The phonetic motivation of stop assibilation

The phoneti c moti vati on of stop assi bil ati on
PaPI 2005
T. A. Hall, Silke Hamann & Marzena Zygis
tahall2@indiana.edu, silke / marzena @zas.gwz-berlin.de
Indiana University, Zentrum für Allgemeine Sprachwissenschaft (ZAS) Berlin
processes whereby (alveolar and dental) stops become sibilant affricates or sibilant
fricatives before high (front) vocoids
Three types differentiated
•
spirantization
•
affrication
•
posteriorization
We measured the duration from the burst of the stop to the beginning of the vowel (the
start of continuous formants). This included both friction and aspiration, called ‘friction
Sample spectrograms for the nonce words Atik and Atjak (by German speaker CG):
t → s /_i
t d → s/_iV
t th → ts tsh/_i
t d → ts dz/_i
t d → tS dZ /_i j
t th → tS tSh /_i
Finnish
Ancient Greek
Korean
Quebec French
Sorbian
West Futuna-Aniwa
8000
8000
td →
0
0
Three properties of stop assibilation
a
t
I
a
Hall & Hamann (to appear)
Logically possible language types
A. Assibilation is triggered by /i/ and /j/
B. Assibilation is triggered only by /j/
C. Assibilation is triggered only by /i/
Occurring language types
Quebec French, Kpando dialect of Gbe
Romanian, West Slavic, Latvian
unattested
D. /t/ and /d/ assibilate
E. Only /t/ assibilates
F. Only /d/ assibilates
Quebec French, Kpando dialect of Gbe
Shona, Dutch, Late Latin
unattested
Generalization s
Assibilation in /ti/ implies the assibilation in /tj/
Assibilation in /di/ implies the assibilation in /dj/
Assibilation of /d/ implies the assibilation of /t/
aTIK
uTIK
iTIK
aTj IK
a DIK
u DIK
iD IK
a Dj IK
Difference between /t/ and /d/:
The friction phase is longer for /t/ than for /d/, independent of the language.
Influence of /i/ versus /j/:
The friction phase of /t, d/ + /j/ is longer than that of /t , d/ + /i/.
Influence of the preceding vowel
for German speakers:
/d/ and /t/ separately
for Polish speakers:
voiced stops
voiceless stops
for /d/: /a/ versus /u/ p < 0.05,
other contexts are not significant.
for /t/: /u/ versus /i/ p < 0.01,
/u/ versus /a/ p < 0.001,
/a/ versus /i/ p not significant.
Results
for /d/: /i/ versus /u/ not significant
other contexts highly significant
for /t/: /i/ versus /u/ not significant
/a/ versus /i/ p<.0.05
/a/ versus /u/ p <.001
i
j
→ a preceding /a/ significantly lengththens the duration of frication (German, Polish) .
4 German and 4 Polish native speakers (2 female, 2 male) read the following nonce
words 10 times.
German carrier phrase was Ich habe ___ gesagt, ‘I have said __’.
Polish carrier phrase was Powiedzialem ___ do ciebie, ‘I said __ to you’.
Capitals indicate stress.
Adj ak
Ud jak
Adj ak
Adj uk
Airflow measurements
15
Exper iment
Atjak
U tja k
Atjak
Atjuk
The x-ray tracings of the Polish segments show a clear difference between vowel and
glide. The glide is articulated with a narrower and also longer constriction. The German
x-ray tracings, however, show only a minimal difference (mainly in tongue root position)
between glide and vowel.
difference between /i/ and /j/ is
highly significant (p < 0.001),
difference between /t/ and / d/ is
highly significant (p < 0.001).
• phonological assibilation only makes sense by appealing to phonetics
• creation of sibilants from stops has its phonetic origin in the brief period of
turbulence (i.e. friction phase) which occurs at the release of the stop into a
following high vocoid.
Ad ik
Ud ik
Idik
Ad jik
jA
for both:
t → [+strident, +continuant] / _ i
t → [+strident] / _ i
t → [+strident, –anterior ] / _ i
Previous studies
Kim (2001)
Atik
Utik
Iti k
Atji k
t
Polish speakers
Featural representation
•
•
•
97.42 ms
German speakers
the trigger is some subset of the high front vocoids (i.e. /i y j/)
the output is a sibilant (either an affricate or a fricative)
the trigger is to the right of the target
of spirantization:
of affrication:
of posteriorization:
1. /j/ is articulated with a longer and/or more narrow area of constriction than /i/. (Thus it
takes longer for the oral pressure to be released when a plosive is followed by a /j/ then
when followed by an /i/)
This explanation is assumed by Maddieson & Emmorey (1985) , who infer a narrower
constriction for /j/ than for /i/ in Amharic, Yoruba, and Zuni from the lower F1 frequency
of /j/.
X-ray tracings for Polish (Wierzchowska 1971) and for German (Wängler 19 58)
/i/
/j/
/i:/
/j/
0
0.340544
67.76 ms
Possible reasons for the difference between /i/ and /j/:
Results
(based on the findings by Clements 1999 and Kim 2001)
•
•
•
Discussion
Measurements
Definition of stop assibilation
20-21 June 2005
aTJA K
uTJA K
iTJAK
aTJU K
a D JAK
u D JAK
i DJA K
a D JUK
There are no minimal pairs such as Atia versus Atja, since a pretest showed that
speakers tend to glide the /i/ in Atia and thus neutralize the contrast.
The two languages differ in so far as German contrasts voicel ess and voiceless
aspirated stops, whereas Polish contrasts voiced and voiceless stops. Therefore, we
expected to find differences in friction duration between the two languages (the friction
duration of German stops to be longer than that of the Polish stops).
ml
10
→ a preceding /u/ leads to a shorter duration of friction (Polish).
5
0
D
Influence of the following vowel
for /d/: /i/ versus /u/ not significant,
for German speakers
other contexts significant
for /t/: all significant
T
tj>ti>dj>di (p<0.001 for all)
Conclusion
for Polish speakers
for /d/: /a/ versus /i/ p < 0.05,
other contexts not significant.
for /t/: /a/ versus /i / p < 0.01,
other contexts not significant.
→ frication duration scale: _i > _u > _a (German)
→ friction duration for following /i/ longer than for following /a/ (Polish)
The phonetic studies support the hypotheses:
Assibilation is more likely in /tj/ than in /ti/, and more likely in /tj/ than in /dj/
Assibilation hierarchy: /tj/ > /ti/ > /dj/ > /di/ (following from the differences in friction
length)
Differences in amount of airflow confirm the hierarchy obtained acoustically.
Bibliography
Chang, S. S. (1999) Vowel dependent VOT v ariation. In: Proceedings of the 14 ICPhS, San Francisco; 1021-1024.
Clements, G. N. (1999) Affricates as noncontoured stops. In: O. Fujimura et al. (eds.) Item, order in language and speech. 271299. Prague: Charles University Press.
Hall, T. A. & S. Hamann (to appear) Towards a typology of stop assibilation . Linguistics.
Keating, Patricia (1980) A phonetic study of a voicing contrast in Polish. Ph.D. thesis Brown University.
Kim, H. (2001) A phonetically based account of phonological stop assibilation. Phonology 18: 81-108.
Maddieson, I. & K. Emmorey (1985) Relationship between seminvowels and vowels: Cross-Linguistic Investigations of Acoustic
Difference and Coarticulation. Phonetica 42: 163-174.
Wängler, H. H. (1958) Atlas deutscher Sprachlaute . Berlin: Akademie- Verlag.
Wierzchowska, B. (1971) Wymowa polska. Warszawa: P.Z.W.S.