Contrastive Syllabification in Blackfoot
It is a common assumption of current phonological theory that syllabification is never used
contrastively among the world's languages (Hayes 1989, Blevins 1995, McCarthy 2003), as contrasts of
the type a.ta vs. at.a or ak.la vs. a.kla have not been convincingly documented. However, it is a matter of
debate whether the absence of such contrasts is simply empirical or whether this absence needs to be
encoded in formal linguistic theory (Morén 1999, McCarthy 2003). As it stands, such contrasts in
syllabification are indirectly predicted by moraic theory (Hyman 1985, Hayes 1989), where contrastive
consonant length is captured by associating long consonants (geminates) with a single mora underlyingly
and short consonants are assumed to be non-moraic underlyingly. Geminate consonants are realised as
ambisyllabic by virtue of the interaction between constraints requiring the preservation of underlying
moras and syllabic constraints requiring (a) that coda consonants be moraic (requiring the underlyingly
moraic consonant to be syllabified in coda position), (b) that onset consonants be non-moraic (eliminating
the possibility of preserving the underlying mora in onset position), and (c) that all syllables have an onset
(requiring that the underlyingly moraic consonant occupy both coda and onset position in order to
preserve its mora). Under this analysis, the a.ta and at.a contrast differs from the a.ta vs. at.ta contrast
only because languages tend to disprefer onsetless syllables, a syllable structure preference which is
presumably distinct from any constraints requiring mora preservation; it is a matter of debate whether any
language would rank the onset constraint lowly enough to result in such a contrast. On the other hand, the
contrast between ak.la and a.kla is subject to no such limitation--both sequences are well-formed in
terms of onset preferences. Moraic theory therefore predicts that languages with either of these contrasts,
and particularly the second, should exist. I argue that Blackfoot, an Algonquian language, fills this
typological gap by providing evidence of contrasts of the second type (ak.la/a.kla).
Blackfoot contrasts consonant length both intervocalically (/ni.na/ `man' vs. /nin.na/ `my
father') and pre-consonantally (/istawásiwa/ `he grew' vs. /istatánsiwa/ `he bragged', /ikstsíksijiwa/ `he
scratched himself' vs. /iksksiwa/ `it was stiff'). While the pa.ta/pat.a contrast is realised in Blackfoot as
geminate/non-geminate contrast (/ni.naa/ vs. /nin.na/), phonotactics, preliminary phonetic measurements,
and orthographic evidence suggest that the pre-consonantal contrasts listed above are indeed syllabified
differently, as follows: /i.stawásiwa/ vs. /is.tatánsiwa/ and /i.kstsíksijiwa/ vs. /ik.ksksiwa/.
I argue that contrastive syllabification in Blackfoot is not specified in the underlying
representation, but is instead a process which is dependent upon the underlying moraic associations of the
segment, following the traditional moraic analysis of geminates. In the examples above, the instances of
long /s/ and /k/ are underlyingly moraic, and are syllabified in coda position, as is warranted by the
preference for coda consonants to be weight-bearing segments. On the other hand, the instances of short
/s/ and /k/ are underlyingly non-moraic, and are syllabified as part of the onset, following from a
dispreference for weightless codas. Contrasts in syllabification on the surface can therefore be simply
accounted for by assuming that the segments in question are differentiated underlyingly by their moraic
association, a concept which is already an integral part of moraic theory.
A consequence of the moraic account is that it makes different predictions than does an account
in which syllabification is specified underlyingly. Note that in the examples given above, the realisation
of length in these contrasts is not consistent: pre-consonantal long /k/ is as long phonetically as the
intervocalic geminate /k/, while pre-consonantal long /s/ is not as long phonetically as the intervocalic
geminate /s/. A possible explanation for this difference involves the syllable contact law (Murray &
Vennemann 1983), which states that the sonority of two segments should decrease over a syllable
boundary. Given that fricatives are more sonorous than stops (Parker 2002), the main difference between
the two sequences /is.ta/ and /ik.sa/ is that the first adheres to the contact law, while the second does not.
Consequently, the sequence /is.ta/ may remain syllabified as is, while the underlyingly moraic /k/ must
become ambisyllabic (/ik.ksa/), and thus eliminate the poor syllable contact. This account parallels the
traditional moraic explanation for the ambisyllabic status of geminates by referring to syllable structure
constraints. Therefore, under the moraic account, we would predict that languages like Blackfoot would
contrast e.g. /i.ksa/ vs. /ik.ksa/; alternatively, in a language where syllable contact is not ranked so highly,
we could expect a contrast between /i.ksa/ and /ik.sa/. However, the moraic analysis makes the prediction
that no single language will contrast all of /i.ksa/, /ik.ksa/, and /ik.sa/, a possibility under syllable theory.

Contrastive Syllabification in Blackfoot
References:
Blevins, Juliette. 1995. The syllable in phonological theory. In J. A. Goldsmith (ed.) The Handbook of
Phonological Theory. Cambridge, MA: Blackwell.
Hayes, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic Inquiry 20 (2):253-306.
Hyman, Larry. 1985. A Theory of Phonological Weight. Publications in Languages Sciences 19. Foris,
Dordrecht.
McCarthy, John J. 2003. Sympathy, cumulativity, and the Duke-of-York gambit. In C. Féry & R. van de
Vijver The Syllable in Optimality Theory. Cambridge: Cambridge University Press, 23-76.
Morén, Bruce Timothy. 1999. Distinctiveness, coercion and sonority: A unified theory of weight. Ph.D.
dissertation, University of Maryland at College Park.
Murray, Robert W. and Theo Vennemann. 1983. Sound change and syllable structure in Germanic
phonology. Language 59 (3): 514-528.
Parker, Stephen G. 2002. Quantifying the sonority hierarchy. Doctoral dissertation, University of
Massachusetts, Amherst.