saghala Cedric Patin

Tone Shift and Tone Spread in Saghala,a Bantu Language of Kenya: 

Tone Shift and Tone Spread in Saghala, a Bantu Language of Kenya Cédric PATIN Laboratoire de Phonétique et Phonologie – UMR 7018 (CNRS/Sorbonne-Nouvelle) Université Paris III

SAGHALA (E 74b): 

SAGHALA (E 74b) pitch-accent system tone shift & tone spread 5 successive rules  Domain Analysis

TONAL PATTERNS: 

TONAL PATTERNS dissyllabic noun roots = a single tonal contrast: (1) Low /ki  tanda/ kitanda bed /²  kuku/ Nguku chicken  (2) High-Low /ma  z@so/ maziso@ eyes   /wa  lu@me/ Balume@ men

TONAL PATTERNS: 

TONAL PATTERNS monosyllabic noun roots are L: (3) /mu  li/ muli body /mi - zi/ mizi cities

TONE SHIFT: 

TONE SHIFT (4) niÄulaÄa I will buy  VS niziÄu@laÄa I will buy them (goats)  (5) nïovu elephant  ilya nïo@vu@ that elephant 

TONE SPREAD: 

TONE SPREAD (6) Nïovu elephant iz"Û nïo@vu these elephants  ilya nïo@vu@ mbwa@a that big elephant  [that - elephant - big] elephant = 3 different tone patterns depending on the preceding word 3rd pattern = H.H VS underlying = L.L  Tone Spread applies after Tone Shift

LENGHT OF THE SPREADING: 

LENGHT OF THE SPREADING last example of (6): ilya nïo@vu@ mbwa@a tone spreads as far as the adjective « big » (mbwaa)  if the demonstrative is deleted, then the entire phrase is low: nïovu mbwaa big elephant 

LENGHT OF THE SPREADING: 

LENGHT OF THE SPREADING 3rd utterance of (6): tone spreads across two syllables ilya nïo@vu@ mbwa@a 2nd utterance of (6): tone only spreads across one syllable iz"Û nïo@vu  the lenght of the spreading vary

LENGHT OF THE SPREADING: 

LENGHT OF THE SPREADING why? HYP : after shifting, spreading stops on the first syllable of the following word: (8) ih"Û mbu@la mbwaa this big nose   ilya mbu@la@ mbwa@a that big nose 

LENGHT OF THE SPREADING: 

LENGHT OF THE SPREADING but: (9) ivilya v@ta@nda vibwaa those big beds  *ivilya v@ta@nda@ v@bwaa

TONE CONTACTS: 

TONE CONTACTS (10) mbuzi@ "Ûzilya z"@luwe zaa@fwa| those dirty goats are dead  [ goat – those – dirty – are dead ] izilya mbu@zi zilu@we@ zaa@fwa| idem [those – goat – dirty – are dead ]

TONE CONTACTS: 

TONE CONTACTS mbuz"Û "Ûzilya z"@luwe izilya mbu@zi zilu@we@ 1st utterance : mbuzi = LH ziluwe = HLL 2nd utterance : mbuzi = HL ziluwe = LHH

TONE CONTACTS: 

TONE CONTACTS 2nd example: izilya mbu@zi zilu@we@ H tone on /mbu@zi/ (which is supposed to be on the syllabe zi) is deleted because of contact with High tone on /izilya@/

TONE CONTACTS: 

TONE CONTACTS spreading may cause tone contacts  ( 2nd tone will be deleted) (11) ilya za@wa@di mbwaa that big gift 

ACROSS WORD BOUNDARIES: 

ACROSS WORD BOUNDARIES (12) ih"Û mbuz"Û this goat  awa@ wana@ wa@lele@ these children are tall  [Dém - children - tall]

ACROSS WORD BOUNDARIES: 

ACROSS WORD BOUNDARIES ih"Û mbuz"Û the tone of the demonstrative does not spread, as expected, onto the following word (cf. iz"Û nïo@vu) no tone contacts + no deletion of lexical tone (cf. ilya za@wa@di)

SUMMARY: 

SUMMARY (13) A] cvcvcv@ cv@cvcv B] cvcvcv@ cv@cvcv@ C] cvcvcv@ cvcv@cv@ D] cvcvcv cv@cv@cv E] cvcvcv cv@cv@cv F] cvcvcv cv@cvcv G] cvcvcv cv@cv@ cv@cv.... tone shifts systematically onto the following syllable tone spreads at least once – after shifting - in A, B, D, E and G tone does not spread in C (1st tone) or F tone spreads twice in G

SUMMARY: 

SUMMARY G (cvcvcv cv@cv@ cv@cv...) = the only situation where the tone spreads across 2 syllables  exception  spreading is not reduced in A or D but extended in G

[C] VS [A]: 

[C] VS [A]

[F] VS [G/D]: 

[F] VS [G/D]

ANALYSIS: 

ANALYSIS an accentual language: no active low tone no tonal distinction on monosyllabic noun roots a single tonal contrast on disyllabic noun roots

ANALYSIS: 

ANALYSIS 5 successive rules and constraints, using Domain Analysis, which deal with all the situations domains have 2 advantages: a. deal in a simple and elegant way with a large variety of phenomena b. dismiss heavy representations with accentual and tonal features

RULE 1: TONE SHIFT: 

RULE 1: TONE SHIFT one syllable to the right: see A – G rule attested in the following eastern Bantu languages : Jita (E 25), Kikuyu (E 51), Daida (E 74a), Nyamwezi (F 22)... I propose that all accents, in nominal phrase, move simultaneously the left boundary of the domain is placed after the tone bearing syllable (14) mbu(z@) goat

RULE 2: TONE SHIFT: 

RULE 2: TONE SHIFT after shifting the accent’s tone spreads on the next syllable rule attested in the following eastern Bantu languages: Bemba (M 42), Gweno (E 65)... domain must be longer than a single syllable (15) i(h@ Nïo@)vu this elephant 

RULE 3: 2nd ACCENT DELETION: 

RULE 3: 2nd ACCENT DELETION 3rd rule claims that if 2 accents appear on a single word, the 2nd is deleted occurs after Tone Shift it is impossible to have 2 left boundary domains on a single word (16) izilya (za@wa@)(d@....  izilya (za@wa@)di....

RULE 4: SPREAD’S EXPANSION: 

RULE 4: SPREAD’S EXPANSION after Spreading, the last syllable of a word can bear a high tone if the first syllable of the next word does not yet bear a tone, this tone will spread on it this rule is also attested in many languages of the area (e.g. Digo (E 73), Nyamwezi) right boundaries of the word and the domain cannot coincide[1] (17) ilya (nïo@vu@) mbwaa  ilya (nïo@vu@ mbwa@)a [1] A constraint proposed by Cassimjee & Kisseberth (1998)

RULE 5: OCP REPAIR: 

RULE 5: OCP REPAIR Saghala suprasegmentals obey the OCP last rule states that any spreading which creates violations of the OCP is undone impossibility to have 2 successive high tones, that is 2 adjacent domains (18) i(hi@ mbu@)(zi@  i(hi@) mbu(zi@

CONCLUSION: 

CONCLUSION presentation: major features of the pitch-accent system of Saghala never been described until now an unusual system: local shifting and local spreading 5 ordered rules & constraints, using Domain Analysis.

ACKNOWLEDGEMENTS: 

ACKNOWLEDGEMENTS Laboratoire de Phonétique et Phonologie (UMR 7018 – CNRS) Fédération de Typologie Linguistique et Changements Diachronique Dave Roberts (corrections) Informants