Learning Center
Plans & pricing Sign in
Sign Out




                                      ROBERT KENNEDY
                                   UNIVERSITY OF ARIZONA
    This paper discusses reduplication in several Micronesian languages and its consequences
for three aspects of phonological theory: the Prosodic Hierarchy, templatic morphology, and
moraic representations. The results have implications for theories of phonological
derivation, representation, and typology.
    THE PROSODIC HIERARCHY. Descriptions of stress in Pohnpeian (Rehg 1993) and
Gilbertese (Blevins and Harrison 1999) suggest a construal of the Prosodic Hierarchy as a
violable principle. Stress in both languages is assigned to moraic elements rather than to
syllables, as in Pohnpeian àdaád ‘sharpen’, dunduné ‘attach’ and Gilbertese
bwa(ábwàro)(ákìna) ‘spread out all over something’. This suggests a foot structure built of
moras, independently of syllable structure. Similar structures are seen in non-Austronesian
languages such as Southern Paiute (Uto-Aztecan) (Cairns 2003) and Banawá (Amazonian)
(Everett 1996).
    In addition, the reduplication paradigms of several Micronesian languages also suggest
prosodic representations using strictly moraic feet. Indeed, there are instances where foot
and syllable boundaries are not perfectly aligned. The Mokilese vowel-initial subpattern is
one case—for example, the foot-sized prefix in an.d-an.dip ‘spit’ spans two syllables.
Furthermore, the regularity of Pohnpeian Quantitative Complementarity depends crucially on
a structure using bimoraic feet, not bisyllabic ones. I therefore suggest that the Prosodic
Hierarchy should be regarded as a robust tendency instead of an inviolable universal, and that
Micronesian prosody offers a rare example of the violability of the principle.
    TEMPLATIC MORPHOLOGY. The reduplicative systems of Pohnpeian (Rehg 1981),
Mokilese (Harrison 1976), Chuukese (Goodenough and Sugita 1980), and Kosraean (Lee
1975) motivate a reconsideration of theories of prosodic morphology. Whereas derivational
templatic models (e.g., McCarthy and Prince 1986) are well-suited to the heavy-syllable
prefixation of these languages, the template has proven especially troublesome for the
Optimality-Theoretic model.
    In Pohnpeian, monosyllabic stems create reduplicative prefixes with the opposite moraic
weight: heavy-light, as in paa-pa ‘weave’, and light-heavy, as in du-duup ‘dive’.
Polysyllables are more complicated; the prefix seems to respond to the weight of the second
syllable, but only if the first stem syllable is heavy. Thus we see to-tooroor ‘independent’
but duu-duupek ‘starved’. An output-driven template model cannot handle this distinction
without reference to moraic foot structure; I show that Quantitative Complementarity is
strictly local, but operates only in stems with even numbers of moras. As a result, the use of
an explicit template is unnecessary.
    In Mokilese, vowel-initial stems create prefixes that extend across syllable boundaries, as
in an.d-an.dip ‘spit’. Often analyzed as a heavy-syllable template, the Mokilese prefix is
shown to pose difficulties for an output-oriented template model unless we also accept the
use of moraic foot structure. As with Pohnpeian, the use of an explicit template is then
    Templatic theory also has nothing to say about the quantitative disparity between
bimoraic prefixes and monomoraic suffixes in Pohnpeian, Mokilese, and Chuukese. Like
Pohnpeian and Mokilese, Chuukese uses a bimoraic prefix, as in s s-st ‘trial’, kuk-kuus

‘use a blanket’, and fif-fini ‘choose’. However, each language has a monomoraic
reduplicative suffix; although it is less productive, its phonological shape is highly consistent,
as in Pohnpeian pika-pik ‘sandy’, Mokilese pirk-rk ‘braided’, and Chuukese meteki-tek
‘painful’ (final consonants are non-moraic). In a templatic model, the relationship between
quantity and position is accidental; however, the monomoraicity of the suffix is shown to
follow independently from the foot structure of each language, in which the final foot is
monomoraic. As a result, an explicit templatic requirement is unnecessary.
    In Kosraean, the reduplicative prefix’s shape varies by the shape of the stem: mono-
syllables receive a light syllable, as in fo-foš ‘to smoke’, but disyllables receive a closed
syllable, as in fur-furok ‘to turn again and again’. An explicit template cannot capture this
predictable distinction. I argue that moraic foot structure can be used to capture the pattern:
monosyllabic stems, which are also monomoraic, allow the prefix and stem in same foot, as
in (fo-foš). Disyllabic stems, which are bimoraic, force the prefix into its own foot, which in
turn must be bimoraic, as in (fur)(furok).
    THE REPRESENTATION OF GEMINATES. A third implication of Micronesian reduplication is
in the representation of moraic consonants. Notably, I discuss the one-root/two-root
opposition of geminate structures in representational theories, and show how both are
necessary. Indeed, both must be possible in the same language: in Chuukese, initial gemi-
nates (fft ‘plant’) must be single segments, while medial ones (fiffini ‘choose’) must be two.
In addition, I discuss the often stipulated claim of final-consonant extrametricality in
Chuukese, Ponapean, and Mokilese, each of which allows any consonant word-finally, but is
much more restrictive of medial codas. I show how extrametrical representations are driven
by the prosodic systems of such languages.
    SUMMARY. I summarize by discussing typological implications. Although I argue for
greater flexibility in the representation of prosody and geminates, which predicts a vast
typology of possible languages, I argue that some typological gaps are predicted even in
overgenerative formal theories. I introduce a typological model called Confluence (Kennedy
2003), which absolves the overpredictive power of generative phonological models like
Optimality Theory.
BLEVINS, JULIETTE, and SHELDON P. HARRISON. 1999. Trimoraic feet in Gilbertese.
   Oceanic Linguistics 38:203–30.
CAIRNS, CHARLES. 2003. Foot and syllable in Southern Paiute. Paper presented at the
   annual meeting of the Linguistic Society of America, Atlanta, GA.
EVERETT, DANIEL. 1996. Syllable integrity. Ms., University of Pittsburgh. Rutgers
   Optimality Archive, ROA-163.
GOODENOUGH, WARD, and HIROSHI SUGITA. 1980. Trukese-English dictionary. Honolulu:
   University of Hawai‘i Press.
HARRISON, SHELDON P. 1976. Mokilese reference grammar. Honolulu: University of
   Hawai‘i Press.
KENNEDY, ROBERT A. 2003. Confluence in phonology: Evidence from Micronesian
   reduplication. University of Arizona Ph.D. dissertation.
LEE, KEE-DONG. 1975. Kusaeian reference grammar. Honolulu: University of Hawai‘i

MCCARTHY, JOHN, and ALAN PRINCE. 1986. Prosodic morphology. Ms., University of
   Massachusetts and Rutgers University.
MCCARTHY, JOHN, and ALAN PRINCE. 1993. Prosodic morphology I: Constraint interaction
   and satisfaction. Ms., University of Massachusetts and Rutgers University.
PRINCE, ALAN, and PAUL SMOLENSKY. 1993. Optimality theory: Constraint interaction in
   generative grammar. RuCCS Technical Report #2. Piscataway: Rutgers Center for
   Cognitive Science.
REHG, KENNETH. 1981. Ponapean reference grammar. Honolulu: University of Hawai‘i
REHG, KENNETH. 1993. Proto-Micronesian prosody. Tonality in Austronesian languages,
   ed. by Jerold A. Edmonson and Kenneth J. Gregerson, 25–46. Oceanic Linguistics
   Special Publication No. 24. Honolulu: University of Hawai‘i Press.

To top