Comparative and Collective Reduplications

The comparative and collective reduplication constructions both target adjective (or stative verb) roots as their working domain. In terms of meaning, the comparative reduplication adds comparison, similar to –er in English, and the collective reduplication makes all the members in a set bear the property of the adjective. In terms of stress distribution, irrespective of a root’s underlying stress position, comparative reduplication forms bear final stress, but collective

reduplication forms bear penultimate stress. ² Many of the examples for these two reduplications also take the ‘stative verb’ prefix ma- (Rau and Dong 2005: 127, 2006: 66).

For the comparative reduplication, when (1a-b) is contrasted with (1c), two things are noticed: firstly, the root-initial vowel is not part of the reduplication; secondly, the root-final consonant is not reduplicated.

(1) comparison in Tao

a. rahet ‘bad’, ma-rahe-rahet ‘worse’ (=T4j) b. -sngen ‘close’, ma-snge-sngen ‘nearer’ (=T4p) c. anáro ‘very long’, a-naro-naro ‘longer’ (=T4a)

So, the left side of reduplicant starts with the first consonant from the root-initial position. As for the right side, although (1a) can be described as ending either with the second vowel or with the final vowel, the forms (1b-c) show that the correct characterization refers to the final vowel. In (1b) there is no “second vowel” and in (1c) it is the third vowel that is relevant. So, the right edge of the reduplicant corresponds to the final vowel. The comparative reduplication is then analyzed as in (2) for the span L|the first C…the last V|R.

2 The stress distribution is not further discussed. As a first approximation, the situation seems to suggest that before morpheme concatenation, the stress associated with either the penultimate or final vowel gets floated and that after linearization, the floated stress reassociates with the relevant vowel according to the reduplicative patterns. Or, the underlying stress gets reduplicated during morpheme concatenation; however, after linearization, the two stresses are conflated into one and that single stress is drawn to the relevant vowel accordingly. (It need not be conflation: perhaps the two stresses further project with the rightmost stress as head.) Also, it is reasonable

(2) comparison for Tao in S&C

Σ (string in the active workspace): adjective root ϛ (initiator of Search): ϛi → ϛj

γ (target of Search): γi = first V; γj = first C δ (direction of Search): δi = L; δj = R

β (beginning point of Search): βi = %; βj = #

in prose: For an adjective root, let the last V precede the first C.

e.g.: # → a → n → a → r → o → % morpheme concatenation>>>

# → a → n → a → r → o → % linearization>>>

# → a → n → a → r → o → n → a → r → o → %

To the collective reduplication, the description that the initial two X-slots are repeated from (3a) is modified to be the initial CV because the very beginning vowel in (3b) is skipped.

(3) collectivity in Tao

a. pía ‘OK, sure’, a-pi-pía ‘all very beautiful, all very good’ (=T4i) b. anáro ‘very long’, a-na-náro ‘all long’ (=T4a)

c. -sngen ‘close’, ma-sé-sngen ‘all near’ (=T4p)

This modification continues when (3c) comes in: the initial C and the initial V do get repeated, but the vowel need not be immediately following the consonant. The velar nasal in (3c) is not involved. Unlike in the comparative reduplication, three rounds of Search are proposed in (4) to deal with this discontinuity between the initial C and V.

(4) collectivity for Tao in S&C

Σ (string in the active workspace): adjective root ϛ (initiator of Search): ϛi → ϛj → ϛk

γ (target of Search): γi = first C; γj = first V; γk = first C δ (direction of Search): δi = R; δj = R; δk = R

β (beginning point of Search): βi = #; βj = γi; βk = #

in prose: For an adjective root, find the first C and the first V which follows this C. Let the identified C precede the identified V, and let the same V precede the same C.

When the initial C and V are indeed adjacent to each other (5a), the effect of the precedence relation added by the Searches i and j (the upper one in (5)) seems to be vacuous. In cases where this condition is not held (5b), the difference between the original precedence relations and the one constructed by the Searches i and j reveals.

(5) schematic derivation of Tao collective reduplication

a. the initial C and V are contiguous b. the initial C and V are near

# → (V →) C → V… # → (V →) C → Cⁿ → V…

morpheme concatenation>>>

# → (V →) C → V… # → (V →) C → Cⁿ → V…

linearization>>>

# → (V →) C → V → C → V… # → (V →) C → V → C → Cⁿ → V…

e.g. a-na-náro e.g. ma-sé-sngen

Table 4 Tao comparison and collectivity data (R&D05 = Rau and Dong 2005; R&D06 = Rau and Dong 2006; R12 = Rau et al. 2012)³

root stem comparative collective R&D R12:

a. anáro ‘very long’ a-naro-naro ‘longer’ a-na-náro ‘all long’

(weight)’ (weight)’

ma-loi-loit ‘dirtier’ ma-lo-lóit ‘all very dirty’

a-pia-pia ‘better’ a-pi-pía ‘all very beautiful, all very

ma-sari-sari ‘darker’ ma-sa-sári ‘all dark’

4 In Rau and Dong (2005: 78), the stress of the root and stem is realized on the ultimate syllable.

5 In Rau and Dong (2005: 66), the stress of the stem is transcribed on the penultimate syllable.

‘close’ ‘close, near’⁶ ‘nearer’ near’

q. tava ‘fat, grease’

matáva ‘vert fat’

ma-tava-tava ‘fatter’ ma-ta-táva ‘all thick, all fat’

The reduplicative span read from (6a-b) stretches from the first X-slot, irrespective of its association with a consonant or a vowel, to the vowel which can be analyzed as being the second or the last depending on seeing from the root-initial or root-final position. Instances like (6c) whose number of vowels is more than two are decisive and regulate that the point to be found is the second vowel. Should the point be the last vowel, total reduplication is incorrectly predicted for (6c).

(6) plurality in Tao

a. kois ‘pig’, koi-kois ‘pigs’ (=T5g)

b. avong ‘shadow, image’ (avo-avong→)avwavong ‘shadows’ (=T5b)

c. inapo ‘ancestor, name given to a deceased ancestor of three generations or above’, (ina-inapo→)ineynapo ‘ancestors’ (=T5e)

To formalize L|the first X-slot…the second V|R, I postulate (7) for the plural reduplication.

(7) plurality for Tao in S&C

Σ (string in the active workspace): noun root ϛ (initiator of search): ϛi → ϛj

γ (target of search): γi = second V; γj = first X δ (direction of search): δi = R; δj = R

β (beginning point of search): βi = #; βj = #

in prose: For a noun root, let the second V precede the first X.

e.g.: # → k → o → i → s → % morpheme concatenation>>> # → k → o → i → s → % linearization>>> # → k → o → i → k → o → i → s → %

Table 5 Tao plurality data (R&D05 = Rau and Dong 2005; R&D06 = Rau and Dong 2006; R12

= Rau et al. 2012) plain

(pla)

reduplicated (red) gloss R&D05: R12:

pla red

a. anak (ana-anak→)ananak⁷ ‘child’ ‘children’ 61 16-17

b. avong (avo-avong→)avwavong ‘shadow, image’ ‘shadows’ 61 40

c. baka baka-baka ‘cow’ ‘cows’ 61 45

d. ciring ciri-ciring ‘words’ ‘words,

language’

61 55-56 e. inapo (ina-inapo→)ineynapo ‘ancestor, name

given to a

j. oned (one-oned→)onewned ‘heart (figurative),

k. soli so-soli ‘taro’ ‘taro (plural)’ 62 229

l. tao tao-tao ‘human, what

the Tao call themselves’

‘someone else, crowd’

61 238-239

m. tapi tapi-tapi ‘board, flat’ ‘planks’ 61 239

n. tokon toko-tokon ‘mountain’ ‘mountains

(plural)’

61 252

o. vato vato-vato ‘stone’ ‘stones’ 61 266

p. zipos zipo-zipos ‘family,

relatives’

‘relatives’ 61 290

2.4 Kinande

In this section, I discuss two reduplicative patterns in Kinande (one for nouns in 2.4.1 and one for verbs in 2.4.2). The main purpose is to demonstrate that the prosodic adjustment of Frampton (2009) is not necessary by proposing a pure rule analysis. This alternative analysis counts onsets, reviving an idea suggested by John A. Goldsmith (mentioned by Mutaka and Hyman 1990: 116 endnote 16). Like Frampton (2009: 133), I base my analysis on the data from Mutaka and Hyman (1990) and Mutaka (2000/1).

I assume with Mutaka and Hyman (1990) that syllabification¹ comes before reduplication,² but I do not use their Morpheme Integrity Constraint or any of its more recent incarnations since only rules are presumed in the present study.³