• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
/r/ deletion

The phonological process of /r/ deletion in Afrikaans happens mainly postvocalically, and in syllable final position, in words with a very high frequency (mainly function words) in casual speech e.g. in hier /ɦir/ [ɦi:] here, as well as in the coda of the prefix ver- /fər/ [fə] VBZ. In this topic we focus on /r/ deletion as it is manifested in the semi-formal productions of news readers of RSG. (See Introduction to phonological processes)

De Villiers and Ponelis (1987) contend that this form of consonant loss is due to English influence via Cape Afrikaans (K-Afr.). Klopper (1981) shows that /r/ loss is sensitive to social factors like religion, social status, ethnic group cohesion and age, and Coetzee (1990) stresses the finding that this variability cannot be described adequately without taking into account the specific type of lexical input, notably word category (e.g. pronouns, adverbs or verbs). Klopper (1981) indicates a definite difference between white speakers of high social economic status and working class users in an informal style but not in readings of word lists. Christian coloured speakers (but not Muslims) from a high social background are less inclined to /r/ loss in word-list readings than white speakers, but /r/ loss happens more readily in informal speech. Seemingly, the former group of speakers are more considerative of "correct" pronunciation in reading tasks than when conversing spontaneously. Combrink and De Stadler (1987) also provide a description and examples of /r/ deletion in Afrikaans.

In the next section an overview of /r/ deletion is provided, and in the last section we concentrate on the extent to which these types of deletion are present in the readings of radio presenters of RSG.

[+]Coetzee's (1990) model

Coetzee (1990) firstly differentiates between /r/ loss that is not restricted to individual speakers of a certain geographical region or social group on the one hand, and, on the other hand, where region or social group is indeed a factor. All the examples as well as labelling are those of Coetzee (1990).

A. Unrestricted r deletion
  1. Locatives and compounds with these locatives as first components: daar, waar, hier
  2. maar
  3. klaar as word and in compounds
  4. haar (pronoun; possibly also in conjunction with ) -self
  5. paar (noun)
  6. weer as adverb
  7. the preposition vir
  8. the prefixes or first components, unstressed: mainly ver- but also per-, ter-, ser-, par-, kor-, for-.
  9. The suffix -er in syllable- and word-final position in rapid speech
  10. The liquid /r/ in coda position
Examples of the above-mentioned ten types are not always given in her description but are scattered throughout her article; for such examples, cf. the following (numbers correspond to those above):
  1. In compounds: daardeur through, waarin wherein, hieroor about this
  2. The word maar but as conjunctive
  3. klaar ready as adverb, occurring in isolation or in compounds like klaarblyklik apparently
  4. In isolation as pronoun haar her; haarself herself (reflexive pronoun)
  5. As second element in 'n paar ... (mense) a few ...(people)
  6. weer in die... again in the
  7. vir my for me
  8. verkoop sell, persent percent, terneergedruk depressed, sertifikaat sertificate, party paert, korporasie corporation, formeel formal
  9. verder further, deurmekaar confused, byvoorbeeld for example, oorlede died
  10. Pietermaritzburg , ander other, Van der Merwe , lekker nice, sommer just as well

B. The restrictions of /r/ deletion to region and social group
  1. Forms preceded by long vowels not included in A: duur expensive, hoor hear
  2. The phoneme /r/ in coda clusters: kinders children, kerk church, werk(s-) work-
  3. Other instances of -er: September September, suster sister

In some of the above-mentioned examples, Coetzee (1990) provides percentages of /r/ deletion, here expressed in terms of rule application probabilities (RAP). It is important to note that these are average values of the three types of structures: monomorphemes, compounds and phrases.

[+]The process of /r/ deletion in the RSG data

This description is limited to a number of notable cases, viz. /r/ loss when preceded by /a/ (restricted to the case of daar(-)). Where possible a comparison is given to data provided by Coetzee (1990).

[+]The phoneme /r/ preceded by long /a/ in daar
  • RSG

    daar /dar/ [da] there: daar, daarin, daarmee, daarna, daarom, daarop, daarvan, daar is, daar kan, daar moet, daar was, daar word. (RAP = 0.32).

  • Coetzee

    daar /dar/ [da] there: daar, daardie, daaroor, daarby, daarnatoe, daar is, daar onder, daar op (die), daar sal, daar word. (RAP = 0.64).

The difference in RAP indices is notable. It is quite likely that this disparity is ascribable to the different types of jargon, namely informal interviews versus more formal radio broadcasts.

In the next section the loss of /r/ in derivations is dealt with.


Here the focus is on one prefix only (ver-), as found in the RSG readings. ver- is the most productive Afrikaans prefix compared to others such as her-, oer-, opper-, weer-.

See the topic on prefixation for more examples of prefixes ending on /r/.

Examples in which /r/ loss was noted in the RSG readings include: verband, verbeter, verby, verdagte(s), verdeeldheid, verduistering, vergader, vergadering(s), verhoging, verjaarsdag, verkeer, verklaar, verklaring, verlede, vermeerder, verminder, vernietig, vervaardiger(s), verwyder. With some of these words (esp. verdagte(s), vergader and vergadering(s)) a high-percentage of cases were present. The degree of productivity of this phonological process is to be observed in the deletion of /r/ in the prefix ver- in verlede verlede: in this case a RAP of 0.54 was found (N = 131).

Coetzee (1990) mentions verdraagsaam, vergader, vergewe, verhewe, verkry, veroorsaak, versigtig, vertaal, vertraag, vertrou, verwantskap.

Coetzee (1990) mentions the view of De Villiers and Ponelis (1987) that the phonetic context -de(r)(s) [_dərs] is notably conducive to /r/ deletion. One should add to their observation the fact that /r/ occurs here in syllable final position (they cite words like beoordelaar(s) , donkerder , duurder , eerder , huurder , lekkerder , prospekteerder , verloorder , woordvoerder ). In a subset of RSG-recordings (N=89) a high RAP of 0.71 was found in the case of woord+voerder spokesperson and in compounds like bevel+voerder , burgemeesters+woordvoerder , hoof+woordvoerder , polisie+woordvoerder . Similarly, bestuurder manager or driver and compounds including bestuurder in taxi+bestuurder , showed high levels of /r/ deletion (RAP = 0.79). Coetzee does not provide a RAP for instances of /r/ deletion in her investigation. The results provided here from the RSG data nevertheless strongly support the assertion made by De Villiers and Ponelis (1987).

[+]The case of eerste

The ordinal eerste /erstə/ first constitutes a special case in which /r/ is frequently deleted, thus phonetically [iəstə]. Given that Coetzee (1990) does provide a RAP for this word, we are able to make a direct comparison with the RSG data.

Coetzee (1990) combines both eers and eerste (both first(ly)) in her calculation, rendering a RAP index of 0.52. In the RSG recordings a RAP of 0.56 was found, underscoring her findings.

The preposition vir for very frequently loses its word-final r, as in the following examples, noted in the productions of the RSG readers: vir +: al, baie, die, drie, groei, honderd, hulle, kalmte, politieke, sy, twee, tydperke, verdediging, verdere, verskeie, vyf (RAP = 0.23). /r/ was deleted twelve times out of 22 instances of vir die for the alone (RAP = 0.54). Coetzee (1990) reports a RAP of 0.14 for her cases with vir +. Comparing the /r/ deletion of a single speaker, as in the RSG case, to that of Coetzee (1990) could possibly be the cause of the higher RAP for the former. Figure 1 demonstrates the loss of [r] in the phrase vir die.

Figure 1: /r/-deletion in the phrase vir die, compared to two other realizations
[click image to enlarge]


During a recent RSG program, Die tale wat ons praat, Rufus Gouws, a well-known Afrikaans lexicographer, dealt in an interview with the use of a dictionary (i.e. woordeboek /vordəbuk/). In the course of the first ten minutes of this interview he used the word woordeboek(e) 77 times, in isolation or as the first constituent of a number of compounds (e.g. woordeboek+didaktiek didactics of dictionaries; woordeboek+kultuur culture of dictionaries). In the first few instances, he pronounced the word woordeboek as [vuərdəbuk], thus in its full form, but thereafter very seldomly; a few times [vuərəbuk] surfaced, but for the rest of the inspected ten minutes only the form with /r/ deletion was heard (a RAP of 0.78). Of general interest here is the observation that as soon as the speaker was at ease with the situation of a radio interview, he departed from the full form of the word. Figure 2 displays his pronunciation of woordeboek with and without [r].

The word's frequency of usage in this context, and, in association with this, the familiarity of the word woordeboek to the lexicogographer-speaker contributed, without much doubt, to this state of affairs.

Furthermore, the data in this case indicates the supremacy of /r/ deletion over /d/ deletion. The latter surfaced only seldomly i.e. as ['vuərəbuk]. This could mean that /r/ deletion is a stronger phonological process than is /d/ deletion (see d-deletion for this phenomenon in Afrikaans).

Compare the following two examples, followed by a visual representation in Figure (2) below:

  1. Full form of woordeboek
  2. Variation with /r/ deletion

Figure 2: Full form of woordeboek and the variation with /r/-deletion
[click image to enlarge]

[+]The weight of following segments

In her study, Coetzee (1990) provides the weights (i.e. RAPs) of the relative influence that segments following /r/ have on /r/ deletion. She nevertheless expresses some caution regarding the reliability of these statistics. These results have been reworked as Figure 3 below.

Figure 3: Rule application probabilities for /r/ deletion in different contexts; pauses and vowels are shown in other colours; consonants in blue. (N = 7195).
[click image to enlarge]

No such information was extracted for the RSG data. These RAP indices are not in support of Clements' Sonority Strength Hierarchy (Clements, 1990), as is indeed the case with some other phonological processes (see Regressive voice assimilation).

  • Clements, George N1990The role of the sonority cycle in core syllabificationPapers in Laboratory Phonology1Cambridge University Press283-333
printreport errorcite