• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Onset: sequences of two consonants
readmore
[+]Possible CC onset clusters

Dutch allows for complex onset consonant clusters consisting of two consonants. Figure 1 displays all attested combinations of CC-clusters that can be found in Dutch. Vertically, all segments are given that are possible as the first element of a CC-cluster; horizontally, all segments are given that are possible as the second element of a CC-cluster. The plus signs refer to clusters that are frequent; the plus signs in parentheses refer to clusters that occur only marginally in Dutch. These clusters are typically found in loanwords. No sign means that this sequence is not allowed in Dutch at all.

Figure 1
Table of all possible Dutch onset CC combinations.
[click image to enlarge]

Disregarding the marginal clusters occurring in loanwords (see figure 2), it becomes apparent that onset consonant clusters of 'native' Dutch words usually consist of an obstruent followed by a sonorant (see right upper part of figure 2). The sonorant is usually either a liquid /r, l/ or the labio-dentalapproximant /ʋ/. Sequences violating the OCP (Obligatory Contour Principle) with respect to place of articulation are prohibited (see shaded cells in figure 2). The occurrence of the coronalnasal /n/ as the second part of an onset cluster is quite restricted; it only follows velars and, in only a tiny set of examples, the labio-dental fricative /f/. Two striking exceptions can be noticed. First, the labio-dental approximant /ʋ/ is the only sonorant that occurs frequently in the first position of a consonant cluster. It can only precede the rhotic consonant /r/. Second, the sibilant /s/ is the only obstruent that frequently co-occurs with other obstruents and that disobeys OCP restrictions.

Figure 2
Table of onset CC combinations found in native Dutch words.
[click image to enlarge]

[+]Examples

Table 1 presents examples of each possible CC-cluster that can be found in Dutch. Infrequent clusters that can only be found in loanwords are marked as such.

Table 1
C1 segment sequence example
/p/ /pl-/ plug /plʏɣ/ plug
/pr-/ prijs /prɛis/ price
in loanwords:
/pj-/ piu /pju/ more, plus
/pn-/ pneumonie /pnø.mo.ni/ [pnø.mo.ˈni] pneumonia
/pf-/ Pfeiffer / Pfeifer /pfɛi.fər/ [ˈpfɛi.fər] (surname, probably of German origin)
/ps-/ psalm /psɑlm/ psalm
/pt-/ Ptolemeus /pto.lə.me.ʏs/ (ancient figure)
/pw-/ pointe /pwɛ̃t/ punch line
/b/ /bl-/ bloem /blum/ flower
/br-/ broek /bruk/ trousers
/t/ /tʋ-/ twee /tʋe/ two
/tr-/ troep /trup/ herd, rubbish
in loanwords:
/ts-/ tsaar /tsar/ czar
/tʃ-/ chip /tʃɪp/ chip
/tw-/ etui /eɪtwi/ etui
/d/ /dʋ-/ dwars /dʋɑrs/ contrary
/dr-/ droom /drom/ dream
/k/ /kʋ-/ kwal /kʋɑl/ jelly fish
/kl-/ klok /klɔk/ watch
/kr-/ krul /krʏl/ curl
/kn-/ kni /kni/ knee
in loanwords:
/kj-/ barbecue /bɑr.bə.kju/ BBQ
/ks-/ xylofoon /ksy.lo.fon/ [ksy.lo.ˈfon] xylophone
/g/ in loanwords:
/gʋ-/ pinguin /pɪŋ.gʋɪn/ [ˈpɪŋ.gʋɪn] penguin
/gl-/ neglige /nɛ.glɪ.ʒeɪ/ [ˌnɛ.glɪ.ˈʒeɪ] neglige
/gr-/ grill /grɪl/ grill
/f/ /fl-/ fluit /flœyt/ flute
/fr-/ framboos /frɑm.bos/ [frɑm.ˈbos] raspberry
/fn-/ fnuiken /fnœy.kən/ [ˈfnœy.kən] to weaken
in loanwords:
/fj-/ fjord /fjɔrd/ fjord
/v/ /vl-/ vlag /vlɑɣ/ flag
/vr-/ vraag /vrax/ question
in loanwords:
/vj-/ vieux /vjø/ old name for 'Dutch cognac'
/vw-/ voile /vwal/ veil
/s/ /sl-/ sla /sla/ lettuce
/sm-/ smaak /smak/ taste
/sn-/ snaar /snar/ string
/sf-/ sfeer /sfer/ atmosphere, mood
/sx-/ schoen /sxun/ shoe
/sp-/ spook /spok/ ghost
/st-/ stoel /stul/ chair
/sk-/ ski /ski/ ski
in loanwords:
/sʋ-/ suite /sʋit/ suite
/sr-/ Sri Lanka /srilɑŋka/ Sri Lanka
/z/ /zʋ-/ zwart /zʋɑrt/ black
/ʃ/ in loanwords:
/ʃʋ-/ sjwa /ʃʋa/ ə, schwa
/ʃl-/ schlemiel /ʃlemil/ shlemiel
/ʃm-/ schmink /ʃmiŋk/ make-up (colloq.)
/ʃn-/ schnabbel /ʃnɑbəl/ gig
/x/ /xl-/ glas /xlɑs/ glass
/xr-/ groen /xrun/ green
/xn-/ gnoe /xnu/ gnu
/m/ in loanwords:
/mw-/ moiré /mwɑr/ moiré pattern
/n/ in loanwords:
/nw-/ nuit /nwi/ night
/l/ in loanwords:
/lj-/ milieu /miljø/ [miˈljø] environment
/r/ in loanwords:
/rw-/ bavarois /ba.va.rwa/ bavarois
/ʋ/ /ʋr-/ wraak /ʋrak/ revenge

As can be seen in figures 1 and 2 and table 1 above, /h/ cannot form part of any cluster in Dutch, i.e. it only occurs as a singleton in syllable onsets. According to Booij (1995:36) this is due to the /h/-onset constraint, which says that "[a] branching onset may not dominate [+aspirated]".

The palatalglide /j/ can only occur as a singleton in syllable onsets or as the second member of a consonant cluster in Dutch.

[+]Combinations of two consonants in word-initial onset clusters

In onset consonant clusters of two consonants the following combinations are theoretically possible:

  • sonorant + obstruent
  • sonorant + sonorant
  • obstruent + obstruent
  • obstruent + sonorant.

[+]Sonorant + obstruent sequences

As can be seen in figures 1 and 2 above, the first combination is strictly ruled out in Dutch as it does not comply with the generally preferred increase in sonority (or, equivalently, the gradual decrease of consonantal strength/complexity) in syllable onsets.

[+]Sonorant + sonorant sequences

The second combination, sonorant + sonorant, is very restricted in Dutch. Apart from the sequence /ʋr/, all other sequences of this type are highly marginal and can only be found in loanwords. The sequence /ʋr/, however, is frequent and occurs in words like wrak /ʋrɑk/ wreck, wraak /ʋrak/ revenge, wroeten /ʋrutən/ to rummage or wringen /ʋrɪŋən/ to wring. Trommelen (1984:105) claims that in word-initial position this sequence is realized as [vr] with a voiced labio-dental fricative and therefore behaves in the same way as other word-initial obstruent + sonorant clusters.

[+]Obstruent + obstruent sequences

The third combination, obstruent + obstruent, is a rather exceptional type due to the fact that in these consonant clusters the sonority does not increase between the two elements of the syllable onset. These clusters can be separated into two types: /s/ + obstruent clusters and obstruent + obstruent clusters in which the first element is not an /s/. Clusters of the latter type, i.e. /pt-/, /pf-/, /ps-/, /ts-/, /tʃ-/ and /ks-/, are very infrequent and are solely found in a small set of loanwords. Generally, the obstruents do not need to agree in continuancy, whereas they must agree in voicing. In the description given here, these obstruent + obstruent sequences have been analysed as consonant clusters, i.e. consonant sequences occupying two syllabic positions. Alternatively, these sequences could be analysed as affricates. However, this analysis has been rejected here due to the fact that none of these obstruent + obstruent sequences can form part of even more complex consonant clusters like in Pflaume /p͜flau.mə/ plum (German cognate of Dutch pruim /prœʏm/).

The second type of obstruent + obstruent clusters, i.e. word-initial /s/ + obstruent clusters, occurs very frequently in Dutch. These clusters behave atypically for a number of reasons. First, as can be seen in the table above, it is only the segment /s/ that can cluster freely with all other (voiceless) obstruents (and almost all sonorants). Second, clusters such as /st-/, /sn-/ or /sl-/ in words like stom /stɔm/ dumb, stupid, sneu /snø/ pitiful or slim /slɪm/ smart illustrate that OCP restrictions regarding place of articulation can be violated by sC(C) - clusters. Third, in word-initial consonant clusters of three consonants the first position can exclusively be occupied by /s/ (see also Onset: sequences of more than two consonants). All of these facts point to a special status of the segment /s/ in any sC(C) - cluster. Phonotactic theories have accounted for the exceptional behaviour of /s/ by postulating a different syllable structure for this type of cluster. In doing so, the /s/ in sC(C) - clusters is allocated to a left appendix position (alternatively called extrametrical or (pre-)marginal position, with differences as to whether it is attached to the onset, syllable node or prosodic word node) or to a preceding syllable (depending on the syllable theory in use).

[+]Obstruent + sonorant sequences

The fourth combination, obstruent + sonorant, is the most frequent type of consonant clusters in Dutch. Obstruent-sonorant-sequences obey the universal preference for an increase in sonority within onset clusters. However, some restrictions can be found nevertheless.

Concerning /Cn-/, /Cl-/ and /Cr-/ onset clusters, a tendency for voiceless obstruents to precede one of the three sonorants can be noticed. Since sonorants are inherently voiced, i.e. they are not explicitly specified with the feature [voice], it is possible to claim a general preference for voice agreement for such sequences.

Furthermore, OCP-effects can be found. First, clusters of two coronals like /*tl-/, /*dl-/ and /*zl-/ are not allowed due to a violation of the OCP with respect to coronal place of articulation. Notice that the sequences /sl-/ and /ʃl-/ are allowed, which gives additional evidence for proposing a different syllabic structure for /s/ʃ + C/ clusters (see also Onset: sequences of more than two consonants). Second, clusters such as /*pʋ-/, /*bʋ-/, /*fʋ-/ and /*vʋ-/ are prohibited as a result of an OCP violation caused by their shared labial place of articulation.

Moreover, the rhotic consonant can cluster relatively freely with obstruents in Dutch, which might be an indication that the rhotic consonant lacks a specification for place of articulation. The only obstruent-rhotic-consonant-sequences that are not allowed are word-initial sibilant-rhotic-consonant-sequences such as /*zr-/ and /*ʃr-/. The sequence /sr-/ occurs only in recent, highly marginal loans (see table 1 above). Trommelen (1984:110-111, citingZonneveld, ms.) claims that there is a filter/constraint active in Dutch that prohibits sibilant-rhotic-consonant-sequences. According to Trommelen, when such a sequence occured in the history of Dutch, the plosive /t/ intruded leading to (semantically related) oppositions such as

Table 2
remmen restrain vs. stremmen coagulate
rekken stretch vs. strekken stretch
siroop syrup vs. stroop treacle
(cf. courant newspaper vs. krant newspaper).
The fact that Dutch does allow /sr-/ sequences in recent loans such as Sri Lanka /sri.lɑŋka/ Sri Lanka can be accounted for by postulating that the /s/ occupies the appendix or extrametrical position in analogy to /sl-/ onset clusters.

Finally, obstruent-glide-sequences are highly restricted. In word-initial onset clusters, the palatal glide follows (almost exclusively voiceless) obstruents only in a small set of loanwords (notice the contrast to word-medial onset clusters involving the diminutive suffix - see extra section below). The velar fricatives cannot be followed by any glide, i.e. /*xj-/, /*ɣj-/, /*xʋ-/ and /*ɣʋ-/.

extra
x

In the examples above we see that obstruent + glide consonant clusters such as /pj-/, /tj-/ or /kj-/ appear only marginally in word-initial syllable onsets. In contrast, in word-medial syllable onsets the very same clusters occur frequently in e.g. diminutives, in which the morpheme /-(t)jə/ is attached to the root and place assimilation applies. Consequently, these clusters can be found in words like boompje /bom.pjə/ tree-DIM small tree, treintje /trɛin.tjə/ train-DIM small train or koninkje /ko.nɪŋ.kjə/ king-DIM little king. These examples illustrate that there are differences in the restrictiveness of phonotactics on the syllable level and phonotactics at the word level.

References
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Trommelen, Mieke1984The Syllable in DutchDordrechtForis
  • Trommelen, Mieke1984The Syllable in DutchDordrechtForis
  • Trommelen, Mieke1984The Syllable in DutchDordrechtForis
  • Zonneveld, Wim1983Lexical and phonological properties of Dutch voicing assimilationvan den Broecke, M., van Heuven, V. & Zonneveld, W. (eds.)Sound Structures: Studies for Anthonie CohenDordrechtForis Publications
printreport errorcite