• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
Word-final sequences of three consonants

This topic is devoted to word-final sequences of three consonants, also in relation to their (word-initial) mirror images. There appear to be severe restrictions on their distribution, while there is also a clear difference between the possibilities in word-final and word-initial position.


There are five word-initial sequences of three consonants (see onset: sequences of more than two consonants); together with their mirror images they are enumerated in the table below:

Table 1: The word-initial sequences of three consonants and their mirror images
Word-initial sequence Mirror image
/skr/ /rks/
/spr/ /rps/
/str/ /rts/
/skl/ /lks/
/spl/ /lps/

Not all mirror images of the word-initial sequences occur in word-final position. The ones that do are /-lks/, /-rks/ and /-rps/. The table below gives an exhaustive list of the words they are part of:

Table 2
Ending in /lks/ Ending in /rks/ Ending in /rps/
daalks /da:lks/ immediately piterbjirks /pitərbjɪrks/ odd, strange korps /kɔrps/ corps
binnenwurks /bɪnənvørks/ on the inside (from binnen#wurk indoor work)
bûtenwurks /butənvørks/ outside (from bûten#wurk outdoor work)
The word daalks is an adapted form of Dutch dadelijk /dadə+lək/, so originally it is a derived form. This also holds for the three words ending in /-rks/. The word korps is a loanword. These sequences therefore do not occur in simplex native words, which is indicative of their scarcity.

On the face of it, the mirror image of /str-/ occurs in words like larts /lats/ lark, l(j)urts /l(j)øts/ lark, koarts /koəts/ fever, and koerts /kuəts/ course. Although in these native words <r> is still written, it has undergone a historical process of deletion (see /r/-deletion in simplex words), so it may no longer be part of the underlying representation of the words in question. This, however, is not the case with /r/ in the loanwords arts /arts/ doctor and erts /ɛrts/ ore, where it is always realized.

An overview of the three-segmental word-final sequences without a word-initial mirror image is provided in (1):

Example 1

Examples of words with a three-segmental word-final sequence without a word-initial mirror image
a. /-lsk/
falsk /fɔlsk/ false
gelsk /ɡɛlsk/ rancid, rank
b. /-ŋks/
flanks /flaŋks/ (in: yn 'e flanks nei towards, in the direction of )
sfinks /sfɪŋks/ sphinx
links /lɪŋks/ left(-hand)
lynks /liŋks/ lynx
c. /-nsk/
minske /me:nskə/ human being
winsk /ve:nsk/ wish
d. /-lts/
elts /ɛlts/ each (one)
folts /folts/ people
kwelts /kwɛlts/ (in: kwelts gean to limp )
lilts /lɪlts/ angry
melts /mɛlts/ milch (in: in melke ko a milch cow )
palts /pɔlts/ Palatinate
Drylts /drilts/ (name of) one of the eleven Frisian cities
e. /-lst/
aalst /a:lst/ absinthe
felst /fɛlst/ bog ore
hulst /hølst/ holly, ilex
f. /-nts/
hoants /vwants/ ruff
mients /miənts/ knot
muonts /mwonts/ monk
mûnts /munts/ monk
wants /vɔnts/ bug
g. /-nst/
aanst /a:nst/ at once, right away
hynst /hi:nst/ stallion
h. /-rst/
barst /bast/ crack, burst
boarst /bwast/ breast
hjerst /jɛst/ autumn
toarst /toəst/ thirst
i. /-xts/
rjochts /rjoxts/ right(-hand)
j. /-fts/
lofts /lofts/ left(-hand)
k. /-kst/
takst /takst/ regular, habitual quantity
drekst /drɛkst/ immediately, at once
tekst /tɛkst/ text
gewikst /ɡəvɪkst/ strong
gewykst /ɡəvikst/ shrewd, smart; robust, sturdy
l. /-pst/
gewûpst /ɡəvupst/ robust, sturdy
[hide extra information]

The words in (1c) and (1g) end in /-ns{k/t}/, which is why the nasal /n/ is not realized here (see vowel nasalization).

The word-final sequences in (1) would have the word-initial mirror images in (2):

Example 2

The three-segmental word-initial miror images of the sequences in (1)

Consonant sequences must be in conformity with the Resolvability Constraint, which holds that the occurrence of a consonant sequence implies the independent occurrence of its constituting parts (see onset: sequences of more than two consonants). A three-consonantal sequence is in accordance with this constraint, if it can be split up into two biconsonantal sequences. So, since the onset sequences /tx-/, /tf-/, /tl-/, and /tn-/ are not allowed, /stx-/, /stf-/, /stl-/, and /stn-/ are not either. The cluster /ts-/ can only precede the close vowel /i/ or the glide /j/ (see the /ts/ Constraint), which implies the non-occurrence of /tsl-/, /tsn-/, /tsr-/, /tsk-/, and /tsp-/. That /s/ is extra-syllabic in /sl-/ and /sn-/ (see extra-syllabic consonants) explains the non-occurrence of /ksl-/ and /ksn-/, for a word can have no more than one extrasyllabic segment at a time. Likewise, the fact that /k/ is extrasyllabic in /kn-/ (see onset: sequences of obstruients and nasals) renders the sequence /skn-/ impossible.

[hide extra information]

/skn-/, however, does occur, although only in the plural nouns sknikel(s)blêden /sknikəl(s)#blɛ:d+ən/ (< sk [ə]nikel(s)blêden < sk [ər]ernikel(s)blêden < sk [ar]nikel(s)blêden) ground ivy and sknieren /skniər+ən/ (< sk [ə]nieren < sk [ər]nieren < sk [ar]nieren) hinges, next to which stand the more common forms skarnikel(s)blêden /skarnikəl(s)#blɛ:d+ən/ and skarnieren /skarniər+ən/.

As to the three-consonantal sequences, there is thus a clear difference between the possibilities in word-initial and word-final position.

There are two asymmetries concerning the sequences in (1). In the first place, the coronal /s/ − which is part of all these clusters, either as final or prefinal element − is preceded or followed by a voiceless plosive, with which it forms a complex segment (see complex segments). The labial /p/, however, does not take part (with the exception of the loanword gotspe /ɡɔtspə/ chutzpah, effrontery). There is not a ready explanation for this asymmetry, though it may not be without significance that /-s/, /-t/, /-st/, and /-sk/ also have the status of a suffix, which /p/ has not.

In the second place, these sequences consist of a sonorant consonant followed by /-ts/, /-st/, /-ks/, or /-sk/. As complex segments, the latter can occur in word-final position. On the one hand, /-sk/ is only preceded by the (coronal) nasal /n/ and /-ks/ by the (velar) nasal /ŋ/. On the other hand, /-ts/ and /-st/ are preceded by /n/ and the liquids /l/ and /r/. The coronal sequences have a wider distribution than those which a non-coronal is part of. Both asymmetries then may have to do with the unmarked status of coronal segments vis-à-vis non-coronals.

Overviews of these sequences are given in (3) and (4) below:

Example 3

Sequences of a sonorant consonant + /sk / and /ks/
Example 4

Sequences of a sonorant consonant + /st/ and /ts/
a. /-nts/ and /-nst/
b. /-lts/ and /-lst/
c. /-lks/ and /-lsk/
d. /-rts/ and /-rst/
[hide extra information]

In the words from (3) and (4a) ending in /-ns{t/k}/ the nasal /n/ is not realized, due to Vowel Nasalization. Since it precedes a dental-alveolar consonant, r is mute in the words from (4d) (see /r/-deletion in simplex words).

[hide extra information]

It is assumed here that words like links and lynks in (3) have the velar nasal /ŋ/ in their underlying representation.

The sequences of an obstruent + /ts/ and /st/ also show an asymmetry. A plosive is followed by /st/ ( /-kst/ and /-pst/), a fricative by /ts/ ( /-fts/ and /-xts/), as exemplified in the table below:

Table 3: Sequences of an obstruent + /st/ and /ts/
/-xts/ ( /*-xst/) rjochts
/-fts/ ( /*-fst/) lofts
/-kst/ ( /*-kts/) takst
/-pst/ ( /*-pts/) gewûpst

Preferably, obstruents in a sequence do not agree in continuancy. Therefore, sequences of two plosives or fricatives are far less likely than plosive-fricative or fricative-plosive sequences (see onset: sequences of two obstruents). This accounts for the observed asymmetry.

The sequences /*-kts/, /*-pts/, /*-fst/, and /*-pts/ do not occur in simplex words, but they do in derived forms, like (do) sjochst /sjoɣ+st/ [sjoxst] you see; you look, (it) dreechst /dre:ɣ+st/ [dre:xst] sturdiest; most thorough, (do) draafst /dra:v+st/ [dra:fst] you run, (it) geefst /ge:v+st/ [ge:fst] healthiest; strongest, (wat) ûnkrekts /un+krɛkt+s/ [uŋkrɛkts] something incorrect, and (wat) benypts /bənipt+s/ [bənipts] something narrow-minded. This means, that the ban on /-kts/, /-pts/, /-fst/, and /-pts/ reflects a Morpheme Structure Constraint.

None of the above long word-final sequences have a high frequency. The Word Constraint, which constrains the size of simplex words, may be held responsible for this.

[hide extra information]

The sequences at hand are prone to simplification. So, the words takst, drekst; aalst, felst; palts and aanst have the variants taks, dreks/drekt; aals, fels; pals and aans. On the other hand, bisegmental word-final sequences ending in /s/ can gain an extra /t/, as in aldergelokst (next to and from aldergeloks /ɔlder+gəlok+s/ fortunately, luckily), selst (next to and from sels /sɛls/ self), and wylst (next to and from wyls /vils/ while).