• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
The phonological domain of Final Devoicing
quickinfo

The subject of this topic is the domain of application of Final Devoicing. Though it is clear that the process applies within the general phonological context of the word, there are also instances where it seems to be conditioned morphologically, that is, by certain suffixes.

readmore

Final Devoicing's domain of application is the word, either a free word or a member of a compound. Examples of the latter are given in (1):

1
Examples of Final Devoicing within the bounds of a compound
syd#line [sit-] sideline ; cf. side [sidə] side
breed#út [bre:t-] spread (out) ; cf. brede [bre:də] broad (inflected)
heech#hawwe [he:x-] esteem highly ; cf. heger [he:ɣər] higher
tiishollich [ti:s-] muddle-headed ; cf. tiizje [ti:zjə] be in a muddle

But matters are not always as straightforward as that. Take the words in (2), in which a voiceless plosive shows up in a word-internal position:

2
Examples of words with a voiceless plosive in word-internal position
abnormaal [(ap)(nɔr)(ma:l)] abnormal
admiraal [(at)(mi)(ra:l)] admiral
objekt [(op)(jɛkt)] object
subordinaasje [(søp)(ɔr)(di)(na:s)(jə)] subordination

In the above words, the voiceless plosives are the coda of the first, left-most syllable. Now, the end of a word always coincides with the end of a syllable, but not the other way around. The right generalization therefore seems to be that the phonological domain of application of Final Devoicinging is the syllable, not the word.

The words in (2) are Latin loans. The prepositions they contain − ab, ad, ob, and sub− function as 'learned' prefixes nowadays. Phonologically speaking, prefixes behave as separate words. This means that the voicelessness of the plosives here is in line with the view that Final Devoicing has the word as its domain of application.

extra

One may doubt whether devoicing is really at stake here at all. In saying so, one may be misled by orthography. Loanwords enter Frisian through mediation of Dutch (see Visser (2000)). This means, that the voiceless plosive in these words is loaned as well.

Word-final voiced obstruents correspond with voiceless obstruents in diminutives (see diminutive nouns, examples of which are given in (3):

3
Examples of voiceless obstruents in diminutives
slab+ke [slapkə] (small) bib ; cf. slabben [slabm̩] bibs
drúf+ke [dryfkə] small grape ; cf. druven [dry:vən] grapes
sys+ke [siskə] siskin ; cf. sizen [si:zn̩] siskins
pad+sje [patsjə] (small) path, track ; cf. paden [pa:dn̩] paths, tracks
each+je [ɪ.əxjə] small eye ; cf. eagen [ɪ.əɣən] eyes
brech+je [brɛxjə] small bridge ; cf. brêgen [brɛ:ɣən] bridges

This is a systematic fact of Frisian. There are nouns which are diminutives in a formal sense only, i.e. they have the characteristics of a diminutive, but there no longer is a synchronic base word which they can be related to. Though there are cases in which the orthography would have it otherwise, the obstruents preceding the diminutive suffix are always voiceless. This is illustrated in (4):

4
Examples of formal diminutives with voiceless obstruents
skroed+sje [skruətsjə] a fragile child
sech+je [sɛxjə] saying, say, tale
ferlech+je [fəlɛxjə] pretext, excuse
babbelegûch+jes [babələɡuxjəs] excuses; fuss
stadich+jes [sta:dəxjəs] slowly

The same phenomenon is seen in diminutive first names, which, as to their formal aspects, link up with diminutive nouns. Obstruents preceding the diminutive suffix show up as voiceless, irrespective of the voicing value of the final obstruent of the base name (see also Visser (2003)). Examples are given in (5):

5
Examples of diminutive names with voiceless obstruents
Wyp+ke [vipkə] ; next to Wibe [vi:bə]
Eap+ke [ɪəpkə] ; next to Eabe [ɪəbə]
Ruerd+ke [ryətkə] ; next to Ruerd [ryət] , cf. Ruerde [ryədə] fyts Ruerd's bike
Doet+sje [duətsjə] ; next to Doede [duədə]
Aach+je [a:xjə] ; next to Age [a:ɣə]
Ych+je [ixjə] ; next to Ige [i:ɣə]

In (3) - (5), the voiceless obstruent is in word-medial, syllable-final position, which seems to argue for the syllable as the phonological domain of application of Final Devoicinging.

These facts, however, are not the whole story. There are many names ending in -y ( [-i]), which is to be considered a kind of diminutive suffix ('pet suffix'), see Hoekstra (1998:86). Some examples are provided in (6):

6
Examples of names ending in -y
Japp+y [japi] ; next to Jabik [ja:bɪk]
Hep+y [he:pi] ; next to Heabeltsje [hɪ.əbəltsjə]
Sipp+y [sɪpi] ; next to Sibbeltsje [sɪbəltsjə]
App+y [ɔpi] ; next to Albert [ɔlbət] or Albertsje [ɔlbətsjə]

The medial obstruent in these names is voiceless. This cannot be accounted for by appealing to its syllable-final position, for [p] is either the onset of the second syllable or it is ambisyllabic.

As to this, also consider the following. Frisian has two suffixes -je ( /-jə/). On the one hand a diminutive suffix, used with nouns ending in /-ɣ(ə)/ (see diminutive nouns), on the other hand a (verbal) inflectional one, used with verbs of the second weak conjugation class (see paradigm of class II). Noun to verb conversion (see N > V) is quite common in Frisian, which gives rise to noun-diminutive and noun-verb pairs, examples of which are given in (7):

7
Examples of noun-diminutive and noun-verb pairs with the diminutive and verb ending in -je
each /ɪəɣ/ eye ; cf. eagen [ɪəɣən] eyes
eachje [ɪəxjə] small eye
eagje [ɪəɣjə] to look, to see
baarch /ba:rɣ/ pig ; cf. bargen [barɣən] pigs
barchje [barxjə] piglet
bargje [barɣjə] make a mess (of/on)
fuorge /fworɣə/ furrow; channel
fuorchje [fworxjə] small furrow, channel
fuorgje [fworɣjə] (make/plough) furrows, channels
herberge /hɛrbɛrɣə/ inn, tavern
herberchje [hɛrbɛrxjə] small inn, tavern
herbergje [hɛrbɛrɣjə] to accommodate

One may assume both suffixes to have the same phonological form. One may also assume that the initial glide /j/ of the suffix -je does not have any devoicing influence on the obstruent by which it is preceded, since regressive voice assimilation in Frisian only induces voicing (see regressive voice assimilation: type 1 and regressive voice assimilation: type 2). So, though the fricatives are in the same syllabic position, they are voiced when part of the verb and voiceless when part of the diminutive, which is an exceptionless pattern. The voicelessness of the obstruent in (6) and (7) then seems to be conditioned by the diminutive suffix as such. If this holds for the above cases, we may assume it to hold for all cases of diminution. Put differently, the systematic voicelessness of an obstruent before the diminutive suffix is conditioned by a morphological and not by a phonological factor.

extra

In Dutch, the diminutive suffix has five allomorphs, which are assumed to derive from the underlying representation /tjə/ (-tje) (see Booij (1995:68-73)). diminution thus proceeds along the lines of an item-and-process approach. In the spirit of this, one might assume the underlying representation of the Frisian diminutive suffix to be /tsjə/ (-tsje), in which case suffix-initial /t/ might play a role in the devoicing of the noun-final obstruent. However, since the allomorphs of the Frisian diminutive suffix all have their own domain, an item-and-arrangement approach is called for (see diminutive nouns). Besides, the diminutive suffix ('pet suffix') /i/ (-y) for names shows the same devoicing effect as the 'normal' diminutive suffixes, but it is highly unlikely to posit /tsjə/ as its underlying representation.

There are more cases of morphological conditioning of the voicelessness of obstruents. Stem-final fricatives which are voiced in underlying representation show up as voiceless before the suffix -lek /-lək/ (see -lik); this is exemplified in (8) (see also Hoekstra (1998:135)):

8
Examples of voiceless fricatives before the suffix -lik
erflik /ɛrv+lək/ [ɛrflək] hereditary ; cf. erve /ɛrv+ə/ [ɛrvə] to inherit
geryflik /ɡəri:v+lək/ [ɡəriflək] comfortable ; cf. gerive /ɡəri:v+ə/ [ɡəri:və] to suit the convenience of
yslik /i:z+lək/ [islək] hideous, dreadful ; cf. izich /i:z+əɣ/ [i:zəx] icy, freezing
pryslik /pri:z+lək/ [prislək] praiseworthy, laudable ; cf. lof#prizing /-pri:z+ɪŋ/ [pri:zɪŋ] eulogy, praise
wyslik /vi:z+lək/ [wislək] wisely ; cf. wizer /vi:z+ər/ [vi:zər] wiser
húslik /hy:z+lək/ [hyslək] domestic; hom(e)y ; cf. huzen /hy:z+ən/ [hyzn̩] houses
freeslik /fre:zə+lək/ [fre:slək] terrible, awful; horrible ; cf. freze /fre:zə/ [fre:zə] fear, fright
beweechlik /bəve:ɣ+lək/ [bəve:xlək] mobile; mov(e)able ; cf. bewege /bəve:ɣ+ə/ [bəve:ɣə] to move
eachlik /ɪəɣ+lək/ [ɪəxlək] attractive ; cf. eagje /ɪəɣ+jə/ [ɪəɣjə] to look nice/good/well

The suffix -lek has the schwa-initial variant -elik ( /-ələk/). Take geryflik [ɡəriflək] comfortable, which is syllabified as [(ɡə)(rif)(lək)], so with the fricative [f] in syllable-final position. In the variant form gerifelik [ɡərifelək], syllabified as [(ɡə)(ri)(fə)(lək)], on the other hand, [f] is in syllable-initial position, where it cannot possibly be subjected to Final Devoicing. Yet, both geryflik and gerifelik have a voiceless fricative. The same holds for words like erfelik [ɛrfələk] [(ɛr)(fə)(lək)] hereditary and iselik [isələk] [(i)(sə)(lək)] hideous, dreadful. This is another indication that the voicelessness of the fricative is triggered by the suffix, whether or not it begins with schwa.

The suffix -enis also forces a stem-final, underlyingly voiced fricative ‒ /v/ in this case ‒ to be voiceless at the surface, see (9):

9
Derivations with the suffix -enis with voiceless [f]
begraffenis /bəɡra:v+ənɪs/ [bəɡrafənɪs] funeral; burial ; cf. begrave /bəɡra:v+ə/ [bəɡravə] to bury
ferjeffenis /fərjɛf+ənɪs/ [fəjɛfənɪs] forgiveness ; cf. the Old Frisian verb urieva to forgive
erfenis /ɛrv+ənɪs/ [ɛrfənɪs] inheritance ; cf. erve /ɛrv+ə/ [ɛrvə] to inherit
drôfenis /drɔ:v+ənɪs/ [drɔ:fənɪs] sadness, sorrow, grief ; cf. drôvens /drɔ:v+əns/ [drɔ:və̃s] sorrow

Final Devoicing thus appears to be conditioned by both phonological and morphological factors: the word ending (phonological) and certain suffixes (morphological).

extra

Admittedly, the pattern in (8) and (9) has a relatively small scope as it only pertains to the suffixes -(e)lik and -enis in combination with the velar and labiodental fricative, respectively. Derivatives like those in (10) are realized with a voiced plosive:

i
Examples of derivations with -(e)lik and -enis with a voiced plosive
ûnheblik [unhɛblək] unmannerly, rude
gemoedlik [ɡəmuədlək] agreeable, pleasant
belidenis [bəli(:)dənɪs] confession (of faith)
ferrizenis [fəri:zənɪs] resurrection
extra

Exceptions to the pattern above are deeglik [de:ɣlək] reliable; sound and mooglik [mo:ɣlək] possible. This also holds for the variants with -elik, viz. degelik and mogelik, which is all the more indicative of the fact that the voicelessness of the fricative in words like geryflik [ɡəriflək] comfortable and yslik [islək] hideous, dreadful is triggered by the suffix, the variants of which behave uniformly with respect to both the regular pattern and the exceptions.

extra

The suffix -inne ( /-ɪnə/) derives the designation of female persons from that of male persons (see -inne). The cases in which -inne is attached to a male denoting noun with a stem-final voiced fricative are the following:

i
Female denoting nouns derived with the suffix -inne
grevinne /ɡre:və+ɪnə/ [ɡre:ˈvɪnə] countess
slavinne /sla:v+ɪnə/ [sla:ˈvɪnə] female slave
bazinne /ba:z+ɪnə/ [ba:ˈzɪnə] female boss
Friezinne /friəz+ɪnə/ [friəˈzɪnə] Frisian woman
reuzinne /rø:z+ɪnə/ [rø:ˈzɪnə] giantess
hartoginne /hartɔɣ+ɪnə/ [hartɔˈɣɪnə] duchess

The distribution of the voiced alveolar and velar fricatives, /z/ and /ɣ/, is such that they neither occur in word-initial position nor in the onset of a stressed word-medial syllable, positions reserved for the voiceless alveolar fricative /s/ and the voiced velar plosive /ɡ/, respectively (see the obstruents: the fricatives). A loanword like psychologysk psychological can be pronounced as [psixo:ˈlo:ɣisk] or as [psixo:ˈlo:xisk], with voiced [ɣ] or voiceless [x] in the onset of the final, unstressed syllable. In psychology psychology, on the other hand, the final syllable is stressed, so the onset can only accommodate [ɡ]: [psixo:lo:ˈɡi]. In (11) above, however, /z/ and /ɣ/ occur in the onset of a stressed word-medial syllable. This exceptional behaviour must be accounted for with an appeal to the suffix -inne, which forces a stem-final fricative to be, or to remain, voiced. This then is another instance of morphologically conditioned voicing behaviour.

extra

The short vowel of the words in (12):

i
geryflik [ɡəriflək] comfortable
yslik [islək] hideous, dreadful
pryslik [prislək] praiseworthy, laudable
wyslik [wislək] wisely
húslik [hyslək] domestic; hom(e)y

links up with the voicelessness of the following fricative, cf. the base forms in (13), in which a voiced fricative is preceded by a long vowel:

ii
gerive [ɡəri:və] to suit the convenience of
iis [i:s] ice < /i:z/ ; cf. izich [i:zəx] icy, freezing
priizje [pri:zjə] to praise
wiis [vi:s] wise < /vi:z/ ; cf. wizer [vi:zər] wiser
hûs [hu:s] house < /hu:z/ ; cf. huzen [hyzn̩] houses

The shortness of the vowel is not a systematic fact, as shown by the words in (14):

iii
freeslik [fre:slək] terrible, awful; horrible
beweechlik [bəve:xlək] mobile; mov(e)able
eachlik [ɪəxlək] attractive

It cannot be coincidental, however, that in deeglik [de:ɣlək] reliable; sound and mooglik [mo:ɣlək] possible, with an exceptionally voiced fricative before -lik, the vowel preceding the fricative is long.

extra

An obstruent affected by Final Devoicing can become part of the underlying representation of words or, put differently, words with a voiced final obstruent may acquire a voiceless final obstruent in the course of time. This is what must have occurred with the words in (15):

i
gebiet /ɡəbiət/ territory; area, district, region
liet /liət/ song
andert /ɔndət/ answer

They have developed out of forms with a final /d/, cf. the Dutch cognates gebied and lied, and the Old Frisian form anderd. As to gebiet and liet, this change has given rise to the new minimal pairs in (16):

ii
gebied(e) /ɡəbiəd/ to order, to command - gebiet /ɡəbiət/ territory; area, district, region
lied(e) /liəd/ to sound, to ring; to lead - liet /liət/ song

Many Frisian place names end in -wert, the etymological meaning of which is terp. Some examples are given in (17):

iii
Ferwert
Holwert
Jorwert

These names have a final /-t/, which is clear from the inhabitant names and adjectives derived from them with the suffix -er ( /-ər/) in (18):

iv
Ferwerter [fɛrvətər] inhabitant of Ferwert; coming from, belonging to Ferwert
Holwerter [holvətər] inhabitant of Holwert; coming from, belonging to Holwert
Jorwerter [jɔrvətər] inhabitant of Jorwert; coming from, belonging to Jorwert

The family names in (19), on the other hand, have retained the original /d/ of werd, as is also the case with the name of the municipality Ferwerderadiel [fə'vɛdəradiəl].

v
Holwerda [holvə(r)da]
Ferwerda [fɛrvə(r)da]
Jorwerda [jɔrvə(r)da]

Finally, the words in (20)

vi
blabs /blabz/ ooze, sludge ; cf. blabze /blabzə/ ooze, sludge
drids /drɪdz/ ooze, sludge ; cf. dridze /drɪdzə/ ooze, sludge

have developed the variants in (21)

vii
blaps /blaps/
drits /drɪts/

As a result, they have the two derivatives with the suffix -ich ( /-əɣ/) in (22):

viii
blabzich/blapsich [blabzəx/blapsəx] oozy, sludgy
dridzich/dritsich [drɪdzəx/drɪtsəx] oozy, sludgy

However, things can also work the other way around. The word held [hɛlt] hero has an underlying word-final /-d/, as appears from the plural helden [hɛldn̩] heroes. But the by now extinct variant helt had a final /-t/, as appears from compounds like helte#dea heroic death and helte#died heroic deed (next to helde#dea and helde#died). The same goes for the word geweld /gəvɛld/ violence, force, cf. geweldich [gəvɛldəx] tremendous, enormous, alongside which there used to stand forms with /t/ ( t). The change from /-t/ to /-d/ is likely to have taken place under the influence of Dutch, which has the cognates held and geweld, both with /-d/.

extra

Next to the noun moard /moəd/ murder there is the regular conversion form moardzje [mwadzjə] to kill, to murder, also in the derivation fermoardzje /fərmwadzjə/ to murder. Historically, the form moartsje [mwatsjə] [moətsjə] scream blue murder ‒ confer the derivative gemoart /ɡə+moət/ /ɡə+mwat/ the screaming of blue murder ‒ also derives from moard. moartsje is likely to have acquired its exceptional /t/ due to its affective meaning. For the /z/ and /s/ in moardzje and moartsje, see /{s/z}/ insertion between /{t/d}/ and /jə/).

extra

The effects of Final Devoicing can be obscured by Regressive Voice Assimilation (see Regressive Voice Assimilation: type 1 and Regressive Voice Assimilation: type 2), as in bloed#bank [bluədbaŋk] blood bank, slûs#doar [slu:zdoər] lock gate, and glês#guod [ɡlɛ:zɡwot] glassware. In isolation, bloed /bluəd/ blood, slûs /slu:z/ lock, and glês /ɡlɛ:z/ glass are realized with a voiceless final obstruent: [bluət], [slu:s], and [ɡlɛ:s].

extra

Unlike the other West-Germanic languages, English does not have Final Devoicing. In this respect then Frisian goes with Dutch and German. There are indications, however, that in the early twentieth century word-final /b/ and /d/ were still realized with a fair amount of voicing, so Final Devoicing must be considered a relatively young phenomenon. In all likelihood, it started after the long vowels, the falling diphthongs and the liquid consonants (see Fokkema (1958)). As yet, it is not a fully-fledged part of the phonology of some mixed Dutch-Frisian dialects, viz. that of the island of Ameland and that of the village of Midsland on the island of Terschelling. See also Van Bree (2003).

References
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Bree, Cor van2003'Auslautverscherping' bij de oude grammatici en in de dialecten en het FriesRuijsendaal, Els, Rutten, Gijsbert & Vonk, Frank J. M (eds.)Bon jours Neef, ghoeden dagh, Cozyn! : opstellen aangeboden aan Geert Dibbets by zijn afscheid als bijzonder hoogleraar Geschiedenis van de Nederlandse grammatica in West-Europese context aan de Katholieke Universiteit Nijmegen op 19 december 2003MünsterNodus Publ.
  • Fokkema, K1958De verzachting van enkele slotconsonanten na lange klinker of tweeklank in het FriesAlbum Edgard BlancquaertTongerenMichiels147-151
  • Hoekstra, Jarich1998Fryske wurdfoarmingLjouwertFryske Akademy
  • Hoekstra, Jarich1998Fryske wurdfoarmingLjouwertFryske Akademy
  • Visser, Willem2000Frjemd wurdt eigener. Oer de âlde Frânske lienwurden yn it FryskIt Beaken62141-218
  • Visser, Willem2003Patroanen yn ferlytsingsnammen.Philologia Frisica anno 2002 : lezingen en neipetearen fan it sechtjinde Fryske filologekongres16263-305
printreport errorcite