressed 

The rule delinks Place from a vowel, and hence this position will be filled in by 

the default rule for schwa. By requiring the presence of a Place specification on 

the preceding consonant, reduction after /h/ is excluded. Since Y * 0, the 

syllable will not be the word-final one, as required. Conditions on the variable 

X will be discussed below. 

The rule will not apply to diphthongs, as required, because diphthongs are 

Examples from Kager (1989: 306).


sequences of [-cons] segments, and hence do not satisfy the structural 

description of the rule. 

As pointed out in the relevant literature, there are cases in which vowel 

reduction takes place in closed syllables. In word-initial position this is 

particularly clear for the sequence /er/, and also for short vowels followed 

by /s/: 

(15) person/person/'person' [parson] 

percent /persent/ 'per cent' [parsent| 

pastoor /pustor/ 'pastor' |pasto:r] 

pastei /pustei/ 'pie' 

[pastei] 

As pointed out by Koopmans-van Beinum (1982), the /r/ has a centrali/.ing 

effect, and will thus further reduction of a vowel to the central vowel schwa. In 

other cases, intuitions differ. In my idiolect, I have to delete the consonant first 

before being able to reduce the vowel. In that way, the vowel becomes 

reducible, being in syllable-final position: 

(16) benzfne /benzina/ 'petrol' [bazina], *[banzina] 

kanton /kunton/ 'canton' [katon], *[kanton] 

porti'er/portir/ 'porter' [pati:r|, *|partiir| 

kwarti'er/kuurtir/ 'quarter' [kuatiir], *[kuarti:r] 

Other native speakers of Dutch (cf. Kager 1989) allow for the phonetic forms 

starred here by me. We also observe here the interdependency of optional 

rules: once the syllable-final consonant has been dropped, the vowel must 

reduce. In other words, phonetic forms like [be/.ina], [pntiir] and [kuton] 

are impossible. Once we have chosen a certain style of speech, the rules 

become obligatory within that style of speech (Booij 1981«: 149ff.): the use 

of one rule implies use of the other. 10 In any case, the exceptional types of 

closed syllables in which reduction is allowed can be specified as an alternative 

condition on the variable X mentioned in the rule. The conditions on X 

observed so far that will further vowel reduction are: in word-initial syllables, 

if X = /r/ preceded by e, or X =/s/." Such variable furthering conditions are 

characteristic of style-governed, optional rules, as shown by Labov (1972). 

As observed in Booij (1981a: 149) and Kager (1989: 282), closed syllables 

in word-internal position allow for vowel reduction on a much larger scale (the 

reducible vowels are in italic): 

(17) identiek'identical' 

compensatie 'compensation' 

anekdote 'anecdote' 

direktéur 'director' 

alimentatie 'alimony' 

" This has also hoen observed by KuCera (1973) for C/.ech and by Hooper (1976) for English. 

1 

The positive effect of /s/ may have to do wilh the fact that /s/ can easily become onset of the 

following syllabic, so that the vowel to be reduced is in syllable luuil position.

6.4. VOWEL REDUCTION 133 

sentimenteel 'sentimental' 

pav/ljóen 'pavilion' 

inlVltréer 'to infiltrate' 

amalgaam 'amalgam' 

adapteer 'to adopt' 

This suggests that for a vowel to be in a word-internal position furthers its 

reduction whatever the value of X. 

Many of the examples given concern /e/. This is no coincidence: the /e/ 

appears to be the vowel that is most susceptible to reduction, to such an extent 

that in many words the underlying /e/ has been replaced with /a/ (cf. Booij 

19826). For instance, the first vowel of Rebecca 'id.' is always pronounced as 

a schwa. Consequently, the schwa may also surface in syllables where the 

Optimal Grid Rule would have assigned secondary stress if the syllable 

contained a full vowel /e/, as in reformatie 'reformation' [raformatsij and 

revolutie 'revolution' [rovolytsij. In other words, in many cases a historical 

/e/ has been replaced with a schwa in the underlying form. 12 In some pairs of 

related words, the /e/ alternates obligatorily with schwa, as in 

(18) juweel 'jewel' Ijyuelj 

juwelfer 'jeweller' [jyualiir], *[jyueli:r] 

genfe 'genius' 

[zjani], *(zjeni] 

geniaal 'genial' 

[yenijal], *[y3nijal] 

mim'ster 'minister' [ministar], ""[minister] 

ministerie 'ministry' [ministen] or (minasterij 

The fact that the schwas are obligatory here also illustrates that some schwas 

have become part of the underlying form of certain allomorphs, since there is 

no regular alternation involved. 

A factor that might play a role in the lack of reducibility of/i/ in suffixes like 

-iicit '-ity' and -iseer '-ize' is that in these cases the /i/ is located at morphological 

boundaries which appear to impede reduction processes (Labov 1972). 

Let us now look at some cases which are more complicated in that there are 

two syllables that are potentially subject to vowel reduction, words of the 

following type: 

(19) fonologi'e 'phonology' 

individu 'individual' 

réparateur 'repairer' 

In these words both the second and the third syllable are unstressed, and both 

unstressed syllables have the same vowel. Yet, as observed in Booij (1981a: 

148) there is a clear preference to reduce the first unstressed syllable. This 

difference must have a structural explanation since the two vowels involved 

are the same. In other words, it cannot be attributed to a difference in 

1 

Native speakers might also differ here as to whether they still have two possibilities, full 

vowel or schwa, or schwa only.

134 CONNECTED SPEECH I WORD I'HONOLOGY 

reducibility of the vowels. The stress representation of, for instance, fonologie 

is as follows: 

(20) f o no lo gie 

(* ) (*) line 1 

* line 2 

The second syllable can be adjoined to the preceding syllable, forming a foot, 

whereas the third syllable cannot combine with another following syllable into 

a foot, and hence remains a stray syllable (adjunction to the right is impossible 

since I assume that Dutch has uniformly left-headed trochaic feet). We thus 

conclude that reduction in the weak syllable of a foot is easier than in a stray 

syllable. Interestingly, this provides evidence for the foot as a prosodie 

category of Dutch (cf. Booij )982c, 1983, Kager 1989: 312-18). 

As pointed out above, reduction in a stray syllable implies that reduction 

also takes place in a weak syllable of a binary foot. So we may assume the 

following hierarchy: 

(21) Formal style: no reduction 

Informal style: 

reduction in adjoined syllables 

Very informal style: reduction in both syllables 

This hierarchy interacts with a second hierarchy concerning vowel reducibility. 

The following hierarchy of increasing reducibility can be established: 

(22) /y, u, 0/ 

N 

IO, 3/ 

/a, u/ 

/e, i/ 

Kager (1989) pointed out that the two hierarchies may interact, and even make 

opposite predictions if the vowel in the adjoined syllable is less reducible than 

the vowel in the stray syllable, as in the following words: 

(23) logopedie /loyopedi/ 'speech therapy' 

epidemfe /epidemi/ 'epidemic' 

desideratum /desideratvm/ 'id.' 

In these cases, the vowel of the stray syllable appears to reduce more easily 

than that of the adjoined syllable. Kager (1989: 315) concluded that the two 

hierarchies interact in the following way (where style III has the highest degree 

of formality, and style I the lowest degree of formality): 

(24) Vowel Reduction Hierarchy 

Adjunct position Stray position 

/e/ style III style III 

/a/ style II style II 

/o,i/ style II style I 

/y,u/ style I excluded

6.4- VOWEL REDUCTION 135 

The use of a style which is lower on the degree of formality implies that the 

reductions allowed for in styles which are more formal also apply (Dressier 

1974, Booij 1981«: 150). So we predict the following possible phonetic 

realizations for a number of crucial cases (in order of increasing informality): 

(25) economi'e /ekonomi/ 'economy': |ekonomi|, [ekanomi), [ekanami], 

*[ekonami| 

adrenaline /adrenalina/ 'adrenalin'; [adrenalins], [adranalina], 

ladranalina], *|adrenalina| 

lögopedi'e /loyopedi/ 'speech therapy'; (loyopedij, [loyopadi), 

lloyapadi], *[loyapedi] 

grammâtika /yrumatika/ 'grammar': [yrumatikaj, (yramatika], 

lyramataka], *|yrumatoka| 

Usually, vowel reduction does not take place in final syllables. There are, 

however, some exceptions, in particular words ending in vowel + /r/, and 

words ending in /i, e/ + consonant: 

(26) motor /motor/ 'engine' (molar) 

kermis /kermis/ 'fair' 

|kermas| 

kennis /ken+nis/ 'acquaintance' [kenos] 

koning /konin./ 'king' 

|konorj| 

monnik /irnnik/ 'monk' 

[rmn.ikj 

i'dem /idem/ 'id.' 

[idam] 

Agnes /axnes/ 'id.' 

(uxnas] 

amen /amen/ 'id.' 

(aman) 

In other words, the condition that the variable Y in the rule of Vowel 

Reduction be non-zero must be relaxed in that Y can be zero if X is a 

consonant and is preceded by /e, i/, or if X is /r/. In words historically ending 

in unstressed /er/, the /e/ is always realized as a schwa, so it must be assumed 

to be present underlyingly, as in peiler 'father' (religious) Ipatar]. This lexical 

replacement of/e/ through schwa can also be observed when stress shift makes 

a syllable with /e/ unstressed, as in the following examples: 

(27) normaliter /normaliter/ 

normalster /normalitar/ 

negeren /neyeran/ 'to ignore' (neye:ran| or |naye:ran] 

negeren /neyaran/ 'to ignore, (lit.) to treat as a negro' 

In the first word, the stress shift is a case of regularization (penultimate is the 

regular pattern), the change in the second word is a case of folk etymology: 

people interpret this verb as a conversion of the noun neger /neyor/ 'negro'. 

We should also be aware of the fact that, since there is no special letter for 

the schwa, the schwa is mostly represented by e, which will induce native 

speakers to interpret the e as a schwa anyway.

136 CONNECTED SPEECH I: WORD I'HONOI.()(, V 

6.5. VOWEL SHORTENING 

Vowels can be shortened in word-initial position under the condition that the 

syllable in which they occur does not bear the main stress of the word. That is, 

secondary stress does not block the shortening: 

(28) /a-u/ paradijs 'paradise' 

amalgaam 'amalgam' 

analyse 'analysis' 

/e-i/ 

/O-D/ 

Amerika 'America' 

banaan 'banana' 

tf'levi'sie 'television' 

telefoon 'telephone' 

g(e)rpformeerd 'reformed' 1 

radeneer 'to reason' 

kolossaal 'enormous' 

prMiti'ek 'politics' 

/i-i/ d/rektéur 'director' 

Except for /a/ the relevant vowels can only be reduced if they head the first 

syllable of a word-initial foot with its weak syllable being headed by schwa. In 

other words, reduction of the second syllable is required, except in the case 

of /a/: 

(29) analyse 'analysis': |anali/o|, [unali/.e], (unalize) 

televisie 'television'; (televi/,i|, [telavizi), [tilovi/.i|, *|tilevi/.i| 

politiek 'polities': [politik], (pnlatikj, *(p3litik] 

monument 'monument': [monyment], *|monymcnt), *[mDnoment| 

In televisie and politick the first vowel can be shortened if the second vowel is 

realized as schwa; in monument the first vowel cannot be shortened because 

the second (high) vowel does not reduce to schwa. 

The rule can be interpreted as a rule that deletes X in word-initial syllables. 

By requiring it to be followed by a consonant, reduction of the first vowel in a 

word like chatitiach /xaotis/ 'chaotic' is correctly blocked: 

(30) Vowel Shortening 

Delete X in the context u,( 0 (Q X X 

I 

I 

|-cons| l+cons] 

where Q stands for zero or more consonants 

Condition: 

o does not bear main stress; if it is not headed by /a/, it forms part ot a 

binary foot with a weak o headed by a schwa. 

" As pointed oui above, the first schwa of Ihis word which is represented by the parenthcsi/' '' ' 

will have to be deleted as well in order to make the shortening possible.

6.6. INTRUSIVK STOPS 137 

In the case of the shortened /i/, which results in [i], the vowel is also lowered 

because there are no short high vowels in Dutch. In other words, this lowering 

is an automatic consequence, and can be taken care of by the relevant 

redundancy rule. 

The rule of Vowel Shortening is related to the rule of Vowel Reduction in 

that application of Vowel Reduction feeds application of Vowel Shortening, as 

illustrated above in (29). 

6.6. INTRUSIVE STOPS 

Intrusive stops illustrate the phenomenon of retiming. As Browman and 

Goldstein (1990) point out, fast-speech processes may be seen as the overlap 

in time and space between the articulatory gestures required for the realization 

«I the individual segments. This does not necessarily imply, however, that 

such processes should not be seen as rules of the language, since languages 

may differ in their fast-speech rules, i.e., in the way in which the phonetic 

requirements imposed by each individual segment on the vocal apparatus in 

the realization of words in connected speech are reconciled. 

The traditional observation is that a stop can be inserted between a noncoronal 

nasal and a following obstruent in the same syllable, a [p] after |m], and 

a |k] after [rj]. Consequently, the phonetic forms of the third-person singular 

present forms of the verbs kammen 'to comb' and kampen 'to fight', and those 

for zingen 'to sing' and zinken 'to sink' become undistinguishable: 

(31) kam -t /kamt/ [kumpt] 

kamp-t /kumpt/ |kumpt| 

zing-t /zirjt/ 

zink-t /zirjkt/ 

l/'Qkt] 

[zirjkl| 

Rather than considering this as a case of segment insertion, it should be 

expressed as a case of retiming. For instance, in the word kamt, due to 

retiming of the melodic tier with respect to the X-tier, the Place feature 

ILabial] is linked to the following [— sonorant] element, which thus becomes 

a contour segment: in its first phase it is a labial stop [pi, and subsequently it is 

u coronal stop [t]. The same applies to a velar nasal plus stop: the feature 

[Dorsal] of the velar nasal spreads, making the first phase of the /t/ sound like a 

Ik]. Thus, the rule reads as in (32) (I presuppose that appendix consonants are 

incorporated into the coda before this rule applies). 

(32) Intrusive Stop Insertion 

[+cons] |+cons| ) a 

[+nas] 

Place 

Place


6.7. /(/-DELETION 

An unstressed /i/ optionally deletes between a consonant and the palatal glide 

/)/ (itself inserted by Homorganic Glide Insertion): 

(33) station /station/ 'station' 

rationed /rationel/ 'rational' 

financieel /finunsiel/ 'financial' 

sociaal /sosial/ 'social' 

religieus /relivi0s/ 'religious' 

piano /piano/ 'id.' 

ambiance /umbiusa/ 'id.' 

ideaal /ideal/ 'ideal' 

radioloog 'radiologist' 

lineaal 'ruler' 

Ariane 'id.' 

alliantie 'alliance' 

|siuspn| 14 

[rasjonel] 

[finunsjell 

Isosjal] 

|reliyj0s] 

| pjano) 

[umbjuso] 

[idijal or lidjal) 15 

[radjolox] 

[Imjal] 

|arjano| 

(uljuntsi) 

The stress condition correctly predicts that the /i/ does not delete in Siam 

/sium/ 'id.' or in Siamees /siames/ 'Siamese'. If the main stress precedes the 

syllable with /i/, there is also no deletion. For instance, deletion cannot take 

place in paria /paria/ 'pariah' *[parja| or Âzic /a/.io/ 'Asia' *[a/J3] (compare 

Aziatisch /a/.iatis/ 'Asian' in which, after glide insertion, deletion of /i/ can 

take place: |azjatis|). 

When the preceding consonant is /s/, application of the rule is more probable, 

corresponding with the fact that of the resulting complex onsets only /sj/ 

is a completely regular onset at the lexical level, and can moreover be 

pronounced as one segment, a postalveolar |J'|. 

Deletion of /i/ does not occur when it is preceded by a complex onset. For 

instance, in Adrianus /adrianvs/ 'id.' it is impossible to get the phonetic form 

[adrjanvsl, presumably because this would create a too complex onset, /drj-A 

The rule can therefore be formulated as (34). 

(34) /i/-deletion: delete X X in the structure 

„( (X) —; ) X 

|+cons| 

+voc 

-back 

+high 

Condition: 

O is unstressed, and is not preceded by main stress. 

4 The tirsl vowel [a] may also he reali/.ecl as |u|, in accordance with the rule of Vowel 

Shortening given above. 

He fore /a/ the /e/ can he reali/eil as an |i|; see also lineaal 'ruler' jlinijal) and UI,;KI/ 'area' 

|anjal|

6.8. /N/-DELETION 139 

The fact that the deletion only takes places in unstressed syllables, which have 

the shortest duration of all syllables, complies with a retiming interpretation of 

this phenomenon: it is only the X-tier that is changed. 

For many speakers, realization of the word station /station/ 'id.' with 

deletion of the /i/ is the only possible pronunciation: [stasJDii] or [statsJDnJ, 

not *|statsipn] or *[stasijsn|. 

6.8. /n/-DELETION 

In standard Dutch, syllable-final /n/s can be dropped after a schwa, except in 

the indefinite article een /an/ 'a': 

(35) Singular nouns 

regen /reysn/ 'rain' 

molen /molon/ 'mill' 

deken /dekan/ 'blanket' 

Plural nouns 

bloem-en /blum-an/ 'flowers' 

plant-en /plum an/ 'plants' 

boek-en /buk-an/ 'books' 

Plural verbs / infinitives 

lopen /lop-an/ 'to walk' (pres. pl./inf.) 

eten /et-an/ 'to eat' (pres. pl./inf.) 

[reyan] or [reya] 

[molan] or [molo) 

|dekan] or [deka] 

[bluman] or [bluma] 

Ipluntan] or [plants] 

[bukan] or [buka| 

[lopan] or [lopa| 

[etan| or [eta| 

It is sometimes assumed that for those speakers who always drop the final /n/ 

of the plural morpheme -en /an/ of verbs and nouns, the underlying form of 

the plural suffix is /D/ instead. However, in Chapter 8 it will be shown that 

even for those speakers an underlying form /an/ is required in order to 

explain why the final schwa of the past-tense suffix may disappear before 

a schwa-initial clitic through Prevocalic Schwa Deletion, but the schwa of 

the plural suffix cannot. For instance, zette het /zêta at/ 'put (past sg.) it' can 

be pronounced as |zetat|, whereas zetten het /zetan at/ 'put (pres. pi.) it' is 

pronounced as [zetanat], and [zctat] is impossible in that case. This can 

only be accounted for if the underlying form of the plural morpheme ends in an 

/n/. 

The /n/ can also be deleted at the end of syllables of other lexical categories 

such as adjectives and particles/prepositions: 

(36) gouden/yûdan/'golden' [youda| 

open /opan/ 'id.' 

|opa| 

boven /bovon/ 'above, upstairs' |bova|

140 CONNECTED SPEECH I: WORD PHONOI O(iV 

Examples of word-internal /n/-deletion are diminutives with a base noun 

ending in -en, and some adverbials:"' 

(37) kuikentje/kceykan-tja/'chicken'(dimin.) [kœykatja] 

molentje /molan-tja/ 'mill' (dimin.) 

[molatja] 

wagentje /uayan-tja/ 'car(t)' (dimin.) [uayatjol 

regentje /reyan-tja/ 'rain' (dimin.) 

|reyotjo| 

openlijk /opan-lak/ 'openly' 

|opalok| 

gezamenlijk /ya-zaman-lak/ 'common' |yazamalak| 

eventjes /evan-tjas/ 'a little while' 

[evatjas] 

As pointed out above, /n/ deletes obligatorily in word-initial syllables if the 

preceding vowel has been turned into a schwa by Vowel Reduction as in 

benzine /benzina/ 'petrol' [bazina]. However, in this position it is also 

possible for the /r/ to delete, as in portier 'porter' |pati:r|, so this might be 

a different rule after all. 

For many speakers, in particular in the western part of the Netherlands, the 

deletion of /n/ is obligatory. This may lead to the conclusion that for those 

speakers the underlying forms of the relevant morphemes end in /a/ rather 

than /an/. Note, however, that this is an incorrect conclusion since there are 

pairs of related words in which the verb is morphologically related to an /n/- 

final word, and the /n/ shows up in non-word-final position, for instance in the 

present plural form and the (phonologically identical) infinitive of the verb: 

(3K) 

regen 'rain' 

baken 'beacon' 

zegen 'blessing' 

teken 'sign' 

open 'open' 

Verb (plural present t» infinitive} 

regenen 'to rain' [reyana(n)] 

afbakenen 'to delimit' [uvbakana(n)] 

zegenen 'to bless' [zeyana(n)] 

tekenen 'to draw' [tekana(n)] 

openen 'to open' [opana(n)] 

Therefore, the words in the left column must still have a final /n/ underlyingly- 

As pointed out by Koefoed (1979), the rule does not apply to the /n/ that 

appears at the end of a verbal stem. This is the case for the first-person singular 

present forms of verbs that have a zero inflectional ending: 

(39) Verbal stem First-person singular present 

teken 'to draw' [tekan], *[teka] 

oefen 'to train' [ufan], *|ufa| 

reken 'to count' [rekan], *[reka] 

This shows again that morphological structure might influence the application 

6 The /n/ also deletes in informal speech before a syllablc-hnal /s/ in frequent words such as 

telkens /tElkans/ 'time and again' lltlk.is], widens /ueyons/ 'because of' |oey.is|, 'A nuirons /smsryans/ 

'in the morning' Isrnnryas], lakens /lakan-s/ 'sheets' |lak,is]. and crocus /crysns/ 

'somewhere' [trps|. The deletion of the nasal before /s/ may be related to the fact that nasals 

rarely occur before fricatives (Padgett 1991, 1992). Note that /n/ does not delete before /t/ as in 

lopend 'walking' /lopand/ [lop-ant], *|lop3t).

6-9- CONCLUSIONS 141 

of rules of connected speech: we have to impose a negative condition on the 

process in that it may not delete /n/s at the end of verbal stems. 

The /n/ also deletes at the end of the first prosodie word of compounds and 

derived words, as expected (cf. Berendsen 1986: 88): 

(40) regen-pak (lit.) 'rain suit' [reyapak] 

open-baar 'public' 

|opoba:r] 

Christen-dom 'Christianity' [kristadom] 

The rule of /n/-deletion can be seen as a rule that deletes the final /n/ of a 

syllable after a schwa at the end of a morpheme that is not a verbal stem (41). 

(41) /n/-deletion 

Delete f+nasl in the context 9 —) 0 ] x 

LcorJ 

Condition: 

X is a morphological boundary but not a verbal stem boundary 

The feature [Coronal) is mentioned in order to avoid deletion of the /m/ which 

also occurs after schwa. This implies that coronal nasals must be specified for 

Place when /n/-deletion applies. 

The requirement that the /n/ is not only syllable-final, but also adjacent to a 

morphological boundary also correctly excludes deletion of the /n/ in present 

participles like volgende 'following, next' with the morphological structure 

[ [[volg] v end]\, e] v , although the /n/ is syllable-final in such cases. The rule 

thus shows that the prosodie and morphological structure of a word may have 

to be referred to simultaneously in a rule (cf. Booij and Lieber 1993). 

6.9. CONCLUSIONS 

A number of specific properties are shared by the rules of connected speech 

discussed in this chapter. They tend to increase the ease of production, and 

thus make words less distinguishable. Ease of production can indeed get 

priority in informal situations in which the speaker can afford to reduce the 

perceptibility of words. Whether they apply also depends on a number of nonphonological 

factors: frequency (i.e. the degree of activation of a word in 

lexical memory), and lexicalization (in the sense of loss of morphological 

transparency). Moreover, speakers differ in the degree to which they allow for 

reduction processes to apply. Phonological conditions sometimes appear to be 

of a gradual nature. 

A third property of rules of connected speech is that although a rule may be 

optional, it may have to apply obligatorily due to the application of another 

rule. In particular, the behaviour of the rules discussed in this chapter support 

Dressler's (1974) hypothesis that the use of rules for less formal or less 

monitored styles of speech implies the use of rules for more formal or more

142 CONNECTED SPI.I.Ml I WORD PHONOl.OCiY 

monitored styles of speech. There also appear to be relations of mutual 

implication between rules of connected speech. 

In the case-of fast-speech rules we saw that they may sometimes be seen as 

cases of retiming rather than as deletion or insertion of segments. 

The rules of connected speech that apply across word boundaries, and that 

are discussed in the next chapter, also exhibit the properties summarized here.

CONNECTED SPEECH II: 

SENTENCE PHONOLOGY 


In this chapter, we deal with rules that (also) apply across word boundaries. An 

important theoretical question is that of how to state the domain of such rules: 

'Are some of these domains to be defined in terms of syntactic information, as 

present in the surface structure of sentences, or can they be stated exclusively 

in terms of prosodie domains?' 1 A related issue is the exact nature of the 

hierarchy of prosodie domains, the so-called Prosodie Hierarchy. 

The domains of rule application for P-rules have been argued to form the 

following hierarchy (Nespor and Vogel I986): 

(1) Prosodie H ici 'an h v 

Syllable (a) 

Foot (F) 

Prosodie Word (w) 

Clitic Group (C) 

Phonological Phrase (()>) 

Intonational Phrase (IP) 

Utterance (U) 

The first three categories have to be available at the lexical level, the higher 

prosodie domains will be constructed on the basis of syntactic structure. 

Evidence for the foot was given above in relation to glottal-stop insertion in 

hiatus position (Section 4.2.3) and the stress-related phenomenon of vowel 

reduction (Section 6.4) 2 

Nespor and Vogel (1986) also proposed the domain of the Clitic Group. 

However, it is doubtful whether we really need this domain (cf. Booij 1988c. 

Zee 1988, Malikouti-Drachman and Drachman 1992). As far as Dutch is 

concerned, it is argued in Chapter 8 that clitics are incorporated into the 

preceding, or Chomsky-adjoined to the following prosodie word, without 

forming a prosodie category of their own. 

The next prosodie category is the Phonological Phrase. As for the construction 

of phonological phrases, it is usually assumed that each word of a lexical 

1 

('I'. Booij (I992u) lor a survey ol' this discussion. 

: Arguments loi the loot as part ol' the Prosodie Hierarchy are given in Poser (I9K'>, 1990), 

McCarthy and Prince (1990), and Im and Mester (1992).

144 CONNECTED SPEECH 11: SENTENCE PHONOLOGY 

category forms a phonological phrase with either the preceding or the following 

function words (determiners, prepositions, conjunctions, auxiliaries). 

According to Selkirk (1986) this is a parameter: English joins function words 

with the following lexical category into a phonological phrase; in Japanese it is 

just the inverse: function words form a phonological phrase with the preceding 

lexical category. For Dutch I will assume that, when possible, function words 

form

7.2. PHONOLOGY ABOVE WORD LEVEL 145 

This structure correctly predicts that a syllable boundary coincides with the 

prefix boundary. Formally, it is a case of Chomsky-adjunction of a syllable to a 

prosodie word, thus creating a new prosodie word node. Thus, it follows that 

Prevocalic Schwa Deletion does not apply, except in very fast or informal 

speech where such boundaries are ignored. This rule requires that the schwa to 

be deleted and the next vowel are dominated by the same prosodie word node. 

This is not the case in structure (2) given the following definition of dominance 

(Chomsky 1986: 7): 

(3) Dominance 

a is dominated by ß only if it is dominated by every segment of ß 

In the example headcm the préfixai schwa (a) is not dominated by the prosodie 

word node (ß) because it is not dominated by the lowest co-node. Hence, the 

rule does not apply, as required. 

Similarly, compounds may be assumed to form prosodie structures of the 

following type in which a prosodie word is Chomsky-adjoined to a preceding 

prosodie word: 

(4) co 

co s 

co w 

Again, rules that have the prosodie word as their domain will not apply to such 

compound structures given the definition of dominance in (3). 

In sum, it seems necessary to weaken the Strict Layer Hypothesis in so far as 

we have to allow for adjunction of prosodie categories to other prosodie 

categories. 

In the next sections of this chapter, the role of the Phonological Phrase and 

the Intonational Phrase are illustrated. It is shown that other levels of representation 

such as the argument structure of the sentence may also play a role in 

sentence phonology. 

7.2. PHONOLOGY ABOVE THE WORD LEVEL 

In Chapter 4 it is repeatedly pointed out that many of the P-rules discussed 

there that apply obligatorily in the domain of the prosodie word—and often 

also in compounds—apply optionally in larger domains such as (compounds

146 CONNECTED SPEECH II: SENTI NCI-. PHONOLOGY 

and) phrases. The same point will be made with respect to /t/-deletion (Section 

7.2.6): this rule exhibits the properties of variable rules in that the rule applies 

above the level of the prosodie word, but with a larger degree of optionality. 

In general, we expect rules of connected speech to apply more frequently in 

smaller domains than in larger domains because words are tied together more 

closely in smaller domains than in larger domains. 

For Dutch, the issue of the prosodie domains of rules above the level of the 

prosodie word is an underresearched area. It is sufficient for present purposes 

to assume that the probability of application of P-rules above the word level 

decreases as the relevant prosodie domain of application becomes larger. In 

other words, I do not assign a particular prosodie domain to P-rules, but 

generally they are restricted to being applied within intonational phrases. 

This implies that they are postlexical rules, and that they can also apply in 

smaller domains than the IP. 

7.2.1. Voice Assimilation 

The rules of voice assimilation presented in Chapter 4 are illustrated there 

mainly by means of compounds, i.e. assimilation across co-boundaries. Voice 

assimilation also applies in larger domains, across ^-boundaries (Loots 1983, 

Menert 1988). The following examples illustrate this: 

(5) ( (dat ik),), (aan de rand),,, (van het bos),,, (bouw)u| (Regressive Assimilation) 

(6) ( (op die manier),» (zal Piet),, (/.akkeiD^),,. 

in that way will Peter fail' 

op die /Dp di/ 'in that' [Dbdi| (Regressive Assimilation) 

Piet zakken /pit zuksn/ 'Peter fail' [pitsukan] (Progressive Assimilation) 

Note that in the sequences bos bouw and Piet zakken the words are separated 

by a ({»-boundary, since they are words of lexical categories which head a 

phonological phrase of their own. 

Loots's (1983) investigation of regressive assimilation clearly indicates that 

the rule has the IP as its domain, and also that the rule applies more frequently 

if the consonants involved belong to the same phonological phrase: there was a 

clear difference in frequency of application of regressive assimilation within 

compounds from that within sentences. Since the prosodie words of a compound 

of course belong to the same phonological phrase, we can state that 

voice assimilation applies more frequently in s than in IPs. 

A rule of voice assimilation which only applies across a prosodie word 

boundary is that of Fricative Voicing that voices prosodie word-final fricatives 

preceded by a sonorant, and followed by a vocoid (cf. Zwaardemaker and


Eijkman 1928: 226, Gussenhoven and Broeders 1976: 140). The following 

examples illustrate this type of voicing: 

(7) pas op 'be careful' [pu/np| 

huisarts 'GP' [hœy/urts] 

was je 'were you' luaZja] 

hoefijzer 'horse shoe' [huvti/,3r| 

twaalf uur 'twelve o'clock' [tualvy:r| 

Note that voicing of fricatives does not occur within prosodie words 

(Gussenhoven 1985). Thus we may get systematic phonetic differences 

between kiesjc [kijjo] 'molar' (dimin.) and kies je |ki3ja] '(lit.) choose 

you', and between wasje (uujjo] 'wash' (dimin. noun) and was je |uu3Jo] 

'(lit.) were you', because the diminutive suffix -je is a cohering suffix, and thus 

forms one prosodie word with its base. The rule can be stated as (8). 

(8) Fricative Voice Assimilation 

[+son) |-son] ), u [+voc] 

|+cont] 

Laryngeal 

Laryngeal 

The frequency of application of rules of sentence phonology such as voice 

assimilation is of course also determined by the performance factors discussed 

in Chapter 6, particularly degree of monitoring and speech rate (Menert 1988). 

There are also individual differences (Slis 1985). 

7.2.2. Nasal Assimilation 

Examples of the application of Nasal Assimilation across word boundaries are 

given in Chapter 4. The examples there are all prepositional phrases, which 

form one phonological phrase. There is no evidence that Nasal Assimilation 

does not apply in the larger domain IP. For instance, in the following 

sentences, Nasal Assimilation seems to be possible for the final nasals of 

haan 'job' /ban/ and boon 'bean' /bon/: 

(9) Wiedebaa[rj] krijgt . . . 

'Who the job gets ..." 

Wie de boo[m] pakt . . . 

'Who the bean takes ..." 

It should be remembered, however, that such observations are based on 

intuitions only, and require experimental underpinnings. Moreover, as pointed 

out by Nolan (1992), assimilations may be only partial. In cases of partial 

Nasal Assimilation, the nasal would be partially Coronal, partially Dorsal or 

Labial, which may be represented as follows (cf. Hayes 1992):

148 CONNECTED SPEECH II: SENTENCE PHONOLOGY 

(10) Nasal Assimilation (postlexical) 

|-son] 

|+nas] Place Place 

Cor 

In other words, whereas nasal assimilation in word phonology spreads the Place 

feature of the following consonant to the preceding underspecified nasal consonant, 

nasal assimilation at the postlexical level only affects coronals. This, by 

itself, could be represented by leaving coronal nasals unspecified until after 

Nasal Assimilation has applied postlexically, provided that the nasal consonant 

assimilates completely. However, if the first stage of the nasal consonant can 

also be coronal, the coronal nasal must be specified as such before spreading of 

the Place feature of the following consonant takes place. The actual stretch of 

alveolar articulation may vary with speech rate and degree of monitoring. 

Moreover, as argued in Section 4.3.2, certain lexical rules already require 

the proper identification of the coronal nasal which therefore cannot be left 

unspecified until the postlexical level. Consequently, we cannot identify the 

lexical and the postlexical processes of nasal assimilation. 

In the case of word-final /n/ preceded by a schwa the /n/ will delete through 

the rule of /n/-deletion instead of being assimilated: 

(11) de kleden (kleda] pakken '(lit.) the carpets take', i.e., 'to take the carpets' 

de kleden | kledo| kloppen '(lit.) the carpets beat', i.e., 'to beat the carpets' 

In the case of nasals before fricatives another phenomenon plays a role. As 

observed by Trommelen (1984: 265) 4 an alveolar nasal may be deleted in fast 

speech before a following fricative. According to Gussenhoven and Broeders 

(1976) nasal deletion is even more general, applying before non-plosives. The 

preceding vowel will then be nasalized and somewhat lengthened: 

(12) on-fatsoenlijk 'indecent' /on-fatsunlak/ [5futsunlok| 

on-zeker 'uncertain' /on-zekar/ 

[5zek3r] 

on-gewoon 'abnormal' /on-yauon/ 

[5v3uon) 

wan-gedrag 'misbehaviour' /uan-yadruy/ 

[uayodruxl 

on-weer 'thunderstorm' /Dn-uer/ 

[5ue:r] 

on-rustig 'unquiet' /on-rvstay/ 

[5rvstox] 

de ton grijpen '(lit.) 'the ton grab', '(lit.) to grab the ton' [tSyreipan] 

de ton pakken '(lit.) the ton take', i.e., 'to take the ton' [tompakan] 

This process does not apply to non-coronal nasals. It may be phonetically 

'explained' in that the sequence nasal + continuant requires two different 

4 Trommelen mentions Zwaardemaker and Eijkman (1928: 230) and Eijkman ( 1955: 107) as her 

sources.


gestures with respect to the degree of stricture of the vocal tract: first, a stop 

and then a continuant. In the case of a nasal followed by a plosive obstruent the 

two consonants can be realized with the same gesture. This may also explain 

why nasals do not co-occur with fricatives in the same coda in Dutch, and that 

this sequence, even when it is a heterosyllabic cluster, is rather rare within 

morphemes, compared to clusters of nasals + plosives, an example being 

kamfer 'camphor'. 5 The restriction of this process to the alveolar nasal is 

the same as that found for other processes of assimilation and reduction in 

connected speech: it is always the alveolar nasal that is subject to such 

processes. 

The rule for this kind of nasal deletion with compensatory lengthening may 

be stated in (13). 6 

(13) Nasal Deletion 

X X X 

[-cons] [+cons] [+cont] 

[+nas] 

Cor 

The rule is a case of delinking-cum-spreading: the features of the nasal 

consonant are all delinked from the X-slot, and the feature [+nasal] is relinked 

to the preceding vowel, thus creating a nasalized vowel. I assume that the 

feature [—consonant] will subsequently spread to the vacated X-position, 

which accounts for the lengthening effect on the preceding vowel. The way 

in which the rule is formulated expresses the insight that such rules are 

basically a matter of retiming: the melodic segments are not deleted, but the 

features are distributed in a different way. Moreover, Eijkman (1955) states 

that the vowel is only optionally lengthened in addition to being nasalized, 

which is understandable in view of the fact that the available articulation time 

is reduced anyway in this kind of speech. 

The rule may also apply word-internally, in particular when a velar nasal 

precedes a velar fricative. For instance, words like congres 'congress' 

/konvres/ and fungeren 'to function' /fvnveran/ can be pronounced with a 

nasal vowel and without the nasal consonant being realized. A special case is 

the pronunciation of the word koningin 'queen' /konirj+m/ as [konsyin] (in 

addition to the standard pronunciation [konirjm])-because the velar fricative 

Ivl that shows up here is synchronically not present in the underlying form of 

the base word koning 'king' /konin/. It is also remarkable that the combination 

velar nasal + stop has survived in a number of native words such as koninklijk 

* See Padgett (1991, 1992) for cross-linguistic evidence concerning this difference between 

fricative and plosives with respect to combinability with other consonants. 

6 This formulation implies that /n/ is not deleted before /!/ which is [-cont], a correct prediction 

as far as 1 know.

150 CONNECTED SPEECH 11: SENTENCE PHONOLOGY 

'royal' [konirjklak], whereas the velar fricative systematically disappeared 

after velar nasals, again suggesting that such clusters have a marked character. 

In western varieties of standard Dutch one may also find /n/-deletion before 

/s/ within lexical morphemes, with concomitant lengthening and nasalization 

of the vowel: 

(14) Hans 'id.' [ho:s] 

gans 'goose' |yo:s] 

kans 'chance' [ka:s] 

ons 'ounce' [5:s] 

This phenomenon can be accounted for by dropping the condition on Nasal 

Deletion that the continuant stand at the beginning of a prosodie word. 

However, as pointed out, the process does not belong to standard Dutch. 

7.2..?. Hiatus rules 

In Chapter 4, we discussed two hiatus rules, Homorganic Glide Insertion (HGI) 

and Prevocalic Schwa Deletion. HGI also applies in compounds, in other s, 

and in IPs (cf. Berendsen and Den Os 1987): 

(15) 4> 

zee[j]arend 'sea eagle' 

toe[u|eigenen 'to appropriate' 

sherry[j]achtig 'sherry-like' 

twee-en-twintig 'twenty-two' [twejantwintax] 

die [j] avond 'that evening' 

IP 

( (Marie),,, [jj (eet niet)$) IP 

'Mary does not eat' 

(Henk) 0 (haalde),,, (de vlo)«, [u] (uit het eten)()>)ip 

'Henk removed the flea from the food' 

Schwa deletion appears to be more restricted at the postlexical level. In 

particular, it does not apply to schwa-final prefixes, except in some lexicalized 

forms such as gereformeerd 'reformed' [yrefbrmeirtl and geloven 'to believe' 

[ylovan): 

(16) b[3]-amen'to agree' *[bam3n] 

b[a]-ogen 'to aim' *[boyon| 

g[o]-opend 'open' *|yopant| 

On the other hand, it may apply within compounds and in noun phrases 

beginning with the determiner de, that is, within (|>s: 

(17) mod[3]-adviseur'fashion adviser' [modatfis0:r] 

d[a] avond 'the evening' 

[davont] 

rod[a] aardbeien 'red strawberries' [rodairtbeijan]

7-2. PHONOLOGY ABOVE WORD LEVEL 15! 

Application of this process is characteristic of casual speech. This is also clear 

from the fact that the preposition te /ta/ is not subject to reduction, unlike the 

phonologically similar article de, presumably because this preposition is only 

used in very formal styles of speech. The absence of reduction in prefixes 

can be related to the observation made in Section 6.4 that morphological 

boundaries impede reduction processes. 

If the schwa deletes, the preceding consonant becomes the onset of the next 

syllable. Thus, the effect of both HGI and Prevocalic Schwa Deletion is that 

the number of onsetless syllables reduces. After a pause, that is, when there is 

no resyllabification, a glottal stop is always inserted before the vowel-initial 

word after the pause (Jongenburger and Van Heuven 1991). 

7.2.4. De gemination 

Degemination is one of the rules that apply obligatorily within prosodie words. 

When two identical consonants come together within a complex word or a 

phrase, one of them may be deleted (or they may be said to become one 

consonant; phonetically it may be the case that the length of a geminate 

consonant is still somewhat larger than that of its single counterpart): 

(18) ver-rassen /ver-rusan/ 'to surprise' [vcrasanj 

pak-kans /puk-kcms/ 'chance of being caught' [pukans] 

ik koop /ik kop/ 'I buy' 

[ikop] 

aan-name /an-nama/ 'assumption' 

[anama] 

in Namen /in naman/ 'in Namen' 

[maman] 

Jan nam /jan num/ 'John took' 

[ junum] 

The rule is the same as that given in Section 4.2.4. The only difference is that it 

is optional as a postlexical rule, and its domain is the IP, as shown by the last 

example of (18). 

7.2.5. Palatalization 

Palatalization affects alveolar consonants before a /j/. Palatalization is clearly 

observable before the personal pronoun je when it joins the preceding prosodie 

word. In other words, the parts anje /anja/ of Spanje /spanja/ 'Spain' and kan 

je /kan ja/ 'can you?' are phonetically identical when je is encliticized. 

(19) had je? 'had you?' |hutja] or [hata] 

was je? 'were you?' [uu3Ja] or [0033] 

kan je? 'can you?' [kunja] or |kuna] 

ben je? 'are you? [btnja] or [bcfia] 

In larger prosodie domains, palatalization is certainly possible in casual and 

fast speech, as in: 

(20) ( (Ik ken) $ (je moeder),), )jp 'I know your mother' [ikenjamudar]


In the case of /n/-final auxiliaries such as kan and ben it is also possible to 

delete the /n/ before /j/: kun je 'can you' [kaja], ben je 'are you' [btjo|. 

The question whether the /j/ after palatalized consonants is still there as a 

separate segment is clearly a matter of speech style and speech rate, in other 

words, a question of timing of the articulatory gestures with respect to the X- 

tier. For instance, the relevant part /sj/ of the representation of was je after 

Palatalization will be as in (21). 

(21) |-son) [+voc] 

I 

Place 

I 

Cor 

I 

Place 

Since the feature [—back] of the /j/ is now linked to the preceding consonant, 

the original X dominating that feature can be deleted in fast speech, while 

leaving the information concerning the place of constriction intact. 

7.2.6. Ill-deletion 

The rule of /t/-deletion is typically one of the processes that occur in fast 

speech, but to a lesser extent also in careful speech. 

If a /t/ in a coda is preceded by an obstruent, and followed by another 

consonant, the /t/ may delete. This is illustrated here first for diminutive nouns 

in which deletion is obligatory: 

(22) klacht-je 'complaint' 

abt-je 'abbot' 

pact-je 'pact' 

[pukjo] 

markt-je 'market' |murkJ3| 

kaft-je 'cover' 

[kafja] 

kast-je 'cupboard' Ikujjo] 

The A/ does not delete, however, when the preceding consonant is a sonorant 

consonant: 

(23) tand-je 'tooth' [tuntjo], *[tanj3) 

hemd-je 'shirt' [hcmtja], *[hemj3J 

hart-je 'heart' [hurtja], *|hurjo| 

kilt-je 'kilt' [kiltjo], *[kilja] 

The rule of /t/-deletion also accounts for the obligatory deletion of the A/ 

before the suffixes -,v and -st: 

(24) echt-st 'most real' [exst] 

echt-s '(something) real' |exs|

licht-st 'lightest' 

licht-s '(something) light' 

J.2. PHONOLOGY ABOVE WORD LEVEL 153 

[lixst] 

|lixs] 

Like diminutive nouns, these forms consist of one prosodie word. 

In compounds the deletion of/t/ is optional, although for many speakers it 

will he obligatory in frequent words such as postkantoor 'post office' (the 

deletable /t/ is in italic): 

(25) vrachr-wagen 'truck' 

markf-plein 'market square' 

herfsf-kleuren 'autumn colours' 

posr-bank '(lit.) Post Office Bank' 

pacf-sluiting 'pact agreement' 

/ichf-baar 'visible' 

ab/-loos 'abbotless' 

on-ach/-zaam 'negligent' 

If the preceding consonant is a sonorant, /t/-deletion is also possible, but then 

the following consonant must be an obstruent; moreover, native speakers may 

differ as to whether /t/-deletion is possible in all cases, and lexicalisation and 

frequency play a role: 

(26) Compounds 

bandbreedte 'band width' 

tandsteen 'tartar' 

tandpasta 'toothpaste' 

Prefixed verbs 

ont-ploffen 'to explode' 

ont-staan 'to arise' 

[bumbreto] 

[tunsten] 

[tumpusta] 

[Dmpbfanl 

[onstan] 

When the obstruent following the sonorant + A/ cluster is /k/, deletion does not 

apply, as in honlkftia^ 'fur collar'. When A/ is preceded by a sonorant, and also 

followed by a sonorant, deletion is impossible: 

(27) kant-lijn 'margin' *[konlcin| 

ont-lopen 'to avoid' *[onlopgn] 

A/-deletion is typically a variable rule in the Labovian sense (Labov 1972) in 

that certain factors further the actual application of the rule. In the case of 

deletion of A/ it was also found that, for English, the less sonorous the 

following segment is, the easier deletion can apply (Guy 1980, Neu 1980). 

The fact that the domain of application also plays a role has already been 

mentioned: within prosodie words deletion is obligatory, which accounts for 

diminutives and for superlative forms. In prefixed words and compounds, A/ is 

separated from the following consonant by a prosodie word boundary, hence 

deletion can more easily be suppressed there. Across phrasal boundaries, A/- 

deletion is possible if both the preceding and the following consonants are 

obstruents, less probable after or before a nasal consonant, and even less 

probable before liquids and glides:


(28) Wint Piet? 'Does Peter win?' [umpit] 

Komt Piet? 'Does Peter come?' (kompit) 

Zakt Piet? 'Does Peter fail?' Izakpit) 

Zakt Marie? 'Does Mary fail?' [zukmari] 

Zakt Ria? 'Does Ria fail' 

*[/ukrija] 

Zakt Jan? 'Does John fail?' *[zukjun) 

Thus, we formulate /t/-deletion as (29). 

(29) /t/-deletion 

Delete [— cont, Coronal) in: ([+cons] —)CODA |+cons| 

Domain: IP 

Condition: at least one of the (+cons) segments must be (—son) 

The condition rules out deletion of /t/ in words like kantlijn 'margin' and 

phrases like Komt Jan'? 'Does John come?' 7 The rule does not distinguish 

between /t/ and /d/, but this is not necessary since /d/ does not occur in the 

relevant context, the coda position. The requirement of a consonant following 

the /t/ implies that, unlike in English, the A/ does not delete at the end of an IP. 8 

As pointed out in Section 6.6, Browman and Goldstein (1990) conceive of 

fast speech rules as retiming of the articulatory gestures, with masking effects, 

rather than interpreting such phenomena in terms of the actual deletion of 

segments. They adduce some evidence that in the case of English /t/-deletion, 

the A/ is indeed masked in perception but not completely absent in terms of 

production. The concept of retiming may also explain why deletion applies 

more readily in smaller domains: since a prosodie word is a smaller domain 

than, for instance, a compound or a phrase in terms of the time dimension, 

reduction of time will have comparatively stronger effects in a prosodie word 

than in a compound or phrase. Nevertheless, in words with obligatory A/- 

deletion, the A/ is completely absent, since we do not perceive any phonetic 

difference between kantje 'cupboard' (dimin.) and kasje 'greenhouse' (dimin.) 

both pronounced as (kujjaj. 

7.3. SENTENCE ACCENT 

As pointed out in Section 5.1, the main stressed syllable of a word is a 

potential locus of sentence accent. The stressed syllable is the 'anchor point' 

7 The 2 sg. pres. form of Dutch verbs ends in /I/, hut is /t/-less it the subject pronoun je follows 

the verb: je komt 'you come', but Köm je'.' 'Do you come?'. The absence of /t/ in the case of the 

inverted word order is clearly not a phonological phenomenon, but a case of syntactically 

conditioned allomorphy, given the constraints on /I/ deletion discussed here. 

K 

However, word-final deletion ol A/ does occur in some non-standard dialects such as thai of' 

Utrecht and Leiden (cf. De Vries et al. 1974). In standard Dutch we find deletion of/t/ in word-final 

position after a vowel in function words like met /nit/ 'nol' and wat /uut/ 'what' which can be 

pronounced as [nil and [uu] respectively in informal language use.

7-3- SENTENCE ACCENT 155 

of the pitch movements that give prominence to certain constituents of a 

sentence. 

The location of sentence accents is determined by semantic and pragmatic 

factors (Selkirk 1984/>, Gussenhoven 1984). For instance, the following sentence 

can have two locations for its sentence accent, depending on which 

background knowledge is assumed: 

(30) a. * 

(Waar is Jan?) Jan is in de kamer. 

'Where is John? John is in the lounge.' 

h. * 

(Wie is in de kamer?) Jan is in de kamer. 

'Who is in the lounge? John is in the lounge.' 

In (30a) the prepositional phrase in de kamer is a constituent that is meant as 

new information, i.e. it carries focus. The non-focused pan is what the speaker 

takes as his starting point. In (30b) the constituent Jan is the constituent with 

focus, and thus it gets a sentence accent. In other words, the basic rule is that 

each focused constituent bears a sentence accent. It is also possible to use 

'narrow focus' accent with a contrastive function as in: 

(31) * * 

Jan is in de kamer, niet in de keuken. 

'John is in the lounge, not in the kitchen.' 

Note that syntactic structure plays a role in that in the prepositional phrase in 

de kamer it is the noun, that is, a member of one of the major lexical 

categories, that bears the sentence accent, not one of the function words in 

or de. Sentence accent on a function word is also possible, if the sentence 

accent has a contrastive function: 

(32) * * 

Het boek ligt niet in, maar onder de kast. 

'The book is not in, but under, the bookcase' 

Here, only the prepositions are [+focus] elements, whereas de kast functions as 

background knowledge. This example also serves to illustrate that function 

words must be provided with a main stress in word phonology for determining 

the potential locus of sentence accent, for example, on the first syllable of the 

preposition onder. 

The relation between focus and sentence accent in Dutch has been investigated 

by Gussenhoven (1984, 1992«). In particular, he showed that a predicate 

may not receive a sentence accent even if it is |+focus]. Gussenhoven distinguishes 

between three types of constituents in a clause: each clause contains a 

predicate (P), zero or more arguments (A), and any number of modifiers (M), 

constituents with an adverbial function. Arguments are noun phrases or prepositional 

phrases for which a predicate is subcategorized in its lexical entry. 

Gussenhoven proposed the following rule for sentence accents:


(33) Sentence Accent Assignment Rule (SAAR) 

If focused, every predicate, argument, and modifier must be accented, with the 

exception of a predicate that, discounting unfocused constituents, is adjacent to an 

argument. (Gussenhoven 1992«: 84) 

This rule can be formulated as an algorithm. First, we determine which parts of 

a clause are [+focus| (indicated by underlining), then we determine the 

domains of sentence accent assignment, and finally we accent the relevant 

constituents (Gussenhoven 1984: 69): 

(34) SAAR Algorithm 

a. Domain assignment: P (X) A -> [P (X) A| 

(ordered) A (X) P -> [A (X) P) 

Y -» [Y] 

b. Accent assignment | ] -» [*] 

In [AP/PA) accent A 

The working of SAAR is illustrated by the following examples: 9 

(35) a. (Has anyone resigned?) 

* 

AP -> [A]P * 

De voor/.itter is afgetreden. 

'The chairman has resigned.' 

h. (What's new?) 

* 

AP -» [AP| * 

De voorzitter is afgetreden. 

r * * * * * * 

AMP —> [A](M||P] 

De voorzitter is inderhaast afgetreden. 

'The chairman has hurriedly resigned.' 

d. (Why can't the chairman do that?) 

* * 

AP -> A [P] 

De voorzitter is afgetreden. 

The examples (35«) and (35/>) have the same sentence accent, although they 

differ with respect to focus. This reflects the ambiguity of this sentence with 

respect to the size of the focused constituent, either part of the sentence, or the 

whole sentence. In other words, if possible, predicate and argument form one 

domain of sentence accent assignment, and it is the argument that then receives 

the sentence accent. In (35c) the focused predicate and the focused argument 

are separated by a focused modifier, and therefore they form separate domains 

of accent assignment, resulting in a sentence with three sentence accents. 

9 Taken from Gussenhoven (1984: 70).


The ordering of the creation of [ AP] domains before [PA] domains accounts 

for the fact that in a sentence like 

(36) * * 

Jan slaat zijn vrouw. 

'John beats his wife.' 

there is a prosodie break after Jan, i.e., it is slaat zijn vrouw that forms a 

domain rather than Jan slaat. 

As pointed out by Gussenhoven (1984: 27), the SAAR also explains the 

difference in accentuation between the following two sentences discussed by 

Schmerling (1976): 

(37) * 

(Have you heard?) Johnson died. 

(38) * 

(Have you heard?) Truman died. 

In the second sentence, it is presupposed that the hearer already knows that 

Truman is very ill. Hence, it is only the predicate that is [+focus|, and therefore 

the sentence accent is located on the verb. This observation also holds for 

equivalent Dutch sentences. 

The sentence accent in the constituent ix afgetreden in sentence (35d) is 

located on af. The location on afgetreden is to be expected since i.v is a 

function word. Afgetreden is the participle of the separable complex verb 

aftreden 'to resign'. These verbs are complex predicates in that they form a 

semantic unit, but yet they are phrases from the syntactic point of view (Booij 

1990«): the two parts can be separated in main clauses, as in: 

(39) Jan belt zijn moeder op. 

'John phones his mother.' 

The sentence accent assignment in the following examples shows that if the 

two parts of the complex predicate are separated, they still form one domain of 

sentence accent assigment (the underlined part is |+focus|): 

(40) (What's Jan doing?) 

* 

a. Jan belt zijn moeder op. 

'Jan phones his mother.' 

* 

h. Jan belt op. 

'John makes a phone call.' 

The location of the sentence accent in (4()b) shows that when a complex 

predicate is accented it is the non-verbal part that receives the accent. 

There are a number of complications in the assignment of sentence accent. 

First, a pronoun with [+focus| forms a domain of its own, not with the


following predicate, as is illustrated by the following examples (Gussenhoven 

1984: 73): 

(41) (What's the matter?) 

* 

a. De gevangenen zijn ontsnapt! 

'The prisoners have escaped.' 

* * 

h. Iedereen is ontsnapt! 

'Everyone has escaped.' 

In (41 h) the argument iedereen does not form a domain together with the 

following predicate. Therefore, the predicate forms a domain of its own, and 

receives its own sentence accent: sentence (4 \h) without sentence accent on 

the predicate is ill-formed. 

Second, SAAR is meant as a generalization for so called eventive sentences. 

Eventive sentences present the proposition 'as a historical development of 

some sort, while a non-eventive sentence could serve as a description of the 

status-quo ('definitional sentence'), or convey information which is potentially 

relevant to the listener ('contingency sentence')' (Gussenhoven 1992a: 103). 

The distinction is nicely illustrated by the two interpretations of the sentence 

De direktie geeft dieven aan 'The management turns thieves in'. It may be a 

general statement on a sign in a shop, as in (42a), or the description of an 

event, as in (42h). In the first case, the sentence is non-eventive, and therefore 

each focused constituent receives its own sentence accent. That is, skipping the 

predicate in the assignment of sentence accent as expressed by the SAAR is 

characteristic of eventive sentences (geeft aan is a discontinuous complex 

predicate here): 

(42) a. * * * 

De direktie geeft dieven aan. 

b. * * 

De direktie geeft dieven aan. 

In addition to SAAR, sentence accent is also assigned by Topicalization. A 

constituent that is preposed by Topicalization to the first position in the 

sentence receives a sentence accent even if it is not [+focus|: 

(43) (What happened to Yvonne?) 

* * 

Yvonne heeft hij gearresteerd. 

Yvonne has he arrested. 

'He arrested Yvonne.' 

In this example, Yvonne is background knowledge, and hence it is [-focus). 

Although the argument Yvonne and the predicate are only separated by a 

|-focus) constituent, they cannot form one domain: the Topic constituent


forms a sentence accent domain of its own, and hence gearresteerd also 

receives sentence accent. 

This effect of Topicalization also explains why the weak forms of pronouns 

are excluded from being topicalized: they cannot bear stress, and hence they 

are also unable to receive sentence accent. In other words, it is not necessary to 

constrain syntactic rules of topicalization in such a way that they do not affect 

weak pronouns: this follows from their prosodie properties.'" 

Attributively used adjectives do not count as constituents for the SAAR. 

Yet, they may function as focus constituents independently from the following 

noun, and then they are accented. When the adjective is preceded by an adverb, 

however, the sentence accent is on the adverb: 

(44) (What did you see?) 

* 

Ik zag een leuk meisje. 

'l saw a nice girl.' 

(45) (What do you think of her?) 

* 

Ik vind haar een leuk meisje. 

'I consider her a nice girl.' 

(46) (What did you see?) 

* 

Ik zag een erg leuk meisje. 

'I saw a very nice girl.' 

In (45) the use of haar 'her' already presupposes the existence of a specific 

female being in the background knowledge of the hearer, and hence it is 

only leuk that is [+focusj. Adjectives also form focus domains when used 

contrastively, as in 

(47) * 

Mijn oudste oom is overleden. 

'My eldest uncle has died.' 

This survey of factors determining the location of sentence accents is by no 

means exhaustive, but presents the basic generalizations; cf. Gussenhoven 

(1984, \992a) for more detailed studies. 

The next issue to be discussed is the actual realization of sentence accents as 

pitch movements. It has been found for English that a sentence accent can be 

'" In the case of weak pronouns, Ihere is an alternative strong form that can be used in topic 

position, with the phonological properties required for being a prosodie word. The same applies to 

tlie determiner her. if it is to carry sentence accent (e.g. in order to express the uniqueness of 

referent of the noun phrase), we cannot use the form |.il|. only the form |htt|, which forms a 

prosodie word of its own. The only exception is

i6o 

CONNECTED SPEECH II: SENTENCE PHONOLOGY 

realized as a fall, a rise, or a fall-rise. These patterns can be interpreted as tonal 

morphemes consisting of contour tones (Gussenhoven 1984, 1988): 

(48) H*L falling tone 

L*H rising tone 

H*LH falling-rising tone 

The asterisk following a tone indicates that this tone is to be linked to the 

syllable with the sentence accent. The following tone or tones are spread 

across the rest of the association domain. Furthermore, there is a copy of the 

last tone of the contour that functions as a boundary tone (indicated by H% or 

L%) at the end of an intonational phrase. The resulting tone patterns are 

illustrated here for sentence (49). 

(49) Leeuwarden wil meer mannen. 

'Leeuwarden needs more men.' 

with both one (Fig. 7.1) and two (Fig. 7.2) sentence accents, and with either 

H*L or L*H as the first sentence accent (from Van den Berg et al. 1992): 

The part of a phonological phrase before the first sentence accent may be 

called the onset, and usually has low pitch. The prosodie boundary is also 

indicated. As Van den Berg, Gussenhoven, and Rietveld (1992) point out. 

(a) 500 

400 

300 

X 200 

o 

100 

L% 

e- w a r d s wilm e-r m un 3 

50 

( 0.0 0.3 0.6 

Time (sec.) 

0.9 1.2 

(b) 500 

400 

300 

X 200 

100 

e- w a r d 3 w i l m e' r m u n 3 

50 

0.0 0.3 0.6 

0.9 1.2 

Time (sec.) 

FIG. 7.1. Contours (H*L L%) AD and (L*H H%) AD on the sentence Leeu*warden wil 

meer mannen

7-4- RHYTHMIC RULES 161 

(a) 500 

400 

300 

K 200 

100 

(b) 500 

400 

300 

£200 

50 

0.0 

I e- w a r d a w i l m e-rm 

0.3 0.6 

Time (sec.) 

«-H 

0.9 1.2 

KM) 

50 

0.0 

w a r d 3 w i l m e-rtn 

0.3 0.6 

Time (sec.) 

0.9 1.2 

FIG. 7.2. Contours (H*L) AD , (H*L%) AD> and (L*H) AD (H*L L%) AD on the sentence 

Leeu*warden wil meer ma*nnen 

there is also a phonetic realization possible in which the two phonological 

phrases in (49) do not each form an association domain (AD), but form only 

one such domain. The effect is that the tone following the starred tone is 

moved to the rest of a string before the next sentence accent (i.e., there is no 

second onset with its own L tone) (Fig. 7.3.) 

7.4. RHYTHMIC RULES 

In Section 5.4. we encountered two types of stress shift, Trochaic Reversal (in 

compounds), and Stress Retraction (cf. Gussenhoven 1983«, 19836; 1984). 

Stress Retraction is a postlexical rule that changes the stress patterns of 

attributively used adjectives and of prepositions, as in: 

(50) Adjectival compounds of the weak-strong type 

[dóod][zi'ek] een doodziek kind 'a critically ill child' 

[straat|[arm] een straatarm mens 'a very poor person'

162 

(a) 5(X) 

400 

300 

CONNECTED SPEECH II: SENTENCE PHONOLOGY 

N 200 

100 

50 

(b) 500 

400 

300 

N 200 

a 

w u r da l m e' r 

0.3 0.6 

Time (sec.) 

L% 

0.9 1.2 

100 

50 

war daw i n a 

0.0 0.3 0.6 

Time (sec.) 

0.9 1.2 

FIG. 7.3. Contours (H*L H*L L%) AD , and L*H H*L L%) AD on the sentence 

Leeu*warden wil meer ma*nnen 

[zélf] [rijzend] 

[nàuw][ge/.ét] 

zèlfrijzend bäkmeel 'self-raising flour' 

een nauwgezet femand 'a scrupulous person' 

(51) Prepositional compounds 

onderaan onderaan de bérg 'at the foot of the mountain' 

bovenop bovenop de schuur 'on top of the shed' 

achterin achterin de tuin 'at the back of the garden' 

Stress shift also occurs in the following cases, referred to as cases of Iambic 

Reversal (Gussenhoven 1984, Kager and Visch, 1988, Visch 1989) in order to 

distinguish it from Stress Retraction. Stress Retraction seems to be obligatory, 

while Iambic Reversal is optional, and its use is also determined by pragmatic 

factors, that is, it can be used as a rhetorical device in speeches in the case of 

attributively used adjectives (Gussenhoven 1983a, 1984): 

(52) Attributively used adjectives 

speciaal spéciaal geval 'special case' 

feodaal feodaal stelsel 'feudal system' 

respectabel respectabel man 'respectable man'

74- RHYTHMIC RULES 163 

(53) Nouns before appositions 

admiraal admiraal de Riiyter 'Admiral de Ruyter' 

kardinaal kardinaal Simonis 'Cardinal Simonis' 

Rotterdam Rötterdam-Züid 'Rotterdam-South' 

These examples show that stress clashes in the domain of the phonological 

phrase can be resolved by shifting the non-primary stress to the left. The 

relevant constituents can indeed be qualified as s if single attributively 

used adjectives (i.e. non-branching APs) are assumed not to form a of their 

own (Nespor and Vogel 1986). Note that it is thus correctly predicted that 

stress shift does not occur when the attributively used adjective is preceded by 

an adverb, and thus is the head of a branching AP: 

(54) [een [zeer speciaal] AP geval) NP 'a very special case' 

een zeer speciaal geval, *een zeer spéciaal geval 

Leftward stress shift is a case of Move * in the domain of , as illustrated here 

for the noun phrase (een) doodziek kind 'a critically ill child': 

(55) ( (dood) w (ziek) u (kind),,)«,, 

* * line 1 

* * * line 2 


(57) ( ( (pe) 0 (ni) a (b3l) a ) M ( (mo) 0 (ment) 0) J „, 

* * line 1 

* * line 2 

* line 3 

( ( (spe) a (si) 0 (al)

8 

CONNECTED SPEECH III: CLITICIZATION 


Clitics are function words such as pronouns, determiners, auxiliaries, particles, 

conjunctions, and prepositions which are phonologically dependent on a 

host word to which they attach, and with which they form a prosodie 

constituent. In addition, they may also have special syntactic distributional 

properties. For instance, the Dutch singular definite article het [het| has a 

weak form which is a clitic: [at|. Its syntactic distribution is determined by 

the rules for the construction of Dutch noun phrases, but phonologically it 

may be dependent on the preceding word in the sentence, its host. This is 

clear from the fact that it can form one domain of syllabification with the 

preceding word, as illustrated here for the sentence Jan kocht het boek 'John 

bought the book': 

(1) syntactic structure: |Jan] N p [kochtJv I't boek] NP 

prosodie structure: (J un )u> ( (' < ox) a (t3t) a ) lo (buk)^ 

The prosodie structure is non-isomorphic to the syntactic structure: the 

determiner (otj depends syntactically on the following noun, and prosodically 

on the preceding verb.' The non-independence of [at] follows from its 

phonological form, since it cannot form a prosodie word of its own. After its 

obligatory incorporation into the preceding prosodie word, resyllabification 

takes place according to the universal CV-rule (cf. Section 3.4.2). This 

process of resyllabification also blocks /n/-deletion when the host word ends 

in /an/: 

(2) Zij kochten 't boek 

(zei) u ( (kDx)„(t3) a (n3t) 0 ) (U (buk)«, 

'They bought the book' 

Due to resyllabification the post-schwa /n/ no longer occurs in a coda, but in an 

onset, and hence it is saved from deletion. 

If their structural descriptions are met, other P-ruIes that apply obligatorily 

within the prosodie word, also apply obligatorily in the host + clitic domain, in 

particular Homorganic Glide Insertion and Prevocalic Schwa Deletion: 

1 

Cf. Klavans (1985) and Nevis (19SX) lor similar observations as lo the non-isomorphism of 

prosodie and syntactic structure in the ease of clitics. Sadock (1991) also argues that two structures 

are necessary in case of clitici/.alion, but he does not qualify one of them as prosodie.

l66 

CONNECTED SPEECH Hi: CLITICIZATION 

(3) Homorganic Glide Insertion (HGI) 

Ik zie 't boek 

(ik) M (sijaOu (buk)« 

'I saw the book' 

Ik doe 't werk 

(ig) M (duu3t) œ (uerk) œ 

'I do the chores' 

(4) Prevocalic Schwa Deletion 

Ik merkte 't direct 

(ik) M (merktat),,, (direkt)^ 

'I noticed it immediately' 

It is also possible to insert an /n/ between the schwa and the next vowel: 

(5) (tk)« (merktanaOo, (direkt) w 

In all cases, the effect is that after incorporation of the clitic, the prosodie word 

fulfils all conditions on well-formed prosodie words. 

These examples also illustrate that cliticization processes form part of the 

connected-speech phenomena. 

8.2. PRONOMINAL AND ADVERBIAL CLITICS 

The Dutch personal pronouns do not have the same distributional properties as 

full noun phrases (although they are syntactically equivalent to full noun 

phrases in that they denote arguments of the verb, and occur in prepositional 

and adverbial phrases), unless they are stressed, that is, bear a sentence accent, 

and have full ('strong') phonological forms. When they are unstressed and/or 

have a reduced ('weak') form, they are to be qualified as clitics. A survey of 

the Dutch system of personal pronouns is given in Table 8.1. The pronouns hen 

and hun in the last row differ in that the first is used as accusative and the 

second as dative. 2 

The full forms can be used in stressed and unstressed positions, with the 

exception of het. When they are stressed they have the same syntactic distribution 

as lexical noun phrases. The third-person pronouns are not only used 

to refer to animate beings, but also for referring to inanimate entities. 

Er is the (suppletive) allomorph for het in prepositional phrases, and 

always occurs before the preposition; 3 it is also the adverbial pronoun for 

'there': 

2 

In substandard Dutch the pronoun hun is also used as Ihe 3 pi. subject pronoun. 

1 The pronoun er and the following preposition are usually written as one word. Cf. Van 

Riemsdijk (I97X) lor a syntactic analysis of er.

8.2. PRONOMINAL AND ADVERBIAL CLITICS 

TABLK 8.1. The Dutch personal pronouns 

1 sg. subj. 

1 sg. obj. 

2 sg. subj. 

2 sg. obj. 

3 sg. subj. masc. 

3 sg. obj. masc. 

3 sg. subj. fern. 

3 sg. obj. fern. 

3 sg. neut. 

1 pl. subj. 

1 pl. obj. 

2pl. 

3 pl. subj. 

3 pl. obj. 

Strong 

ik /ik/ 

mij /mei/ 

JU /jei/ 

jou /pu/ 

hij /hei/ 

hem /hem/ 

zij /zei/ 

haar /har/ 

het /het/ 

wij /uei/ 

ons /nns/ 

jullie /jvli/ 

zij /zei/ 

hen /hen/, hun /hvn/ 

Weak 

ik /ik/, /ak/, 'k /k/ 

me /ma/ 

je /ja/ 

je /ja/ 

ie /i/ 

'm /am/ 

ze /za/ 

'r /ar/, d'r /dar/ 

't /at/, M 

we /ua/ 

ze /za/ 

ze /za/ 

(6) Het ligt er. 

it lies there 

'It's lying there.' 

Ik lig [er op] PP /*op het 

I lie it on 

'I'm lying on it.' 

Er has the full forms /er/ and the weak form /ar/. In contrast to the personal 

pronouns, the pronouns het and er cannot be used in stressed positions, not 

even in their full forms. Instead, the corresponding demonstrative pronouns dat 

/dut/ and daar /dar/ have to be used. The weak form of daar, /dar/, also 

functions as a stylistic variant of er. 

The weak forms of these pronouns cannot be derived from the corresponding 

full forms by means of a productive phonological reduction rule. The 

general rule of Vowel Reduction (Section 6.4.) does not apply to diphthongs. 

Yet, the full forms of mij, jij, :ij, and wij have corresponding weak forms 

with schwa. Also, a vowel cannot reduce after /h/ or in word-initial position. 

Nevertheless, the reduced form of hem is [am], and ik has [ak] as one of its 

weak forms. Consequently, we have to list the weak forms as such in the 

lexicon. This means that they are 'special clitics' (cf. Zwicky 1977) in that 

they have opaque phonology. In addition, they also have special syntactic 

properties, as we see below, which means that they are also special clitics for 

syntactic reasons. 

The conclusion that the weak forms have to be listed as lexical items is 

corroborated by the observations made by Berendsen (1986) and Zwart (1992) 

that in some contexts only the weak forms can be used. For instance, the weak

168 CONNECTED SPEECH III: CLITICIZATION 

forms of the second-person singular and third-person plural pronouns, ye and ze 

may have a generic interpretation, unlike the corresponding full forms: 

(7) Ze/*zij zeggen zoveel. 

'People say a lot.' 

Je/*jij moet eerlijk zijn 

'One has to be honest.' 

Furthermore, the strong third-person plural pronouns zij, hen, and Hun can only 

be used to refer to animate entities or abstract entities like institutions, but not 

to other kinds of concrete entity. Idiomatic expressions often only allow for the 

use of the weak forms, as in: 

(8) Ben je/*jij gek? 

'Are you crazy?' 

Smeer 3m/*hem! 

'Clear off!' 

Inversely, certain contexts require the strong form to be used, for instance after 

the preposition volgens 'according to', which requires prominence of its 

complement-noun phrase: 

(9) volgens mij/*me 'according to me' 

The weak form je is the only permissible second-person singular pronoun in 

pseudo-reflexive verbs such as zich vergissen 'to make a mistake' and zieh 

schamen 'to be ashamed': 

(10) Je vergist je/*jou. 

'You are making a mistake.' 

Schaam je / *jou! 

'Shame upon you!' 

Pronominal clitics also exhibit syntactic properties that distinguish them 

from lexical NPs (Model 1991: ch. 10, Zwart 1992). For instance, adverbs such 

as gisteren 'yesterday' cannot stand before clitics, as illustrated in (11). Dutch 

clitics also exhibit 'clitic climbing', see (12): 

(11) dat gisteren Jan/*ie het boek gelezen heeft 

that yesterday John/he the book read has 

'that yesterday, John/he read the book' 

(12) *dat ik de afwas Marie heb zien doen 

that I the dishes Mary have see do 

'that I saw Mary doing the dishes' 

dat ik 't Marie heb zien doen 

that I it Mary have see do 

'that I saw Mary doing it' 

Furthermore, clitics cannot be conjoined:

8.2. PRONOMINAL AND ADVERBIAL CLITICS 169 

TABLE 8.2. Dutch singular possessive pronouns 

f:' 

•' 

Strong form 

mijn /mein/ 

JOUW /J3U/ 

zijn /zein/ 

haar /har/ 

Weak form 

m'n /man/ 

je /ja/ 

z'n /zan/ 

'r /ar/, d'r /dar/ 

(13) Ik zag hém en haar 

*Ik zag am en ar 

'I saw him and her' 

Dutch singular possessive pronouns also have weak forms that only occur in 

unstressed position (Table 8.2). The possessive pronoun for 'its' only has the 

weak form /zan/. Again, the weak forms cannot be derived by rule, and have to 

be lexically listed. As was the case for personal pronominal clitics, the clitic 

forms can be used only in certain idiomatic expressions: 

(14) in je eentje/*in jouw eentje 

'on your own' 

op z'n best/* op zijn best 

'at best' 

The adverbs er 'there', daar 'there', and eens 'once' also have clitic forms 

(Table 8.3). The use of the weak form is again lexicalized, as in *Eensles even 

kijken 'Let's see.' 

TABLE 8.3. Dutch adverbial clitics 

er 'there' 

daar 'there' 

eens 'once' 

Strong form 

er /er/ 

daar /dar/ 

eens /ens/ 

Weak form 

er /ar/, d'r /dar/ 

d'r /dar/ 

es /as/ 

8.2.1. Prosodie integration 

The question now arises how clitics are incorporated into prosodie structure. 

Except for ie l\l and ik /ik/ all weak pronouns and adverbs have schwa as their 

only vowel (if any: there are also vowelless clitics, /k/ and A/). This implies 

that they do not form prosodie words of their own (Section 3.6), but at most 

syllables, because a prosodie word requires at least one syllable with a full 

vowel. The fact that they are not prosodie words also explains why they cannot

170 CONNECTED SPEECH III: CLII ICI/. ATION 

be stressed, since the domain of word-stress rules is the prosodie word. 

Moreover, syllables with schwa never bear word stress. 4 

Can we assume then that clitics form directly part of phonological phrases, 

without being dominated by a prosodie word node? This would be in conflict 

with the Strict Layer Hypothesis (Selkirk 1984/>, 1986, Nespor and Vogel 

1986) which says that 

The prosodie categories are ordered in a hierarchy and in phonological representations 

they are strictly organized into layers according to that hierarchy i.e. prosodie 

constituents of the same category are not nested. (Selkirk 1986: 384) 

This hypothesis excludes clitics, which are only syllables, from being directly 

dominated by a phonological phrase node. This means that clitics must either 

be Chomsky-adjoined to an adjacent prosodie word, as proposed in Neijt 

(1985), or incorporated into that prosodie word, that is, Chomsky-adjoined 

to a foot. The difference between the two solutions can be represented as 

follows: 

(15) u>- adjunction w 

o w 

I 

Clitic 

F-adjunction 

GO 

I 

. F 

Logically, the mirror images of these adjunctions are also possible. However, 

it appears that in Dutch, proclitics are Chomsky-adjoined to prosodie 

words, and enclitics are Chomsky-adjoined to feet. Proclitics cannot adjoin to 

the following foot since a Dutch foot is a trochee, and hence cannot begin with 

a syllable with schwa, because it cannot function as the head of a foot. 

Enclitics, on the other hand, can adjoin to the preceding foot without violating 

the canonical structure of the left-headed feet of Dutch. Thus they exhibit the 

maximal degree of prosodie integration. In sum, proclitics have the same 

The fact that Dutch clitics do not form prosodie words also excludes an analysis analogous to 

that proposed by Nespor and Vogel for Italian clitics in which clitics are prosodie words that are 

dominated, together with their host word, by a node of the prosodie category Clitic Group (cf. 

Booij I988f).

8.2. PRONOMINAL AND ADVERBIAL CLITICS 17! 

prosodie status as schwa-containing prefixes, and enclitics have the same 

prosodie status as suffixes. 

Chomsky-adjunction still violates the Strict Layer Hypothesis in that a 

prosodie node is dominated by another one of the same category, but this is 

a very restricted type of violation. In the case of adjunction of a syllable to a 

prosodie word, we also skip one prosodie level. 5 

In principle, clitics can function both as enclitics and as proclitics. The only 

exceptions are the clitic ie and its allomorph die, and the pronominal clitic der 

/dor/ 'her' which are obligatory enclitics: that is, they require a host word on 

their left. Moreover, function words that contain a full vowel do not have to 

cliticize at all, since they can form prosodie words of their own. A clitic can be 

only a proclitic in sentence-initial position, and only an enclitic in sentencefinal 

position. 

In the case of schwa-initial clitics, there is clear evidence that these clitics 

can be incorporated into the preceding prosodie word, as illustrated here for 

the clitic /at/ 'it' in sentence-final position: 

(16) (Jan) ziet 't 

prosodie structure: ( (zi) a (t9t) a ) œ 

'John sees it.' 

The basic observation here is that the clitic /at/ induces obligatory resyllabih'cation 

of the preceding word (the coda consonant /t/ of ziet becomes onset of 

the next syllable), because a syllable cannot begin with a schwa. Since the 

prosodie word is the domain of syllabification, the incorporation of the clitic 

into the preceding prosodie word predicts this obligatory resyllabitication. 

Another rule that applies in host + clitic combinations is /n/-insertion: after 

schwa, In/ can be inserted to resolve the hiatus that arises when the clitic 

begins with a vowel (note that hiatus is resolved obligatorily in prosodie words 

only): 

(17) ik zette-n-et'I put it' [iksetanat] 

(hij) wilde-n-et 'he wanted it' 

(uildanat] 

wilde-n-ie 'wanted he' 

(uildani) 

wilde-n-ik 'wanted I' 

[uildanik] 

(het) gekke-n-is 'the strange thing is' [vekanis] 

The fact that /n/-insertion also applies before vowel-initial clitics with a full 

vowel (is* ie, ik) suggests that they can also be incorporated into the preceding 

prosodie word. Note that /n/-insertion does not apply between a clitic and a 

lin Mild Mesler (I'W2) have Mrgucd on the basis of Japanese thai skipping of one level should 

he permitted. So they allow for untooted syllables thai are dominated directly by the prosodie word 

node. In such a theory, cohering suffixes and enclitics can he incorporated into the preceding 

piosodic word without being Chomsky-adjoined to a foot. However, this approach does not explain 

why there is a difference in the degree of prosodie integration between proclitics/prefixes and 

enclilics/sutlixcv 

6 As will be shown below, auxiliaries like is also behave as clitics.


following word, as in je eet lekker 'you eat nicely' (*[janet]). Instead, the 

vowel-initial word after the clitic begins with a glottal stop: [ja ?etj. In this 

respect proclitics exhibit the same behaviour as unstressed prefixes with 

respect to Prevocalic Schwa Deletion (cf. Section 7.1). In other words, the 

domain of /n/-insertion is the prosodie word. This supports the hypothesis that 

proclitics do not form one prosodie word with a following word, but are rather 

Chomsky-adjoined to them. Unlike Prevocalic Schwa Deletion, /n/-insertion 

only applies postlexically. 

The resyllabification induced by the clitic bleeds /n/-deletion (cf. Section 

6.8), a rule that deletes /n/ after schwa. For instance, the sentence ze deden 't 'they 

did it' has the phonetic form [za dedanatj. It is impossible to first apply /n/- 

deletion, and then, after prosodie incorporation, Prevocalic Schwa Deletion, 

which would result in the wrong phonetic form [dedal]. This is an impossible 

pronunciation for the sequences (zij) deden 't '(they) did it'. These facts 

follow straightforwardly if the rules of connected speech such as /n/-deletion 

apply after the proper prosodie structure of a sentence has been created. The 

resyllabification induced by clitics puts the final /n/ in onset position, thus 

making it immune to /n/-deletion that only applies to /n/ in coda position. 

In the case of the clitic ie, it is also possible to avoid the hiatus within the 

prosodie word by choosing the allomorph die instead of ie. The same applies to 

the clitic er for which the variant d'r can be used. However the hiatus may be 

resolved, the fact that it must be resolved somehow shows that these vowelinitial 

function words indeed form a prosodie word with the preceding word, 

obligatorily when it is ie, optionally in the case of the schwa containing clitics, 

ik and is. When vowel-initial clitics are not encliticized, a glottal stop is 

inserted before the vowel. 

As pointed out above, when pronominal clitics occur in sentence-initial 

position, they can only cliticize to the following word. This also applies to 

clitics that consist of one consonant only (/k/, /t/) and clitics that begin with a 

consonant: 

(18) Er [ar] staat een paard in de gang. 

'There stands a horse in the corridor.' 

(19) 't gaat [at xat] goed/'t gaat [txat] goed. 

'It goes well.' 

(20) Ik zal [ik sol] komen/'k zal [ksal] komen. 

'I will come.' 

(21) We eten [ua etan] lekker. 

'We eat nicely.' 

In cases where the clitic is a proclitic, it is not incorporated into the following 

prosodie word, as is proved by the fact that the phonological rules discussed 

above that apply within the domain of the prosodie word, do not apply to such 

clitic + host combinations. (Recall the definition of dominance given in section 

7.1!) For instance, Prevocalic Schwa Deletion does not apply obligatorily to


the schwa of we /V3/ in (21), since the sequence schwa-vowel is not dominated 

by the prosodie word node. 7 This follows from Chomsky-adjunction of the 

clitic to the following prosodie word. In the case of consonantal clitics this 

results in clusters such as /tx/ and /ks/ at the phonetic level, which we do not 

find at the lexical level. 

When pronominal clitics can in principle take either the preceding or the 

following word as their host, it is enclisis that takes precedence over proclisis 

in the case of vowelless clitics. 8 This is clear from minimal pairs such as zaik 

eten versus zal keten which can be perceptually distinguished (examples from 

Gussenhoven 1985): 

(22) (Dat) zal'k eten (zalk) a (e) 0 (t3n) CT 

'(lit.) that will I eat' i.e., 'that I will eat' 

(Jan) zal keten (zal) 0 (ke)„(t3n) 0 

'John will fool' 

Another indication that consonantal clitics take a host on their left is that 

they cannot be used when the preceding word ends in a consonant; in those 

cases, the variant with preceding vowel has to be used: 

(23) Ik pak *[t]/[3t] 'I fetch it.' 

Dat rol *[k|/|ik] 'That I roll.' 

This constraint shows that there is a co-occurrence restriction between clitics 

and the preceding phonological material, which suggests that they form a unit 

with that material. 

Clitics with a vowel (e.g., er, ik, je, ze, and we) can in principle also be 

integrated into the following prosodie structure, that is, their phonological 

structure permits them to acts as proclitics. Evidence that schwa-initial clitics 

may function as enclitics is presented above. On the other hand, as pointed out 

by Gussenhoven (1985), there is also evidence that they do not always form 

part of the preceding prosodie word. This evidence concerns the process of 

Fricative Voice Assimilation discussed in Section 7.2.1: a prosodie word-final 

fricative may be voiced after a sonorant if the following word begins with a 

vowel, for example, in the compound hui\z]arts 'GP' and the phrase lee[\] ik 

'(lit.) live I'. The rule can be stated as in (24). 

(24) Fricative Voice Assimilation 

[+son] [-son] )„, [+voc] 

[+cont] 

Laryngeal 

Laryngeal 

'However, in casual speech the word-final schwa of function words may be deleted as in we eten 

A>3 elan/ [uetan] 'we eat', and de avond /da avand/ (davont) 'the evening'. 

8 Gussenhoven (1985) also argued that Dutch prefers enclisis to proclisis.


By interpreting this process as the spreading of the class node Laryngeal rather 

than the feature [+voice] we correctly predict that no other segment with a 

laryngeal specification can be inserted between the fricative and the following 

vowel. Thus, we account for Gussenhoven's (1985) observation that glottal 

stops cannot be inserted before the vowel-initial word after application of 

voicing since a glottal stop is crucially specified on the Laryngeal tier. 

Gussenhoven (1985) concluded that clitics such as er do not necessarily 

belong to the preceding prosodie word, since Fricative Voice Assimilation may 

apply in relevant phrases. For instance, geef er 'give her' may be realized 

either as [yefor] or as [yevar]. In the first case, er is encliticized, in the second 

case it is not. 

Evidence for consonant-initial clitics concerning their (optionally) functioning 

as enclitics is provided by the fact that they are subject to contraction 

processes in which it is always a host word on the left with which they are 

contracted. For instance, the complementizer dat 'that' contracts with a 

following clitic, dropping its final A/: 

(25) dat 'k [dok] 

dat je 

l du j o I 

dat ze |dusa] 

dat we [duua] 

Nevertheless, this does not necessarily mean that these words also encliticize 

when they do not contract. As we saw above, and as we will also see in Section 

8.3, there is evidence that schwa-containing clitic words may be either 

enclitics or proclitics. In sum, we do not have to specify whether a Dutch 

pronominal clitic is an enclitic or a proclitic, except for ie, die, and der. That 

is, the latter are directional clitics. y Hence, ie does not have the same distribution 

as its strong counterpart hij, since the clitic cannot occur in sentenceinitial 

position where no left host word is available. Therefore, this clitic must 

be subcategorized for occurring with a prosodie word on its left (Booij and 

Lieber 1993). I0 It induces obligatory resyllabirication if the host ends in a 

consonant, which follows from the incorporation analysis proposed above: 

(26) dat-ie komt 'that he comes' (do) a (ti) CT (kDmt) 0 

komt-ie? '(lit.) comes he' i.e., 'does he come?' (kom) a (ti) 0 

8.2.2. Cliticization and Final Devnicing 

Since Final Devoicing of obstruents in codas is a lexical word-level rule, and 

prosodie integration with concomitant resyllabification takes place at the post- 

9 Nespor and Vogel (I98f>) also make a distinction between non-directional and directional 

clitics. 

'" Similar clitics with a prosodie subcategorization frame are found in Serbo-Croation, cf. Zee 

and Inkelas (1990).


syntactic level, it is predicted that word-final obstruents are voiceless even 

when they shift to onset positions. This indeed appears to be the case: 

(27) vond-ik /vond ik/ 'found I' phonetic form (von) 0 (tik) 0 

vond-ie /vond i/ 'found he' phonetic form (von) 0 (ti) 0 

vond et /vond at/ 'found it' phonetic form (vDn) a (tat) 0 

However, there appears to be variation here (Berendsen 1983, 1986; Booij 

19856) in that for certain frequent verbs the variants with voiced obstruents are 

also found, but only in cases where there is an underlying stem-final voiced 

obstruent. This suggests that certain verb+clitic combinations are lexically 

stored." Consequently, the resyllabification induced by the attachment of the 

clitic will bleed the word-level rule of Final Devoicing: 

(28) heb-ik 'have I' 

lexical clitic: /heb ik/ (he) 0 (bik) 0 

postlexical clitic: /hep ik/ (he) 0 (pik) 0 

Less frequent verbs with stem-final voiced obstruents such as verbind 'to 

connect', verwond 'to injure', and vermoord 'to murder' do not surface with 

a voiceless obstruent before the clitic: 

(29) verbind-ik [varbintik], *[varbmdik] 

verwond 'r [varuontar], "[varuondar] 

vermoord 'm [varmoirtam], *[varmo:rdam] 

This supports the lexicalization hypothesis for frequent verbs like hebben 'to 

have'. As we will see in Section 8.4, there are more cases of lexical listing of 

verb + clitic combinations in Dutch. 12 

A special property of ie is that it never occurs after a stem-final /d/. For 

instance, the only possible phonetic realization ofvond-ie 'found he' is [vDnti]; 

*[vondi] is impossible. This restriction is related to the fact that the allomorph 

/di/ always surfaces as [ti] after an obstruent: 13 

(30) keek-[i], keek-[ti], *keek-[di] 'watched he' 

gaf-fi], gaf-[ti], *gaf-[di] 'gave he' 

In this respect /di/ differs from other /d/-initial function words which can either 

voice a preceding voiceless obstruent or be devoiced themselves after voiceless 

obstruents, as discussed in the next section. The weak form of the pronoun 

haar, /dar/, is also obligatorily subject to devoicing after an obstruent. In other 

" A similar variation is found for combinations of 'to have' and clitics in German, cf. Wiese 

(1996). Note also that in Middle Dutch, which had no standard orthography, such frequent verb + 

clitic combinations were written as one word, e.g. hebbic 'have I' (cf. Van Kerckvoorde 1993: 

14-17 for a survey). 

12 

In Booij and Rubach (1987) evidence for the existence of lexical attachment of clitics is 

provided for Dutch and Polish. 

3 This observation was made in Zonneveld (1983) and Gussenhoven (1985). Berendsen (1983) 

accounts for these facts as follows: the clitic l\l is derived from an underlying form that begins with 

a dental fricative. A special rule converts the sequence alveolar stop + dental fricative into a [t]. 

This rule is ordered before the rule that derives [i] from this underlying form.

176 CONNECTED SPEECH III: CUTICIZATION 

words, /d/-initial personal pronominal clitics are obligatorily subject to incorporation, 

and hence devoice by a rule of devoicing presented in Section 8.3. 

8.3. CLITICIZATION OF OTHER FUNCTION WORDS 

Cliticization also applies to determiners, and to a number of /d/-initial function 

words. The three Dutch determiners are de, het, and een: 

(31) een/an/ (indef., sg.) 

het /het/, weak form /at/ (def., sg., n.) 

de /da/ 

(def., sg., masc., fern.; def., pi.) 

In the case of het, the strong form has to be used in stressed position. The other 

determiners have only one form, which is weak since the only vowel is schwa. 

If necessary, they can carry contrastive stress, as in Dit is dé winkel voor 

hoeken 'This is thé shop for books', Dat is één mogelijkheid 'That is a 

possibility.' 

As shown in Section 8.1, the weak forms of determiners can be encliticized 

when there is a potential host word on the left. Otherwise, they will be 

Chomsky-adjoined to the following prosodie word. 

A number of function words with initial /d/ including de have the special 

property that they have two possible phonetic realizations: the normal rule of 

Regressive Assimilation across prosodie word boundaries may apply, which 

means that /d/ causes voicing of the preceding word-final obstruent, or /d/ 

devoices itself. For instance, in the prepositional phrase op die manier 'in that 

way', the sequence op die can be pronounced as either [obdi], the expected 

form, or as [opti]. This pattern is found for the following words: 

(32) de /da/ 'the', dit /dit/ 'this', deze /deza/ 'these', dat /dat/ 'that', die /di/ 

'that, which', daar /dar/ 'there', d'r /dar/ 'there, her', dan /dan/ 'than, 

then' 

Zonneveld (1983) proposed to explain this pattern hy assuming that these 

words are optional enclitics. If they are incorporated into the preceding 

word they are devoiced. 

Encliticization, that is, the prosodie integration of function words into a left 

host, can be represented as in (33). 

A function word may form an co of its own, for instance the demonstrative 

(33) o> 

I 

F

8.3. CLITICIZATION OF OTHER FUNCTION WORDS 177 

die. When the syllable node that it dominates is adjoined to the last foot of the 

preceding prosodie word, the dominating prosodie nodes will be erased automatically, 

because a syllable cannot simultaneously belong to two feet or 

prosodie words. 

The consequence of these alternations is that in a sentence like Ik geef er 

niet om 'I do not care about it' the word sequence geef er may have four 

different phonetic realizations given the fact that the clitic /ar/ has an allomorph 

/dar/ which may be subject to devoicing (Gussenhoven 1985: 187): 

(34) geef er: [xefar], [xevar], [xeftar], [xevdar] 

The second form has a |v] derived by Fricative Voice Assimilation. Note that 

this rule does not apply to sequences such as gaf-ie 'gave he' with the 

obligatory enclitic ie, since there is no prosodie word boundary after the 

fricative. Thus, as pointed out in Gussenhoven (1985), [yavi] is impossible. 

The personal pronoun die, the allomorph of ie, has to be marked as being 

obligatorily encliticized, hence it always surfaces as [tij. Note that we 

observed above that the allomorph ie is also an obligatory enclitic. The 

determiner de always surfaces as [ta] after a word-final /t/, which means 

that it is obligatorily encliticized in that position: 

(35) (Ik) lees de (krant) 'I read the paper' [lesta], [lezda] 

(Hij) leest de (krant) 'He reads the paper' [lesta], *[lezda] 

When encliticization does not apply, the function words are prosodically 

integrated into the following phonological phrase. This analysis implies that, 

for clitics with more than one potential host word, encliticization is optional, 

and they may also procliticize to the following word. Except for de and d'r 

these /d/-initial function words contain a full vowel, and therefore do not 

necessarily require a host word. 

The rule that devoices the first consonant of these encliticized function 

words is given in (36). 

(36) Word-internal Devoicing 

[—son] [—son] 

[-voice] 

Domain: co 

| + voice ] 

Note that this rule does not apply to obstruents preceded by the clitics /k/ and 

A/ when used as proclitics before a voiced obstruent as in 'k ben [kbcn| 'I am' 

or 't begint [tbayint] 'it begins'. This supports the idea that such consonantal 

proclitics are not integrated into the following prosodie word, but are rather 

Chomsky-adjoined to it. 

Prepositions always contain at least one full vowel, except te Aa/ 'at'. 

Hence, they do not require a host word. Yet, it appears that they do encliticize

178 CONNECTED SPEECH 111: CLITICIZATION 

optionally, as is clear for vowel-initial prepositions like op 'on', in 'in', aan 

'to', and onder 'under'. The relevant phenomena are Prevocalic Schwa Deletion, 

/n/-insertion, and Fricative Voice Assimilation. For instance, the 

sequence fietste in in the sentence Jan fietste in Polen 'John cycled in 

Poland' has the following possible phonetic realizations: 

(37) (fietste-inJu, Prevocalic Schwa Deletion: [fitstin] 

/n/-insertion: 

[fitstamn] 

(fietste),,, (in),,, 

[fitsta ?m] 

Fricative Voice Assimilation can be observed in prepositional phrases with a 

postposition (the underlying form of bos ends in an /s/): 

(38) het bos in '(lit.) the wood into' i.e., 'into the wood' [adbozin] 

The vowel-initial auxiliary is is also optionally encliticized to the preceding 

word. If not, we get a glottal stop before the vowel: 

(39) Het gekke is 'The strange thing is' 

/n/-insertion: 

[atxckanis] 

Prevocalic Schwa Deletion: [atxckisj 

No encliticization: [atxeka ?is| 

The conjunction en 'and' has the weak form [an] in number terms below 100 

which are formed by co-ordination. For instance, in Mee-en-twintig '(lit.) two 

and twenty' ('twenty-two') we get this weak form of en. This schwa-initial 

weak form also appears to be incorporated into the preceding prosodie word, 

and this triggers Homorganic Glide Insertion: the phonetic form in this case is 

[tuejantumtox]. 

8.4. CONTRACTION 

Certain sequences of function words can be contracted: dat 'that' + personal 

pronoun, auxiliary/modal verb + pronoun, and certain auxiliaries preceded by a 

pronoun: 

(40) dat + pronoun 

dat 'k -» [dak l 

dat je -» [düja] 

dat ze -» [dasaj 

dat we —» [doua] 

The regularity involved here is that it is the final HI of dat that is deleted before 

the clitic. We also find even stronger forms of contraction like [duj] for datje 

and [duu| for dat we. 

In the case of pronouns it is the weak form 'k that induces contraction, 

causing deletion of the final consonant of the first word:

(41) modal verb/auxiliary + pronoun 

wil 'k 'want I' —> [uik] 

zal 'k 'will I' -> [zak] 

kan 'k 'can I' —> (kurjk] 

moet 'k 'must I' -> [muk] 

mag 'k 'may I' —> [mak] 

heb 'k 'have I' -» [hek] 

8-4- CONTRACTION 179 

In the case of je, the final consonant of the first word can be deleted, and 

even the final schwa of the pronoun. Such forms are typically sub-standard 

Dutch: 

(42) wil je (uijaj, [uij] 

zul je [zYJo], [ZYJ] 

kan je [kujo], [kaj| 

moetje [muja], [muj| 

mag je (no contraction) 

heb je [hejal, [hej| 

ben je [bejs|, [bej] 

The generalization that can be made is that the first of two adjacent consonants 

deletes: 

(43) [+cons] -> 0 /(... — [+cons] . . . )„, 

This formulation of the rule presupposes that the second word is encliticized to 

the preceding word. The rule is lexically governed: it only applies to a specific 

set of verbs + pronouns. Alternatively, one may list these contracted forms as 

such in the lexicon. After application of consonant deletion, it is also possible 

to delete the final schwa. Without consonant deletion, deletion of the schwa 

would lead to ill-formed word-final consonant clusters like -nj and -tj. 

As pointed out by Gussenhoven (1985), forms of the verbs hebben 'to have' 

and zijn 'to be' contract with preceding pronouns: 14 

(44) ik heb 'I have' [ikep] 

'k heb 'I have' [kep] 

je hebt 'you have' U e P t ] 

hij heeft 'he has' 

[helft] 

ze heeft 'she has' [zeft] 

we hebben 'we have' [ueban] 

ze hebben 'they have' [zebon] 

The disappearance of the /h/ is predictable: in such clitic + verb combinations 

the /h/ will be preceded or followed by another consonant, whereas it is 

impossible for the /h/ to occur in consonant clusters. For instance, the form 

/zheft/ for ze heeft is phonotactically impossible. 

14 According to Gussenhoven (1985) the verb hoeven 'to need' may also be subject to contraction, 

for instance, in the idiolect of the author of that article.

l80 

CONNECTED SPEECH III: CLITICIZATION 

Note that contraction does not apply to all /hAinitial words. For instance, ye 

houdt 'you hold' cannot be pronounced as [jout]. 

The only complication left is that in hij heeft 'he has' it is the diphthong that 

survives whereas normally the vowel of the clitic pronoun is dropped. If the 

diphthong of hij were deleted, the resulting contracted form would be homophonous 

with the phonetic form of heeft 'has'. 

The possibility of contraction in combination with the use of weak forms of 

pronouns leads to a number of alternative phonetic realizations of word 

sequences that may be quite distant from their written forms, as in: 

(45) Ik heb het hem (gezegd) '(lit.) I have it him told' ('I told it to him') 

[ikhcpatam], [ikheptam], [ikepotam], [ikeptam], fkcpatom], [keptam] 

The verb form is 'is' can appear in the form /s/, but only after function 

words, which suggests that it is to be interpreted as a form of contraction in 

function-word sequences. 15 

(46) Dat is jammer 'That's a pity' [dots], [das] 

Jan is ziek 'John's ill' *[jans], [jams] 

Hij is ziek 'He's ill' 

[hcis] 

Zij is rijk 'She's rich' 

[zeis] 

Contracted forms with /s/ can also undergo further contraction, that is, they 

may also exhibit reduction in the preceding function word, as in: 

(47) Dat's (jammer) 'That's a shame' [das] 

Dit's (lekker) 'This is nice' [dis] 

Wat's dat? 'What's that? 

[wazdat] or [wastat] 

The rule of consonant deletion (43) takes care of these contraction cases as 

well. As will be clear again from the data presented here, contraction is a 

lexically governed process. For instance, we do not find consonant deletion in 

a lexical word: 

(48) (De) kat is (ziek) [katis], *[kats], *[kas] 

Since contraction is restricted here to pronouns + is, it is possible to list the 

contracted forms in the lexicon, although certain phonological generalizations 

about the contraction patterns can be made, as shown above. 

" Contraction is not possible when there is a movement of deletion site in the sequence of 

function words, as in: 

(a) Ik weet niet wie zij / is '1 do not know who she is' 

in which the sequence ùj is cannot be realized as [zeis]. This suggests sensitivity to the presence of 

traces and also that contraction should not be interpreted as a prosodie rule since prosodie rules are 

not sensitive to the presence of empty elements in syntactic (surface) structure (Nespor and Vogel 

1986) 

However, as pointed out in Gussenhoven (1985: 182) contraction is also impossible when the 

movement site follows the sequence of function words. This point is supported by the fact that in 

the English equivalent of (a), with SVO word order, contraction of she is to she's is also 

impossible, although the trace of who follows the sequence of function words. This matter 

therefore deserves further investigation.

ORTHOGRAPHY 


At first sight, it may seem rather strange to deal with the orthography of a 

language at the end of a book on its phonology rather than at the beginning. 

The reason behind this organization of the book is that the orthography of 

Dutch cannnot be understood properly without knowledge of the phonological 

system of Dutch as described in the preceding chapters. In particular, the 

syllable structure of words plays a crucial role in the orthographic representations 

of vowels and consonants. Moreover, Dutch does not always represent 

the phonetic form of words. Instead, some more abstract level of phonological 

representation is reflected by the spelling, and this more abstract level appears 

to be only definable in terms of the typology of phonological rules developed 

in Chapters 4 to 7. 

9.2. THE CORRESPONDENCE BETWEEN SOUNDS AND LETTERS 

The first problem for Dutch orthography is the lack of a sufficient number of 

letters for its sounds. This applies in particular to vowels: there are only five 

vowel letters (;', u, e. o, a) for the thirteen Dutch vowels, the diphthongs not 

included. This has the effect that there is no one-to-one correspondence 

between sounds and letters. The lack of one-to-one correspondence is 

increased by the fact that the history of words also plays a role in their 

spellings. This does not only apply to loan-words that often keep their original 

spelling, but also to native words, the spelling of which may reflect an older 

stage of Dutch, as will be illustrated below. The historical background of 

Dutch orthography is discussed in detail in Booij et al. (1979). 

The spelling of the consonants is more straightforward. Complications only 

arise because there are two digraphs, ch for /x/, and ng for /rj/, and because of 

the etymology effect. A survey of the spelling of the Dutch consonants is given 

in Table 9.1. 

Ambisyllabic consonants, that is, intervocalic consonants occurring after a 

short vowel, are spelled as geminates, but digraphs cannot be geminated (this 

also applies to vowel digraphs):

182 OK lHOOKAPHY 

TABLE 9.1. 'I'he x/>fllinf> 

of 'consonants 

Consonant 

/f/ 

/v/ 

/s/ 

/Jj/ 

/3j/ 

M 

/Y/ 

/P/ 

/b/ 

A/ 

/d/ 

/k/ 

/ku/ 

As/ 

/m/ 

/n/ 

/nj/ 

/O/ 

/I/ 

/r/ 

/o/ 

/j/ 

Spelling 

f 

v 

s 

ch 

g 

P 

b 

t 

d 

k 

kw 

ks 

m 

n 

"J 

ng 

1 

r 

w 

j 

Special cases 

ph (/7/ïoenix 'id.') 

v (veertig 'forty', 

vijftig 'fifty') 

c (cent 'id.') 

ç (facade 'id.') 

se (.scene 'id.') 

sch (suffix -isch as in logisch 'logical') 

7. (zestig 'sixty', zeventig 'seventy' 

ch (chef 'id.') 

sh (.v/iilling 'id.') 

s (analyse 'analysis', fuse 'phase') 

g (horloge 'watch') 

j (y'ury 'id.') 

gh (yojç/iurt 'id'.) 

th (thee 'tea') 

ddh (BoeMha 'Buddha') 

c (café 'id.') 

ch (i/iristen 'Christian') 

qu U/m>tient 'id.') 

qu (quiz 'id.') 

cc (accent 'id.') 

x (tari 'id.') 

gn (campa^rte 'campaign') 

n before velar consonants (ba/ik 'id.') 

uw after ie (niewn' 'new') and ee (leeuw 'lion') 

i after vowels (moo; 'beautiful') 

ij in proper names (Boo/) 'id.') 

y (rovaal 'generous') 

ill (ta;7/iet 'bankrupt') 

(D 

kikker /kikar/ 'frog' 

jekker /jckar/ 'coat' 

hutten /hvtan/ 'huts' 

botten /bDtan/ 'bones' 

bakker /bukar/ 'baker'

9-2. CORRESPONDENCE BETWEEN SOUNDS AND LETTERS 183 

but: 

lachen /luxon/ 'to laugh' 

/.ingen //.irpn/ 'to sing' 

The spelling of short vowels is straightforward (Table 9.2). However, the 

spelling of long vowels is complicated (Table 9.3). The generalization is that 

long vowels are spelled as single letters in open syllables (including word-final 

position), and as geminates in closed syllables, that is, in syllables in which the 

vowel is followed by at least one consonant. However, oe and eu cannot be 

geminated because they are digraphs. Moreover, the geminate form of / is ic 

rather than /;'. In the case of /e/, it is also spelled as a geminate ee in word-final 

position in order to avoid confusion with the schwa which is spelled as e in that 

position. In the case of/i/ the geminate spelling ie is used in open syllables as 

well, namely in native words. 

TABLE 9.2. 'l lic spelling n/' slmrl wntr/.v 

Short vowels 

/i/ 

/e/ 

/Y/ 

hl 

/u/ 

Vowel Spelling 

i 

e 

u 

o 

a 

Long vowels before a consonantal digraph are spelled as geminates, 

although from the phonological point of view they may stand in an open 

syllable. Therefore, a word like- \-ooi-hcm /voxom/ 'smart' is spelled with 

oo. The reason is that before consonantal digraphs a single vowel letter is 

interpreted as a short vowel, as pointed out above. 

The schwa is usually spelled as c. Exceptions are words like monnik 'monk' 

and suffixes such as -ig /ay/ where / is used, and -lijk, where ij stand for the 

schwa. In the indefinite determiner ecu /on/, the schwa is represented by the 

geminate ee. In geographical names such as Dokkiini 'id.' the schwa is spelled 

as ;/. Since the schwa patterns with long vowels, we do not find consonantal 

geminate letters after letters standing for schwa, as illustrated by the following 

words: 

(2) bezemen 'to sweep' /bezomon/ 

monmken 'monks' /mnnokon/ 

Dokkumcr 'of Dokkum' /dnkomor/ 

The three diphthongs of Dutch are spelled as sequences of two letters (Table 

9.4). Complications arise, because etymology plays a role in their spelling. The 

diphthong /ti/ is spelled as ei when it derives historically from Proto-Germanic 

/ui/, and as // when it derives from long /i/. Thus, we get orthographical

184 ORTHOGRAPHY 

TABLE 9.3. The spelling of long vowels 

Vowel 

Spelling 

/i/ 

i (in open syllables, e.g. liter 'litre') 

ie (in closed syllables and in native words, e.g. Piet 'Pete', P/'eter 

'Peter') 

y (in loan-words, e.g. nrythe 'myth') 

ij (only in byzonder 'special') 

/y/ u (in open syllables and before /w/, for example, vuren 'fires', «w 

'your') 

mi (in closed syllables, e.g. v««r 'fire') 

/u/ 

oe 

ou (in loan-words, e.g. dowane 'customs') 

/e/ 

e (in open syllables, e.g. beter 'better') 

ee (in closed syllables and at the end of a (prosodie) word, e.g. beel 

'bite', trofee 'trophy', meewarig 'compassionate') 

é (in loan-words, e.g. café 'id.') 

er (in loan-words, e.g. diner 'dinner') 

ai (in loan-words, e.g. container 'id.') 

ae (in loan-words, e.g. /4esopus 'id.') 

M 

eu 

oe (in loan-words, e.g. «edeem 'oedema') 

lol 

o (in open syllables) 

00 (in closed syllables) 

eau (in loan-words, e.g. bureau 'desk') 

01 (in geographical names, e.g. O/rschot 'id.') 

au (in loan-words, e.g. auto 'car') 

/a/ 

a (in open syllables) 

aa (in closed syllables) 

minimal pairs like leiden 'to lead' versus lijden 'to suffer'. The same applies to 

the diphthong /ou/ which is spelled as ou when it derives from the sequence 

/ol/, and ou or au otherwise. Sometimes, the off-glide [u] at the end of the 

diphthong is also represented in the spelling. 

The consequence of this system is that many orthographical forms have to 

be memorized by the writer of Dutch since for many words, their orthographic 

form is not predictable from their phonological form. 

Sequences of vowel letters may cause problems as to their proper interpretation. 

For instance, the sequence ei occurs in both gein /yein/ 'fun', where it 

stands for a diphthong, and in geind /yaind/ 'collected', where it stands for the 

sequence schwa + /!/. Therefore, Dutch spelling uses diaeresis, two dots above 

a vowel letter, to indicate the proper interpretation, as in gein versus geïnd. 

Another example is reëel /reel/ 'real' versus reeën /rean/ 'deer' (pi.) (cf. 

Wester 1985 for detailed discussion of the diaraesis system). A related 

principle concerns the spelling of /i/ before schwa: when /i/ is spelled as ie

9-3- DEGREES OF ABSTRACTNESS IN SPELLING 185 

TABLE 9.4. The spelling of diphthongs 

Diphthong 

lt\l 

/ou/ 

/cey/ 

Spelling 

ei (e.g. ei 'egg') 

ij (e.g. //' 'water') 

ou (e.g. no« 'now') 

au (e.g. au 'ouch!') 

ouw (e.g., gouw 'province') 

auw (e.g., nauw 'narrow') 

ui (e.g. ui 'onion') 

eui (in loan-words like fautfwil 'armchair') 

eu (e.g. in neuron 'id.') 

in word-final position, and does not bear word stress, it is reduced to i. Thus we 

get the following difference: 

(3) genie 'genius' genieën (pi.) 

versus 

olie 'oil' oliën (pi.) 

A final complication is the use of the diacritic '. This diacritic is used to 

represent the genitive suffix /s/ after a stem ending in s. Thus we get: 

(4) Jans boek 'Jan's book' (= the book of Jan) 

versus 

Jans' boek 'Jans's book' (= the book of Jans) 

When the plural suffix /s/ is added to a word ending in a single vowel letter, the 

corresponding letter s is preceded by the diacritic ' in order to guarantee the 

correct phonetic interpretation of the vowel letter as standing for a long vowel: 

kano's 'canoes', pyjama's 'pyjamas', rabbi's 'id.', reçus 'receipts'. 

9.3. DEGREES OF ABSTRACTNESS IN SPELLING 

Dutch orthography abstracts away from the effects of certain phonological 

rules. For instance, the spelling of the word hond 'dog' represents the underlying 

form /hond/ instead of the phonetic form [hont]. Traditionally, this kind 

of abstractness is described as the 'principle of uniformity' which says that 

morphemes should always have the same orthographical form. However, as we 

will see below, this principle is not applied consistently in Dutch orthography. 

Crucial for a proper characterization of the degree of abstractness of Dutch 

spelling is the rule typology as presented in the preceding chapters (cf. Booij 

1987ft). The first generalizations are: (a) the effects of the rules of connected 

speech including those of voice assimilation are never represented in the 

spelling; (b) the effects of the morpholexical rules of word phonology are 

always represented in the spelling.

186 ORTHOGRAPHY 

For instance, Dutch spelling abstracts away from the effects of the rules of 

Regressive and Progressive Voice Assimilation, whether they apply in compounds 

or in phrases. Also, the insertion of schwa in coda clusters is never 

represented. On the other hand, the allomorphy of the diminutive suffix is 

always represented in Dutch spelling. This makes sense since there are 

sometimes two allomorphs, as in the case of bloempje 'little flower' versus 

bloemetje 'bunch of flowers'. In other words, by representing the effects of 

morpholexical rules in the spelling, it is guaranteed that the phonological form 

of a word can be recovered from its orthographical form. 

As far as the effects of the P-rules of word phonology are concerned, things 

are more complicated because these effects are partially represented in the 

spelling. 

The effects of Final De voicing are only represented orthographical ly for 

underlying /z/ and /v/, that is, they are spelled as s and ƒ respectively when 

devoiced. In the case of/b/, /d/ and /y/ it is always the underlying form that is 

spelled: 

(5) hl dief/div/[dif]—dieven [divan] 'thief/thieves' 

/z/ kaas /kaz/ |kas|—kazen |kazan] 'cheese/cheeses' 

/b/ tob /tob/ [top]—tobben [tsban] 'to toil' (l pers. sg./pl.) 

/d/ zaad /zad/ [zat)—zaden [zadan] 'seed/seeds' 

/y/ vlag /vldv/ |vlux|—vlaggen [vluyan] 'flag/flags' 

A related complication is that /v/ and /z/ are even spelled as .v and / in pasttense 

forms of verbs with a voiced fricative in stem-final position, although 

these fricatives are always voiced in that position: 

(6) beefde /bev+da/ [bevda] 'shivered' (sg.) 

raasde /ra/.+da/ (razdoj 'raged' (sg.) 

In the case of Nasal Assimilation, its effect on the place of articulation of 

nasal consonants is not represented for the palatal and the velar nasal, but it is 

represented for the bilabial nasal: 

(7) [m] damp [dump) 'id.' 

versus 

[n] Spanje [spanjDJ 'Spain', [rjl bank [bunk] 'id.' 

The effects of the rule of Degemination are not represented in the spelling, 

except when we would get a geminate consonant at the end of a (prosodie) 

word. Compare: 

(8) voedde /vud+da/ |vuda] 'fed' (pasi) 

achtte /uxt+ta/ [axtoj 'considered' (past) 

wordt /uord+t/ [uDrt| 'becomes' 

versus 

eet /et+t/ [etj 'eats' 

gevoed /ya+vud+d/ lyovut] 'fed' (part.) 

(iets) vies /vi/,+s/ |vis| '(something) dirty'

9-4- SPELLING AND HIGHER-ORDER LINGUISTIC STRUCTURE 187 

In the last example, the /z/ is devoiced, and hence spelled as .v. This would give 

rise to the letter sequence ss, which is subsequently subject to degemination in 

the orthographical sense. On the other hand, the sequence dt is kept, although it 

is also subject to degemination in the phonological sense, and realized as [t], 

because Dutch orthography does not represent the effect of Final Devoicing on 

/d/. In other words, -dt represents the underlying form of the phonological 

derivation. 

As far as the hiatus rules of word phonology are concerned, a difference has 

to be made between Prevocalic Schwa Deletion and Homorganic Glide Insertion. 

The effects of the first rule are always represented in spelling, the effects 

of the second rule never. We do find intervocalic glides spelled in words like 

koi'ien [kujan] 'cows' and vlooien (vlojanj 'fleas', but this is precisely because 

these glides are unpredictable, the predictable glide in these cases being the 

[v]. Table 9.5 summarizes the way in which the effects of the P-rules of word 

phonology are represented orthographically: 

TABLE 9.5. .Vnnrv of orthographical representation of P-rule 

effects 

P-rule Effect represented Effect not represented 

Final Devoicing for /v, z/ for /b, d, y/ 

Laryngeal Spreading 

always 

Nasal Assimilation for [m| for Iji.rj] 

Degemination word-linal position word-internal position 

Prevocalic Schwa Deletion always 

Homorganic Glide Insertion 

always 

9.4. SPELLING AND HIGHER-ORDER LINGUISTIC STRUCTURE 

Dutch orthography also provides syntactic and semantic information. Capital 

letters mark the beginning of sentences and proper names. Punctuation is used 

to mark the boundaries between certain syntactic constituents (e.g. clauses and 

appositional and parenthetical constituents), and spacing is used to mark word 

boundaries. Compounds are always written as one word. That is, the fact that 

compounds are one word from the grammatical point of view is represented in 

the spelling. This is different from English where many compounds are spelled 

with word-internal spaces. This means that in English spelling, spacing often 

reflects prosodie structure rather than grammatical structure (recall that compounds 

consist of two or more prosodie words). However, there is also a 

tendency among Dutch writers to write compounds with internal spaces 

between the prosodie words. 

Hyphenation, the use of hyphens in word-splitting, on the other hand, is 

based on the prosodie structure of words, not on their morphological structure

l88 

ORTHOGRAPHY 

(as is the case in English) as is illustrated by the hyphenation of the word 

morphemic and its Dutch counterpart: 

(9) English: morphem-ic 

Dutch: mor-femisch or morfe-misch 

That is, in Dutch the hyphen has to coincide with a syllable boundary. In the 

case of consonantal geminates and consonantal digraphs after short vowels the 

hyphen is inserted in the middle of the sequence, except for the digraph ch: 

(10) baker ba-ker 'nurse' 

bakker bak-ker 'baker' 

zanger zan-ger 'singer' 

lachen la-chen 'to laugh' 

In sum, the spelling conventions of Dutch illustrate that orthography 

represents a lot more than just the spoken forms of words.

REFERENCES 

ABBREVIATIONS 

CLS 

IULC 

J L 

JoP 

LB 

Lg. 

LI 

LIN 

NLLT 

Ntg. 

Ph. 

Sp. 

YM 

Papers from the nth Regional Meeting of the Chicago Linguistic Society 

Indiana University Linguistics Club 

Journal of Linguistics 

Journal of Phonetics 

Leuvense Bijdragen 

Language 

Linguistic Inquiry 

Linguistics in the Netherlands 

Natural Language and Linguistic Theory 

De Nieuwe Taalgids 

Phonology (Yearbook) 

Spektator, tijdschrift voor Neerlandistiek 

Yearbook of Morphology, edited by Geert Booij and Jaap van Marie 

ANDERSON, J. (1986), 'The English Prosody /h/', in D. Kastovsky and A. Szwedek (eds.), 

Linguistics across Historical and Geographical Boundaries (Berlin: Mouton de 

Gruyter), 799-809. 

ANDERSON, S. R. (1974), The Organisation of Phonology (New York: Academic Press). 

— (1979), 'Notes on the Development of Phonological Theory', in E. Fischer- 

J0rgensen, J. Rischel, and N. Thorsen (eds.), Proceedings of the 9th Int. Congress of 

Phonetic Sciences, Copenhagen, 1979 (Copenhagen: Institute of General Linguistics), 

133-42. 

ARCHANGELI, D. (1984), Underspecification in Yawelmani Phonology and Morphology, 

MIT Ph. D. dissertation (reprinted New York: Garland Press 1988). 

— (1991), 'Syllabification and Prosodie Templates in Yawelmani', NLLT 9: 

231-84. 

ARONOFF, M. (1976), Word Formation in Generative Grammar (Cambridge, Mass.: MIT 

Press). 

BACKHUYS, K.-J. (1989), 'Adjectival Compounds in Dutch', in H. Bennis and A. van 

Kemenade (eds.), LIN 1989 (Dordrecht: Foris), 1-10. 

BECKMAN, M. E. (1986), Stress and Non-stress Accent (Dordrecht: Foris). 

— and PIERREHUMBERT, J. (1986), 'Intonational Structure in English and Japanese', 

Ph., 3: 255-309. 

BERENDSEN, E. (1983), 'Final Devoicing, Assimilation, and Subject Clitics in Dutch', 

in H. Bennis and W. U. S. van Lessen Kloeke (eds.), LIN 1983 (Dordrecht: Foris), 

21-30. 

- (1986), The Phonology of Cliticization (Dordrecht: Foris). 

— and Os, E. DEN (1987), 'Glide Insertion: Domains, Speech Rate and Phonetic 

Prominence', in F. Beukema and P. Coopmans (eds.), LIN 1987 (Dordrecht: Foris), 

13-20.

ICO 

REFERENCES 

BERENDSEN, .E. and ZONNEVELD, W. (1984), 'Nederlandse schwa-invoeging op z'n 

Deens', Sp., 14: 166-9. 

BERG, B. VAN DEN (1958), Foniek van het Nederlands (Den Haag: Van Goor Zonen). 

1470), 'Het woordaccent van afleidingen met het prefix on-', Ntg. Van 

Haeringen-issue, 1-18. 

BERG, R. VAN DEN, GUSSENHOVEN, C., and RIETVELD, T. (1992), 'Downstep in Dutch: 

Implications for a Model', in Dogherty and Ladd (1992), 335-58. 

BtxMj, G. E. (1977), Dutch Morphology. A Study of Word Formation in Generative 

Grammar (Dordrecht: Foris). 

(1980), 'De onderliggende vorm van de velaire nasaal', LB, 69. 1-10. 

(I981«), Generatieve fonologie van hel Nederlands (Utrecht/Antwerpen: Het 

Spectrum). 

(1981/)), 'Rule Ordering, Rule Application and the Organization of Grammars', 

in Dressier et al. (1981), 45-56. 

(1982«), 'Lexicale fonologie en de organisatie van de morfologische component', 

Sp., 12: 169-88; English translation in E. Gussmann (ed.), Kulcx und the 

Lexicon (Lublin: Catholic University of Lublin, 1987), 43-65. 

(1982ft), 'Fonologische en fonetische aspecten van klinkerreductie', Sp., 11: 

295-301. 

— (1982r), Hiërarchische fonologie. Inaugural lecture (Amsterdam: Vrije 

Universiteit). 

(1983), 'Principles and Parameters in Prosodie Phonology', Linguistics, 21: 

249-80. 

(1984), 'Syllabestructuur en verkleinwoordvorming in het Nederlands', Glot, 7: 

207-26. 

— (1985a), 'Coordination Reduction in Complex Words: A Case for Prosodie 

Phonology', in Van der Hulst and Smith (1985), 143-60. 

(1985ft), 'Lexical Phonology, Final Devoicing, and Subject Pronouns in 

Dutch', in H. Bennis and F. Beukema (eds.), LIN 1985 (Dordrecht: Foris), 21-6. 

(1986), 'Form and Meaning in Morphology: The Case of Dutch "Agent 

Nouns" ', Linguistics, 24: 503-17. 

( 1987«), 'Clitisatie in het Nederlands', Sp.. 16: 219-28. 

(1987/7), 'The Reflection of Linguistic Structure in Dutch Spelling', in Ph. A. 

Luelsdorff (ed.), Orthography and Phonology (Amsterdam: Benjamins), 215-24. 

(1988«), 'Polysemie en polyfunctionaliteit bij de-nominale woordvorming'. 

Sp., 17: 268-76. 

(1988ft), 'On the Relation between Lexical Phonology and Prosodie Phonology', 

in P. M. Bertinetto and M. Loporcaro (eds.), Certamen Phonologicum 

(Torino: Sellier), 63-76. 

-(1988e), 'Review of Nespor and Vogel ( 1986)',.//., 24: 515-25. 

- ( I989a), 'On the Representation of Diphthongs in Frisian', JL, 25: 319-32. 

(1989ft), 'Complex Verbs and the Level Ordering Hypothesis', YM 

(Dordrecht: Foris), 21-30. 

(1990a), 'The Relation between Morphology and Syntax: Separable Complex 

Verbs in Dutch', YM 1990, (Dordrecht: Foris), 45-63. 

(1990ft), 'Dutch Phonology and Morphology', in F. Aerts and Th. van Els 

(eds.), Dutch LinguistK .v (Washington, DC: Georgetown University Press), 38-49.

REFERENCES 

ICI 

- (1992a), 'Review of Inkelas and Zec (1990)'. Ph., 9: 145-52. 

— (1992/j), 'Compounding in Dutch', Rivisla ai Linguistica, 4: 37-59. 

— (1992e), 'Lexical Phonology and Prosodie Phonology', in Dressler et al. 

(1992), 49-62. 

— (1993), 'Against Coronal Undcrspecification: The Case of Dutch', in H. 

Demeersseman and Th. de Jong (eds.), Jaarboek Corpusgebaseerde Woordanalyse 

1993 (Amsterdam: Vakgroep Taalkunde, Vrije Universiteit), 11-20. 

— HAMANS, C., VERHOEVEN, G., BALK, F., and VAN MINNEN, C. M. (1979), Spelling 

(Groningen: Wolters-Noordhoff). 

and LIEBER, R. (1993), 'On the Simultaneity of Morphological and Prosodie 

Structure', in S. Hargus and E. Kaisse (eds.). Studies in Lexical Phonology (San 

Diego: Academic Press), 23-44. 

and RUBACH, J. (1984), 'Morphological and Prosodie Domains in Lexical 

Phonology', Ph., 1: 1-28. 

— (1987), 'Postcyclic versus Postlexical Rules in Lexical Phonology'. /./, 

18: 1-44. 

- (1991), 'Lexical Phonology', in Bright (1991), ii, 293-6. 

BRIGHT, W. (ed.) (1991), International Encvclopedia of Linguistics, 4 vols. (New York: 

Oxford University Press). 

BROKCKE, M. VAN DEN, HEUVEN, V. VAN, and ZONNEVELD, W. (eds.) (1983), Sound 

Strut-lures. Studies for Antonic Cohen (Dordrecht: Foris). 

BROSELOW, E. ( 1991 ), 'The Structure of Fricative-Stop Onsets', Department of Linguistics, 

SUNY at Stony Brook, MS. 

— (1992), 'Parametric Variation in Arabic Dialect Phonology', in E. Broselow, M. 

Eid, and J. J. McCarthy (eds.). Perspectives on Arabic Linguistics IV (Amsterdam: 

Benjamins), 7-45. 

BROWMAN, Ch., and GOLDSTEIN, L. (1990), 'Tiers in Articulatory Phonology, with some 

Implications for Casual Speech', in Kingston and Beekman (1990), 341-76. 

CAIRNS, Ch. (1988), 'Phonotactics, markedness, and lexical representations', Ph.. 5: 

209-36. 

CAMMKNCJA, J. and REENEN, P. T. VAN (1980), 'Review of Trommelen and Zonneveld 

1979, G/wr, 3: 183-9. 

CARSTAIRS, A. (1988), 'Some Implications of Phonologically Conditioned Suppletion', 

YM 1988 (Dordrecht: Foris), 67-94. 

CHOMSKY, N. (1986), Barriers (Cambridge, Mass.: MIT Press). 

- and HALLE, M. (1968), The Sound Pattern of English (New York: Harper & 

Row). 

CLEMENTS, G. N. (1985), 'The Geometry of Phonological Features', Ph., 2: 225-52. 

( 1989), 'On the Representation of Vowel Height', Department of Linguistics, 

Cornell University, MS. 

— (1990), 'The Role of the Sonority Cycle in Core Syllabification', in Kingston 

and Beekman (1990), 283-333. 

and HIIMK, E. V. (1993), 'The Internal Orgam/ation of Speech Sounds', CRNS, 

Paris and Department of Linguistics, Ohio State University MS. 

— and KKYSER, S. J. (1983), CV Phonology. A Generative Theory of the S\lUihle 

(Cambridge, Mass.: MIT Press).

192 REFERENCES 

COHEN, A. (1958), 'Het Nederlands diminutiefsuffix, een morfonologische proeve', 

Ntg., 51:40-5. 

— and 'T HART, J. (1967), 'On the Anatomy of Intonation', Lingua, 19: 177-92. 

— EBELING, C. L., FOKKEMA, K., and HOLK, A. F. G. VAN (1959), Fonologie van het 

Nederlands en hel Fries, 2nd edn. ('s Gravenhage: Martinus Nijhoff [1961]). 

COHN, A. C. (1989), 'Stress in Indonesian and Bracketing Paradoxes',NLLT, 7: 167-216. 

COLLIER, R. (1989), 'On the Phonology of Dutch Intonation', in F. J. Heyvaert and F. 

Steurs (eds.), Worlds behind Words (Leuven: Leuven University Press), 245-60. 

— and 'T HART, J. (1981), Cursus Nederlandse intonatie (Leuven: Acco). 

COLLINS, B., and MEES, I. (1981), The Sounds of English and Dutch (Leiden: 

Universitaire Pers). 

CoMBRiNK, J. G. H., and DE STADLER, L. G. (1987), Afrikaanse Fonologie (Johannesburg: 

Macmillan). 

DAVIS, S. (1991), 'Coronals and the Phonotactics of Non-adjacent Consonants in 

English', in Paradis and Prunet (1991), 49-60. 

DEBROCK, M. (1978), 'Is the Fortis-Lenis Feature Really Redundant in Dutch?', LB, 67: 

457-72. 

DEKEYSER, X. 1978. 'Some Considerations on Voicing with Special Reference to 

Spirants in English and Dutch', in J. Fisiak (ed.). Recent Development in Historical 

Phonology (Berlin: Mouton de Gruyter), 99-122. 

DOGHERTY, G., and LADD, D. R. (eds.) (1992), Papers in Laboratory Phonology II. 

Gesture, Segment, Prosody (Cambridge: Cambridge University Press). 

DRESSLER, W. U. (1974), 'Methodisches zu Allegro-regeln', in W. U. Dressler and F. 

Mareä (eds.), Phonologica 1972 (München: Fink Verlag), 219-34. 

- (1977), 'Phono-morphological Dissimilation', in Dressler et al. (1977), 41-8. 

- LUSCHÜTZKY, H., PFEIFFER, O. E., and RENNISON, J. R. (1992), Phonologica 1988 

(Cambridge: Cambridge University Press). 

PFEIFFER, O. E., and HEROK, T. (eds.) (1977), Phonologica I97f> (Innsbruck: 

Institut für Sprachwissenschaft). 

RENNISON, J. R., and DOGIL, G. (eds.) (1981), Phonologica 19X0 (Innsbruck: 

Institut für Sprachwissenschaft). 

EEFTING, W. Z. F. (1991), 'The Effect of "Information Value" and "Accentuation" on 

the Duration of Dutch Words, Syllables, Segments', Journal of the Acoustical Society 

of America, 89: 412-24. 

EIJKMAN, L. P. H. (1955), Phonetiek van het Nederlands (Haarlem: De Erven Bohn). 

EWEN, C. J. (197H), 'The Phonology of the Diminutive in Dutch: A Dependency 

Account', Lingua, 45: 141-73. 

— (1982), 'The Internal Structure of Complex Segments', in Van der Hulst and 

Smith (1982), 27-68. 

FIDELHOLTZ, J. (1975), 'Word Frequency and Vowel Reduction', CLS, II: 200-13. 

GOLDSMITH, J. (1990), Autoxegmental and Metrical Phonologv (Oxford: Basil 

Blackwell). 

GUSSENHOVEN, C. (1978)i 'Het Nederlandse diminutief-suffix: schwa-insertie nader 

bekeken', Ntg., 71: 206-11. 

- (1980), 'Review' of Trommelen and Zonneveld (1979), Gramma, 4: 174-83. 

-(I983a), 'Stress Shift and the Nucleus', Linguistics, 21: 303-39; repr. in 

Gussenhoven (1984), 291-332.

REFERENCES 

IÇ3 

- (1983/J), 'Stress Shift as a Rhetorical Device', Linguistics, 21: 603-19; repr. in 

Gussenhoven (1984), 332^18. 

— (1984), On the Grammar and Semantics of Sentence Accents (Dordrecht: Foris). 

- (1985), 'Over de fonologie van Nederlandse clitica', Sp., 15: 180-200. 

(1988), 'Adequacy in Intonational Analysis', in Van der Hulst and Smith 

(1988), 95-122. 

(1991), 'Tone Segments in the Intonation of Dutch', in Shannon and Snapper 

(1991), 139-55.' 

(1992a), 'Sentence Accents and Argument Structure', in I. G. Roca (ed.), 

Thematic Structure, its Role in Grammar (Berlin: Foris), 79-106. 

— (1992/j), 'Intonational Phrasing and the Prosodie Hierarchy', in Dressier et al. 

(1992), 89-100. 

(1992c), 'Illustrations of the IPA: Dutch ', Journal of the International Phonetic 

Association, 22: 45-7. 

and BROEDERS, A. (1976), The Pronunciation of English. A Course for Dutch 

Learners (Groningen: Wolters-Noordhof-Longman). 

and BRKMMLK, R. H. ( 1983), 'Voiced Fricatives in Dutch: Sources and Presentday 

Usage', North Western European Language Evolution, 2: 55-71. 

and RIETVELD, A. C. M. (1992), 'Intonation Contours, Prosodie Structure, and 

Preboundary Lengthening', JoP, 20: 283-303. 

GUSSMANN, E. (1992), 'Resyllabification and Delinking: The Case of Polish Voicing', 

LI, 23: 29-56. 

GUY, G. R. (1980), 'Variation in the Group and the Individual: The Case of Final Stop 

Deletion', in Labov (1980), 1-36. 

HAAS, W. DE (1985), 'Ritmische accentverschuiviing in samenstellende afleidingen in 

het Nederlands', Glot, 8: 129-58. 

— (1986), 'Partial Syllabification and Schwa Epenthesis in Dutch', Gramma, 10: 

143-62. 

- (1991), 'De notie "stress minimalisatie" en klemtoonaantrekking', Sp., 20: 

245-72. 

HAERINGEN, C. B. VAN (1958), 'De plaats van de ie, oe en uu in het Nederlandse 

klinkerstelsel', repr. in id., Gramarie (Assen: Van Gorcum), 1962, 280-7. 

— (1955), 'Is tat juist, op tie manier?', Taal en Tongval, repr. in id., Gramarie 

(Assen: Van Gorcum, 1962), 207-16. 

HAI.L, T. A. (1989), 'Lexical Phonology and the Distinction of German (ç] and [x]', Ph., 

6: 1-18. 

— ( 1992), Syllable Structure and Syllahle-related Processes in German (Tübingen: 

Niemeyer). 

(1993), 'On Coronal Underspecification', in H. van der Hulst and J. van de 

Weijer (eds.), Proceedings of the First Holland Institute of Generative Linguistics 

Phonology Conference (Leiden: Holland Institute of Generative Linguistics, forthcoming). 

HALLE, M. (1991), 'Phonological Features', in Bright (1991) iii, 207-11. 

— and CLEMENTS, G. N. (1983), Problem Book in Phonology (Cambridge, Mass.: 

MIT Press). 

— and KENSTOWIC/, M. (1991), 'The Free Element Condition and Cyclic versus 

Non-cyclic Stress', LI, 22: 457-501.


HALLE, M. and LADEFOGED, P. (1988), 'Some Major Features of the International 

Phonetic Alphabet', Lg., 64: 577-82. 

— and VERGNAUD, J.-R. (1987), An Ksxay un Stress (Cambridge, Mass.: MIT 

Press). 

HARCIUS, S. (1988), The Lexical Phonology of Sekani (New York: Garland). 

HARRIS, J. (1987), 'Non-structure-preserving Rules in Lexical Phonology: Southeastern 

Bantu harmony', Lingua, 73: 255-92. 

— (1989), 'Towards a Lexical Analysis of Sound Change in Progress', JL, 25: 

35-56. 

'T HART, J. (1969), 'Fonetische steunpunten', Nig., 62: 168-74. 

- and COHEN, A. (1973), 'Intonation by Rule: A Perceptual Quest ',JoP, 1: 309-27. 

— and Con IKK, R. (1975), 'Integrating Different Levels of Intonational Analysis'. 

JoP, 3: 235-55. 

and COHEN, A. (1990), A Perceptual Study of Intonation. An E\peri 

mental-Phonetic Approach to Speech Melody (Cambridge: Cambridge University 

Press). 

HAYES, B. (1981), A Metrical Theor\ of Stress Rules (Bloomington, Ind.: IULC; repr. 

New York. Garland Press, 1984). 

- (1982), 'Extrametricality and English stress', /./, 13: 227-76. 

- (1984), 'The Phonology of Rhythm in English', LI, 15: 33-74. 

- (1986), 'Inalterability in CV-Phonology', Lg., 62: 321-53. 

(1989), 'Compensatory Lengthening in Moraic Phonology', LI, 20: 253-306. 

(1990), 'Diphthongisation and Coindexing', Ph., 1: 31-72. 

(1992), 'Comments on Chapter 10', in Dogherty and Ladd (1992), 280-6. 

HEEROMA, K. (1959), 'De plaats van de ie, oe en uu in het Nederlandse klinkersysteem', 

Ntg., 52: 297-304. 

HIRMANS, B. (1992), 'On the Representation of Quasi-long vowels in Dutch and 

Limburgian', in R. Bok-Bennema and R. van Hout (eds.), LIN 1992 (Amsterdam: 


Hi i VEN, V. VAN ( 1986), 'Phonetic Correlates of Stress Patterns in Dutch (Compound) 

Adjectives', in F. Beukema and A. Hulk (eds.), LIN 1986 (Dordrecht: Foris), 79-88. 

(1987), 'Stress Patterns in Dutch (Compound) Adjectives: Acoustic Measurements 

and Perception Data', Phonetica. 44: 1-12. 

— and Hoos, A. (1991), 'Hiatus Deletion, Phonological Rule or Phonetic Coarticulation?', 

in F. Drijkoningen and A. van Kemenade (eds.), LIN 1991 (Amsterdam: 


HOI.KSI.MA, J., and ZONNEVELD, R. VAN (1984), 'Een autusegmentele theorie van hel 

Nederlandse woordaccent', Sp., 13: 450-72. 

Hooi'iR, J. B. (1972), 'The Syllable in Generative Phonology', Lg., 48: 525-40. 

--(1976), An Introduction to Natural (ienerative Phonology (New York: 

Academic Press). 

HULST, H. VAN DER (1980), "On the Formulation of Phonological Rules', in S. Daalder 

and M. Gerritsen (eds.), LIN 1980 (Amsterdam: North Holland), 10- 19. 

— (1984), Syllable Structure ami Stress in Dutch (Dordrecht: Foris). 

(1985), 'Ambisyllabicity in Dutch', in H. Bennis and F. Beukema (eds.), LIN 

1985 (Dordrecht: Foris), 57-66.

REFERENCES 195 

— and SMITH, N. (eds.) (1982), The Structure of Phonological Representation*. 

(Dordrecht: Foris). 

— (eds.) (1985), Advances in Non-linear Phonologv (Dordrecht: Foris). 

— (eds.) (1988), Autosegmental Studies in Pilch Accent (Dordrecht: 

Foris). 

HYMAN, L. 1985. A Theory of Phonological Weight (Dordrecht: Foris). 

INKELAS. S., and ZEC, D. (eds.) (1990), The Phonology-Syntax Connection (Chicago: 

University of Chicago Press). 

ITÔ, J. (1986), 'Syllable Theory in Prosodie Phonology', University of Amherst, Mass. 

Ph. D. dissertation. 

- (1989), 'A Prosodie Theory of Epenthesis', NLLT 7: 217-60. 

— and MKSTER, R. A. (1992), 'Weak Layering and Word Binarity', Linguistics 

Research Center, University of California at Santa Cruz MS. 

IVERSON, G., and SALMONS. J. (1992), 'The Place of Structure Preservation in German 

Diminutive Formation', Ph., 9: 137-44. 

JACOBS, H., and WEIJER, J. VAN DE ( 1992), 'On the Formal Description of Patalisation', in 

R. Bok-Bennema and R. van Hout (eds.), LIN 1992 (Amsterdam: Benjamins), 125- 

35. 

JANSON, T. (1986), 'Cross-linguistic Trends in the Frequency of CV Sequences', Ph., 3: 

179-95. 

JoNGLMUKcii.K, W., and HEUVEN, V. VAN (1991), 'The Distribution of (Word-Initial) 

Glottal Slops in Dutch', in F. Drijkoningen and A. van Kemenade (eds.), LIN 1991 

(Amsterdam: Benjamins), 101-10. 

KAGIK, R ( 1989), A Metrical Theory of Stress and Deslressing in English and Dutch 


— and Visen, E. (1988), 'Metrical Constituency and Rhythmic Adjustments', Ph., 

5: 21-71. 

-and ZONNEVELD, W. (1986), 'Schwa, Syllables and Extrametricality', The 

Linguistic Review, 5: 197-222. 

KEATING, P. (1991), 'Coronal places of articulation', in Paradis and Prunet (1991), 29-48. 

KENSTOWIC/. M.. and RUBACH, J. (1987), 'The Phonology of Syllabic Nuclei in Slovak', 

Lg., 63: 463-97. 

KERCKVOORDB, C. VAN (1993), An Introduction to Middle Dutch (Berlin: Mouton de 

Gruyter). 

KINGSTON, J., and BECKMAN, M. E. (eds.) (1990), Papers in Laboratory Phonology I 

Hctut'cn the Grammar and Physics of Speech (Cambridge: Cambridge University 

Press). 

Kii'ARSKY, P. (1968), 'Linguistic Universals and Linguistic Change', in E. Bach and R. 

Harms (eds.), l/niversals in Linguistic Theory (London: Holt, Rinehart and Winston), 

170-202. 

— (1982), 'From Cyclic Phonology to Lexical Phonology', in Van der Hulst and 

Smith (1982), i, 131-75. 

- (1985), 'Some Consequences of Lexical Phonology', Ph., 2: 85-138. 

KLAVANS, J. (1985), 'The Independence of Syntax and Phonology in Cliticization', Lg., 

61: 95-121. 

Km HM u, G. A. T. (1979), 'Paradigmatische invloeden op fonetische processen', in T. 

Hoekstra and H. van der Hulst (eds.). Morfologie in Nederland (Clot-special), 51-70.


KOHLER, K. J. (1990), 'Segmental Reduction in Connected Speech in German: Phonological 

Facts and Phonetic Explanations', in W. J. Hardcastle and A. Marchai (eds.), 

Speech Production and Speech Modelling (Dordrecht: Kluwer), 69-92. 

— (1991), 'The Phonetics/Phonology Issue in the Study of Articulatory Reduction', 

Phonetica, 48: 180-92. 

Koou, J. G. (1977), 'Schwa Insertion in Dutch: Phonology or Morphology?', in Dressier 

et at. (1977), 65-70. 

— and NIET, M. VAN DER (1985), 'Stress, Stress Shift, and Morphology: The Case 

of Dutch -haar', in Van der Hulst and Smith (1985), 101-15. 

KOOPMANS-VAN BsiNUM, F. J. (1969), 'Nog meer fonetische zekerheden', Ntg., 62: 

245-50. 

— (1980), Vowel Contrast Reduction. An Acoustic and Perceptual Study of Dutch 

Vowels in Various Speech Conditions (Amsterdam: Academische Pers). 

(1982), 'Akoestische en perceptieve aspecten van klinkercontrastreductie en de 

roi van de fonologie', Sp., 11: 284-94. 

KutERA, H. (1973), 'Language Variability, Rule Interdependence, and the Grammar of 

Czech', LI, 4: 499-521. 

LABOV, W. (1972), Sociolinguiltic Patterns (Philadelphia: University of Pennsylvania 

Press). 

— (ed.) (1980), Locating Language in Time and Space (New York: Academic- 

Press). 

LADD, R. D. (1986), 'Intonational Phrasing: The Case for Recursive Prosodie Structure', 

Ph., 3: 311^0. 

— (1992), 'An Introduction to Intonational Phonology', in Dogherty and Ladd 

(1992), 321-34. 

LADEFOGED, P. (1982), A Course in Phonetics, 2nd edn. (New York: Harcourt, Brace, 

Jovanovich). 

LAHIRI, A., and EVERS, V. (1991), 'Palatali/.ation and Coronality', in Paradis and Prunet 

(1991), 79-100. 

— and KOOREMAN, J. (1988), 'Syllable Weight and Quantity in Dutch', Papers 

from the West Coast Conference on Formal Linguistics, 1: 217-28. 

LANGEWEG, S. J. (1986), '"Roots", "Words" and the Stress Behaviour of Non-native 

Words in Dutch', in F. Beukema and A. Hulk (eds.), LIN 1986 (Dordrecht: Foris), 

151-60. 

- (1988), 'The Stress System of Dutch', University of Leiden Ph. D. dissertation. 

LEVIN, J. (1985), 'A Metrical Theory of Syllabicity', MIT, Ph. D dissertation. 

LIBERMAN, M., and PRINCE, A. (1977), 'On Stress and Linguistic Rhythm', LI, 8: 249-336. 

LIEBER, R. (1989), 'On Percolation', YM 1989 (Dordrecht: Foris), 95-138. 

LINDAU, M. (1978), 'Vowel Features', Lg., 54: 541-63. 

LOMBARD:, L. (1991), 'Laryngeal Features and Laryngeal Neutralization', University of 

Massachusetts at Amherst Ph. D. dissertation. 

LOOTS, M. (1983), 'Syntax and Assimilation of Voice in Dutch', in Van den Broecke 

Van Heuven, and Zonneveld (1983), 173-82. 

MCCARTHY, J. J. (1981), 'A Prosodie Theory of Non-concatenative Morphology', LI, 12: 

373-118. 

- (1986), 'OCP-effects: Gemination and Anti-gemination', LI, 17: 207-63. 

— (1988), 'Feature Geometry and Dependency: A Review', Phonetica, 43: 84-1 OH.

REFERENCES 

IQ7 

— and PRINCE, A. (1990), 'Foot and Word in Prosodie Morphology', NLLT, 8: 

209-83. 

and TAUB, A. (1992), 'Review' of Paradis and Prunet (1991), Ph., 9: 363-72. 

MALIKOUTI-DRACHMAN, A., and DRACHMAN, G. (1992), 'Greek Clitics and Lexical 

Phonology', in Dressier et al. (1992), 197-206. 

MARLE, J. VAN (1980), 'The Stress Patterns of Dutch Simplex Words: A First Approximation', 

in Zonneveld el al. (1980), 89-128. 

— (1982), 'Een niet-generaliserende analyse van schwa-deletie', Sp., 11: 326-41. 

— (1985), On the Paradigmatic Dimension of Morphological Creativity 


MARTIN, W. (1968), 'De verdoffing van gedekte en ongedekte e in niet-hoofdtonige 

positie bij Romaanse leenwoorden in het Nederlands', Ntg., 61: 162-81. 

MENERT, L. (1988), 'Effects of Speech Rate and Experimental Method on Voice 

Assimilation in Dutch', in P. Coopmans and A. Hulk (eds.), LIN 1988 (Dordrecht: 

Foris), 121-7. 

MUSTER, A., and ITÔ, J. (1989), 'Feature Predictability and Underspecification: Palatal 

Prosody in Japanese Mimetics', Lg., 65: 258-93. 

MODEL, J. (1991), Grammatische analyse. Syntactische verschijnselen van het 

Nederlands en Engels (Dordrecht: ICG Publications). 

MOHANAN, K. P. (1991), 'On the Bases of Radical Underspecincation', NLLT, 9: 

285-326. 

MOULTON, W. G. (1956), 'Syllabic Nuclei and Final Consonant Clusters in German', in 

M. Halle (ed.), For Roman Jakobson (The Hague: Mouton), 372-81. 

— (1962), 'The Vowels of Dutch: Phonetic and Distributional classes'. Lingua, 

11: 294-313. 

NEUT, A. (1985), 'Clitics in Arboreal Phonology', in Van der Hulst and Smith (1985), 

179-92. 

— (1990), 'Prosodie Structures and Phonetic Findings. The Case of Equally 

Stressed Adjectives', in R. Bok-Bennema and P. Coopmans (eds.), LIN 1990 

(Dordrecht: Foris), 113-22. 

and HEUVRN, V. VAN (1992), 'Rules and Exceptions in Word Stress', in R. Bok- 

Bennema and R. van Hout (eds.), L1N 1992 (Amsterdam: Benjamins), 185-96. 

and ZONNEVELD, W. ( 1981 ). 'De aantrekkingskracht van -baar', Clot, 4: 215-28. 

NESPOR, M., and VOGEL, I. (1986), Prosodie Phonology (Dordrecht: Foris). 

NEU, H. (1980), 'Ranking of Constraints on /t,d/ Deletion in American English', in 

Labov (1980), 37-54. 

NEVIS, J. (1988), Finnish Particle Clitics and General Clitic Theory (New York: 

Garland). 

NOLAN, F. (1992), 'The Descriptive Role of Segments: Evidence from Assimilation', in 

Dogherty and Ladd (1992), 261-80. 

Noon:n(X)M, S. G. (1972), 'Production and Perception of Vowel Duration. A Study of the 

Durational Properties of Vowels in Dutch', University of Utrecht Ph. D. dissertation. 

— and COHEN, A. ( 1976), Spreken en verstaan. Een inleiding tot de experimentele 

fonetiek, 2nd edn. (Assen: Van Gorcum, 1984). 

ODDEN, D. (1990), 'Phonology and its Interaction with Syntax and Morphology', Studies 

in the Linguistic Sciences, 20: 69-105. 

(1991), 'Vowel Geometry', Ph., 8: 261-89.


PADGETT, J. (1991), 'Stricture in Feature Geometry', University of Massachusetts at 

Amherst Ph. D. dissertation. 

— (1992), 'Stricture and Nasal Place Assimilation', Linguistics Research Center, 

University of California at Santa Cruz MS. 

PARADIS, C., and PRUNET, J. F. (1989), 'On Coronal Transparency', Ph., 6: 317-48. 

— (eds.) (1991), The Special Stains oj'Coronals. Internal anil Internal Evidence 

(San Diego: Academic Press). 

PESETSKY, D. (1985), 'Morphology and Logical Form', LI, 16: 193-246. 

POSER, W. J. (1989), 'The Metrical Foot in Diyari', Ph., 6: 117-48. 

— (1990), 'Evidence for Foot Structure in Japanese', Lg., 66: 78-105. 

POSTAL, P. M. (1968), Aspect* oj Phonological Theory (New York: Harper and Row). 

PRINCE, A. (1983), 'Relating to the Grid', LI, 14: 19-100. 

PULLEYBLANK, E. G. (1989), 'The Role of Coronal in Articulator Based Features'. ( 7 S, 

25: 379-93. 

REENEN, P. TH. VAN (1986), 'The Vocalization of/!/ in Standard Dutch, a Pilot Study of 

an Ongoing Change', in F. Beukema and A. Hulk (eds.), LIN I9H6, Dordrecht: Foris, 

189-98. 

RHMSDIJK, H. C. VAN (1978), A Case Study in Syntactic Markedness (Dordrecht: Foris). 

RiETVhLu, A. C. M. (1983), 'Syllaben, klemtonen en de automatische detectie van 

beklemtoonde syllaben in het Nederlands', Katholieke Universiteit Nijmegen, 

Ph.D. dissertation. 

- and LOMAN, H. G. (1985), '/H/, stemhebbend of stemloos'. (Hot. 8: 275-85. 

ROCA, I. (1986), 'Secondary Stress and Metrical Rhythm', Ph., 3: 341-70. 

—— forthcoming. Generative Phonology (London: Routledge). 

RUBACH, J., and Boou, G. E. (1985), 'A Grid Theory of Stress in Polish'. Lingua, 66: 

281-319. 

- (1990a), 'Syllable Structure Assignment in Polish', Ph., 7: 121-58. 

- (1990A), 'Edge of Constituent Effects in Polish', NLLT, 8: 427-63. 

SADOCK, J. M. (1991), Autoleiical Syntax. A Theory of Parallel Grammatical Kepir 

sentalions (Chicago: Chicago University Press). 

SAGEY, E. (1986), 'The Representation of Features and Relations in Non-linear Phonology'; 

MIT Ph.D. dissertation (repr. New York: Garland; 1990). 

SAPIR, E. (1917), 'Review' of C. C. Uhlenbeck. lit'/ passieve karakter van het verbum 

transitivum of van het verbum aclionis in de talen van Noord-Amerika', International 

Journal of American Linguistics, 1: 82-6. 

Sou IN, B., and STERIADE, D. (1986), 'On Geminates', LI, 17: 691-744. 

SCHVII.KI INC,, S. F. (1976), Aspect* of English Sentence Sires* (Austin, Tex.: Texas 

University Press). 

SCHOUTEN, M. E. H. (1981), 'Het verschil tussen hot en hou, een vergeefse speurtocht', 

Ntg., 74: 537-46. 

SCHULTINK, H. (1964), 'De bouw van nieuwvormingen met her-'. Tijdschrift voor 

Nederlandse Taal- en Letterkunde, 80: 151-84. 

- (1979), 'Reacties op "stress clash'", Sp., 8: 195-208. 

- (1980), 'Boundaries, word classes and the accentuation of derived words in 

Dutch', in Zonneveld, Van Coetsem, and Robinson (1980), 205-22. 

SELKIRK, E. O. (1982), 'The Syllable', in Van der Hulst and Smith (1982), part II: 

337-84.

199 

— ( 1984«), 'On the Major Class Features and Syllable Theory', in M. Aronott and 

R. Oehrle (eds.), Language Sound Structure (Cambridge, Mass.: MIT Press), 107-36. 

— (1984/7), Phonology and Synta\. I hi' Relation between Sound and Structure. 

Cambridge, Mass.: MIT Press. 

- (1986), 'On Derived Domains in Sentence Phonology', Ph., 3: 371-405. 

SHANNON, TH. F. ( 1991 ), 'On the Relation between Morphology and Syllable Structure: 

Universal Preference Laws in Dutch', in Shannon and Snapper (1991), 173-205. 

— and SNAPPER, J. P. (eds.) (1991), The Berkeley ('(inference on Dutch Linguistics 

1989 (Lanham, Md.: University Press of America). 

Sus, I. H. (1985), 'The Voiced-Voiceless Distinction and Assimilation of Voice in 

Dutch'. Katholieke Universiteit Nijmegen, Ph.D. dissertation. 

— and COHEN, A. (1969), 'On the Complex Regulating the Voiced-Voiceless 

Distinction', Language and Speech, 12: 80-102, 137-55. 

and HEIIGTEN, M. VAN (1989), 'Voiced-Voiceless Distinction in Dutch Fricatives', 

in H. Bcnnis and A. van Kemenade (eds.), LIN 1989 (Dordrecht: Foris), 

123-32. 

SIOOTWEG, A. (198K), 'Metrical Prominence and Syllable Duration', in P. Coopmans 

and A. Hulk (eds.), LIN 1988 (Dordrecht: Foris), 139-48. 

SLUIJTER, A. (1992), 'Lexical Stress and Focus Distribution as Determinants of 

Temporal Structure', in R. Bok-Bennema and R. van Hout (eds.), LIN 1992 

(Amsterdam: Benjamins), 247-60. 

SIIYTKKS, W. A. M. (1992), 'Representing Diphthongs', Katholieke Universiteit 

Nijmegen, Ph. D. dissertation. 

SMITH, N. (1976), '-Aar', LB, 65: 485-96. 

— BOI.OC.NESI, B., LEEUW, F. VAN DER, RUTTEN, J., and Wir. H. DE (1989), 'Apropos 

of the Dutch Vowel System, 21 Years On', in H. Bennis and A. van Kemenade (eds.), 

LIN 1989 (Dordrecht: Foris), 133^42. 

SPENCER, A. (1991), Morphological Theory (Oxford: Blackwell). 

STIKIADE, D. (1982), Greek Prosodies and the Nature of Syllabification, MIT, Ph.D. 

dissertation (repr. New York: Garland Press. 1988). 

- (1987), 'Redundant Values'. CLS Parasession, 339-62. 

-(1991), 'Syllables', in Bright (1991), iv, 106-11. 

SIKI.IKSTKA, N., and (ÏRAAI-, TJ. in (1979), 'Het kenmerk [+kort] bij hoge gespannen 

vocalen', Taalkundig Bulletin van liet Nederlands Instituut Groningen, 9: 40-7. 

SIUÏTERHEIM, C. F. P. (1978), 'Enkele grepen uit de sjwa-problematiek',G/t>f, 1: 213-34. 

TIIKSMA, P. M. (1985), Frisian Reference Grammar (Dordrecht: Foris). 

T(K)RN, M. C. VANDEN (1982), 'Tendenzen bij de beregeling van de verbindingsklank in 

nominale samenstellingen', Ntg., 75: 24-33, 153-60. 

lors, (i. A. J. ( 1974), ' Assimilation of Voice in Dutch. A Generative Approach', LB 63: 

143-51. 

TROMMELEN, M. ( 1983). 'Phonological Properties of the Dutch Velar Nasal', in Van den 

Broecke, Van Heuven, and Zonneveld (eds.), 259-69. 

(1984), The Syllable in Dutch (Dordrecht: Foris). 

— and ZoMvii viii). W. ( 1979), Inleiding in de generatieve fonologit (Muiderberg: 

Coutinho). 

( 1987), 'On ie, uu and tie: Dutch Vowel Length Re-examined', in G. de

20O 

REFERENCES 

Haan and W. Zonneveld (eds.), Formal Parameters of Generative Grammar, III, 

Yearbook 1987 (Dordrecht: ICG Printing), 71-82. 

(1989a), Klemtoon en metrische fonologie (Muiderberg: Coutinho). 

(19896), 'Stress, Diphthongs, r in Dutch', in H. Bennis and A. van 

Kemenade (eds.), LIN 1989 (Dordrecht: Foris), 143-52. 

VENNEMANN, TH. (1972), 'Rule Inversion', Lingua, 29: 209-42. 

VERHOEVEN, G. (1974), 'Onregelmatigheid van klankveranderingen als gevolg van 

lexicale geleidelijkheid', Sp., 4: 29-41. 

VIJVER, R. F. H. E. VAN DE (1993), 'A Main Stress First Approach to Metrical Stress 

Rules', University of Leiden MA thesis. 

VISCH, E. A. M. (1989), A Metrical Theory of Rhythmic Stress Phenomena (Dordrecht: 

Foris). 

VOGEL, I., and SCALISE, S. (1982), 'Secondary Stress in Italian', Lingua, 58: 213-42. 

VRIES, J. W. DE (1975), Lexicale morfologie van het werkwoord in modern Nederlands 

(Leiden: Universitaire Pers). 

et al. (1974), 'De slot-t in consonantclusters te Leiden: een sociolinguistisch 

onderzoek'. Forum der Letteren, 15: 235-49. 

WANG, W. (1968), 'Vowel Features, Paired Variables and the English Vowel Shift', Lg., 

44: 694-708. 

WESTER, J. (1985), 'Language Technology as Linguistics: A Phonological Case Study of 

Dutch Spelling', in H. Bennis and F. Beukema (eds.), L1N 1985 (Dordrecht: Foris), 

205-12. 

WIESE, R. (1996), The Phonology of German (Oxford: Clarendon Press). 

WIJK, N. VAN (1939), Phonologie. Een hoofdstuk uit de structurele taalwetenschap ('s 

Gravenhage: Marti nus Nijhoff). 

WILLIAMS. E. (1981), 'On the Notions "lexically related" and "head of a word'", LI, 

12: 245-74. 

WISSING, D. P. (1982), Algemene en Afrikaanse generatiewe fonologie (Johannesburg: 

Macmillan). 

YIP, M. (1991), 'Coronals, Consonant Clusters, and the Coda Condition', in Paradis and 

Prunet (1991), 61-78. 

ZEC, D. (1988), 'Sonority Constraints on Syllable Structure', Stanford University Ph.D. 

dissertation. 

— and INKELAS, S. (1990), 'Prosodically Constrained Syntax', in Inkelas and Zee 

(1990), 365-78. 

ZONNEVELD, R. M. VAN (1985), 'Word Rhythm and the Janus Syllable', in Van der Hulst 

and Smith (1985), 133-42. 

ZONNEVELD, W. (1978), A Formal Theory of Exceptions in Generative Phonology 


— (1983), 'Lexical and Phonological Properties of Dutch Voicing Assimilation', 

in Van den Broecke, Van Heuven, and Zonneveld (1983), 297-312. 

and TROMMELEN, M. (1980), 'Egg, Onion, Ouch. On the Representation of 

Dutch Diphthongs', in Zonneveld et al. (1980), 265-92. 

COETSEM, F. VAN, and ROBINSON, O. W. (eds.) (1980), Studies in Dutch Phonology 

(The Hague: Nijhoff). 

ZWAARDEMAKER, H., and EIJKMAN, L. P. H. (1928), Leerboek der Phonetiek (Haarlem: De 

Erven Bohn).

REFERENCES 201 

ZWART, J. W. (1992), 'Notes on Clitics in Dutch', paper presented at the ESF workshop 

on clitics, Lund, 22 May 1992. 

ZWICKY, A. M. (1972), 'On Casual Speech', CLS, 8: 600-15. 

(1977), On Clitics (Bloomington, Ind.: IULC).

INDEX 

Afrikaans 2 

ambisyllabicity 32. 131, 181 

anterior l l 

apical l l 

appendix 26-9, 49, 60, 121, 127, 128 

approxinianl 8, 10. 25, 38, 43 

Arabic 98 

aspirated K) 

bilabial X 

bracketing paradox 

casual speech 125 6 

Chomsky-adjunction 145, 170-73, 177 

circumfix 48 

clitic 47-9, 165-80 

climbing 168 

directional 174 

special 167 

Clitic Group 143, 170 n. 

coda 23 

constraint 40 3 

compensatory lengthening 149 

complex segment 39 n., 41, 128 

compound 49-52 

adjectival 118-19 

tonnai 1 16 

nominal 115 18 

prosodie 59 

prepositional 1 19 

verbal 117 

Compound Stress Rule I 15-23 

conjunction reduction 50-2 

Consonant Cluster Condition 44-5 

constraint: 

cross-over I OK 

cross-syllabic 44-5 

sequential 43-6 

pansyllabic 42-3 

continuant 10 

contoid 10 

contour segment 137 

contraction 174, 178-80 

( 'oronal 11 

C V rule 30-1 

Czech 132 n. 

/d/-deletion 91-2 

degemination 68-9, 151, 186 

determiner 48. 165-6, 176 

diaeresis 184 

diminutive 15, 48. 57-8, 69-73, 140, 147, 

152, 154. 186 

/d/-insertion 74—5 

diphthong 5. 7, 18-19, 33-4, 44, 87, 90-1, 

106, 131, 167, 183 5 

diphthongi/.ation 15. 100 

distributed 1 1 

Dominance 145 

Dorsal 11-12 

downstep I 19 

/d/-weakening 90-2 

Elsewhere Principle 59. 71 n.. 75. 104. 123 

enclisis. sec enclitic 

enclitic 170-9 

English 44, 46, 50 n., 108-10. 124 n., 

130 n., 132 n., 144, 153-4, 180 n. 

extrametricality 46 n., 102 105 

Extrametricality Rule 102 

last speech rule 124, 137 

feature: 

binary 9 

geometry 9 10 

laryngeal 10 

major class 9 

manner 10 

monovalent 10 n. 

percolation 76, 124 

privative 10 n., I 1 n.. 33, 79 n. 

supralaryngeal 10-12 

Final Devoicing 32-3, 55-7. 59-60. 92-3, 

174-5 

focus 155-9 

foot 98, 101. 134, 143 n. 

French 6, 40 n.. 99-100, 103 

frequency effect 130 

Fricative Voice Assimilation 146-7, 173^4, 

177-8 

Frisian 2 

gapping 50 J 

geminate 28 9. 58. 6.3 n.. 182-3, 186 

partial 127 

German 1-2, 27, 65, 94, 175

204 INDEX 

Germanic 99-100, 103-04 

glide 13, 33, 66-7 

glottal stop insertion 65, 151 

grid 98 

Hammock Principle 105, 108, 164 

hiatus rule 65-8 

Homorganic Glide Insertion 65-7, 91, 120, 

138, 150-1, 165-6, 178, 187 

hyphenation 187-8 

Iambic Reversal 162-3 

/iAdeletion 138-9 

Indonesian 50 n., 72 

Intermediate Phrase 144 n. 

Intonational Phrase 144-7 

intrusive stop 137 

Italian 108 n., 170 n. 

Japanese 98, 144, 171 n. 

Labial 12 

labiodental 11 

laminai 11 

laryngeal 10 

Laryngeal Spreading 62-4, 187 

lateral 10, II n. 

Latin 74, 78, 99, 103^* 

Learned Vowel Backing 18, 54, 77 

level ordering 124 

lexical diffusion 89 

Lexical Phonology 54 

licensing 32-3, 60, 63 

parasitic 36 n., 63 

linking 127-8 

Linking Constraint 32 n., 63 n. 

linking phoneme 51, 86 

Main Stress Rule 99-105, 109-10, 112-13, 

I 15, 120, 124 

masking effect 125, 154 

Maximal Onset 31 

Minimal Rhyme Constraint 25, 26, 31, 94 

monitoring 125, 126, 141, 147 

Morpheme Structure Condition 34, 43, 45 

MP-rule, see morpholexical rule 

nasal: 

coronal 64-5, 139-^tl, 148-9 

velar 35-6, 40, 43^t, 80, 92 

Nasal Assimilation 57, 64-5, 71, 78, 147-8, 

186 

Nasal Deletion 149 

native 75-6, 86-93 

/n/-deletion 139^1, 148-50, 152, 165, 172 

/^-insertion 166, 171, 178 

non-native 75-6, 78, 80-6 

Noun Phrase Stress Rule 117-18, 123 

nucleus 23 

Obligatory Contour Principle 14 

obstruent 9, 10 

onset 23, 26 

constraint 35-9 

Optimal Grid Principle 99, 102, 105-6, 108, 

112, 116-17, 120, 124, 133 

Optimal Grid Rule, see Optimal Grid 

Principle 

palatalization 95, 151-2 

past participle 63 

past tense 61-2, 91-2 

Phonological Phrase 143-6 

pitch accent 97 

Place Constraint 42 

Polish 50 n., 60, 108, 175 

prefix 48-51 

boundary 30, 145 

native 120-3 

non-native 119-20 

prepositions 119, 122, 177-8 

Prevocalic Schwa Deletion 49, 54, 57, 62, 

67-8, 145, 150-1, 165-6, 172, 178, 187 

proclisis, see proclitic 

proclitic 170^t, 177 

Progressive Voice Assimilation 59-60 

Prosodie Hierarchy 143 

prosodie incorporation 170 

P-rule, see automatic phonological rule 

quantity sensitive, sec syllable weight 

Regressive Voice Assimilation 58-61, 146, 

176, 186 

resyllabification 55-6, 165, 171-2, 174 

retiming 137, 139, 142, 149, 154 

rhyme 23 

constraint 33-5 

rule: 

allophonic 93-5 

automatic phonological 57 

cyclic 56-7, 77, 82-3, 109-10 

morpholexical 54, 57, 58, 69-93 

postlexical 126 

typology 185-6 

variable 126, 132, 146, 153 

Sanskrit 128 

schwa 5, 19-20, 24, 26, 46 n., 47-9, 74, 86 

-deletion 72, 89, 125-6, 128-30 

-epenthesis 124, 127-8 

-insertion 71, 84 

Schwa Restriction 99 

segment structure rule 12 

sentence accent 154-61, 166 

Sentence Accent Assignment Rule 156-9

separable complex verb 114, 157 

Serbo-Croatian 174 

sonorant 9-10 

Sonority Ranking Hierarchy 25 

Sonority Sequencing Generalization 24, 27, 

129 

Spanish 108 

speech rate 125-6 

Stray Consonant Adjunction 27 n. 

stress 96 

grid 98 

main 99-104 

primary 97 

secondary 97-9, 101 n., 105-8 

clash 99, 109, 115 

Stress Erasure Convention 109, 112-13, 

124 

Stress Retraction 161-2 

Strict Cycle Condition 57 

Strict Layer Hypothesis 144-5, 170-1 

structure preservation 94 n., 126 n. 

suffix 48-51 

cohering 49-51, 111-12 

competing 74 

cyclic 110, 113-14 

native 111-14 

non-cohering 49-51, 111-12 

non-cyclic 112 

non-native 75-6, 110 

stress-bearing 113 

stress-neutral 111-13 

stress-shifting 114-15 

syllabicity 24 

syllabification 29-33 

continuous 35 

direction of 31 

syllable 22-6 

contact 31 n., 45-6, 86 

stray 134 

weight 101 n., 106-8, 153 n. 

INDEX 205 

Syllable-final Devoicing, see Final 

Devoicing 

/t/-deletion 72-3, 152^ 

tone: 

boundary 160 

contour 160 

Trochaic Reversal 99, 107, 116-17, 161 

truncation rule 85 

underspecification 12-13, 74 n. 

Uniform Applicability Condition 32, 36, 60, 

63 

Utterance 143 

Velar Nasal Constraint 36 

via rule 79, 89 

vocoid 13, 25, 33 

voice, see laryngeal 

voice assimilation, see Fricative Voice 

Assimilation; Progressive Voice 

Assimilation; Regressive Voice 

Assimilation; Word-internal Devoicing 

vowel: 

deletion 84 

duration 5, 97 

height 16-18 

length 5, 13-16 

lengthening 18, 72, 80-3, 87-8, 93^ 

marginal 6, 16, 17 

nasal 6 

nasalized 148-50 

reduction 14, 85, 87, 106-7, 109, 113, 

120, 125-6, 130-5, 167 

shortening 81 n., 136-7 

word: 

grammatical 30 

prosodie 27-9, 37, 47-52, 111, 131 

Word-internal Devoicing 177 

word-level rule 55, 94-5, 174-5

The Phonology of Dutch

Create successful ePaper yourself

Delete template?

Save as template?