Authenticity Reduplication

In Kinande, nouns are composed of augment, prefix, and stem (in that order) and classified into twenty different classes (Mutaka and Hyman 1990: 74-75). When a noun undergoes the reduplication at issue, it acquires the meaning of “‘a real X’…normally with a positive

1 Mutaka and Hyman’s (1990: 89) syllable template is: (N)CV1. The nasal is optional and the number of vowel can be more than one. For the following analysis, I assume that each and every vowel is the nucleus of a syllable and that there are three syllable types: V, CV, or NCV.

2 According to Odden and Odden (1985), Kíhehe, another Bantu language, has nasal onset incorporation (p.498), glide formation (p.499), and glide epenthesis (ibid.) take place before reduplication. The application of the three processes determines if some pre-stem contents get reduplicated.

3 Some caveats for all the Kinande data sets are in order here. Since the only underlying consonant cluster in Kinande is nasal plus consonant (Mutaka and Hyman 1990: 75), I restore a glide into its vowel counterpart in the string of CGV. For items where there is not enough information to speculate their vowel melody, a capital U or I is used as a place holder for surface w or j, respectively. All the palatal glides in the sources are changed from y to j. No tones are marked in the following data. Because of tone distribution, some items lexically end with vowel geminates (represented as Vː) according to Mutaka and Hyman (1990: 115 endnote 6). I keep their representations but leave open its interaction with reduplication.

connotation” (Mutaka and Hyman 1990: 76); I will call the pattern “authenticity reduplication”.

Interestingly, augments never get reduplicated, but their presence is necessary for the authenticity reduplication to occur (Mutaka and Hyman 1990: 81-83). To facilitate the upcoming discussion in this section, the active workspace (Σ) prefix and stem together is enclosed in a pair of square brackets. Three representative examples repeated from table 6 (at the end of this section) are given in (1). Three descriptions are observed from this set: (1a) shows triplication, (1b-d) reduplication, and (1e) nothing. Within (1b-d), there are two sub-phenomena: either only the noun stem is reduplicated (1b), or the stem and the prefix together are reduplicated (1c-d).

(1) authenticity reduplication for Kinande nouns

a. ‘dog’ e-[m-bUa] >>> e-m-bUa~mbUa~mbUa >>> e-m-bwa~mbwa~mbwa (=T6a) b. ‘insect’ a-[ka-huka] >>> a-ka-huka~huka >>> a-ka-huka~huka (=T6n)

c. ‘chicken’ e-[n-goko] >>> e-n-goko~ngoko >>> e-n-goko~ngoko (=T6k) d. ‘cucumber’ e-[ri-oli] >>> e-ri-oli~rioli >>> e-rj-oli~rjoli (=T6s)

e. ‘bride’ o-[mu-heruki̧] >>> --- >>> --- (=T6gg)

The key to making a generalization for the authenticity reduplication lies in the assumption that syllabification takes place before reduplication. Once syllabification is taken into consideration, a certain pattern emerges for (1a-d), specifically paying attention to the number of onsets.

(Some speculation on (1e) is mentioned later.) Triplication happens when there is only one onset. When the number of onsets is bigger than one, reduplication happens. The inclusion of the prefix can be explained by positing that everything which follows the penultimate onset is

duplicated.

So, after syllabification, if the number of onsets is one, what is repeated starts from the first timing slot to the last one. If, after syllabification, the number of onset is bigger than one, the right limit is still the last timing slot, but the left limit needs to change. Although (1c-d) are ambiguous about the left limit to be the first onset or the penultimate onset, (1b) tells us that the correct reference is the penultimate onset. The penultimate onset and the penultimate C are interchangeable for (1b,d), but the same is not true for (1c)—the penultimate onset is ng together but the penultimate C is g only. To have a common expression for (1b-d) is the reason why onset is referred to here. When the penultimate onset is considered against only (1b), it means either the second onset searching from string-initial rightward or the second onset searching from string-final leftward. The former alternative is wrong once (1c-d) enter into the picture. Not only are the two sub-rules different on what to repeat, they are also different on how many times the repetition occurs. When there is only one onset, two repetition happens and leads to triplication; when there are more than one onset, only one repetition takes place.

Since it takes (at least) two rounds of Search to add a precedence relation which may effect reduplication after linearization, and triplication is deduced to be the post-linearized result of two added precedence relations, I formalize triplication as four rounds of Search in (2).

(2) authenticity for Kinande in S&C

Σ (string in the active workspace): noun stem plus prefix (but after augment prefixation) syllabification in prose: For a noun complex with one onset,

add two same precedence relations which make the last X precede the first X.

in prose: For a noun complex with more than one onset, add a precedence relation which makes the last X precede the penultimate onset.

The word ONSET in (2bγ) is small-capitalized to indicate that its use is not part of the vocabulary in Samuels (2010, 2011). (Recall from 2.1.2, the target of Search contains two sets: {first, second, stressed} and {X, C, V, foot}.) Triplication is achieved in (2a) by four rounds of Search.

What happens to be the case is that both the odd numbered Searches and the even numbered Searches each look for the same point. Four partial derivations are provided in (3). After syllabification, only (3a) has the number of onsets as one and only one, so this instance undergoes triplication (2a). The remaining, due to the number of onsets, have reduplication (2b) applied. The reason that I can generalize across the surface partial reduplication (3b) and the surface total reduplication (3c-d) is attributed to the reference of the penultimate onset (in addition to syllabification, of course). Syllabification and the penultimate onset together are

also the reason why sometimes the prefix portion is not reduplicated (3b) but sometimes it is (3c-d). All that matters is to identify the two objects: the last X-slot and the penultimate onset, if the number of onset is bigger than one—the surface variations are simply descriptive.

(3) four partial derivations

a. ‘dog’ b. ‘insect’

e-[m-bUa] a-[ka-huka]

syllabification>>>

e-[mbU.a.] a-[ka.hu.ka.]

authenticity reduplication>>>

e-[mbU.a.] a-[ka.hu.ka.]

linearization>>>

embUambUambUa akahukahuka other phonological processes>>>⁵

embwambwambwa ---

c. ‘chicken’ d. ‘cucumber’

e-[n-goko] e-[ri-oli]

syllabification>>>

e-[ngo.ko.] e-[ri.o.li.]

authenticity reduplication>>>

e-[ngo.ko.] e-[ri.o.li.]

linearization>>>

engokongoko eriolirioli other phonological processes>>>

--- erjolirjoli

Cases like (1e) are the reason why the Morpheme Integrity Constraint was postulated

(Mutaka and Hyman 1990: 83). As they “are relatively rare and are often borrowings or have

5 “…rules of gliding or vowel coalescence take place.” (Mutaka and Hyman 1990: 78)

compound stems” (ibid.), their indifference to the reduplication in question might be related to the layers of Kinande lexicon. Because any such speculations which resort to how information is structured in the lexicon (whether in the model of Lexical Phonology or not) await further studies based on well-designed experiments, I leave this issue for further study.

Table 6 Kinande authenticity reduplication data (M&H90 = Mutaka and Hyman 1990; M0/1 = Mutaka 2000/1)⁶

gloss UR IR SR M&H90:

a. ‘dog’ e-[m-bUa] e-m-bUa~mbUa~mbUa e-m-bwa~mbwa~mbwa 80

b. ‘belly’ e-[n-da] e-n-da~nda~nda e-n-da~nda~nda 80

c. ‘cow’ e-[n-de] e-n-de~nde~nde e-n-de~nde~nde 80

d. ‘wedding’ e-[n-dUa] e-n-dUa~ndUa~ndUa e-n-dwa~ndwa~ndwa 80

e. ‘at home’ Ø-[e-ka] Ø-e-ka~eka~eka Ø-e-k~ek~eka 82

f. ‘cabbage’ e-[sUa] e-sUa~sUa~sUa e-swa~swa~swa 81, 82

g. ‘child’ o-[mu-ana] o-mu-ana~muana o-mw-ana~mwana 78

h. ‘arm’ o-[ku-boko] o-ku-boko~boko o-ku-boko~boko 76

i. ‘sheep’ e-[m-buli] e-m-buli~mbuli e-m-buli~mbuli 79

j. ‘grain’ e-[m-buto] e-m-buto~mbuto e-m-buto~mbuto 79

k. ‘chicken’ e-[n-goko] e-n-goko~ngoko e-n-goko~ngoko 79

l. ‘back’ o-[mu-gongo] o-mu-gongo~gongo o-mu-gongo~gongo 76

m. ‘leg’ o-[ku-gulu] o-ku-gulu~gulu o-ku-gulu~gulu 76

n. ‘insect’ a-[ka-huka] a-ka-huka~huka a-ka-huka~huka 76

o. ‘killer’ o-[mu-iti] o-mu-iti~muiti o-mw-iti~mwiti 78

p. ‘woman’ o-[mu-kali̧] o-mu-kali̧~kali̧ o-mu-kali̧~kali̧ 76

q. ‘bowels’ e-[bi-la] e-bi-la~bila e-bi-la~bila 77

r. ‘village’ o-[mu-longo] o-mu-longo~longo o-mu-longo~longo 76

s. ‘cucumber’ e-[ri-oli] e-ri-oli~rioli e-rj-oli~rjoli 75, 78

t. ‘watch’ e-[Ø-saha] e-Ø-saha~saha e-Ø-saha~saha 81

u. ‘girl’ o-[mu-sika] o-mu-sika~sika o-mu-sika~sika 76

v. ‘market’ e-[Ø-soko] e-Ø-soko~soko e-Ø-soko~soko 81

w. ‘lion’ e-[Ø-soro] e-Ø-soro~soro e-Ø-soro~soro 81

6

x. ‘towels’ e-[Ø-sume] e-Ø-sume~sume e-Ø-sume~sume 81

y. ‘grinder’ o-[mu-so-i] o-mu-so-i~musoi o-mu-sw-i~muswi 77, 105 z. ‘grinding place’ e-[ki-so-er-o] e-ki-so-er-o~soero e-ki-sw-er-o~swero 117

aa. ‘stick’ a-[ka-ti] a-ka-ti~kati a-ka-ti~kati 77

bb. ‘head’ o-[mu-tue] o-mu-tue~mutue o-mu-twe~mutwe 75, 77

cc. ‘room’ e-[ki-umba] e-ki-umba~kiumba e-kj-umba~kjumba 75, 78

dd. ‘snake’ e-[n-zoka] e-n-zoka~nzoka e-n-zoka~nzoka 79

ee. ‘playing place’ e-[ki-es-er-o] e-ki-es-er-o~sero ?e-kj-es-er-o~sero⁷ 117

ff. ‘sleepiness’ o-[tu-gotseri] --- --- 83

gg. ‘bride’ o-[mu-heruki̧] --- --- 83

hh. ‘bathing place’ e-[ki-nab-ir-o] --- --- 104

ii. ‘butterflies’ e-[bi-nIurugunzu] --- --- 83

jj. ‘playing place’ e-[ki-sat-ir-o] --- --- 104

kk. ‘kind of tree’ e-[ki-tembekali̧] --- --- 83