AFLA 9 Cornell University April 26, 2002 Rhythmic vowel deletions and syllabification of Maga Rukai and Tsou Tien-Hsin Hsin Institute of Linguistics, Academia Sinica 1. Introduction Rhythmic vowel deletion (RVD) is a widely attested process that deletes vowels in alternating syllables. 2. RVD of Macushi Carib (Kager 1997) Kager 1997: there is no unitary phenomenon of RVD, but two varieties of deletion instead -- gradient and categorial, depending on the rhythmic structure it occurs. Gradient deletion (reduction): no opacity because deletion is incomplete with respect to its metrical context (see for example Macushi Carib). Categorial deletion: causing opacity since vowel deletion destroys its own context (see e.g., South-eastern Tepehuan). (1) RVD of Macushi (From Kager 1997: 466 taken from Hawkins 1950 and Abott 1991) 2σ /πατα/ (π τα :) place /πε-π ν/ (π π ν) no, never 3σ /πιριπι/ (πρι :).(πι :) spindle /υπατα/ ( πα :).(τα :) my place 4σ /ερεπαµ / ( ρε :).(πµ :) I arrive /πιρ τ -r / (π ρ :).(ρτ ) gunshot 5σ /υ-µαναρι-ρ / (µα :).(νρι :).(ρ :) my cassava grater /ψ-ακινα-τ?-τον/ (ψ κι :).(ντ? :).(τ something to comb him?ν) with 6σ /υ-ωαναµαρι-ρ / (ωα :).(νµα :).(ρρ : my mirror ) 1
7σ /αριµαρακαψαµ / (ρι :).(µρα :).(κψα : dogs ).(µ :) Macushi: Vowels in odd-numbered syllables, counting from left to right, are deleted (Hawkins 1950). A short, unstressed vowel is reduced to open transition before stops, and may be completely lost before voiced consonants (Abott 1991 in Kager 1997: 465). (2) a. /wαναµαρι/ (ωνα:).(µρι :) mirror b. /υ-ωαναµαρι-ρ / (ωα :).(νµα :).(ρρ :) my mirror (3) [Cv.Cv] [C.Cvv] (4) Rhythmic vowel deletion in Macushi is phonologically incomplete, occurring in free variation with vowel reduction. Reasons: i. Rhythmic vowel deletion in Macushi is an optional process, as reduced vowel and zero seem to be in mutually interchangeable realization; ii. if deletion were complete, it would violate otherwise unviolable phonotactic laws of consonant clusters; iii. Rhythmic Vowel deletion does not trigger resyllabification, since a number of phonological processes, such as nasal assimilation and vowel lengthening process that applies to open syllables, do not take place after vowel deletion. (5) MAX: Every segment in the input has a correspondent in the output. MAX does not require a correspondent of an input segment to retain the feature content of that input segment it only asserts that something be identifiable in the output as a realization of the input vowel, e.g. an open transition (Kager 1997: 469-70). (6) FT-FORM: (cover constraint defining the iamb as a foot of the shape (LH) or (H)) FTBIN: Feet are binary under moraic or syllabic analysis. WSP: Heavy syllables are prominent within the foot.) RHTYPE=IAMB: Feet are right-headed. 2
PK-PROM: Peak (x) > Peak (y) if x > y (7) REDUCE: Weak syllables dominate no vocalic features. (accounting for vowel reduction/ or the loss of vocalic features in weak syllables) (8) PARSE-σ: All syllables must be parsed by feet. (9) ALL-FT-R: All feet must stand at the right edge of PtWd. ((8) >> (9) to create an iterative foot parsing) (10) IDENT: If a segment is specified as [αf] in the input, then it must be specified as [αf] in the output. (11) MAX, REDUCE, FT-FORM, PARSE-σ >> ALL-FT-R >> IDENT (12) Input: /piripi/ MAX REDUCE FT-FORM PARSE-σ ALL-FT-R IDENT a. (p.ri:).(pi:) * * b. (pi:).(r.pi:) **! * c. (pi:).(ri:).(pi:) *,*!* d. p.(r.pi:) *! ** e. (p.ri).(pi:) PK-PROM! * * f. (pi.ri:).(pi:) *! * g. (pr.pi:) *! * (13) More generally, the prediction is that if RVD is foot-based, then it must be gradient. Conversely, if RVD is categorial, then it cannot be foot-based (Kager 1997, p.472). 3
3. RVD of Maga Rukai and Tsou (14) RVD of Maga Rukai UR Positive Negative Gloss a. /καπαρ / κπαρ ι-καπρ palm b. /ταβαλαναν / τβαλναν ι-κ-ταβλανν arm c. /ρ γ ρ γ / ργ ργ ι-κ-ρ γρ γ mountain d. /ταµαλ σ κ σ τµαλσ κσ κ ι-κ-ταµλ σκ σκ blue κ / e. /µαχα/ µχαα ι-κ-µαχαα eye (15) UR Finite Non-finite Negative Imperative Gloss a. /πιρι/ υ-πριι πιρι ι-πιριι πιρψ-αα choose b. /συλατ / υ-σλατ συλτ ι-συλτ συλτ-αα draw, write c. /ριγυ/ υ-ργυυ ριγυ ι-ριγυυ ριγω-αα know d. /δ ναι/ υ-δνεε δ νε ι-δ νεε δ νψ-αα lick e. /l Ν l Ν / u-λν λν λ Νλ Ν ι-λ Νλ Ν l Νl Ν-αα aim (16) RVD of Tsou (from Tsuchida 1976) UR Gloss a. /τοποσ / τποσ book (or something written) τ-µ-οπσ to write (AF) b. /σε?ε-ν τ ηi/ σ?ε-ντ η-α to cut (off with a bolo) (GF) σ-µ-ε?-ν τη to cut off (with a bolo) (AF) c. /τιριηi/ τριη-α to hang with a string (GF) τ-µ-ι ρηι to hang with a string (AF) d. /σοπορο/ σ-µ-οπρο to lay a mat (AF) συ-σπορ to lay a mat to sit (AF) e. /-πυροηο/ τ-µ-α?-πυρηο to fall down from the cliff (AF) σ-µ-ο-προη to jump down (AF) 4
4. Is RVD of Maga and Tsou gradient? (17) Characteristics of categorial deletion of Maga RVD: a. There is no secondary stress in Maga and Tsou that coincides with the alternating deletion pattern. b. Evidence from other phonological processes shows that syllable-governed phonology refers exclusively to the output syllabification: i) Lengthening of di-syllabic roots, ii) Stress assignment refers to output syllabification. (18) Lengthening of di-syllabic roots to maintain minimal length: a. /µαχα/ µχαα eyes b. /κυχυ/ κχυυ head louse c. /γιλι/ γλιι younger sibling (19) a. σιτο peanut not στοο / σιτοο b. β κ pig not βκ / β κ (20) Stress assignment refers to the output syllabification: Stress of Maga and Tsou invariably falls on the penult syllable of the surface forms. (21) Stress pattern of Maga Rukai 2σ /µαχα/ µχα α face /βυσυ/ βσυ υ bow /β κ / β κ pig 3σ /βαρατι/ βρα Τι sweet potato /Ν διϖι/ Νδι ϖι mouth 4σ /ρ γ ρ γ / ργ ργ mountain /κυλυβυνυ/ κλυ βνυ tortoise /βαλ βαλ / βλ βλ bamboo 5σ /ταβαλαναν / τβαλνα ν arm /τακασυλυ υ/ τκασλυ υ shrimp 5
(22) Stress pattern of Tsou 2σ /οκο/ ο κο child χφ. οκο -σι his child 3σ /τοποσ / τπο σ book /τιριηi/ τρι η-α to hang with a string (GF) 4σ /τ-vµ-ι ριη/ τ-µ-ι ρηι to hang with a string (AF) /µισιπινι/ µσι πνι to wedge (23) a. /r g r g / (r.g ).(r.g ) b. /µισιπινι/ (m.si).(p.νι) (24) Conflicting foot parsing: Rhythmic vowel deletion -- iterative iamb feet Stress assignment -- final trochaic foot 5. A Metrical Analysis (cf. Idsardi 1992, Halle and Idsardi 1995) (25) Line 0 mark projection (Project) Project a line 0 element for each syllable head. (26) Edge-Marking parameters (Edge) Place a {left, right} parenthesis to the {left, right} of the {left, right}-most element in the string. (27) Iterative Constituent Construction parameter (ICC) Insert a {left, right} parenthesis for each pair of elements. (28) Head Location parameter (Head) Project the {left, right}-most element of each constituent onto the next line of the grid. (29) Parameter settings for RVD (Level 1): Edge: LLL ICC: R Head: R 6
X * (30) Syncope: V --> Ο / (31) Parameter settings for stess assignment (Level 2): Edge : RRR ICC : L Head : L (32) Level 1 Project * * * * ρ γ ρ γ Edge: LLL ( * * * * ρ γ ρ γ ICC: R (* *) * *) ρ γ ρ γ * * Head: R ( * *) * *) ρ γ ρ γ Syncope Ο Ο Level 2 Project Edge: RRR ICC: L Head: L Output * * ρ γ ρ γ * *) ρ γ ρ γ (* *) ρ γ ρ γ * (* *) ρ γ ρ γ ργ ργ 7
(33) Level 1 Project * * µ α χ α Edge: LLL ( * * µ α χ α ICC: R (* *) µ α χ α * Head: R ( * *) µ α χ α Syncope Ο Level 2 Project Edge: RRR Lengthen ICC: L Head: L Output * µ χ α *) µ χ α **) µ χ α α ( * *) µ χ α α * ( * *) µ χ α α µχα α 8
6. The syllabification of Maga Rukai and Tsou (34) Attested consonant clusters of Maga Rukai x = clusters appearing both word-initially and internally (x) = clusters appearing only word-internally C2 π β ϖ µ τ δ Τ σ χ ν λ ρ κ γ Ν C1 π (x) (x) x x x (x) x (x) (x) x x x x x x x β (x) x x (x) x x (x) x x x (x) x ϖ (x) (x) (x) (x) (x) x x x (x) (x) x µ x x (x) x (x) (x) x x x x x x x x x τ x x x x (x) x x x x x x x x δ (x) x (x) x x x x (x) x x (x) x x x x (x) x Τ x x x (x) x (x) x (x) x x x x x (x) σ x x x x x x x (x) x x x x x x (x) χ x x x x x (x) (x) (x) (x) x (x) x ν (x) (x) (x) (x) (x) (x) (x) (x) (x) λ x x x x (x) (x) (x) x x (x) x (x) (x) (x) ρ x x x x x (x) x (x) x (x) x (x) (x) x x x κ x x x x (x) x x (x) x x x x x (x) (x) (x) γ (x) (x) x (x) x (x) (x) (x) x x x Ν (x) (x) x x (x) (x) (x) (x) (x) x (x) (x) 9
(35) Attested consonant clusters of Tsou (adapted from Wright 1996: 35) C2 π φ ϖ µ τ χ σ ζ ν κ Ν? η C1 π x x x x (x) x x x x x φ x x x x x x x x ϖ x x (x) (x) (x) x x (x) (x) µ x x x (x) x x x x x x x x x x τ x x x x x x x x x x χ x x x (x) x x x x x x x σ x x x x x (x) x x x x ζ x (x) ν (x) (x) x (x) x x x x (x) x (x) κ (x) x x x (x) x x (x) (x) x Ν x x x x x (x) (x) (x) (x) (x) x? x (x) x x x x x (x) (x) (x) (x) η x x x x x x x x x x x (36) Similarities: a. Consonant clusters appear word-initially and -medially, but not word-finally; b. A consonant cluster contains at most two consonants; c. Almost any segment can appear as either the first or the second member in a cluster. (37) Differences: a. With few exceptions (clusters involving laryngeals and nasals, plus the cluster χζ), homorganic obstruent clusters (including geminates) are disallowed in Tsou, either at word-initial or medial position. b. Homorganic clusters and geminates are allowed in Maga, but geminates are not found in word-initial position. 10
6.1. Syllabification of Tsou Evidence from reduplication: (38) a. πυζυ fire πυ-πυζυ fire place b. φαηε fir φα-φαηε fir forest c. παι rice plant πα-παι rice field (39) a. φσυψυ bow φσυ-φσυψυ hunting bows b. ηχυψυ hill µα-ηχυ-ηχυψυ hills (40) a. νανηια friend, friendly να-να-νηι-νηια friendly to all b. να?ϖαµα father and son να-να-?ϖα-?ϖαµα father and sons c. νατ?οηαψσα two brothers να-να-τ?ο-τ?οηαψσα many brothers (Tung 1964: 167, 497) (41) Ladefoged 1994, Wright and Ladefoged 1997 : In a stop-stop cluster of Tsou, both members are fully released. (42) The syllable structure of Tsou σ R N X X X 11
(43) Restriction of homorganicity of Tsou a. OCP violation b. No-Release in Linked-Structure * X X * X X Place Place Place Place α α α 6.2. Syllabification of Maga Rukai (44) The word-medial syllable structure of Maga σ R N X X X (45) The word-initial syllable structure of Maga σ σ R N X X X α β (if α and β are not homorganic) 12
7. Conclusion Reference Abott, M. 1991. Macushi, in D. C. Derbyshire and G. K. Pullum (eds.), Handbok of Amazonian Languages, III. Berlin: Mouton de Gruyter, 23-160. Halle, Morris and William Idsardi. 1995. General Properties of Stress and Metrical Structure. The Handbook of Phonological Theory, ed. by John Goldsmith, 403-443. Cambridge, Mass.: Blackwell. Idsardi, William. 1992. The Computation of Prosody. Cambridge, Mass.: Ph.D. dissertation, MIT. Hawkins, W. N. 1950. Patterns of Vowel Loss in Macushi (Carib), International Journal of American Linguistics 34: 233-41. Kager, Réne 1997. Rhythmic Vowel Deletion, in Derivations and Constraints in Phonology, Roca Iggy (ed.). Oxford: Clarendon Press, 463-499. Ladefoged, Peter. 1994. The Links Between Theory and Fieldwork in Phonetics. University of California Working Papers in Phonetics, v. 88: 13-28. Tsuchida, Shigeru. 1976. Reconstruction of Proto-Tsouic Phonology. Study of Languages and Cultures of Asia and Africa, Monograph series No. 5: Tokyo. Tung, T.-H. 1964. A Descriptive Study of the Tsou Language, Formosa. Taipei: Institute of History and Philology: Special Publication: No. 48. Wright, Richard. 1996. Consonant Clusters and Cue Preservation in Tsou. Ph.D. dissertation, Los Angeles: UCLA. Wright, Richard and Peter Ladefoged. 1997. A Phonetic Study of Tsou. Bulletin of the Institute of History and Philology, 68.4: 987-1028. Tien-Hsin Hsin Institute of Linguistics Academia Sinica thsin@gate.sinica.edu.tw 13