The combination /rn/ (with tap /r/ is not common at any level of derivations. In relatively productive compounds and the like, when morpheme-final /r/ comes into contact with following morpheme-initial /n/ the sequence may be retained: /-rajbur=na-/ 'to mock' (compound with /=na-/ 'to see').
However, in some of the more frozen combinations we appear to have a rule by which /rn/ is converted into /n/ First, in demonstrative (including interrogative) pronouns, we find a MDu ending /-ni/ in opposition to a FDu form /-r-ngi/ as in /ya:-wi-ni/ 'here they (MDu) are' vs. /ya:-wi-r-ngi/ 'here they (FDu) are' (Table 7-1, Chapter 7). Similar endings occur with independent personal pronouns (Table 6-1, Chapter 6). The M morpheme elsewhere shows up as /na-/ /ni-/ or /nu-/ depending on the particular context (inflectional vs. derivational noun-class prefix, intransitive vs. transitive pronominal prefix, etc.), opposed to F morpheme /ngara- / /ngi-/ or /ngu-/ It seems reasonable, on a fairly abstract analysis, to represent /-r-ngi/ as consisting of a special Nonsingular morpheme /-r-/ related to Nonsingular /-rV-/ in pronominal prefixes; by analogy, we could take MDu /-ni/ as surface form of a sequence //-r-ni// with Nonsg /-r-/ and M /-ni/ If so, we need a rule by which //rn// → /n/.
The other instance of this rule is now even more frozen. There is a verb /-w2u-na-/ 'to make large campfire', which has the same (very unusual) suffixal paradigm as /=na-/ 'to burn' (trans.), and which can thus be regarded as a compound thereof. It is quite possible that the compound initial is historically identical to /wurg/ 'bushfire or grass fire', although synchronically this relationship is at best opaque. (The /w/ of /wurg/ seems to be /w1/ rather than /w2/ to judge by the verb /=agurganyji-/ 'to set fires in bush', which seems to be from /-aG=w1urg-ga-nyji-/ In any event, if /-w2u=na-/ is indeed to be treated as a compound of /=na-/ we probably need a rule of the type //rn// → /n/.
| (P-24) | rn-Contraction |
| rn → n | |
| in the combination of Nonsg /-r-/ with M /-ni/ and in the compound /-w2u=na-/ from //-w2urg=na-// |
There is also a compound /-w2u=nagi-/ '(fire in oven) to be burning well', related to /=nagi-/ 'to burn' (intrans.). Essentially, /=nagi-/ functions as an irregular intransitivised (mediopassive) of /=na-/ 'to burn' (trans.). It is thus not surprising that /-w2u=na-/ has a counterpart /-w2u=nagi-/ both showing P-24. However, both /-w2u=na-/ and /-w2u=nagi-/ are intransitive, and they are not semantically related in precisely the same way the uncompounded stems are.