Fifth Ultrafest Conference — Haskins Laboratories New Haven, CT

Transcription

Fifth Ultrafest Conference — Haskins Laboratories New Haven, CT
Fifth Ultrafest Conference — Haskins Laboratories
New Haven, CT 19 March - 21 March 2010
Articulatory Grounding of Southern Salentino Harmony Processes
1
1,3
1
1,2
Mirko Grimaldi , Andrea Calabrese , Francesco Sigona , Luigina Garrapa , Bianca Sisinni
1
1
Centro di Ricerca Interdisciplinare sul Linguaggio (CRIL), Salento University, Lecce (Italy)
2
Department of Linguistics—University of Padova (Italy)
3
Department of Linguistics—University of Connecticut (USA)
1. The underlying stressed vowel system
in Southern Salentino
i
ɔ
Ɛ
u
4. Research questions
 What is the articulatory nature of the F1 differences that are observed in the
stressed mid vowels before high vs. non-high vowels?
 What is the articulatory process producing the F1 differences? Is there just a
single process or are there more processes?
 How can we interpret these facts in phonological terms?
a
9. Results
11. Conclusion

9. 1 The FRONT mid vowel before –i and –u
TRICASE
MIGGIANO


5. Hypotheses
Fig. 1: F1 and F2 formant chart of the five Southern Salento vowels produced by 36 male
speakers, and Kernel curve analysis (the most appropriate measure for cases where little is known
about the process under study). The mid stressed vowels /e/, /o/ are differentiated according to the
following vowel (cf. Grimaldi, 2009).
2. The dialects investigated articulatorily
TRICASE (T)
200
i
i
u
e1
400
e2
e3
600
a
o1
e2
e3
o2 o3
800
e1 = e/__ i
e2 = e/__ u
e3 = e/__ e, a
o1 = o/__ u
o2 = o/__ i
o3 = o/__ a, e
u
e1
400
o1
F1
F1
200
o2 o3
600
a
800
1000
3000 2500 2000 1500 1000 500
F2
1000
3000 2500 2000 1500 1000
F2
500
Fig. 2-3: Vowels produced by one 34-year-old male speaker of the Tricase dialect.
Ellipses on data, confidence level = 68,8% (on the left), and ellipses on centroids,
confidence level = 95% (on the right).
MIGGIANO (M)
i
e1
u
e2
o3
F1
e3
i
400
o1
o2
600
F1
400

Benus, S. & A. Gafos (2006), "Articulatory characteristics of
Hungarian „transparent‟ vowels", Journal of Phonetics, 35.
271-300.
Calabrese, A. (2008), "On the Feature [ATR] (and [Tense]) in the
Romance languages", in Pennello, N.
et alii, (eds.):
Miscellanea in onore di Laura Vanelli . Udine:
Forum, 103-115.
Calabrese, A. (1998), "Metaphony revisited", Rivista di
Linguistica, 10. 7-68.
Davidson, L. (2006), "Addressing phonological questions with
ultrasound", Clinical Linguistics and Phonetics, 19.
619–633.
Gick et al., (2006), Low vowels and transparency in Kinande
vowel harmony, Phonology: 23: 1-20
Grimaldi, M. (2009), "Acoustic correlates of phonological microvariations. The case of unsuspected highly diversified metaphonetic processes in a small area of Southern Salento (Apulia)", in Danièle Torck and W. Leo Wetzels (eds.): Romance Languages and
L i n g u i s t i c
T h e o r y
2006. Amsterdam / Philadelphia: John Benjamins. ISBN:
9789027290922.
Ladefoged, E. I. & P. Maddieson (1996), The Sounds of the
World’s Languages. Oxford: Backwell.
Lilienthal, J. (2009), "The articulatory and acoustic impact of
Scottish
/r/
on
the
preceding
vowel
onset". In Proceedings of Interspeech 2009, 6-10 September 2009,
Brighton, UK: 2819-2822.
Lindau, M. (1975), UCLA Working Papers in Phonetics, 30. Phonetics Laboratory, Depart- ment of Linguistics, UCLA, Los Angeles, California.
Walker, R. (2005), Weak triggers in vowel harmony, Natural Language & Linguistic Theory, 23.917-989.
9.2 The BACK mid vowel before –u
6. Methodology
TRICASE
 Ultrasound (US) system APLIO Toshiba 25fps .
 Recording of the tongue movements
 2 male speakers who, after a preliminary acoustic analysis, have
MIGGIANO
shown to possess the exemplified harmony processes and their
microvariation.
 The stimuli were embedded in the frame sentence: I say ____ now
 10 repetitions for each vowel type: 5 in open syllable
(e.g., ['fi:.li] „threads‟) e 5 in closed syllable (e.g., ['frid.di]
„cold-M. PL.‟).
e1
e2
e3
o1
o3 o
600
2
a
a
800
800
1000
3000 2500 2000 1500 1000
F2
500
1000
3000 2500 2000 1500 1000
F2
 Tokens containing the mid vowels were
e1 = e/__ i
e2 = e/__ u
e3 = e/__ e, a
o1 = o/__ u
o2 = o/__ i
o3 = o/__ a, e
u
differentiated according to the final vowels
(cf. 1 and 2).
 The F1 of stressed mid vowels lowers when the following syllable contains a high vowel.
 The “lowered” F1 of the stressed mid vowels does not reach the F1 values
of unstressed high vowels: mid vowels become mid-high vowels.
 Asymmetry between the front mid vowel /ɛ/ and its back
counterpart /ɔ/:
 F1 of /ɛ/ lowers before both unstressed high vowels
(cf. 1b-c vs. 1a).



['llɛva]
['llevi]
['llevu]
‘(s/he) removes’ (496 Hz in T, 538 Hz in M)
‘(you) remove’ (407 Hz in T, 405 Hz in M)
‘(I) remove’
(472 Hz in T, 490 Hz in M)
 F1 of /ɔ/ lowers only before unstressed posterior high vowel
 The randomised stimuli were presented to the
informants on a computer screen.
7. [+ATR] vs. [-ATR] vowels in the Tricase dialect
The tongue root advances in [e]:
[e] [+ATR]
[ɛ] [-ATR]



['bbɔna]
['bbɔni]
['bbonu]
‘good-F.SG.’
‘good-M.PL.’
‘good-M.SG.’
(512 Hz in T, 539 Hz in M)
(500 Hz in T, 547 Hz in M)
(465 Hz in T, 446 Hz in M)
 F1 of /ɛ/ lowers more before the unstressed front high vowel than
before the back high vowel (cf. 1b vs. 1c above)
(2) The mid vowels become [+high] when
followed by the unstressed high back vowel –[ʊ]
i
ɛ
a
ɔ
u
High
+
–
–
–
+
Low
–
–
+
–
–
Back
–
–
+
+
(1)
*
[ *
N
X1
+
Rounded
–
–
–
+
+
ATR
+
–
–
–
+
[-consonantal]
*
N
*
[ *
N
*
N
X1
X2
X2
[-consonantal]
[-consonantal]
Tongue Body Tongue Root Tongue Root Tongue Body
Tongue Body
[-consonantal]
Tongue tip
Tongue tip
Tongue root
Tongue root
['mɛ:te] ‘(s/he) reaps’
['me:ti] ‘(you) reap’
8. Analysis
 The inferential statistic analysis of the tongue contours was computed
with “Smoothing Spline ANOVA” (SS-ANOVA)
 SS-ANOVA already used by Davidson (2006) and Lilienthal (2009).
Final unstressed back high vowel are
[-ATR]:
*
N
|
X
[-cons, +high, +back]→ [-ATR]/ ____
[-low]
[+ATR]
[-high]
[+high]
However, the constraint *[+Back, +ATR] blocks the
application of this rule to back vowels.
[-low]
(3) The superficial stressed vowel system
in the dialects of Tricase and Miggiano
(Southern Salento)
i
e
u
I
ʊ
Tongue Body
ɔ
Ɛ
[+high]
(cf. 2c vs. 2a-b)
(2) a. /bbɔna/
b. /bbɔni/
c. /bbɔnu/
The mid vowels become [+ATR] when followed by
the unstressed high front vowel –i:
500
3. The acoustic data: synthesis
(1) a. /llɛva/
b. /llɛvi/
c. /llɛvu/
10. Phonological interpretation
Assumptions:
Vowels feature assignments:
Fig. 4-5: Vowels produced by one 54-year-old male speaker of the Miggiano dialect.
Ellipses on data, confidence level = 68,8% (on the left), and ellipses on centroids,
confidence level = 95% (on the right).
Our research represents the first
attempt
to
study
an
Italian harmony process articulatorily and
to use the data that are thus obtained to
validate theoretical hypotheses (see for the
Hungarian harmony Benus, Gafos, 2006).
References
200
200
The US system could represent a fruitful
resource in order to integrate phonetic and
phonological levels.
In this way, phonological models can validate and refine their predictive
power.

 Ladefoged & Maddieson (1996): Height contrasts in Romance mid vowels
are due to different positions of the tongue body with no involvement of the
tongue root.
 Calabrese (1988, 2008),Walker (2005), a.o: These contrasts are due to the
advancement/non-advancement of the tongue root, i.e. to different
specifications of the feature [Advanced Tongue Root] (where [+ATR]
implies an advancement of the tongue root with respect to its neutral
position ([-ATR]).
 Calabrese (2008), following Lindau (1975): The feature [+ATR] is
responsible for the tongue body convexity characterising the mid-high
vowels in Romance languages. Specifically, the advancement of the TR
causes raising/fronting of the TB.
 Additional hypothesis: The anterior dislocation of the tongue body due to
the feature [+ATR] may be problematic in the case of the back vowels.
There is a markedness constraint *[ +Back, +ATR].
This study provides evidence in favor of
the use of US in analyzing the
articulatory grounding of phonological
phenomena.
In particular:
[+Back]
a