• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
The dorsal obstruents

This section deals with the distribution of the four dorsal obstruents, viz. the fricatives /ɣ/ and /x/ and the plosives /ɡ/ and /k/. Of these, /ɡ/, /ɣ/, and /x/ have a complementary distribution, so they are not distinctive. The voiceless plosive /k/, however, can occur in all positions where the other dorsals can, so it stands in opposition to them all. This means that all dorsal obstruents can be assigned phonemic value.


The dorsal obstruents /ɡ/, /ɣ/, and /x/ have a complementary distribution. The voiced plosive /ɡ/ occurs in the onset of a word-initial and a stressed word-medial syllable, the fricatives /ɣ/ and /x/ in all other positions, viz. in the syllable coda, in the onset of an unstressed word-medial syllable (in native words for the most part a schwa syllable, in loanwords also a syllable with a full vowel). So, the strong plosive occurs in the strong onset position, while the weaker fricatives occur in the weak coda position.

[hide extra information]

The complementary distribution of /ɡ/, /ɣ/, and /x/ leads Cohen et al. (1959:116-117) to posit one phoneme with a threefold realization, a so-called archiphoneme, denoting a 'neutral, voiced, dorsal (velar) constriction'. As such it stands in opposition to the voiceless dorsal plosive /k/.

For the behaviour of /ɡ/, /ɣ/, and /x/ in word-initial and word-final clusters, see onset: singleton consonants, consonant sequences in general, onset: sequences of obstruent and nasal, extra-syllabic consonants, word-final clusters of a liquid and an obstruent, word-final clusters of two obstruents, word-final clusters of two fricatives, words ending in /-ɣd/, and words with a non-coronal as final segment of a four-positional word-final sequence. In the table below, some examples of the occurrence of single /ɡ/, /ɣ/, and /x/ are listed:
Table 1
/ɡ/ /ɣ/ /x/
gean /ɡɪən/ to go dreech /dre:ɣ/ thorough wach /vax/ watchful
begean /bə+ɡɪən/ /(bə)(ɡɪən)/ to commit; to walk on drege /dre:ɣ+ə/ /(dre:)(ɣə)/ thorough (inflected form) wache /vax+ə/ /(va)(xə)/ watchful (inflected form)
figuer /fiɡyər/ /(fi)(ɡyər)/ figure kollega /kole:ɣa/ /(kol)(le:)(ɣa)/ colleague kiche /kɪxə/ /(kɪ)(xə)/ to cough
figuerlik /fiɡyər+lək/ /(fi)(ɡyə(r))(lək)/ figurative noga /no:ɣa/ /(no:)(ɣa)/ nougat psychologysk /psixoloɣisk/ /(psi)(xo:)(lo:)(ɣisk)/ psychological
psychology /psixoloɡi/ /(psi)(xo.)(lo.)(ɡi)/ psychology sago /sa:ɣo:/ /(sa:)(ɣo)/ sago Dachau /daxɔu/ /(dax)(xɔw)./ Dachau
pagina /paɣina/ /(pa:)(ɣi)(na)/ page
figuratyf /fiɣyrativ/ /(fi)(ɡy)(ra)(tif)/ figurative
Of the dorsal obstruents, /ɡ/ has the most limited distribution — it is the special case —, which is captured by the two positive constraints below:
a. ω(ɡ…), b. ω((…).n('ɡ…)) (where n ≧ 1)

Due to their complementary distribution, /ɡ/, /ɣ/, and /x/ cannot be distinctive. Since the dorsal voiceless plosive /k/ can occur in all positions where the other dorsals as a whole can, it stands in opposition to them all, as exemplified in the table below, which shows examples of phonological opposition between /k/ and the other dorsals:

Table 2
a. /k/ - /x/ b. /k/ - /ɡ/ c. /k/ - /γ/
kik /kɪk/ whimperkich /kɪx/ cough kear /kɪər/ time; turn - gear /ɡɪər/ done, cooked reak /rɪək/ rick - reach /rɪəɣ/ cobweb(s)
rak /rak/ reach, leg - rach(e) /rax/ to rant and rave komm(e) /kom/ to come - gom /ɡom/ gum weak /vɪək/ soft(hearted) - weach /vɪəɣ/ wave
ikel /ikəl/ acorn - ychel /ixəl/ bright girl kâns /kɔ:ns/ chance - gâns /ɡɔ:ns/ a lot slak /slak/ snail, slug - slach /slaɣ/ blow, stroke
gûk(je) /ɡuk/ to look - gûch /ɡux/ scorn kat /kɔt/ cat - gat /ɡɔt/ hole boek /bu:k/ book - bûg(e) /bu:ɣ/ to bend; to bow
kêst /kɛ:st/ clause - gêst /ɡɛ:st/ yeast alk /ɔlk/ auk - alch /ɔlɣ/ alga
klei(e) /klai/ to complain - glei /ɡlai/ inflamed, angry wurk /vörk/ work - wurch /vörɣ/ tired
kriem(e) /kriəm/ to turn (around) - griem(e) /ɡriəm/ to make a mess (of) slûk /sluk/ straight - slûch /sluɣ/ sleepy
draak /dra:k/ kite - drag(e) /dra:ɣ/ to bear; to wear
ik /Ik/ I - ich /ɪɣ/ edge
rikel /rikəl/ (he-)dog - rigel /riɣəl/ line
Murk /mörk/ proper noun - murch /mörγ/ (bone) marrow
In (a,b) we see genuine minimal pairs, since /k/ and /x/ and /k/ and /ɡ/ differ in one aspect, viz. ±cont ( /k/ vs /x/) and ±voice ( /k/ vs /ɡ/). In (c), on the other hand, there are near-minimal pairs, for /k/ and /ɣ/ differ with respect to both ±cont and ±voice. But despite their limited distribution and notwithstanding the fact that this is somewhat less clear for /ɣ/ than it is for /x/ and /ɡ/, all dorsal obstruents can be considered to have phonemic status in Frisian.

[hide extra information]

As noted in all Frisian grammars, nouns ending in /-ɣ/ display a systematic opposition between [ɣ] and [x] , when they function either as a verb (of the second weak conjugational class) or as a diminutive, both containing the suffix -je (infinitive, imperative, all plural persons present tense; diminutive). This difference is also reflected in the spelling. Some examples are given in the table below:

Table 3
eagje [ɪ.əɣjə] to look nice; to gaze - eachje [ɪ.əxjə] little eye          
bargje [barɣjə] to make a mess (of) - barchje [barxjə] little pig
swolgje [swolɣjə] to guzzle, to gobble - swolchje [swolxjə] little pull
The above words are morphologically complex, so they have no bearing on the issue of the phonemic status of the dorsal obstruents.

[hide extra information]

For the most part, interjections ((wild) oaths) like got! gosh!, goodness!, gut! gosh!, goodness, gadferdarry! darn!, damn!, and godferdomme! goddamned! are realized with the voiceless dorsal [x]: [xɔt], [xøt], [xatfərdari], and [xɔtfərdomə] (though [ɡɔtfərdomə] occurs as well). One is inclined to think of Dutch influence here. But the emotional content of interjections may also manifest itself in a marked phonological configuration, in this case word-initial [x]. Take the final /r/ of the prefix fer-. It is not realized before a consonant, as in fergean /fər+gɪən/ [fəgɪ.ən] to decay, to perish and fergunne /fər+gønə/ [fəgønə] to begrudge, to resent. In ferdomme! damned! and fergeemje/fergemy damned!, however, /r/ is always realized: [fərdomə], [fərɡe:mjə]/ [fərɡe:mi]; in ferdomme, it may even be lengthened: [fər:domə] (see also godferdomme and gadferdarry above). In the same vein, the interjection tomme!(< gottomme < goddomme < godferdomme) damned! has an aspirated [t] ( [thomə]), whereas aspiration does not play a role elsewhere in the phonology of Frisian.

[hide extra information]

In the southern part of the language area, a simplex word like the numeral njoggen /njoɣən/ nine is pronounced as [njoɡən], with the dorsal plosive [ɡ]. The same holds for plural forms like dagen [da:ɡən] days, eagen [ɪ.əɡən] eyes, and wegen [ve:ɡən] roads; ways, for inflected adjectives like drege [dre:ɡə] difficult, laborious, and inflected verbs, like drage [dra:ɡə] to carry; to wear; to bear (infinitive, all plural persons present tense). So far, only native words have been presented, in which [ɡ] stands between a full vowel and schwa. It is in loanwords that [ɡ] also shows up between two full vowels, as in noga [no:ɡa] nougat and lego [le:ɡo:] Lego. In native verbs of the second weak class, like seagje [sɪəɡjə] to saw (infinitive, imperative, all plural persons present tense), [ɡ] is between a full vowel (a centring diphthong) and the suffix –je, which is vowel-initial at the underlying level.

The other Frisian dialects all have the fricative [ɣ] here. In the southern dialect therefore /ɣ/ seems to undergo strengthening in intervocalic position. Now, /ɡ/ might be assumed to be part of the underlying representation of these words. But this is not a viable option. First, this would render the distribution of the dorsal obstruents for Frisian as a whole much harder to state. Second, it leads to wrong surface forms, such as ik draach [*dra:k] I carry; wear; bear, it is dreech [*dre:k] it is difficult, laborious, and geseach [*ɡəsɪ.ək] sawing, deriving from /dra:ɡ/, /dre:ɡ/, and /ɡəsɪəɡ/, respectively, by means of Final Devoicing (see final devoicing: the process. As in the other dialects, these forms are pronounced as [dra:x], [dre:x], and [ɡəsɪ.əx], with final [–x] deriving from /-ɣ/. All in all, it is only the surface sequence [ɣ-vowel] which is prohibited in the southern dialect.

[hide extra information]

In hartoginne / /hartɔɣ+ɪnə/ duchess, the voiced dorsal fricative /ɣ/ shows up in the onset of a stressed word-internal syllable: [hartɔˈɣɪnə]. This unexpected realization seems to be tied up with the suffix -inne, which always bears primary stress. Compare forms like bazinne /ba:z+ɪnə/ [ba:'zɪnə] mistress; lady of the house and Friezinne /friəz+ɪnə/ [friə'zɪnə] female Frisian, which are realized with the voiced coronal fricative /z/. Like /ɣ/, /z/ neither occurs in word-initial position nor at the beginning of a stressed word-medial syllable. The suffix -inne, however, seems to have the power to license the voiced dorsal and coronal fricatives in this non-canonical position.

[hide extra information]

From the previous two notes, one may gain the impression that the constraints on the occurrence of voiced fricatives are not of equal strength. While the ban on the occurrence in the onset of a stressed word-medial syllable can be violated, the one on the occurrence in word-initial position cannot. The latter therefore seems to be stronger than the former. In general, this seems to be true. But when taking all dialects into account, the ban on the word-initial occurrence appears to be a little less strict. First, the by now virtually extinct Frisian dialect of the island of Schiermonnikoog (Skiermûntseach) used to systematically have the voiced velar fricative /ɣ/ in word-initial position (in the course of time it turned into the voiced plosive /ɡ/). Second, the western variant of the Frisian dialect of the island of Terschelling (Skylge) has both the voiced plosive /ɡ/ and the voiceless fricative /x/ in word-initial position. Third, in the southeast of Fryslân, at the border of the language area, the voiceless fricative /x/ occurs in word-initial position, though not systematically so. In Cohen et al. (1959:117) it is termed an 'individuall variant', which is on the brink of extinction. Dyk (2008:33-35) considers this word-initial /x/ as the result of language contact or, put differently, as an intrusion from non-Frisian dialects.

The distribution of the dorsal obstruents also manifests itself in the adaptation of Dutch loan words; if the latter have [ɣ-] or [x] in word-initial position or as the onset of a stressed word-medial syllable, they are invariably realized with initial [ɡ], as shown in the examples in the table below:

Table 4
Dutch Frisian Translation
[ɣ]laciaal [ɡ]lasiaal glacial
si'[ɣ]aar si'[ɡ]aar cigar
[x]loor [ɡ]loar chloride
[x]aos [ɡ]aos chaos
ar'[x]ief ar'[ɡ]yf archives

[hide extra information]

In the following non-native Dutch words, /ɣ/ is in the onset of an unstressed word-internal syllable: a'dagium adage, col'lega colleague, collegi'aal amicable, legi'oen legion, 'lego Lego, legu'aan iguana, rigo'reus rigorous, 'sago sago , 'pagina page, and psycho'logisch psychological. As loan words in Frisian, /ɣ/ is realized as either the plosive [ɡ] or as the fricative [ɣ]. The realization as a plosive is surprising.

In native derivatives like bûging /bu:ɣ+ɪŋ/ bend, curve (from bûg(e) to bend) and driging /dri:ɣ+ɪŋ/ threat (from driig(je) to threaten), in which /ɣ/ occurs in the same context, it is invariably the underlying fricative /ɣ/ which shows up.

Frisian has quite a number of trisyllabic surnames ending in -inga, like Andringa, Elzinga, Hettinga, Kamminga, and Wybinga. They have primary stress on the antepenultimate and secondary stress on the final syllable, as in [ˈɔndrɪŋˌɡa] (Andringa) and are realized with the plosive [ɡ]. The secondary stress on the final syllable may favour the plosive realization.

[hide extra information]

In loan words like diagnoaze diagnosis, diagnostysk diagnostic, and paragnost psychic, medium, the cluster of dorsal obstruent plus /n/ is in the onset of the syllable with primary stress, where the obstruent is realized as the plosive [ɡ]: /ˌ(di)(ja)'(gno.ə)(zə)/, /ˌ(di)(ja)'(gnɔs)(tisk)/, and /ˌ(pa:)(ra)'(gnɔst)/. In magnetron microwave oven and ynkognito incognito, the obstruent is in the coda of the syllable, where it is realized as the fricative [ɣ]. The syllable in question bears secondary stress (magnetron) or main stress (ynkognito): /ˌ(maɣ)(ne:)'(tron)/ and /(iŋ)'(kɔɣ)(ni)ˌ(to:)/. Stress seems to be the decisive factor for the difference in syllabification here (and, indirectly, the difference in realization). A syllable with primary stress seems to 'attract' a complex onset, while an unstressed or secondarily stressed syllable seems to require a simplex one. But this cannot be the whole story. In words like magnaat magnate, tycoon, magneet magnet, and stagnear(je) to stagnate, all of which have primary stress on the final syllable, the dorsal fricative is part of the coda, hence it is realized as [ɣ]: [(maɣ)'(na:t)], [(maɣ)'(ne:t)], and [(staɣ)'(njɛr)((jə))]. This cannot be due to the shortness of preceding [a], cf. paragnost above. A supplementary explanation is called for.

  • Cohen, Antonie, Ebeling, C.L., Eringa, P., Fokkema, K. & Holk, A.G.F. van1959Fonologie van het Nederlands en het Fries: Inleiding tot de moderne klankleerMartinus Nijhoff
  • Cohen, Antonie, Ebeling, C.L., Eringa, P., Fokkema, K. & Holk, A.G.F. van1959Fonologie van het Nederlands en het Fries: Inleiding tot de moderne klankleerMartinus Nijhoff
  • Dyk, Siebren2008Mûljearring: in oersjochUs wurk: tydskrift foar Frisistyk571-43