Phrasal and Boundary Tones in Shingazidja -...
Transcript of Phrasal and Boundary Tones in Shingazidja -...
Phrasal and Boundary Tones in Shingazidja
Cédric PATIN STL (UMR 8163 – Université Lille 3 & CNRS)
Conférence en l’honneur d’Annie RIALLAND
Paris – 3-4 juillet 2014
2
Ne parle pas de trop de choses ! Anonyme.
July 3, 2014, sur une péniche.
Introduction
3
Introduction
4
Announcement of a course by Firmin Ahoua: (http://www.als.rutgers.edu/Ahoua.html)
Introduction
Annie lead the field in Tone-Intonation interface (in African languages)
see for instance her recent works with Martial Embanga Aborobongui, with Laura Downing on Intonation Phrases in Chichewa and Embosi, her works on the African ‘lax’ question prosody…
Purpose of this talk: to demonstrate that Intonational Phrases are associated with a H% in Shingazidja.
5
Introduction
Shingazidja: Bantu language [G44a] spoken on Grande Comore (Comoros)
one of the five Comorian languages, along with Shindzuani [G44b], Shimwali [G44c], Shimaore [G44d] and Shikombani.
Data = mostly recorded in Paris with my first informant, Ibrahim Barwane (‘Western’ dialect – Moroni [M]), and in Lille with my current informant, Said Mohamed (‘Washili’ dialect – [W]), except:
‘Southern’ dialect [S]: Said Bacarzme (Fumbuni) – 07/08 & 03/09
‘Northern’ dialect [N]: Mohamed Miraddji Lihoma (Mbeni) – 07/08 & 03/09
6
Outline
Introduction
Background on Phonological Phrases in Shingazidja
Background on Intonational Phrases in Shingazidja
Boundary Tones in Shingazidja
(Is there any) Phrasal Tones in Shingazidja (…yet)?
7
8
This talk builds on a joined work with Kathleen O’Connor > O’Connor & Patin (under revision)
Background on Phonological Phrases in Shingazidja
9
Phonological Phrases: Background Rules identified and discussed by Tucker & Bryan (1970),
Cassimjee & Kisseberth (1989, 1992, 1993, 1998), Philippson (1988, 2005), Patin (2007, 2008, 2009)
1. ‘Unbounded’ tone shift (evidence for phonological phrase boundaries) blocked by a following underlying tone bearing unit1:
(1) a. wa-leví drunkards (2) b. i. wa-levi wá-raru three drunkards ii. wa-levi wa-íli two drunkards iii. wa-leví pia all drunkards c. i. ha-wono má-βaha he saw (some) cats ii. ha-wono wa-lévi he saw (some) drunkards
10
Phonological Phrases: Background 2. The tone shift leads to the deletion of every even-
numbered tone (OCP)
(2) ha-wono wa-lévi wa-il!í
1PER-see 2-drunkard 2-two ‘he saw two drunkards’
(3) ha-wono mí-pira mi-!íli
1PER-see 4-ball 4-two ‘he saw two balls’
11
Phonological Phrases: Background In (4), the tone shifts from the verb to the last word of the
sentence.
(4) ha-wono n-dovu m-ɓíli 1PER-see 10-elephant 10-two ‘he saw two elephants’
However, the tone shift rule cannot apply eveywhere. In (5), the tone of the subject cannot shift to the verb.
(5) m-limadjí ha-rem!é paha 1-farmer 1PER-beat (5-)cat ‘a farmer beat a cat’
12
13
(5’) ( m̩-limadjí )Φ ( ha-rem!é paha )Φ 1-farmer 1PER-beat (5-)cat ‘a farmer beat a cat
MLIMADJI HAREME PAHA
50
150
60
80
100
120
140
Pitc
h (H
z)
Time (s)0 1.17964
M
Phonological Phrases: Background Phonological phrase boundaries – (…)Φ – block the shift of
the tone.
A PhP boundary separates two coordinated nouns – (N) (& N) –, adverbs from a following VP – (Adv) (VP) –, a noun from a non restrictive relative – (N) (Relnon restr. ) –, etc.
Phonological phrases roughly correspond to XPMAX
(6) ( tsi-nika wa-nɖu má-pes!á )Φ 1SG.PER-give 2-person 6-money ‘I gave money to people’
(Cassimjee & Kisseberth 1992)
14
15
(6’) ( tsi-nika wa-nɖu má-pes!á )Φ 1SG.PER-give 2-person 6-money ‘I gave money to people’
TSI NI KA WA NDRU MA PE SA
60
130
80
100
120
Pitch
(Hz)
Time (s)0 1.41215
N
Phonological Phrases: Background
The augment, as in other Bantu languages, is preceded by a PP boundary…
(7) ( ha-niká )Φ ( ye=ɲ-uŋgu $n-dz!íro )Φ 1PER-give AT9=9-pot 9-heavy he gave the heavy cooking pot
…except when it cliticises to a preceding element.
(8) ( ha-nik’=é )Φ $( ɲ-uŋgu $n-dz!íro )Φ 1PER-give=AT9 9-pot 9-heavy he gave the heavy cooking pot
16
S
Phonological Phrases: Background
The phrasing is also conditionned by Focus in Shingazidja
In (9), the focalised verb phrases separately from the object
(9) a. ( [ŋg]am-andzo $tʃáy )Φ < /tʃaí/ 1SG.PRE-like 9.tea
‘I like tea’
b. ( [ŋg]am-andzó )Φ ( tʃa!í )Φ & ‘I LIKE tea’
17
M
Background on Intonational Phrases in Shingazidja
18
Intonational Phrases: Background
Consider (7’):
(7’) $( ha-niká )Φ ( ye=ɲ-uŋgu $ $n-dz!íro )Φ 1PER-give AT9=9-pot 9-heavy
‘he gave the heavy cooking pot’
In (x), the tone of the word ‘pot’ does not shift up to the final syllable of the sentence > Nonfinality
(7’’) $*ha-niká ye=ɲ-uŋgu n-dziró
19
NB. A tone can appear on the last syllable of a sentence – e.g. (x)
Intonational Phrases: Background
Nonfinality has been claimed to be the clue for Intonational Phrases, following F. Cassimjee & C. Kisseberth :
"Non-finality affects the last element in an intonational phrase" (Cassimjee & Kisseberth 1993)
"Nonfinality refers to the Intonational Phrase" (Cassimjee & Kisseberth 1998:127)
Also in Philippson (2005), and Patin (2007, 2008)
20
Intonational Phrases: Background
These last years, however, I felt more and more uncomfortable with this analysis (e.g. Patin 2010)
The main reason: in my data, Nonfinality, when it occurs elsewhere than in the utterance-final position, is always (?) associated with…
a Contrastive (Focus?) interpretation
a Mid (or downstepped H?) tone on the final vowel
21
Intonational Phrases: Background
(10) a. [ ( tsi-wono m-lévi ya-ha-w!á )φ ]I 1SG.PER-see 1-drunkard 1REL.PER-15-fall ‘I saw a drunkard that fell’ b. [ ( tsi-wóno )φ ]? [ ( m-leví )φ ( ya-ha-w!á )φ ]I ‘I SAW a drunkard that fell’
22
M
23
(10’) a. [ ( tsi-wono m-lévi ya-ha-w!á )φ ]I b. [ ( tsi-wóno )φ ]? [ ( m-leví )φ ( ya-ha-w!á )φ ]I ‘I saw / SAW a drunkard that fell’
M
Intonational Phrases: Background
NB: if the tone is on the last syllable, the vowel is lengthened:
(11) [ ( leóo )φ ]? [ ( ha-lim!í)φ ]I < /leó/ today 1PER-cultivate ‘TODAY, I have cultivated’
24
leo: halimi60
180
100
150
Pitc
h (H
z)
Time (s)0 1.308
M
Intonational Phrases: Background
From another side, Nonfinality is not observed where an Intonational Phrase boundary is expected
• e.g. no Nonfinality in the end of non-restrictive relative clauses (Patin 2010)
(12) wa-limadjíˑ $| $wa-faɲ-á-o $ $hazí $| &2-farmer 2REL.PER-do job
wa-l!émewa 2PER-to be tired ‘Farmers, who work, are tired.’
25
Except if….
M
26
(12’) wa-limadjíˑ $| $wa-faɲ-á-o $ $hazí $| 2-farmer 2REL.PERS-do job
wa-l!émewa 2PER-to be tired ‘Farmers, who work, are tired.’ *An op5onal rule raises a trigger when it occurs
on a stressed syllable, before a H
* M
Intonational Phrases: Background > Nonfinality may not be the proper clue to the
identification of Intonational Phrases (evidence for the Utterance phrase level?).
NB. other phenomena were claimed to operate at / be clues for the Intonational Phrase level (they will not be discussed here): Initial Accent Deletion (the initial tone of a Phonological Phrase
is optionally deleted if the preceding Phonological Phrase ends in a High) – Cassimjee & Kisseberth 1992
Tone liaison in the Northern dialect (the final tone of a non-final Phonological Phrase spreads to the first syllable of the following Phonological Phrase) – Patin 2009
Penultimate lengthening (while irregular) in the Northern dialect (the penultimate syllable of the utterance is long) – Patin 2009
27
Boundary tones
28
Boundary tones
In this section, I will discuss the results of a research conducted on appositives in Shingazidja
Building on these results, I will claim that Nonfinality is not the proper clue for Intonational Phrasing
Our proposal (NB. ≠ our talk at Bantu5): Intonational phrasing is signaled by a H% if non-final (and probably by a L% if final, while it’s far from being clear)
29
Restrictive apposition
No prosodic boundary between anchor and appositive – tone shifts from anchor to appositive
(13) Djumwa $m-léví$ | $ha-ɾem!é $pah[a] &
Juma 1-drunkard 1PER-hit 5.cat
‘Juma the drunkard hit a cat’
30
Djumwa mléví haremé pah(a)
60
140
80
100
120
Pit
ch (
Hz)
Time (s)
0 1.54
(13’) Djumwa $m-léví$ | $ha-ɾem!é $pah[a] & Juma 1-drunkard 1PER-hit 5.cat ‘Juma the drunkard hit a cat’
31
W
Restrictive apposition
No prosodic break between the apposition and the rest of the clause
(14) tsi-nika $Djúmwa $m-levi$ $ɲ-!úmɓa &
1SG.PER-give Juma 1-drunkard 10-house
‘I gave Juma the drunkard a house’
32
Non-restrictive apposition
A prosodic break separates the anchor from the appositive
The tone on the anchor does not shift onto the appositive
(15) Djumwá | m-leví | ha-ɾem!é pah[a] Juma 1-drunkard 1PER-hit 5.cat ‘Juma, a drunkard, hit a cat’
Appositive always followed by a pause, may be preceded by a pause
33
Djumwá mleví haremé pah(a)
70
150
80
100
120
140
Pit
ch (
Hz)
Time (s)
0 2.177
(15’) Djumwá | $ m-leví $| ha-ɾem!é $pah[a] &Juma 1-drunkard 1PER-hit 5.cat ‘Juma, a drunkard, hit a cat’
34
W
(15’’) Djumwá | $ m-leví $| ha-ɾem!é $pah[a] & Juma 1-drunkard 1PER-hit 5.cat ‘Juma, a drunkard, hit a cat’
Djumwá mleví haremé paha
60
140
80
100
120
Pit
ch (
Hz)
Time (s)
0 1.844
35
W
Non-restrictive apposition
The appositive is also followed by a prosodic boundary
The tone on the appositive fails to shift rightward
(16) tsi-nika$ $Djúmwá | $m-leví$ | $ɲ-umɓ!á &1SG.PER-give Juma 1-drunkard 10-house ‘I gave Juma, the drunkard, a house.’
Question: What type of boundary (PPh or IPh) is associated with apposition?
36
Non-‐restrictive apposition
Prosodic boundaries • In other languages IPh boundaries are associated with (the
right edge of) apposition (see Selkirk 2005, and references therein)
• Recall that in Shingazidja, IPh boundaries were claimed to be indicated by extraprosodicity
However…
(17) Djumwá | $m-limadjí | ha-ɾem!é $pah[a] &Juma 1-farmer 1PER-hit 5.cat ‘Juma, a farmer, hit a cat’
37
Djumwá mlimadjí haɾemé pah[a]
60
130
80
100
120
Pit
ch (
Hz)
Time (s)
0 1.652
(17’) Djumwá | $m-limadjí | ha-ɾem!é $pah[a] & Juma 1-farmer 1PER-hit 5.cat ‘Juma, a farmer, hit a cat’
38
W
Non-‐restrictive apposition
Is it a PPh boundary and not an IPh boundary (Patin & O’Connor’s 2012 analysis)? • Problem: PPh boundaries are not ‘spontaneously’ associated with
pauses • Problem: If an appositive is a CP, it should be associated with an
IPh, not a PPh. This is the case for non-restrictive relatives in other Bantu languages (e.g. Mbochi – Beltzung et al. 2010; Zulu – Cheng & Downing 2007))
Conclusions • The boundary is an IPh boundary • Extraprosodicity is not the main indicator for IPhs
39
Non-‐restrictive apposition
Our proposal: (Non-final) IPh boundaries are indicated by H%
• A sequence of PPh results in downstepping
(18) [ ( ye= m-limadjí )ɸ$ ( $ha-nik!á )ɸ& AT1=1-farmer 1PER-give
( e=m-lev!í )ɸ $( e=ɲ-umɓ!á )ɸ ]ɪ & AT1=1-drunkard AT9=9-house
‘The farmer gave the house to the drunkard’
40
ye=mlimadji hanika e=mlevi e=nyumba
80
150
100
120
140Pi
tch (H
z)
Time (s)0 2.017
(18’) [ ( ye= m-limadjí )ɸ ( ha-nik!á )ɸ& AT1=1-farmer 1PER-give
( e=m-lev!í )ɸ$ ( e=ɲ-umɓ!á )ɸ ]ɪ& AT1=1-drunkard AT9=9-house
‘The farmer gave the house to the drunkard’
41
W
Non-‐restrictive apposition
Final tones of non-restrictive appositives are not downstepped
So the right boundary of an appositive is an IPh and not a PPh
> IPh boundaries are indicated by H%
42
43
Djumwá mleví haremé pah(a)
70
150
80
100
120
140
Pit
ch (
Hz)
Time (s)
0 2.177
(153) Djumwá | $ m-leví $| ha-ɾem!é $pah[a] & Juma 1-drunkard 1PER-hit 5.cat ‘Juma, a drunkard, hit a cat’
W
Non-‐restrictive apposition
Further evidence for H% and IPh boundaries
• In stacking, the final tones of subsequent appositives are not downstepped (or the downstep is minimal)
(19) Djumwá | m-leví | ɗaɓá $| ha-ɾem!é $paha &Juma 1-drunkard 5.idiot 1PER-hit 5.cat ‘Juma, a drunkard, an idiot, hit a cat’
44
45
Djumwámleví
ɗaɓá haɾemé pah[a]
70
150
80
100
120
140
Pit
ch (
Hz)
Time (s)
0 2.648
(19’) Djumwá | m-leví | ɗaɓá $| ha-ɾem!é $paha &Juma 1-drunkard 5.idiot 1PER-hit 5.cat ‘Juma, a drunkard, an idiot, hit a cat’
W
46
(12) wa-limadjíˑ $| $wa-faɲ-á-o $ $hazí $| 2-farmer 2REL.PERS-do job
wa-l!émewa 2PER-to be tired ‘Farmers, who work, are tired.’
Also (recall):
M
Non-‐restrictive apposition Additional evidence:
• The final tone of a NRA is much higher than the final Hs of the PPs inside
(20) $Djumwá || $haβ(ah)á | $[y]a-tsó-na $ $hazí || &
$Juma now rel(pas)1-neg-have job $&
$ha-ɾem!é $paha &
$1(pas)-hit $5.cat #
$‘Juma, at this time without work, hit a cat’&
47
Djumwá haβ(ah)á (y)atsóna hazí haɾemé pah[a]
70
150
80
100
120
140
Pit
ch (
Hz)
Time (s)
0 2.205
48
(20’) $Djumwá || $haβ(ah)á | $[y]a-tsó-na $ $hazí || & $Juma now rel(pas)1-neg-have job &
$ha-ɾem!é$ $paha &$1(pas)-hit $5.cat #$‘Juma, at this time without work, hit a cat’&
W
49
Additional evidence (Manus & Patin 2011)
W
Non-‐restrictive apposition
Further evidence for IPh boundaries • The beginning of the appositive may be subject to Peak Delay,
which usually occurs at the beginning of sentences • Peak Delay: a tone that should appear on the first or second syllable
shifts to the subsequent syllable
(21) $Aɓuɗú $| $ye=twaɓiɓu $m=ó=m-ra[y]a=ní $| & Abudu AT9=9.doctor 1[.of]=LOC=quarter=in m-tãã́lːamu $w=[h]ahé | $zé=m-ɓwaɗe $ &1-expert 1(.of)=POSS3 AT10=10-illness z-á=n-tsi $ z-a=djuwá | ŋg’=e $djiɾani $ w-á=hangú &10-of=10-country 10-of=5.sun is= AT1 neighbor 1-of=POSS1 ‘Abudu, a local doctor, an expert on tropical medicine, is my neighbour.’
50
Non-‐restrictive apposition
Interim summary for IPs • Non-restrictive appositives are separated from the HC by prosodic
boundaries • The boundary following the appositive is an IPh, as evidenced by
the presence of a H% • The boundary preceding the appositive is an IPh, as evidenced by
Peak Delay
Question: Is the IPh of the appositive embedded in the IPh of the HC or is it prosodically independent • Embedding: [ Anchor [ NRA ]IPh VP ]IPh • Independence: [ Anchor ]IPh [ NRA ]IPh [ VP ]IPh
51
Non-‐restrictive apposition
Evidence for embedding (NB. consistent with the syntactic analysis) • The apppositive exhibits a H% but the anchor does not
(22) Djaná $ | $Djumw!á | $[ha]nika $m-l!ímadj’=e $| &yesterday Juma 1PER-give 1-farmer=AT1 m-leví$ $|| $e=ɲ-umɓ!á &1-drunkard AT9=9-house ‘Yesterday, Juma gave a farmeri, the drunkardi, the house’
• If the anchor were at the end of an IPh, we would expect it to exhibit a H%
52
53
Djaná Djumwá [ha]nika mlímadj’=e mleví e=ɲumɓá 70
175
80
100
120
140
160
Pitc
h (H
z)
Time (s)0 2.527
(22’) Djaná $ | $Djumw!á | $[ha]nika $m-l!ímadj’=e $| &yesterday Juma 1PER-give 1-farmer=AT1 m-leví$ $|| $e=ɲ-umɓ!á &1-drunkard AT9=9-house ‘Yesterday, Juma gave a farmeri, the drunkardi, the house’
W
Conclusion
Appositives are separated from the HC by prosodic boundaries
The boundaries are IPh boundaries
H% is the main indicator of IPh boundaries, rather than
extraprosodicity/Nonfinality
The IPh is embedded within the IPh of the HC, rather than being
prosodically separate
54
55
Thank you !