• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Final Devoicing
quickinfo

Take the bare adjective goed [ɡuət] good. It ends in [t], whereas goede [ɡuədə] good (inflected) has [d]. This kind of alternation between a voiceless and a voiced realization holds for all obstruents, the only class of consonants in Frisian for which voicing has distinctive value. The devoicing of obstruents in word-final position, called Final Devoicing, is the subject of this section.

readmore

Take the bare adjective goed [ɡuət] good. It ends in [t], whereas goede [ɡuədə] good (inflected) has [d]. This kind of alternation between a voiceless and a voiced realization holds for all obstruents, as shown by the examples in the table below:

Table 1: Examples of the alternation between a voiceless and a voiced realization of obstruents
With a voiceless obstruent With a voiced obstruent Translation
slaab [sla:p] slabben [slabm̩] bib-bibs
goed [ɡuət] goede [ɡuədə] good-good (inflected)
dief [diəf] dieverij [diəvərɛj] thief-theft
tsiis [tsi:s] tsizich [tsi:zəx] cheese-cheesy
dreech [dre:x] dregens [dre:ɣə̃s] difficult-difficulty
extra

Due to the fact that /ɡ/ does not occur in word-final position (see the obstruents: the plosives), there are no word pairs showing the alternation [k] ~ [ɡ].

In principle, this alternation can be accounted for in two ways:

  • by assuming underlying representations ending in a voiceless obstruent, which shows up as voiced in intervocalic position (for instance: /ɡu.ət/ [ɡu.ət], /ɡu.ət+ə/ [ɡu.ədə]);
  • by assuming underlying representations ending in a voiced obstruent, which shows up as voiceless in word-final position (for instance: /ɡu.əd/ [ɡu.ət], /ɡu.əd+ə/ [ɡu.ədə]).

On the one hand, it seems quite 'natural' for a voiceless obstruent to become voiced when occurring between two vowels. Although it does not constitute a general phonological pattern in Frisian, there are examples of words which show a dialectal alternation between an intervocalic, word-medial, voiceless and voiced obstruent, see the following table:

Table 2: Examples of words with a voiceless or a voiced obstruent in intervocal position
With a voiceless obstruent With a voiced obstruent Translation
skûtel /skutəl/ skûdel /skudəl/ dish
skûteldoek skûdeldoek dish cloth
skûtelwaskje skûdelwaskje wash the dishes
wantputtings /-pøtɪŋs/ wantpuddings /-pødɪŋs/ futtock shroud
lapperel /lapərɛl/ labberel /labərɛl/ dirt road, sloping upwards against a sea wall
appelear(je) /apəljɛr/ abbelear(je) /abəljɛr/ to object, to protest
rappelemint /rapələmɪnt/ rabbelemint /rabələmɪnt/ reprimand, rebuke
stipelear(je) /stipəljɛr/ stibelearje /stibəljɛr/ to support

On the other hand, that an obstruent is realized as voiceless in word-final position seems to be equally 'natural', for this holds for quite a few (also unrelated) languages. At first sight then both accounts above seem to have good credentials.

Obstruent voicing in intervocalic position is not obligatory in Frisian or, put differently, there is not a ban on intervocalic voiceless obstruents. Suppose, alternating word-final obstruents are assumed to be voiceless in underlying representation. This is unproblematic as far as simplex words are concerned. However, in case the words are made longer through inflection or derivation, a problem crops up, for the obstruents in question surface as voiceless in some words and as voiced in others, which is exemplified in this table:

Table 3: Examples of words with a voiced or a voiceless obstruent in longer forms
With a voiceless obstruent With a voiced obstruent
leat [lɪət] ~ leaten [lɪətn̩] (off)shoot - (off)shoots lead [lɪət] ~ leaden [lɪədn̩] lead - lead (made of lead)
akkoart [akoət] ~ akkoarten [akoətn̩] agreement - agreements akkoard [akoət] ~ akkoarden [akoədn̩] chord - chords
wach [vax] ~ wache [vaxə] watchful - watchful (inflected) slach [slax] ~ slaggen [slaɣən] blow, stroke - blows, strokes

To account for the alternation at hand then some words with voiceless final obstruents would have to be marked for their final obstruent showing up as voiced in intervocalic position. This is a laborious and non-generalizing approach.

Assuming the alternating word-final obstruents to be voiced in underlying representation does not meet with this problem. There is a ban on the occurrence of word-final voiced obstruents in surface representation. Word-final obstruent devoicing therefore can be accounted for by the following, maximally general constraint:

word-final obstruent constraint
a word-final obstruent shows up as voiceless
Underlyingly voiceless word-final obstruents are affected by the devoicing process in a vacuous way.
extra

Final Devoicing is inherently contradictory, for whereas it is a process of strengthening ‒ voiceless obstruents are the strongest, least vowel-like type of consonant ‒ it occurs in word-final position, in which consonants are prone to lenition.

extra

The effects of Final Devoicing are partly reflected in the spelling.

An underlyingly voiced plosive is always written with the spelling sign for the voiced variant (as is also clear from the first table of this topic): slaab bib and goed good, though realized with final [p] and [t], are written with final b and d, in accordance with the underlying representations /sla:b/ and /ɡuəd/.

An underlyingly voiced fricative, on the other hand, is written with the spelling sign for the voiced variant in intervocalic and with the one for the voiceless variant in word-final position (see also the first table of this topic): dief thief ~ dieverij theft, tsiis cheese ~ tsizich cheesy, dreech difficult ~ dregens difficulty, with the spelling sign pairs f~ v, s~ z, and ch~ g, respectively.

The same difference between plosives and fricatives is made in Dutch orthography, albeit that the velar fricative which alternates between a voiceless and a voiced realization is always written with the spelling sign for the voiced variant, viz. g.

See the following topics for more on the process:

References
    printreport errorcite