View
217
Download
0
Tags:
Embed Size (px)
Citation preview
Distribution and Timing of Glottalized Nasals in Athabaskan
Sharon Hargus
University of Washington
February 11, 2005
2
Outline
• Background
• Distribution and timing of glottalization in Athabaskan
• Witsuwit’en and Deg Xinag– Distribution of glottalized nasals– Timing of glottalized nasals
• Conclusions
3
Glottalized sonorants: timing possibilities
• Ladefoged and Maddieson 1996: 109: Laryngeal constriction can be ‘centered at’ the oral closure or can occur ‘at the beginning or the end’ of a nasal consonant
• Plauché 1998:140: ‘creaky voice and often full glottal closure preceding, simultaneous to, and following the sonorant are found as acoustic cues for glottalization of sonorants’
• Kingston 1990: implies that glottalized sonorants vary much more in timing than glottalized stops
4
Glottalized sonorants: timing preferences
• Silverman 1997:98: with glottalized nasals, ‘leftward laryngealization is preferred to rightward laryngealization’– ‘…non-modally phonated sonorant consonants are
realized with laryngeal gestures phased to the early portion of the supralaryngeal configuration.’ (p. 106)
– But Silverman was only discussing prevocalic glottalized nasals.
5
Glottalized sonorants: distributional preferences
• V__– Steriade 1999:102 ‘a preglottalized segment…will
depend for optimal identification of its laryngeal category on…a preceding vowel or sonorant’
– Blevins 2004:95: ‘word-initial neutralization of sonorant pre-glottalization is common, while word-final neutralization of pre-glottalization is rare’
6
Interaction of timing, distribution• In Yowlumne (Newman 1944:15), glottalized
sonorants can ‘never appear initially in a word or in a syllable that follows a closed syllable.’
• Steriade 1999: since glottalized sonorants are pre-glottalized, exclusively post-vocalic distribution < timing– Therefore, Licensing by Cue is more explanatory than
Licensing by Prosody
7
Timing and distribution in Yowlumne
• Howe and Pulleyblank 2001: Steriade’s analysis is wrong on two counts– Yowlumne glottalized sonorants not
restricted to coda position – Yowlumne glottalized sonorants are not
always pre-glottalized (Plauché 1998)
8
Timing and distribution in Yowlumne
___# ___C ___V
V___ amaaxam’ ‘and perhaps’
p’um’na ‘a full-blooded Indian’
c’oo’woo ‘work’
postglottalized postglottalized preglottalized
9
• In fact, distribution determines timing– If postvocalic and
• prevocalic (onset), then preglottalized• not prevocalic (coda), then postglottalized
• What about post-vocalic restriction?– plausibly, ‘the implicational relation between the feature of
glottalisation and the postvocalic position is…non-arbitrary, grounded in but semi-independent from phonetic properties governing the production and perception of glottalisation.’ (Howe and Pulleyblank 2001:63)
Timing and distribution in Yowlumne
10
Timing and distribution in Yowlumne
• Plauché’s explanation for Vn vs. nV– ‘optimize recoverability of the formant transitions
into a following vowel’ (139)
• Cf. Steriade/Blevins– Vn optimizes recoverability of the laryngeal
contrast (so preferred over Vn)
11
Interaction of timing, distribution
• Languages surveyed in Howe and Pulleyblank 2001
Cross-linguistically, timing is semi- independent of distribution
___ ]
Kwak’wala post-glottalized
Oowek’yala post-glottalized
Coatlán-Loxicha Zapotec
post-glottalized
Kashaya post-glottalized
Kutenai post-glottalized
Yowlumne post-glottalized
Sm’algyax pre-glottalized
Montana Salish pre-glottalized
Lai pre-glottalized
12
Interaction of timing, distribution
• ‘There is a correlation between syllabic position and the patterns of glottal timing’ (Howe and Pulleyblank 2001:76)
Some timing < distribution
[__V
Kwak’wala pre-glottalized
Sm’algyax pre-glottalized
Montana Salish
pre-glottalized
Lai pre-glottalized
Nuu-chah-nulth
pre-glottalized
Yowlumne pre-glottalized
13
Theoretical implications
• If timing determines distribution– phonology is phonetically motivated– supports Integrated model (no phonology-
phonetics interface, no phonetic component of grammar)
• If timing independent of distribution– phonology separate from, mapped to phonetics– supports Modular model
14
Outline
• Background
• Distribution and timing of glottalization in Athabaskan
• Witsuwit’en and Deg Xinag– Distribution of glottalized nasals– Timing of glottalized nasals
• Conclusions
15
Distribution of glottalization in Athabaskan
1. Proto-Athabaskan, Ahtna, Hupa– ejectives: pre-vocalic, post-vocalic– glottalized sonorants: post-vocalic
2. Dena’ina– ejectives: pre-vocalic, post-vocalic
3. Deg Xinag, Gwich’in, Han, N. Tutchone, S. Tutchone, Tanacross, Kaska, Tagish, Tahltan, Witsuwit’en
– ejectives: pre-vocalic– glottalized sonorants: post-vocalic
4. Sekani, Chilcotin, Dene Suine, Dogrib, Slave, Tsuut’ina, Navajo
– ejectives: pre-vocalic
(stem syllables)
16
Four types of Athabaskan languages
PA-type Dena’ina Deg Xinag-type
Sek-type
T’V
VT’ * *
R’V * * * *
VR’ * *
Maddieson 1984: ‘if a language has any laryngealized sonorants it also has glottalic or laryngealized stops. 19/20 95.0%’
17
Reflexes of *T’, *R’
PA Dena’ina Deg Xinag Sekani
T’V/v *ts’n ‘bone’ ts’n -t’n -ts’nè
VT’ *-wt’ ‘belly’
-vt’ -vt -bt
vT’ *u:t’ ‘scab’ it’ et ut
VR’ *-t’a:n’ ‘leaf’
-t’un -t’on’ -t’ò
vR’ *qn’ ‘fire’ qn qn’ kn
PA reconstructions from Leer 1987; V = full vowel, v = reduced vowel
18
Timing of T’ in Athabaskan
• T’V: consistently post-glottalized (many instrumental studies)
• VT’: in Ahtna (Siri Tuttle, p.c.), glottalization optional; if present, pre-glottalized
19
Timing of R’ in Athabaskan
• In Proto-Athabaskan: pre-? • Kingston to appear:‘contrastive laryngeal articulations in post-vocalic sonorants are
often pronounced at the beginning of or before their oral constriction…If the glottalic articulation were timed in this way relative to the oral constriction in glottalic sonorants in PA, i.e. if */VR’/ were pronounced [V’R], then the glottalic articulation would already overlap with the preceding vowel. Not only would the vowel coarticulate enough with the sonorant’s glottalic articulation for that articulation to shift readily to the syllable nucleus, but the pronunciation of /VVR’/ would be indistinguishable from that of /VR/ and /VR’/, and these sequences could not contrast.’
20
Timing of R’ in Athabaskan
• In Tututni (Golla 1976): post-glottalized• In Hupa (Golla 1970, 1977; Gordon 1995): post-
contrast with pre-– ‘an aspectual contrast between heavy and light stems…is
signaled by differences in the timing of the creak relative to the sonorant...in light stems...root-final creaky voiced nasals realize their creak on the end of the nasal (i.e. as post-glottalized nasals) while, in heavy stems, the creak is realized at the beginning of the nasal (i.e. as pre-glottalized nasals)’ (Gordon 1995: 18 ff.)
– ‘pre-glottalized sonorants underlyingly precede a vowel, while post-glottalized sonorants underlyingly precede a consonant or word boundary.’
21
Deg Xinag
Witsuwit’en
Map from Krauss (to appear)
22
Impressionistic auditory observation
• Deg Xinag glottalized nasals are pre-glottalized
• Witsuwit’en glottalized nasals are post-glottalized
23
Research question
• Can the impression of post-glottalization in Witsuwit’en, pre-glottalization in Deg Xinag be acoustically verified, preferably in a quantitative way?
24
Why would this be important to know?
• Basic research for the description of these languages
• Theories of role of phonetics in phonology rely crucially on such phonetic information
25
Outline
• Background
• Distribution and timing of glottalization in Athabaskan
• Witsuwit’en and Deg Xinag– Distribution of glottalized nasals– Timing of glottalized nasals
• Conclusions
26
Witsuwit’en nasals
• [n], [m]– [tn] ‘sheet of ice’– [m] ‘chunk of ice’
• [n’], [m’]
• no nasal vowels
27
Witsuwit’en glottalized nasals
n’ nit’n’ ‘he’s working (not for wage)’
ntsn’ ‘downhill’
c’qaq tsan’ ‘apron’
dyin’ ‘wild potato’
blenen’ ‘half of it’
m’ hat’m’ ‘it’s little’
hat’um’ ‘it’s really little’
from Hargus (to appear)
28
Wit. [n’]: pre- or post-glottalized?
• [’n], [’m] in Kari 1975, Lake Babine Band 1977
• [n’], [m’] as /n/, /m/– Other clusters allowed word-finally
• sqy’ ‘blood’
• -tw’ ‘hop’
– Some instances < /n-/ or /m-/:
29
Wit. glottalized nasals < suffixation
to ‘water’ bto - possessive suffix
ts’lm ‘packed lunch’
bts’lm’ ‘his packed lunch’
ia ‘hire her’ yia ‘she hired her’
- perfective/ optative (durative)
yin ‘he sees him’
yinn’ ‘he saw him’
30
Further phonological evidence for post-glottalization in Witsuwit’en
• /n-n/ (*[nn])• Degemination
– -ni human plural/inanimate• nn ‘to the side’• nni ‘people to the side’
• Epenthesis– Ny- ~ n- second person singular possessive prefix
• uzi ‘name’, nyuzi? ‘your name’• -le ‘hand’, nle ~ nyle ‘your hand’• -nin ‘face’, nynin, *nnin, *nin ‘your face’
• Distribution of optional variants – N- neutral directional prefix (optional)
• yen ~ nyen ‘across’• nq, *nnq, *nnq ‘uphill’
31
Witsuwit’en [n’-n]
• [n’-n] does not degeminate or epenthesize– [nit’n’] ‘he’s working’– [niwest’n’ni bi hbdli] ‘disability pension’ (lit.
‘those who do not work are taken care of with it’)
• Suggestive of [n’n] as /nn/, not /nn/
32
Deg Xinag nasals
• No nasal vowels
• Three-way place contrast: [m n ]• Three-way laryngeal contrast (final position):
33
Deg Xinag glottalized nasals
voiceless voiced glottalized
m -- dmzeg~ dmeg ‘spring/ summer rabbit’
vandlzm’ ‘it goes fast’
n čon ‘rain’ eq čon ‘wet fog, misty rain’
čon’ ‘it rained’
ðG ‘it dried’ -loG ‘fingernail’
-q’ ‘husband’
34
Evolution of final voicing distinction
• [n] /n/, [n] /nV/– possessed suffix *-e (Leer to appear) >
• - / V__ : te ‘water’, -te ‘water’ (psd.)• [+vd] / C__
ek ‘dog’, -leg ‘dog’ (psd.)
– dčn ‘coffin’, -dčn ‘coffin (psd.), stick, stem of woody plant’
35
DX [n’]: pre- or post-glottalized?
• Some [n’] < /n-/– - durative perfective/optative
• ntlanh ‘I’m looking at it’• natlan’ ‘I looked at it’
36
Summary
• Witsuwit’en, Deg Xinag– both have final [n], [n’] contrast– both have (morpho)phonological evidence for
post-glottalization– both contrast final [n’] with medial [n], [n]– But Witsuwit’en [n’] post-glottalized, Deg Xinag
[n’] preglottalized?
37
Outline
• Background
• Distribution and timing of glottalization in Athabaskan
• Witsuwit’en and Deg Xinag– Distribution of glottalized nasals– Timing of glottalized nasals
• Conclusions
38
Acoustic properties of n’: Columbian Salish
Columbian Salish /nmmal/ ‘lukewarm’
(postglottalized nasal)
‘strong, almost periodic, low frequency pulses’
‘quite turbulent airflow’
‘in both cases, the laryngealization…culminates in a glottal stop’
39
Acoustic properties of n’: Montana Salish
Montana Salish /smú/ ‘mare’
‘could be regarded as preglottalized’ < ‘strong glottal constriction at the beginning of the nasal’
‘complete glottal stop followed by a nasal with what appears to be modal voice’
‘nasal which is almost entirely creaky voiced’
‘In both cases there is an epenthetic separating the first two consonants in the initial cluster.’
40
Acoustic properties of n’: Hupa• Gordon 1995 18 ff.:
– Pre-glottalized: • Vn: ‘glottalization is not realized as a complete glottal stop, but
rather as creak on the end of the preceding vowel and on the beginning of the sonorant.’ or
• Vn ‘the preglottalized nasal may also be voiced...[with] the vowel preceding final nasal is glottalized. Full glottal closure is only achieved for a very brief period of time immediately prior to the beginning of the nasal.’ or
• Vn: the glottal closure is complete and the nasal following the glottal stop is voiceless or
• Vt: ‘often lose their glottal stop and instead have an oral release’
– Post-glottalized: ‘nasals [are] typically quite short and abruptly truncated by the glottal stop. Glottalization also typically spills over onto the end of the nasal.’
41
• Plauché 1998– main cue: creaky voice (20-100 ms.)
• V.R’V: apx. 50 ms., possibly overlapping with preceding vowel about 10 ms.
• VR’.: 20-80 ms; 50-100 if word-final
– secondary cues• bandwidth (narrower for R’ than R)
• amplitude (lower for R’ than R)
Acoustic properties of n’: Yowlumne
42
• In other languages (Plauché 1998)– lower pitch on R’ than R (e.g. Zapotec)– shorter duration of R’ than R (Lai)
• shorter duration of V/__R’ vs. /__R (Lai)
Acoustic properties of n’ (other languages)
43
Acoustic properties of n’ (other languages)
• ‘There is obviously room for further language-specific variation in the way that these oral and laryngeal gestures are related to each other, but the documentation is not yet very extensive.’ (Ladefoged and Maddieson 1996:111)
44
Acoustic properties of [n’] in Witsuwit’en and Deg Xinag
• Materials: word list recordings of [n’] made for study of effect of final glottalization on voice quality (some results for Witsuwit’en in Hargus 2005)
• Speakers– 8 Witsuwit’en (2 male, 6 female)
– 7 Deg Xinag (2 male, 5 female)
• 4-6 sets/speaker• 4 repetitions/token
45
Sample sets
• Witsuwit’en– ye ‘louse’
– ye ‘boy’ (vocative)
– nyen ‘across’yen’ ‘bridge’
• Deg Xinag– va ‘his sister-, brother-in-law’
– va ‘its grease’von’ ‘half of it’
– don ‘it’s hairy, furry’
– vdoon ‘his chest’
46
Method
• Observation of [n’]– Witsuwit’en: 157 tokens– Deg Xinag: 173 tokens– Developed criteria for different types of [n’]
• Classification of each [n’] according to type
• Number of types/speaker
• Number of speaker-types/language
47
Criteria for timing classification
• Pre-glottalized if– laryngealization starts before or simultaneous with nasality
– or nasal C follows laryngealization
• Post-glottalized if laryngealization starts after nasality• Difficulties
– variable realization of laryngealization
– determining onset of nasalization• nasality on V
• widely spaced glottal pulses over [n]
48
Pre-glottalized [n’] in Deg Xinag
gehon’ ‘he ate’ (AJ)
49
Cf. [n] in Deg Xinagdne ‘he says’ (AJ)
50
Pre-glottalized [n’] in Deg Xinag
gehon’ ‘he ate’ (LH)
51
Cf. [n] in Deg Xinagdne ‘he says’ (LH)
52
Pre-glottalized [n’] in Witsuwit’enyen’ ‘bridge’ (MA)
53
Cf. [n] in Witsuwit’enyqaninzin’ ‘he wanted it’ (MA)
54
Post-glottalized [n’] in Deg Xinag
con’ ‘it rained’ (JD)
55
Cf. [n] in Deg Xinagdne ‘he says’ (JD)
56
Post-glottalized [n’] in Witsuwit’en
btsan’ ‘its excrement’ (LM)
57
Cf. [n] in Witsuwit’enc’qaninzin’ ‘he wanted something, was hunting/trapping’ (LM)
58
Results: Deg Xinag
AJ ED HM JD KH LH RD
post- (%)
0 4 0 83 35 4 19
pre- (%)
100 96 100 17 65 96 81
total 25 24 24 24 23 26 27
59
Results: Witsuwit’en
AJ BM HM KN LM MA MF SM
post 100 25 70 95 100 18 100 96
pre 0 75 30 5 0 72 0 4
total 17 16 20 20 16 18 24 26
60
Witsuwit’en vs. Deg Xinag [n’]
• Witsuwit’en: post-glottalized (6/8 speakers)
• Deg Xinag: pre-glottalized (6/7 speakers)
61
Why timing differences?
• < Morphological differences? No.– In both languages, some instances result from suffixation of .
• < Distribution differences? No. – Witsuwit’en post-, Deg Xinag pre- [n’] both restricted to coda position
• < Place contrasts?– If place contrasts cued at formant transitions, then Vn’ preferred to V’n
(Silverman, Plauche)– No. Both languages contrast nasals of other places of articulation (but
not many such contrasts).
• < Differences in timing of nasality?– Witsuwit’en /Vn’/ as [VŒ]?– Deg Xinag /Vn’/ as [Vn]?
62
Some reflexes of *Vn and *Vn’
*Vn *vn *Vn’ *vn’
Aht Vn vn Vn’ vn’
DX Vn vn Vn’ vn’, Vn’
ST, KS VŒ vn VŒ vn’
Tg Vn, VŒ vn Vn’, VŒ
vn’
Tc Vn, VŒ vn Vn’, VŒ
vn’
Wit Vn vn Vn vn
Sek VŒ vn VŒ vn
Gal Vn Vn VŒ Vn’
63
Glottalization timing differences
• < nasality timing difference?– VŒ preferred over VŒVŒ (Witsuwit’en)
• Vn < PA? (Deg Xinag)
64
Conclusions
• Timing does not follow from distribution– Howe and Pulleyblank 2001 survey– Witsuwit’en vs. Deg Xinag coda [n’]Language-specific phonetics
• But onset n’: exclusively pre-?
65
Conclusions
• Need to test perceptually based explanations for phonological phenomena such as Licensing by Cue – What is the relative importance of laryngealization
vs. place for correct identification/discrimination?
66
Acknowledgements
• Thanks to Witsuwit’en and Deg Xinag speakers who participated in these studies.– Witsuwit’en: Alfred Joseph, Helen Michell, Kathryn
Naziel, Beatrice Morris, Lillian Morris, Margaret Austin, Mabel Forsythe, Stanley Morris
– Deg Xinag: Alta Jerue, Edna Deacon, Hannah Maillelle, James Dementi, Katherine Hamilton, Lucy Hamilton, Ray Dutchman
• Thanks to National Science Foundation for funding for research on Deg Xinag (OPP 0137483)
• Thanks to RA Julia Miller for initial help with Deg Xinag sound files.