This rule applies only to a derived demonstrative form meaning '(on) this/that side'. The derived form involves reduplicating the final /CV/ of the demonstrative pronoun or adverb which serves as input. The Centripetal suffix /-ala/ is added. There is also a parallel but less common form with Centrifugal /-ali/ For further details and exx. see §7.21 (also §6.11).
Consider these input forms: /ya:-ni/ 'this (ANA class)' and adverb /ya:-ji/ 'here'. The relevant derived forms are //ya;-ni-ni-ala// → /ya:-ni-ni:-'la/ and //ya:-ji-ji-ala// → /ya:-ji-ji:-'la/
These forms present no difficulties; we just reduplicate the final /CV/ add /-ala/ and have the surface form produced regularly by VV-Contraction P-49. However, some difficulties arise in insuring that human nonsingular forms are correctly generated.
Consider input forms /ya:-wi-ni/ (MDu), /ya:-wi-r-ngi/ (FDu) and /ya:-wu-ru/ (Pl) 'these'. The derived forms are /ya:-wi-ni-ni:-'la/ /ya:-wi-r-ngi-ngi:-'la/ and /ya:-wu-ra-ra:-'la/ These forms force us to specify more precisely what gets reduplicated and also whether the -Rdp- segment precedes or follows the relevant input syllable.
In the FDu form, note that the /-r-/ morpheme is not included in the -Rdp- segment. This is a Nonsingular morpheme. Only the F morpheme (Feminine) gets reduplicated.
In the MSg form, /-ni-/ is arguably underlying //-r-ni-// with the same Nonsingular //-r-// and M //-ni-// which is related to other M morphemes in verbal and nominal morphology with /n/ not /n/ A rule //rn// → /n/ though unproductive, could possibly be recognised (see P-24, below). If so, we must argue that this contraction rule precedes Final Reduplication, since in the relevant MSg form we get /-ni-ni:-/ instead of /-ni-ni:-/
The Pl form is even more tricky. From input /ya:-wu-ru/ we might expect output //ya:-wu-ru-ru-ala// which would produce surface /ya:-wu-ru-ra:-'la/ but the attested form is /ya:-wu-ra-ra:-'la/ Note that the third morpheme /-ra-/ shares surface vowel quality with the following syllable. However, the latter in turn has acquired its surface vowel quality through VV-Contraction involving the initial vowel of the Centripetal suffix, with //ua// becoming /a:/ In generative terms, then, it is probably best to order Final Reduplication P-3 after VV-Contraction P-49. If this is done, we must locate the -Rdp- segment before the identical segment of the input, and specify that the vowel of the -Rdp- segment is short regardless of the length of the input vowel.
| (P-3) Final Reduplication |
| Rdp → -C1V1- // D- _____ -C1V1(:)- |
| where D- is a demonstrative root |
An alternative would be to take //ya:-wu-ru-ru-ala// as the underlying form after all, with P-3 followed by an ad hoc assimilation rule. This would require the assimilation rule, but not P-3 itself, to follow VV-Contraction P-49. Another alternative, for those linguists not favorable to generative rules, would be a surface-to-surface schema deriving the reduplicated form directly from the unreduplicated surface form (this would be relatively easy in this instance).