4 Verbal Derivation
4.3 Combinations of verbal extensions
Often more than one verbal extension appears on a single verb base. In the follow-ing sub-sections, some generalizations over the respective order of morphemes will be given (§4.3.1), followed by a discussion of the derivation of pluractionals by means of combining the reciprocal/associative and the causative (§4.3.2) and a discussion of the shape of applicativized causatives (§4.3.3).
4.3.1 Morpheme order
When several extensions appear in a verbal base, their respective ordering is subject to several restrictions. The unproductive extensions appear closest to the root and follow the ordering illustrated in Figure 4.1.
Positional Impositive Extensive Separative Tentive
Figure 4.1: Order of unproductive verbal extensions
The following examples illustrate the attested combinations of unproductive extensions:15
(95) a. Positional and impositive:
gundamika‘bend over/down (tr.)’
b. Extensive and impositive:
lambalɪka‘put to bed’
c. Extensive and separative:
nyagalʊka‘get well (health)’
The more productive extensions follow the unproductive ones. The passive (including the fossilized passive) always occupies the last position.
(96) a. Fossilized passive:
kʊbɪlwa (°kʊb-ɪl-ʊ-a) ‘suffer’
ʊmɪlwa (°ʊm-ɪl-ʊ-a) ‘be thirsty (for)’
agɪlwa (°ag-ɪl-ʊ-a) ‘diminish (by); lack’
tumukɪlwa (°tum-ʊk-ɪl-ʊ-a) ‘be(come) short on; late for’
15The verbpangalatʊla‘destroy by taking part after part out’ (cf. pangʊla‘dismantle’) has the sequence -al-at-ʊl, which resembles the combination of extensive plus tentive plus separative.
It is unclear if this is a chance resemblance or a case of three unproductive extensions.
4.3 Combinations of verbal extensions b. Productive passive:
bɪɪkɪligwa (°bɪɪk-ɪl-igʊ-a) ‘be put for’
meleligwa (°mel-ɪl-igʊ-a) ‘owe’
bombeleligwa (°bomb-ɪlɪl-igʊ-a) ‘be weeded’
manyisigwa (°many-ɪsi-igʊ-a) ‘be taught’
fumusigwa (°fum-uk-i-igʊ-a) ‘be announced’
ɪmikigwa (°ɪm-ɪk-igʊ-a) ‘be respected’
saamikɪsigwa (°saam-ɪsi-ɪl-igʊ-a) ‘be made exile to’
A causative -i, either the short causative1 -ior the last segment of a split-up long causative2-ɪs-i, normally occupies the last position within the base, unless it is followed by the passive.16
(97) fulasania (°fulal<an>i-a) ‘hurt each other’
ʊlɪkɪsania (°ʊl-ɪkɪs<an>i-a) ‘sell each other sthg.’
Apart from these generalizations, the ordering of the productive extensions in Nyakyusa requires a dedicated study of its own, given the high number of logi-cal possibilities and the question of how morpheme order, meaning and syntax relate to each other. As Hyman (2002) points out, in many Bantu languages the relative order of certain verbal extensions follows a default pattern, which can have both a compositional reading (morpheme order reflecting semantic scope) and a non-compositional one, while the opposite order exclusively receives the compositional reading. Further, Lusekelo (2013) indicates that the relative posi-tion of the applicative in Nyakyusa may be linked to the semantic role of the argument it licenses.
Lastly, a few cases of doubled verbal extensions are attested in the data. In most cases it is unclear what the semantic and syntactic functions of these are.
Doubling of a verbal extension may serve the purpose of fulfilling the require-ment for both a default morpheme order and a compositional order at the same time (Hyman 2002),
(98) a. Two applicatives:
lɪɪlanila ‘eat together with sb.’ <lya ‘eat’
lwɪlanila ‘fight with each other for’ <lwa ‘fight’
b. Two reciprocals/associatives:
sopanilana ‘throw to each other’ <sopa ‘throw’
16See §4.2.2 for the process of spirantization induced by the causative1-i, and §4.3.3 for the formation of applicativized causatives.
c. Two causatives:
tiimɪsyanisya‘make each other herd’17 <tiima ‘herd’
4.3.2 Complex derivations: pluractional
The combination of the reciprocal and causative extensions often gives a plurac-tional reading. The range of possible meanings includes re-iteration, intensifica-tion or the involvement of multiple subjects or objects (also cf. Schumann 1899:
79). This combination is used on transitive bases and verbs of motion (99a, 99b), the only attested exception being the intransitive sulumania ‘afflict, be sorry’.
With verbs denoting ‘to return’, this combination gives a cyclic reading. When the short causative -iis used, sometimes spirantization takes place. This seems not to be predictable and is probably a function of time depth and lexicalization.
Concerning the length of the vocalic segment in the reciprocal, see §4.2.5.
Botne (2008: 86) and Gray (n.d.) observe a similar pluractional function of an-y- / an-i- in neighbouring Ndali and Kisi G67, respectively, and Kisseberth (2003: 557) for Makhuwa P30 givesú-hókól-an-yáán-ih-a ‘to go and come back the same day’. All these suggest that the combination of the causative and the reciprocal yielding pluractionality might have a wider distribution in Bantu.
The intransitive counterpart to the pluractional has the shape -anik and can be analysed as consisting of the reciprocal -an and neuter -ɪk extensions. It is used on transitive as well as intransitive bases (99c).
(99) a. Pluractionals derived from transitive verbs:
buuta ‘cut; slaughter’ >buutania ‘cut into pieces’
>buutanika ‘break into pieces (intr.)’
joba ‘speak to/about’ >jobesania ‘dispute about’
>jobanika ‘speak much’
lʊnga ‘add spice; >lʊngaania ‘join, connect (tr.)’
put together’ >lʊngɪsaania ‘join, connect (tr.)’
>lʊngaanika ‘be confused’
menya ‘break; chop’ >menyania ‘chop into pieces’
>menyanika ‘be chopped into pieces’
nyamba ‘throw’ >nyambania ‘scatter’
>nyambanika ‘disperse, be scattered’
pinya ‘bind; detain; fix’>pinyania ‘splice; tie together’
>pinyanika ‘be spliced; tied manifold’
17This example was elicited on the basis of Lusekelo (2012).
4.3 Combinations of verbal extensions b. Pluractionals derived from verbs of motion:
buja ‘return (to)’ >bujɪsania ‘go & return (same day)’
>busania ‘go & return (same day)’
gomoka ‘return; reprove’ >gomosania ‘go & return (same day)’
kɪnda ‘pass’ >kɪɪsania ‘pass by’
c. Pluractionals derived from intransitive verbs:
jeeta ‘turn pale’ >jeetanika ‘faint’
lɪla ‘cry; sound; mourn’ >lɪlanika ‘complain’
tʊʊja ‘pant; breathe out’ >tʊʊjanika ‘pant heavily’
4.3.3 Applicativized causatives
Applicativized causatives have a special form -(ɪ)kɪsi / -(ɪ)kɪfi. The alternations in vowel height as described in §4.2.1 apply. When they are derived from a causativized base subject to spirantization (see §4.2.2) -kɪsi/ -kɪfiis suffixed to the underlying non-causativized base, with /k/ replacing the base-final conso-nant. The fricative is /f/ if the replaced consonant is a labial, and is /s/ elsewhere.
In other words, it is the fricative that causative spirantization would produce.
(100) buja ‘return’ >busya ‘return (tr.)’ >bukɪsya ‘return (tr.) + appl’
fulala‘be hurt’ >fulasya‘hurt’ >fulakɪsya‘hurt + appl’
oga ‘bathe’ >osya ‘baptise’ >okesya ‘baptise + appl’
pyʊpa‘get warm’ >pyʊfya ‘warm up’ >pyʊkɪfya ‘warm up + appl’
If the causativized base is derived by causative1 -ifollowing non-spirantizing consonants or if it is derived by causative2 -ɪsi, a suffix -ɪkɪsiis attached to the non-causativized base:
(101) saama‘migrate’ >saam-y-a‘transfer’ >saam-ikɪsy-a‘transfer + appl’
ʊla ‘buy’ >ʊl-ɪsi-a ‘sell’ >ʊl-ɪkɪsy-a ‘sell to/for/at’
This uncommon phonological realization of applicativized causatives has been noticed from the first treatments of Nyakyusa on. Meinhof (1966) as well as Schu-mann (1899) mention this and EndeSchu-mann (1900) presents an attempt at a purely phonological explanation. Unfortunately, Endemann does not discuss cases of the suffixing of -ɪkɪsiafter nasals, which cannot be accounted for by his approach.
In his later grammar sketch, he further gives a rather curious explanation in which he tries to link this phenomenon to distal/itiveka- (Endemann 1914: 51).
The examples given by Botne (2008: 76) suggest a comparable formation in Ndali. Wolff (1905: 63) describes a similar phenomenon for neighbouring Kinga, where applicativized causativizes take the shape -ihitsa, although without replac-ing the final consonant.
Alhough non-transparent from a synchronic point of view, the morphophonol-ogy of applicativized causatives finds a diachronic explanation in a sequence of analogy formations, as Hyman (2003) plausibly illustrates (Berger 1938: 266f de-velops a parallel interpretation). In this scenario, the point of departure would have been a stage in which infixing of the applicative, together with a cyclic ap-plication of spirantization, took place (thus e.g. sook-a>soos-i-a>soos-el-i-a>
*soos-es-i-a), followed by despirantization of the root-final consonant (> sook-es-i-a). See Nyamwezi F22 (Schadeberg & Maganga 1992: 20–22) for a comparable case. In the next stage, despirantization to /k/ was generalized; note that spiranti-zation leads to a merger of the six non-labials affected. In the case of non-labials, one possible interpretation of the sequence /kɪs/ would be that the final conso-nant spirantized in the first place was being post-posed. This re-analysis was then extended to cases of labials, yielding -kɪfi. Once established, this pattern of applicatived causativizes surfacing as /kɪ{s, f}i/ was extended to non-spirantizing consonants (introducing what Hyman labels an “extra k”) and thus fully gener-alized.
The main source for Hyman’s interpretation is the chronolect described by Schumann (1899), in which causative1-iis the most productive of the two deriva-tions. The present data show that Nyakyusa has gone one step further, extending this pattern to causative2-ɪsiand thus generalizing the requirement that any ap-plicativized causative must surface with the sequence /kɪ{s, f}/ (and respective vowel alternations).18