• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
Degemination: the process

This topic deals with degemination as a phonological process, viz. with the question which consonant of a sequence of two identical ones remains. It is argued that degemination is best viewed as the contraction of two identical consonants.


The degemination of a sequence of two identical consonants might proceed in either of the following three ways:

Figure 1

[click image to enlarge]

Frisian does not have double consonants: when a sequence of two identical consonants arises, the left-hand and the right-hand one end up in the syllable coda and the syllable onset, respectively. Since the onset is the position typical for consonants, the left-hand consonant is likely to delete, leaving the right-hand one. In most cases, this also yields a good syllable contact.

If deletion proceeded the other way around, the remaining left-hand consonant could also end up in the syllable onset, due to resyllabification. The latter process, however, is much more common when the consonant in question remains part of the same word than when it has to move from one word to the other. Therefore, if degemination is analyzed as consonant deletion, this is best considered as the deletion of the left-hand consonant.

The above considerations have no bearing on degemination in word-final position, for which the contraction analysis provides the only reasonable account. It might be assumed then that degemination is to be looked upon as the contraction of two adjacent consonants onto one phonological position and that the position in which the one consonant ends up can be relegated to general principles of syllabic affiliation.

[hide extra information]

Past participles of verbs of the first weak class are formed by adding the suffix /-d/ or /-t/ to the stem, depending on the voice specifation of the stem-final segment (see paradigm of class I and progressive voice assimilation: the past tense of the weak verbs of the first conjugation class). So, in case the stem ends in /d/, the participle ends in a sequence of two identical voiced plosives ( /-dd/), see the examples in (1):

Example 1

Examples of weak past participles ending in -/dd/
(ik haw) wjud /vjød+d/ [vjøt] (I have) weeded
(ik haw) besteed /bəste:d+d/ [bəste:t] (I have) spent (on)

Though there is no underlying voicing conflict in (1), the surface forms end in [-t]. There are three possible scenario's for arriving at this single final [-t]:

  • The underlying sequence /-dd/ is subject to degemination, after which the remaining final /-d/ undergoes final devoicing: /-dd/ [-d] [-t].
  • Final Devoicing affects the right-most /-d/, resulting in a voicing conflict between /d/ and /t/, which is resolved by the devoicing of the /d/, followed in turn by degemination: /-dd/ [-dt] [-tt] [-t].
  • The facts of Regressive Voice Assimilation force us to assume that obstruents in a sequence do not have their own, separate voice specifications, but that they, instead, share one and the same specification, so that the sequence as a whole can be voiced and devoiced in one swoop (see final devoicing: obstruent sequences). An independent piece of evidence for this analysis is that obstruent sequences cannot be broken up by schwa epenthesis (see schwa insertion in word-final sequences). Not only does the latter hold of underlying clusters, but also of the ones resulting from inflection and derivation. The derivation then can proceed as follows. Final Devoicing affects the cluster /-dd/ as a whole, after which the resulting voiceless cluster is subjected to degemination: /-dd/ [-tt] [-t]. This scenario then seems to be the most plausible one.
[hide extra information]

Degemination can be fed by Final Devoicing, the effects of which can be undone by Regressive Voice Assimilation (see Regressive Voice Assimilation: type 1), which in its turn also feeds degemination. This might be at stake in compounds like those in (2):

Example 2

bloeddruk /bluəd#drøk/ blood pressure
syddyk /sid#dik/ side-road
wrâlddiel /vrɔ:d#diəl/ continent (lit. world part)

For syddyk /sid#dik/ side-road, for example, there are two scenarios for arriving at the surface form [sidik], depicted here:

Table 1
(a) /sid#dik/ /sidik/ [sidik]
(b) /sid#dik/ /sit#dik/ /sid#dik/ [sidik]

In (a), the two adjacent /d/ s of the underlying representation are affected by degemination straight away, whereas in (b) two processes apply − final devoicing ( /sid#dik/ /sit#dik/) and regressive voice assimilation ( /sit#dik/ /sid#dik/) − before degemination can operate. Final Devoicing and Regressive Voice Assimilation are independently motivated phonological processes, so (b) involves no extra 'costs'. Scenario (a), however, consists of fewer steps than (b), for which reason it is to be preferred. This is also in line with the assumption that phonological processes apply wherever and as soon as they can.

[hide extra information]

Degemination also affects clusters of /s/ + a voiceless plosive, in either order (see Complex segments as single units).

[hide extra information]

Degemination does not affect a sequence of two adjacent sonorant consonants, if the right-hand one heads a syllable or, put differently, if it is syllabic (see The phonological behaviour of syllabic sonorant consonants).