• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
Schwa deletion as a synchronic process: how to deal with hiatus in syntactic configurations

The ordering of words in syntax may result in a configuration of schwa + vowel, viz. one of vocalic hiatus. But whereas word-internal hiatus is strictly forbidden in Frisian, hiatus between words is not (though it may be a less preferred option). This topic argues that though schwa deletes here, it does not delete fully, thus preventing some phonological processes from applying.


If a word ending in schwa is followed by a vowel-initial word, word-final schwa may, but it need not delete. This means that in the sentences in (1) the combinations of words have two possible realizations.

Example 1

Examples of sentences with a combination of words which display vocalic hiatus
a. hoe let komt de sinne op at what time will the sun rise? [sɪnə ʔop] [sɪn op]
b. se hawwe de flage út they have hung out the flag [fla:ɣə ʔyt] [fla:ɣ yt]
c. de earste kear the first time [də ʔɪəstə] [d ɪəstə]
d. de oare kear some other time; next time [də ʔoərə] [d oərə]
[hide extra information]

According to Tiersma (1979:139), noun and adjective-final schwa does not delete as easily as does verb-final schwa. As to nouns, this seems to be contradicted by (1a,b) above.

The deletion of the schwa of the definite article de /də/ results in an unaffiliated consonant, viz. /d/. Incorporation into the following word saves /d/ from the fate of deletion. At the same time, the vowel-initial word acquires an onset, which is an improvement of its phonological structure. So, two birds are killed with one stone. The syllabification of the combination of the article de the and the inflected adjectives earste first and oare other then seems to be as indicated in (2):

Example 2

The syllabification of the combination of the definite article and the adjective in (1)

On closer scrutiny, however, these representations are untenable. The transcription [(doə)(rə)] suggests that d'oare and the verb doare dare are fully homophonous. But they are not, for there is a short period of silence between [d] and [oə] in d'oare, which is absent in doare. The fact that de oare is a combination of two words seems to have influence on its pronunciation.

This means that a different analysis of schwa deletion is called for here. The segmental (phonetic) content of schwa may be assumed to have been deleted, whereas its structural (phonological) position has remained, in line with the approach put forward in Berendsen (1986:75-84), in order to explain the perceptual difference between Dutch sentences like 'k ( /ək/) aas op een mooi Delfts blauw bord I have my eye on a beautiful Delft-blue plate and kaas op een mooi Delfts blauw bord cheese on a beautiful Delft-blue plate. It yields a more insightful analysis, if it is assumed that schwa is not fully deleted here.

The same analysis can be applied to cases like those in (1a,b) above and to the examples of pairs of non-homophonous words and phrases in (3):

Example 3

Pairs of non-homophonous words and phrases
(dat gers) meande er /mɪən+də ər/ (that grass) mowed he
(gers)meander /mɪən-d+ər/ (lawn) mower
(dat) helle er (op) /hɛl+ə ər/ (that) collected he
heller /hɛl+ər/ shriller; brighter
(hy) helle it (op) /hɛl+ə ət/ he collected it
(hy) hellet (op) /hɛl+ət/ he pulls in the fishing-rod

The words (gers)meander (3a), heller (3b) and hellet (3c) are not homophonous with the word combinations meande er (3a), helle er (3b), and helle it (3c). The latter are realized with a short period of silence between the words they consist of. Moreover, whereas the words (gers)meander and heller can be realized with a syllabic /r/ ( [r̩]), this is not possible with the phrases meande er and helle er. This is an indication that verb-final schwa cannot have been fully deleted here. The latter prevents /d/ and /l/ from becoming the onset of /-ər/, whilst it is a prerequisite for a syllabic sonorant consonant that it have an onset (see the onset condition).

The above may also shed light upon two other facts of connected speech. Take, in the first place, Regressive Voice Assimilation, the weaker version of which affects word-final fricatives (see regressive voice assimilation: type 2), illustrated by the examples in (4):

Example 4

Examples of Regressive Voice Assimilation affecting word-final fricatives
hast dy les al ynhelle? /lɛs ɔl/ [lɛz ʔɔl] have you already made up for that lesson?
ik trof ien /trof iən/ [trov ʔiən] I met someone

Riemersma (1979:67) notes that there is no assimilation in case the fricative precedes a word-final schwa, see the examples in (5):

Example 5

Examples of the non-application of Regressive Voice Assimilation with schwa-final words
hast dy flesse al omspield /flɛsə ɔl/ [flɛs ɔl] [*flɛz ʔɔl] have you already rinsed out that bottle?
se treffe ien /trɛfə iən/ [trɛf iən] [*trɛv ʔiən] they meet someone

Something similar, in the second place, is at stake with Final Devoicing, a process affecting word-final obstruents (see final devoicing), as shown by the examples in (6):

Example 6

Examples of Final Devoicing affecting noun-final obstruents
doe wie dat gesnob oer /ɡəsnob uər/ [ɡəsnop uər] then all that eating sweets was over
wat is der noch fan dat guod oer? /gwod uər/ [gwot uər] what has been left of that stuff?

There is no devoicing if the fricative precedes a word-final schwa, see the examples in (7):

Example 7

Examples of the non-application of Final Devocing with schwa-final words
se stieken de grutte dobbe oer /dobə uər/ [dob uər] [*dop uər] they crossed the big water hole (lit.); meaning: they crossed the Atlantic Ocean in order to emigrate to Canada or the United States
ik haw neat foar sa'n fodde oer /fodə uər/ [fod uər] [*fot uər] I do not want to give anything for such rubbish

These cases also yield to an analysis in which the segmental content of schwa deletes, whereas its structural position remains. The latter, in effect, causes the fricative and the vowel of the following word in (5) to remain non-adjacent, which prevents Regressive Voice Assimilation from applying; it also causes the voiced obstruent in (7) not to end up in word-final position, so that Final Devoicing cannot apply either.

  • Berendsen, Egon1986The Phonology of CliticizationUniversity of UtrechtThesis
  • Riemersma, Tr1979Sylabysjerring, nazzeljerring, assymyljerringLjouwertKoperative Utjowerij
  • Tiersma, Pieter M1979Aspects of the phonology of Frisian based on the language of GrouMeidielingen fan de stúdzjerjochting Frysk oan de Frije Universiteit yn Amsterdam4