• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Phonotactics at the word level

In the topic on the phonotactics at the syllable level the segmental restrictions and co-occurrence restrictions connected to the individual positions within a syllable were presented. The same basic restrictions apply to each syllable of more complex words; however, they are accompanied by additional constraints that regulate syllable contacts. In order to describe the phonotactic restrictions at a higher prosodic level, i.e. the word level, it is crucial to know how words are syllabified in Dutch.

[+]The Sonority Sequencing Generalization

Following a traditional account of syllables (Selkirk 1982; Booij 1995, Blevins 1995, see Goldsmith 2011 for a historical overview), a nucleus or peak of the syllable can only be filled by a vowel in Dutch (or by a nasal in some dialects). The nucleus can be preceded by up to three onset consonants, whereby the first one is restricted to /s/. The nucleus can be followed by up to two coda consonants plus maximally three coronal consonants, the so-called appendix, if the syllable is word-final. Consonant sequences in onsets and codas are regulated by the Sonority Sequencing Generalization, which is given below:

Sonority Sequencing Generalization
Between any member of a syllable and the syllable peak, a sonority rise or plateau must occur (Blevins 1995:210).
In other words, complex onsets with rising sonority contour, e.g. /pl-, tr-, kn-/, and complex codas with a falling or a level sonority contour, e.g. /-mp, -rt, -lk, -rn/, are preferred.

[+]Morphologically simplex words

In morphologically simplex or underived words with more than one syllable each word-medial consonant is assigned to a particular syllable as a result of the interaction of the following constraints:

Syllable Contact Constraint or the Maximal Onset Constraint
According to the syllable contact constraint, the contact between two syllables must be optimal, which implies that the end of the first syllable must be more sonorous than the beginning of the second syllable (Murray and Vennemann 1983; Clements 1990). Thus, a syllable contact as in (1a) is optimal since sonorants are more sonorous than stops. Furthermore, a syllabification such as /-.bl-/ would be preferred to /-b.l-/ since the syllable contact in the latter example is less optimal (cf. also (1b)). If the two consonants are of similar or equal sonority, variation in the syllabification can be observed (1c). The examples in (1d) illustrate that an additional factor might play a role - if the resulting onset cluster results in a cluster that cannot be found word-initially (or only word-initially in exceptional loanwords), informants tend to prefer splitting up the cluster. However, for all examples in (1b-d), variation in syllabification can be found.

a. aarde /ar.də/ earth
      groente /xrun.tə/ vegetables
b. iglo /i.xlo/ igloo
      duplo /dy.plo/ Duplo
c. klooster /klos.tər/ or /klo.stər/ cloister
      buuste /bys.tə/ or /by.stə/ bust
d. bauksiet /bɑuk.sit/ bauxite
      hypnose /hyp.no.zə/ hypnosis

As an alternative to the Syllable Contact Constraint we may postulate the Maximal Onset Constraint, which assumes that syllabification in Dutch proceeds from right to left. As a result, in any word-medial sequence of consonants, as many consonants as possible are assigned to the onset - as long as the resulting onset is a well-formed structure. This constraint yields the same syllabification for the examples in (1a) and (1b). The two sequences /-rd-/ and /-nt-/ do not qualify as well-formed onsets in Dutch and must be split by a syllable boundary. In contrast, the sequences /-xl-/ and /-pl-/ do occur as well-formed onsets in Dutch - compare glas /xlɑs/ glass and plak /plɑk/ slice. For the examples in (1c), only the second syllabification /-.st-/ would be allowed since /-st-/ is a well-formed onset in Dutch - compare stok /stɔk/ stick. The syllabification of the examples in (1d) would formally result in onsets such as /-.ks-/ and /-.pn-/; however, if these consonant sequences were regarded as unacceptable onset clusters in Dutch, a syllabification like /-k.s-/ and /-p.n-/ would be preferred instead. This fact might account for the observed variation in syllabification among speakers.

B-class vowel restriction or the Minimal Rhyme Constraint
All B-class vowels /ɑ, ɛ, ɪ, ʏ, ɔ/ must be followed by a coda consonant, i.e. they cannot occur in an open syllable (Moulton 1962, Van Oostendorp 1995, 2000). Translated to a length-based account of Dutch segments, in which B-class vowels are vowels occupying one position (in contrast to A-class vowels and diphthongs, which occupy two positions), the Minimal Rhyme Constraint(Booij 1995) requires rhymes to comprise at least two positions.

The interaction of the Syllable Contact Constraint / Maximal Onset Constraint and the B-class vowel restriction / Minimal Rhyme Constraint leads to the following syllabification of words containing B-class vowels:

Casper /kɑs.pər/ (name)
magneet /mɑx.net/ magnet
acne /ɑk.ne/ acne
paprika /pɑp.ri.ka/ (also /pa.pri.ka/ ) bell pepper, capsicum
deksel /dɛk.səl/ lid
hamster /hɑm.stər/ hamster

Ambisyllabic consonants
The fact that B-class vowels can only occur in closed syllables clashes with the Maximal Onset Constraint in words like:

emmer /ɛmər/ bucket
bakker /bɑkər/ baker
modder /mɔdər/ mud

The word-medial consonant must be processed as a coda consonant due to the B-class vowel in the first syllable; however, it must also be processed as the onset of the second syllable due to the constraint that enforces maximal onsets. One way to solve this problem is to postulate that the word-medial intervocalic consonant is ambisyllabic, i.e. it belongs to both syllables, as illustrated in the following figure for the word emmer bucket:

Figure 1
[click image to enlarge]

Additional evidence for the special status of the intervocalic consonant comes from the fact that the voiced stop /d/ in modder mud does not devoice although it is assigned to a coda position in the word and the domain for final devoicing in Dutch is the syllable.

[+]Morphologically complex words

Compounds and morphologically complex words behave slightly differently in their patterns of syllabification. Consider the following examples (cf. Booij 1995:29):

a. handappel [[hand][appel]] /hɑnd.a.pəl/ [ˈhɑntɑpəl] (hand) apple
b. werkloos [[werk][loos]] /ʋɛrk.los/ [ˈʋɛrklos] unemployed

The compound in (4a) consists of the two parts hand and appel. The first part ends in a consonant cluster and the second one begins with a vowel. Due to the onset maximization principle one would expect that the final /d/ in hand was parsed into the the onset position of the first syllable of appel. However, this is not what we find. Instead, the voiced stop is parsed into the coda of the first part and undergoes final devoicing. So, both parts of the compound hand and appel `form two different domains of prosodification, and hence each corresponds with a prosodic word because the syllables in a domain of syllabification form one prosodic word' (Booij 1995:30).

A similar effect can be seen in the adjective in (4b). Here, the suffix -loos is added to the noun werk. Since the vowel in werk is a B-class vowel, it needs to be followed by a coda consonant. However, following the onset maximization principle, the final consonant /k/ could easily be parsed into the onset position of the following syllable, thereby resulting in a well-formed Dutch onset cluster /kl-/. Again, this is not what we find. Instead, the velar stop is parsed into the complex coda of the first part.

What can be seen from the examples above is that each constituent of a compound forms a prosodic word of its own and phonotactic constraints apply within each prosodic word domain. Furthermore, Dutch provides a number of suffixes that form independent domains of syllabification and independent prosodic words of their own. What those suffixes have in common is that they contain at least one full vowel and are usually consonant-initial (with the exception of -achtig -like).

[+]Phonotactic constraints on possible words

As we have seen before, there is a whole family of conditions that constrain what a possible Dutch word can look like. First, phonotactic conditions at the syllable level govern whether a sequence of a vowel and consonant(s) make up an acceptable syllable in Dutch. Second, examples such as handappel and werkloos illustrate that prosodic word conditions have to be taken into account. A prosodic word contains minimally one full vowel and cannot begin with a schwa. Furthermore, the boundaries of prosodic words always coincide with syllable boundaries. Third, if a word contains more than two syllable peaks (nuclei), syllabic sequential constraints come into play. Principles like Syllable Contact / Maximal Onset Constraint and the B-class vowel restriction / Minimal Rhyme Constraint distribute the intervocalic consonants to the respective coda and onset positions of two consecutive syllables. Lastly, phonological rules can apply in the domain of the prosodic word, e.g. the process of degemination, which deletes one of two successive identical consonants that might be the result of a morphological process as illustrated in (5):

grootte [[groot]te] /ɣrot+tə/ [ˈɣrotə] size
  • Blevins, Juliette1995The Syllable in Phonological TheoryGoldsmith, John A., Riggle, Jason & Yu, Alan C. L. (eds.)The Handbook of Phonological TheoryBlackwell Publishers
  • Blevins, Juliette1995The Syllable in Phonological TheoryGoldsmith, John A., Riggle, Jason & Yu, Alan C. L. (eds.)The Handbook of Phonological TheoryBlackwell Publishers
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Clements, George N1990The role of the sonority cycle in core syllabificationPapers in Laboratory Phonology1Cambridge University Press283-333
  • Goldsmith, John A2011The syllableGoldsmith, John A., Riggle, Jason & Yu, Alan C. L. (eds.)The Handbook of Phonological TheoryWiley-Blackwell164-196
  • Moulton, William G1962The vowels of Dutch: phonetic and distributional classesLingua11294-312
  • Murray, R.W. & Vennemann, T1983Sound change and syllable structure in Germanic phonologyLanguage59514-528
  • Oostendorp, Marc van1995Vowel Quality and Phonological ProjectionTilburg UniversityThesis
  • Oostendorp, Marc van2000Phonological ProjectionNiemeyer
  • Selkirk, Elisabeth1982The syllableThe structure of phonological representations2337-383
printreport errorcite