Dept. of German Studies and Linguistics Humboldt...
1
Corpus 1. ny nu (+ e ) Îu nu wo H non-H # non-H # H H 2. ny nu (+ e ) nya a me H non-H # H # non-H non-H 3. ny nu (+ e ) nya m a H non-H # H # H H 4. ma ma (+ e ) nya m a H H # H # H H 5. a me (+ e ) Îu a be non-H non-H # non-H # non-H non-H 6. ma ma (+ e ) Îu a be H H # non-H # non-H non-H Abstract Tone languages such as Mandarin Chinese (Xu, 1999) or the Bantu language Chichewa (Downing, 2003) display tonal properties beyond the lexical level when some part of the utterance is in focus. Ameka (1992) describes the Gbe language Ewe to utilize morpho-syntactic means to signal focus without making any reference to prosody. The question arises whether there are typological commonalities with regard to prosody across different types of tone languages when signaling focus. Möhlig (1971) makes reference to ‘expressive prosodemes’ in Ewe which serve to emphasize a word or phrase or intensify the main meaning. We tested for these ‘expressive prosodemes’ by comparing the phonetic realization of the high-toned /-é/ focus marker (FM) in subject focus condition and in ex-situ object focus utterances to see if the language also makes use of prosodic (durational or pitch) cues to highlight or structure information. While duration appears to mark phrasal boundaries after the focussed constituents, F0 measurements of controlled tonal patterns were inconclusive as of now (and thus are not discussed here). We selected 6 tonal structures for further investigation and restricted our analysis to the following 3 conditions: 1. S V O (out-of the blue) (36 utterances) 2. S + FM V O (subject focus) (36 utterances) 3. O + FM S V (object ex-situ focus) (39 utterances) Our (preliminary) results indicate that there are no pitch cues that aid in signaling focus or structuring information. Durational cues however appear to play a significant role in structuring the information. Prosodic Edge Marking in Ewe Stefanie Jannedy (With Support of Ines Fiedler - B1) Dept. of German Studies and Linguistics SFB 632-D3 Humboldt University Berlin Tones (2): /H/ and /non-H/ [H, M, L, rising, falling] Syllable Structure: V, CV, CCV Vowels (7): / e i o ø u () / + nasalized variants (no contrastive length) Consonants (28): A: / b d b dz v z h / B: / p t k kp ts f s x / C: / m n l r w y / Results Linear Mixed Effects Model Anova on the duration of the focus marker in [S+FM ] vs. [ O+FM] utterances shows a significant effect (p.< .05, df=1, F=15.77) indicating that the focus marker in the ex-situ object is produced reliably longer than in the [ S+FM ] condition. Theoretical Implications There appears to be evidence for Möhlig’s ‘expressive prosodemes’ in the speech of this one speaker: • lengthening can have the perceptual effect of emphasis. In [S+FM] V O case: phrasing into two separate intermediate level phrases? • lengthening can have the perceptual effect of the restructuring of sentences: In [O+FM] S V case: phrasing into two separate intonational level phrases? Questions 1. Why redundant marking via use of prosody when having a) a specific marker for focus and b) syntactic movement 2. Can the results be interpreted to mean that Subject focus is ex- situ, too? Conclusions 1. There appears to be evidence for Möhlig’s ‘expressive prosodemes’ in the speech of this one speaker: lengthening can have the effect of the percept of increased salience resulting in the interpretation of emphasis. (Planned: recordings of same material with more speakers) 2. Our findings (that S and O in focus position are ex-situ) seem to support the Narrative Hypothesis (Fiedler & Schwarz, to appear): It suggests that historically, focus constructions in Ewe can be regarded as bi-clausal, consisting of an NP as first and a narrative clause as second clause. Comparison S Focus and O Focus (ex-situ) The durations were automatically calculated from the segmentation file. Mean durations with standard error bars were plotted: [S + FM] V O [O + FM] S V Acknowledgements: Our sincerest gratitude goes to Yugo Fujii and Jörg Dreyer at the ZAS for their help and time during the recording sessions. Yugo Fujii and Lars Marstaller helped with Praat scripts for the analyses. Bei Wang provided us with a script written by Yi Xu that cut the time of the analyses in half. Malte Zimmermann, Robin Hörnig und Shravan Vasishth took time for discussion of the data. The SFB-632 provided financial assistance for the study. Last but not least, our sincerest gratitude is extended to our Ewe speaker K. D. without whom this study would not have been possible. Linguistic Facts of Ewe Procedure Recordings: Recordings were made at 44 kHz in a sound proofed room onto DAT or hard disk. Speaker: 45 year old educated native speaker of the Anglo dialect of Ewe. Materials: The material was controlled for tonal co-occurrences as well as sonority for better tracking of the F0. Analysis: Duration and pitch analyses were carried out in Praat. Results Duration Measurements For each utterance, segment durations were calculated. For each identical condition, the mean and the standard error was computed and graphed. Note that the S is graphed in green, the V in blue and the O in grey. The focus marker on the S or O is plotted in red while the final vowel final vowel of the utterance is graphed in yellow: S V O Figure 3a: ny nu Îu nu wo. Figure 3b: ny nu nya m a Figure 3c: ny nu nya a me Figure 3d: ma ma nya m a Figure 3e: ma ma Îu a be Figure 3f: a me Îu a be Observations 1. utterance final vowel (yellow bar) generally lengthened when O onset is a consonant 2. word initial vowel of O (dark grey bars) indicates lengthening of vowel when following another vowel Interpretation 1. Final lengthening at right edge of utterance (prosodic phrase?) 2. Vocalic onset elongated in context of preceding vowel Observations 1. Focus marker /- e / in [ S + FM ] and [O + FM] is lengthened (red bar) relative to other vowels 2. utterance final V is lengthened (yellow bars) relative to other vowels 3. duration of focus marker on average longer in [O + FM] (right panel) case than in [ S + FM ] (left panel) case 4. no lengthening on last vowel of subject in [O + FM] S V cases Interpretation 1. In the [ S + FM ] case, the S builds a separate domain leaving the V and the O in a second domain. [ S + FM ] [ V O ] 2. In the [O + FM ] case, syntactic movement additionally marked by lengthening of FM: [O + FM ] [S V] 3. tendency for final lengthening at right edge of utterance (prosodic phrase) References Ameka, F. (1992) Focus Constructions in Ewe and Akan: A Comparative Perspective. In Collins, Ch. & Manfredi, V. (eds.) Proceedings of the Kwa comparative syntax workshop, pp. 1-25. Cambridge. Jannedy, S. & Fiedler, I. (ms) Prosody or Focus Marking in Ewe. Humboldt University Berlin, draft. Fiedler, I. & A. Schwarz. (to appear). Focus or Narrative Construction? In Aboh, E., Hartmann, K. & Zimmermann, M. (eds.) Focus Strategies: Evidence from African Languages, Berlin: de Gruyter. Möhlig, W. J. (1971) Zur Psosodologischen Struktur des Standard-Ewe. In Six, V., Cyffer, N., Wolff, E., Gerhardt, L. & H. Meyer-Bahlburg (eds.) Afrikanische Sprachen und Kulturen: ein Querschnitt. Deutsches Institut für Afrikaforschung, Hamburg. Figure 1: Map of Ghana &Togo showing where Ewe is spoken Figure 2: Sample of Praat Labeling file (upper panel: sound pressure waveform; middle panel: spectrogram and F0 curve; lower panel: segmentation marks with segment labels. 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 m m m a a e a a o ny 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 n u u o e ny o n u d 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 m m m a a a a o e ny 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 ewe3_3dd = d a a m e e d u gb e [n=7] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 a a gb e e e m u d [n=9] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 m m a a e d e u a gb [n=7] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 a a a gb e e m m u d [n=6] [n=7] [n=6] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 a a e e m ny o n u ny [n=6] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 ewe3_1dx ny ny o n u e a a m e [n=8] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 ny ny ny o n u e a a m o [n=6] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 m o a e ny ny o n u a [n=6] [n=6] 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 1 ny o n u e d u u o n [n=7] 0 0,05 0,1 0,15 0,2 0,25 0,3 1 ny o o n n u n u n u n n n d n n=5 0 0,05 0,1 0,15 0,2 0,25 0,3 1 ny ny o o n u a m a n=6 0 0,05 0,1 0,15 0,2 0,25 0,3 1 ny ny n o u a a m e n=5 0 0,05 0,1 0,15 0,2 0,25 0,3 1 m m m a a a a o ny n=8 0 0,05 0,1 0,15 0,2 0,25 0,3 1 m m a a a u gb e d n=6 0 0,05 0,1 0,15 0,2 0,25 0,3 1 a a m e u d gb e n=6 Primary use of class C-Consonants to avoid contour tones due to tone-sandhi. Ewe is a Gbe language spoken in Ghana & Togo Ex-Situ Object Focus 1. nu wo + e ny nu Îu 2. a me + e ny nu nya 3. m a + e ny nu nya 4. m a + e ma ma nya 5. a be + e a me Îu 6. a be + e ma ma Îu
Transcript of Dept. of German Studies and Linguistics Humboldt...
Corpus
1. nynu (+ e ) Îu nuwo H non-H # non-H # H H
2. nynu (+ e ) nya ameH non-H # H # non-H non-H
3. nynu (+ e ) nya maH non-H # H # H H
4. mama (+ e ) nya ma H H # H # H H
5. ame (+ e ) Îu abenon-H non-H # non-H # non-H non-H
6. mama (+ e ) Îu abeH H # non-H # non-H non-H
AbstractTone languages such as Mandarin Chinese (Xu, 1999) or the Bantu languageChichewa (Downing, 2003) display tonal properties beyond the lexical levelwhen some part of the utterance is in focus. Ameka (1992) describes the Gbelanguage Ewe to utilize morpho-syntactic means to signal focus without makingany reference to prosody. The question arises whether there are typologicalcommonalities with regard to prosody across different types of tone languageswhen signaling focus. Möhlig (1971) makes reference to ‘expressiveprosodemes’ in Ewe which serve to emphasize a word or phrase or intensifythe main meaning. We tested for these ‘expressive prosodemes’ bycomparing the phonetic realization of the high-toned /-é/ focus marker (FM) insubject focus condition and in ex-situ object focus utterances to see if thelanguage also makes use of prosodic (durational or pitch) cues to highlight orstructure information. While duration appears to mark phrasal boundaries afterthe focussed constituents, F0 measurements of controlled tonal patterns wereinconclusive as of now (and thus are not discussed here).
We selected 6 tonal structures for further investigation and restricted ouranalysis to the following 3 conditions:
1. S V O (out-of the blue) (36 utterances)
2. S + FM V O (subject focus) (36 utterances)
3. O + FM S V (object ex-situ focus) (39 utterances)
Our (preliminary) results indicate that there are no pitch cues that aid insignaling focus or structuring information. Durational cues however appear toplay a significant role in structuring the information.
Prosodic Edge Marking in EweStefanie Jannedy
(With Support of Ines Fiedler - B1)Dept. of German Studies and Linguistics
SFB 632-D3Humboldt University Berlin
Tones (2): /H/ and /non-H/ [H, M, L, rising, falling]
Syllable Structure: V, CV, CCV
Vowels (7): / e i o ø u () /+ nasalized variants(no contrastive length)
Consonants (28):
A: / b d b dz v z h /
B: / p t k kp ts f s x /
C: / m n l r w y /
Results
Linear Mixed Effects Model Anova on the duration of the focus markerin [S+FM ] vs. [ O+FM] utterances shows a significant effect
(p.< .05, df=1, F=15.77)indicating that the focus marker in the ex-situ object is produced
reliably longer than in the [ S+FM ] condition.
Theoretical Implications
There appears to be evidence for Möhlig’s ‘expressive prosodemes’in the speech of this one speaker:
• lengthening can have the perceptual effect of emphasis.In [S+FM] V O case:phrasing into two separate intermediate level phrases?
• lengthening can have the perceptual effect of the restructuringof sentences:In [O+FM] S V case:phrasing into two separate intonational level phrases?
Questions
1. Why redundant marking via use of prosody when havinga) a specific marker for focus andb) syntactic movement
2. Can the results be interpreted to mean that Subject focus is ex-situ, too?
Conclusions
1. There appears to be evidence for Möhlig’s ‘expressiveprosodemes’ in the speech of this one speaker: lengthening canhave the effect of the percept of increased salience resulting in theinterpretation of emphasis.(Planned: recordings of same material with more speakers)
2. Our findings (that S and O in focus position are ex-situ) seem tosupport the Narrative Hypothesis (Fiedler & Schwarz, to appear):It suggests that historically, focus constructions in Ewe can beregarded as bi-clausal, consisting of an NP as first and a narrativeclause as second clause.
Comparison S Focus and O Focus (ex-situ)The durations were automatically calculated from the segmentationfile. Mean durations with standard error bars were plotted:
[S + FM] V O [O + FM] S V
Acknowledgements:Our sincerest gratitude goes to Yugo Fujii and Jörg Dreyer at the ZAS for their help andtime during the recording sessions. Yugo Fujii and Lars Marstaller helped with Praatscripts for the analyses. Bei Wang provided us with a script written by Yi Xu that cut thetime of the analyses in half. Malte Zimmermann, Robin Hörnig und Shravan Vasishthtook time for discussion of the data. The SFB-632 provided financial assistance for thestudy. Last but not least, our sincerest gratitude is extended to our Ewe speaker K. D.without whom this study would not have been possible.
Linguistic Facts of Ewe
Procedure
Recordings: Recordings were made at 44 kHz in a sound proofed room onto DAT or hard disk.
Speaker: 45 year old educated native speaker of the Anglo dialectof Ewe.
Materials: The material was controlled for tonal co-occurrences aswell as sonority for better tracking of the F0.
Analysis: Duration and pitch analyses were carried out in Praat.
Results Duration MeasurementsFor each utterance, segment durations were calculated. For eachidentical condition, the mean and the standard error was computedand graphed. Note that the S is graphed in green, the V in blue andthe O in grey. The focus marker on the S or O is plotted in red whilethe final vowelfinal vowel of the utterance is graphed in yellow:
S V O
Figure 3a: nynu Îu nuwo. Figure 3b: nynu nya ma
Figure 3c: nynu nya ame Figure 3d: mama nya ma
Figure 3e: mama Îu abe Figure 3f: ame Îu abe
Observations1. utterance final vowel (yellow bar) generally lengthened when O
onset is a consonant2. word initial vowel of O (dark grey bars) indicates lengthening of
vowel when following another vowel
Interpretation1. Final lengthening at right edge of utterance (prosodic phrase?)2. Vocalic onset elongated in context of preceding vowel
Observations1. Focus marker /- e / in [ S + FM ] and [O + FM] is lengthened (red
bar) relative to other vowels2. utterance final V is lengthened (yellow bars) relative to other vowels3. duration of focus marker on average longer in [O + FM] (right panel)
case than in [ S + FM ] (left panel) case4. no lengthening on last vowel of subject in [O + FM] S V cases
Interpretation1. In the [ S + FM ] case, the S builds a separate domain leaving the V
and the O in a second domain. [ S + FM ] [ V O ]2. In the [O + FM ] case, syntactic movement additionally marked by
lengthening of FM: [O + FM ] [S V]3. tendency for final lengthening at right edge of utterance (prosodic
phrase)
References
Ameka, F. (1992) Focus Constructions in Ewe and Akan: A Comparative Perspective. In Collins, Ch.& Manfredi, V. (eds.) Proceedings of the Kwa comparative syntax workshop, pp. 1-25. Cambridge.
Jannedy, S. & Fiedler, I. (ms) Prosody or Focus Marking in Ewe. Humboldt University Berlin, draft.
Fiedler, I. & A. Schwarz. (to appear). Focus or Narrative Construction? In Aboh, E., Hartmann, K. &Zimmermann, M. (eds.) Focus Strategies: Evidence from African Languages, Berlin: de Gruyter.
Möhlig, W. J. (1971) Zur Psosodologischen Struktur des Standard-Ewe. In Six, V., Cyffer, N., Wolff,E., Gerhardt, L. & H. Meyer-Bahlburg (eds.) Afrikanische Sprachen und Kulturen: ein Querschnitt.Deutsches Institut für Afrikaforschung, Hamburg.
Figure 1: Map of Ghana &Togoshowing where Ewe is spoken
Figure 2: Sample of Praat Labeling file (upper panel: sound pressurewaveform; middle panel: spectrogram and F0 curve; lower panel:segmentation marks with segment labels.
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
m m ma a e a aony
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
n u uo e ny o n u d
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
m m ma a a ao e ny
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
ewe3_3dd
=d
a am e e d u gb e
[n=7]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
a agb e e em ud
[n=9]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
m ma a e d eu a gb
[n=7]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
a a agb e e m m ud
[n=6]
[n=7] [n=6]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
a ae em ny o n u ny
[n=6]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
ewe3_1dx
ny nyo n u e a a m e
[n=8]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
ny
ny nyo n u e a am o
[n=6]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
m o a e ny nyo n u a
[n=6]
[n=6]
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
1
ny o n u e d u u on
[n=7]
0
0,05
0,1
0,15
0,2
0,25
0,3
1
ny o onn
un
un
un
nn
dn
n=5
0
0,05
0,1
0,15
0,2
0,25
0,3
1
ny nyo on u a m a
n=6
0
0,05
0,1
0,15
0,2
0,25
0,3
1
ny nyno u a a m e
n=5
0
0,05
0,1
0,15
0,2
0,25
0,3
1
m m ma a a aony
n=8
0
0,05
0,1
0,15
0,2
0,25
0,3
1
m ma a au gb ed
n=6
0
0,05
0,1
0,15
0,2
0,25
0,3
1
a am e ud gb e
n=6
Primary use of classC-Consonants toavoid contour tones dueto tone-sandhi.
Ewe is a Gbe languagespoken in Ghana & Togo
Ex-Situ Object Focus
1. nuwo + e nynu Îu 2. ame + e nynu nya3. ma + e nynu nya4. ma + e mama nya5. abe + e ame Îu 6. abe + e mama Îu
