Penn Linguistics Colloquium
February 27, 2004
Deriving Coda Conditions through the Generalized Local Conjunction of markedness constraints
Laurie Woods
New York University
I.Introduction
1.1In languages that allow codas at all, many limit the consonants that can fill this position. Very often, these limitations are sensitive to two familiar kinds of markedness phenomena: place markedness and sonority.
1.2Place Markedness: Coronal segments are less marked than labial or dorsal segments.
Lardil (Australian, Pama-Nyungan, Tangic) and Finnish are examples of a language whose restrictions on coda segments reflect place markedness. In these languages,acoda consonant can have only Coronal place or else no place specification of its own at all. (Prince and Smolensky 1993)
1.3Coda Sonority: Coda segments of greater sonority are less marked than those of less sonority. (Clements 1992, Murray and Venneman 1983)
Fanti (Niger-Congo, West-Sudanic, Kwa) is an example of a language whose restrictions on coda segments reflect sonority considerations. In this language, only nasals and approximants are permissible codas. (Welmers 1946)
1.4There are also languages whose restrictions on coda segments reflect BOTH place markedness and sonority considerations.
Pitta-Pitta (Australian, Pama-Nyungan, Karnic, Palku) does not allow word-final codas. Word medially, however, only coronal sonorants are possible. Non-coronal nasals must be homorganic to the following onset. The coronal stop (e.g. /t/) is also not allowed in coda position. (Blake 1979)
II.Proposal
I propose to account for a wide range of coda conditions through the Generalized Local Conjunction (Gafos and Lombardi 1999) of two sub-hierarchies of familiar markedness constraints.
III.Constraints
3.1Place Markedness
Coronals Labials, Dorsals
*[lab/dor] > *[cor][1]
3.2Associational Harmony (Prince and Smolensky 1993)
M/d M/t … M/i M/a
*M/a > *M/i > … *M/d > *Mt
3.3Coda Sonority
C/approximant C/nasal C/obstruent
*C/obstruent > *C/nasal > *C/approximant
IV.Constraint Conjunction
4.1Local Conjunction
A constraint C1 and a constraint C2 can be conjoined into the constraint C1& C2, which is violated when there is some domain of type D in which both C1 and C2 are violated. (Smolensky 1995)
4.2Generalized Local Conjunction
Given two constraint hierarchies C = C1 > C2 > …Cn and D = D1 > D2 > … Dm, their generalized local conjunction CD is defined by the rankings:
For every i,j,k,l: if Ci >Cj Ci&Dk > Cj& Dl
Else if i=j and Dk > Dl Ci&Dk > Cj&Dl (Gafos and Lombardi 1999)
When two hierarchies of two or more constraints are conjoined, the question of which hierarchy heads the operation becomes crucial: the process is not commutative.
If C = [C1 > C2] and D = [D1 > D2], C D ≠ D C. The resulting hierarchy of the first operation is [C1& D1 C1& D2C2& D1C2& D2]. The resulting hierarchy of the second operation differs in the ranking of the middle two constraints. It is [C1D1> C2D1 > C1 D2 > C2D2]. When a hierarchy heads the GLC operation it will be said to have priority over the second hierarchy.
Other proposals in the literature define similar operations (Aissen 1998, Artstein, 1998, cited in Gafos and Lombardi 1999). These proposals differ from the GLC in that, given the hierarchies C and D, above, they do not assume that C2& D1 and C1& D2 are ranked with respect to one another, nor do they assume that either hierarchy would be given priority.
I will employ the GLC in my analysis. The data introduced in section 6 suggest that this operation is correct.
4.3Generalized Local Conjunction of Place and Coda Sonority Sub-Hierarchies
(1)Place has priority
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] >
*C/obst&[cor] > *C/nasal&[cor] > *C/approx&[cor]
(2)Sonority has priority
*C/obst&[lab/dor] > *C/obst&[cor] > *C/nasal&[lab/dor] >
*C/nasal&[cor] > *C/approx&[lab/dor] > *C/approx&[cor]
V. Deriving Coda Conditions
5.1 Lardil Coda Condition: A coda consonant can have only Coronal place or else no place specification of its own at all. (Prince and Smolensky 1993)
5.2 Place has priority
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] > Max-IO
*C/obst&[cor] > *C/nasal&[cor] > *C/approx&[cor]
(1) Lardil
/Naluk/ / Dep-IO / Ident(place) / C/&[lab/dor] / Max-IO / C/&[cor]a. Naluk / *!
b. Nalu t / *! / *
c. Naluk A / *!
d. Nalu <k> / *
(2) Lardil
/Nalut/ / Dep-IO / Ident(place) / C/&[lab/dor] / Max-IO / C/&[cor]a. Nalut / *
b. Naluk A / *!
c. Nalu <t> / *!
In these tableaux, C/&[lab/dor] stands for the first three conjoined constraints in the sub-hierarchy; C/&[cor] stands for the last three conjoined constraints in the sub-hierarchy.
5.3 Pitta-Pitta Coda Condition (word-medial)
Licit word-medial codas are:
(a) homorganic nasal plus stop, (b) homorganic lateral plus stop, or (c) apical nasal or lateral or rr (Blake 1979, p. 188)
5.4 Place has priority
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] >
*C/obst&[cor] > FAITH > *C/nasal&[cor] > *C/approx&[cor]
(3) Pitta-Pitta
/kim.pa/ / *C/PLACE / FAITH / MaxPlace-IO / *C/nasal&[cor] / *C/approx&[cor]a. kim.pa
\ |
[lab] / *
b. kim.pa
| |
[lab][lab] / *!
c. kin.pa
| |
[cor][lab] / *! Ident-IO (plc) / *
d. kip.pa
\ |
[lab] / *! Ident-IO (man)
e. kimA.pa / *!
(Dep-IO)
f. ki<m> pa / *!
(Max-IO)
(4) Pitta-Pitta
/yan.ka/ / *C/PLACE / FAITH / MaxPlace-IO / *C/nasal&[cor] / *C/approx&[cor]a. yan.ka / *
c. yaN.ka
\ |
[dor] / *!
d. yanAka / *!
(Dep-IO)
e. ya<n>ka / *!
(Max-IO)
In these tableaux, *C/PLACE stands for the first four conjoined constraints of the sub-hierarchy.VI. Implications
(1) If FAITH is ranked above either sub-hierarchy, it would yield the condition that all segments are allowed in coda position.
6.1 Place has priority
RANKINGCONDITION
(2)*C/obst&[lab/dor] > FAITH > *C/nasal&[lab/dor] > *C/approx&[lab/dor] >*C/obst&[cor] > *C/nasal&[cor] > *C/approx&[cor] / Labial and dorsal obstruents are banned. All other segments are allowed
(3)
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > FAITH > *C/approx&[lab/dor] >*C/obst&[cor] > *C/nasal&[cor] > *C/approx&[cor] / Labial and dorsal obstruents and nasals are banned. All coronals and labial and dorsal approximants are allowed.
(4)
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] > FAITH *C/obst&[cor] > *C/nasal&[cor] > *C/approx&[cor] / All labial and dorsal segments are banned. All coronals are allowed.
(5)
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] > *C/obst&[cor] > FAITH >*C/nasal&[cor] > *C/approx&[cor] / All labial and dorsal segments and coronal obstruents are banned. Coronal nasals and approximants are allowed
(6)
*C/obst&[lab/dor] > *C/nasal&[lab/dor] > *C/approx&[lab/dor] > *C/obst&[cor] > C/nasal&[cor] > FAITH > *C/approx&[cor] / Only coronal approximants are allowed. All other segments are banned.
6.2 Coda-Sonority has priority
RANKINGCONDITION
(7)*C/obst&[lab/dor] > FAITH >*C/obst&[cor] >
*C/nasal&[lab/dor] > *C/nasal&[cor] > *C/approx&[lab/dor] > *C/approx&[cor] / Labial and dorsal obstruents are banned. All other segments are allowed
(8)
*C/obst&[lab/dor] > *C/obst&[cor] > FAITH >
*C/nasal&[lab/dor] > *C/nasal&[cor] >
*C/approx&[lab/dor] > *C/approx&[cor] / All obstruents are banned. All nasals and approximants are allowed.
(9)
*C/obst&[lab/dor] > *C/obst&[cor] >
*C/nasal&[lab/dor] > FAITH >*C/nasal&[cor] >
*C/approx&[lab/dor] > *C/approx&[cor] / All obstruents and labial and dorsal nasals are banned. Coronal nasals and all approximants are allowed.
(10)
*C/obst&[lab/dor] > *C/obst&[cor] >
*C/nasal&[lab/dor] > *C/nasal&[cor] > FAITH >
*C/approx&[lab/dor] > *C/approx&[cor] / All nasals and obstruents are banned. All approximants are allowed.
(11)
*C/obst&[lab/dor] > *C/obst&[cor] >
*C/nasal&[lab/dor] > *C/nasal&[cor] >
*C/approx&[lab/dor] > FAITH >*C/approx&[cor] / Only coronal approximants are allowed. All other segments are banned.
(12) If FAITH is ranked below either sub-hierarchy, it would yield the condition that all segments (with place) in coda position.
Rankings (2) and (7) and Rankings (6) and (11) are the same.
Rankings (3) and (4) and Rankings (10) and (11) differ only on the set of labial and dorsal approximants: /V/, labial dental approximant, /Â/, velar approximant, /;/, velar lateral approximant, and /w/, labial and dorsal approximant
6.3 Typology of Coda Conditions and languages that attest them[2]
# / Condition / RankingNumber(s) / Language(s)
1 / All segments are allowed. / (1) / ENGLISH
2 / Labial and dorsal obstruents are banned. All other segments are allowed. / (2), (7) / SPANISH, GALICIAN,
SAWERU
(West-Papuan)
3 / Labial and dorsal segments are banned. All coronals are allowed. / (3), (4) / LARDIL, FINNISH
4 / All labial and dorsal segments and coronal obstruents are banned. Coronal nasals and approximants are allowed. / (5) / PITTA PITTA, WATJARRI (Pama-Nyungan, Wadjari)
JAFFNA TAMIL
5 / All segments except coronal approximants are banned. / (6), (10), (11) / ITALIAN
6 / All obstruents are banned. All nasals and approximants are allowed. / (8) / FANTI,
GUMBAYNGGIR
(Pama-Nyungan, Gumbaynggiric)
7 / All obstruents and labial and dorsal nasals are banned. Coronal nasals and all approximants are allowed. / (9) / WARGAMAY (Pama-Nyungan, Dyirbalic)
8 / All segments with place are banned. / (12) / JAPANESE
References
Aissen, J. 1998. Markedness and subject choice in Optimality Theory. Ms., UC Santa Cruz.
Artstein, R. 1998. Hierarchies. Ms., Rutgers University, NJ.
Asher, R.E. 1985. Croom Helm Descriptive Grammers: Tamil. Croom Helm, Dover, NH
Blake, B. 1979. Pitta Pitta. In R.M.W. Dixon and Barry J. Blake, (eds.) Handbook of Australian Languages, Volume I. Amsterdam: John Benjamins B.V.: 183-242
Clements, G.N. 1992. The Sonority Cycle and syllable organization. In W. Dressler, H. Luchutzky, O. Pfeiffer and J. Rennison, (eds.) Phonologica 1988: proceedings of the 6th International Phonology Meeting. New York: Cambridge University Press: 63-76
Dixon, R.M.W. 1981. Wargamay. In R.M.W. Dixon and Barry J. Blake, (eds.) Handbook of Australian Languages, Volume II Amsterdam: John Benjamins B.V.: 1-143
Douglas, W. 1981. Watjari. In R.M.W. Dixon and Barry J. Blake, (eds.) Handbook of Australian Languages, Volume II Amsterdam: John Benjamins B.V.: 197-271
Eades, D. 1979.Gumbaynggir. In R.M.W. Dixon and Barry J. Blake, (eds.) Handbook of Australian Languages, Volume I. Amsterdam: John Benjamins B.V.: 245-361
Harris, James W. 1983. Syllable Structure and Stress in Spanish: A Nonlinear Approach. MIT Press, Cambridge, MA
Ito, J. 1988. Syllable Theory in Prosodic Phonology. New York: Garland
Eades, D. 1979.Gumbaynggir. In R.M.W. Dixon and Barry J. Blake, (eds.) Handbook of Australian Languages, Volume I. Amsterdam: John Benjamins B.V.: 245-361
Gafos, A. and L. Lombardi. 1999. Consonant transparency and vowel echo. NELS 29
Lombardi, Linda. 2002. Coronal epenthesis and markedness. In Phonology 19: 219-251
Loving, Richard. 1973. “Awa” in The Languages of the Eastern Family of the East New Guinea Highland Stock, Howard McKaughan, ed. University of Washington Press, Seattle
Murray, R.W. and T. Venneman. 1983. Sound change and syllable structure in Germanic phonology. Language 59: 514-528
Nikiema, E. 1992. More than coda conditions in Italian phonology. In C. Laeufer and T. Morgan, (eds.) Theoretical Analyses in Romance Linguistics. Philadelphia: John Benjamins: 3-18
Parker, Steve. 1989. The Sonority Grid in Chamicuro phonology in Linguistic Analysis19:3-58.
Prince, A. and P. Smolensky. 1993. Optimality Theory: Constraint Interaction in Generative Grammar. Ms. Rutgers University and University of Colorado at Boulder
Rugueira Fernandex, X. L. 1996. “Galician” in Journal of the International Phonetic Association 26(2): 119-122
Smolensky, P. 1995. “On the internal Structure of the constraint component Con of UG.” Handout of talk given at UCLA, April 7, 1995.
Sulkala Helena and Merja Karjalainen. 1992. Finnish. Routledge, New York
Welmers, William Everett. 1946. A Descriptive Grammar of Fanti. Linguistic Society of America, Baltimore, Maryland.
[1] I am not considering Pharyngeal place, because of the controversy surrounding the markedness of Pharyngeals, but whether Pharyngeals are the least marked place (Lombardi 2002) or whether they are the most sonorant segments (Parker 1989), the existence of languages like Awa (Loving 1973) that allow only /// and /h/ in coda position is consistent with my analysis.
[2] References for languages: Spanish: Harris (1983); Galician: Regueira Fernandez (1996); Saweru: Donohue (p.c.); Lardil: P&S (1993); Finnish: Sulkala & Karjalainen (1992); Pitta-Pitta: Blake (1979), Watjarri: Douglas (1981); Jaffna Tamil: Asher (1985); Italian: Nikiema (1992); Fanti: Welmers (1946); Gumbaynggir: Eades (1979); Wargamay: Dixon (1981); Japanese: Ito (1988)