elicitation and documentation of tone and documentation of tone what is the place of quantitative...

Elicitation and documentation of tone

What is the place of quantitative analysis in the study of tone?

THIS PPT CONTAINS A SUBSET OF THE ILLUSTRATIONS

Bert Remijsen, University of Edinburgh

Master Class, 4th International Conference on Language Documentation and

Conservation

1

•  Does the second syllable have a high tone pattern or a rising one?

2

Figure. Waveform, spectrogram and F0 trace of the word /dawa/ ‘lots (quantity)’ in Kakua. Data from Katherine Bolanos, cited in Hyman (2014:553).

What is the place of the quantitative angle in the study of tone?

3

Low

gìk-ánɪ́ á-lɛ̀ŋ kɪ̀ kɛ̂ɲ PRON:P-DEM PAST-take:2SG PREP place:S.DEM ‘Somebody has taken them gradually in this place.’

High gìn-ánɪ́ á-lɛ́ŋ bùul PRON:S-DEM PAST-beat:APPL.2SG drum:S ‘You have used it to beat the drum.’

Even in a system with just Low vs. High, the alignment matters:

Figure 1: Schematic representations of Low and High tone configurations on CV syllable, preceded by a High target and followed by a Low target. 4

Even in a system with just Low vs. High, alignment matters: Fu

ndam

enta

l fre

quen

cy (F

0)

5

Xu & Sun (2002) on the speed of f0 change

Figure from Xu & Sun (2002:1402), showing the difference between response time and overall time of the f0 change.

6

Xu & Sun (2002) on the speed of f0 change

Table from Xu & Sun (2002:1405), showing the difference between response time and overall time of the f0 change.

Glissando threshold (Rossi 1971, 1978; Greenberg & Zee 1978; ‘t Hart, Collier & Cohen 1990): the smaller the time domain is small, the greater an f0 change needs to be in order to be heard as a pitch change.

7

A perceptual source of time pressure: the glissando threshold

8


Figure from ‘t Hart, Collier & Cohen (1990:32). The glissando threshold as a function of duration, meta-analysis of several experimental studies.

Glissando threshold (Rossi 1971, 1978; Greenberg & Zee 1978; ‘t Hart, Collier & Cohen 1990): the smaller the time domain is small, the greater an f0 change needs to be in order to be heard as a pitch change.

Formula from ‘t Hart, Collier & Cohen (1990) – for a given duration (D), the glissando threshold, in semitones (ST) per second = 0.16 / D2

Example – a change from 150 to 135 Hz (1.83 ST). Perceived as a pitch contour over 100 ms, but as a level pitch over 50 ms.

9


10

Low

gìk-ánɪ́ á-lɛ̀ŋ kɪ̀ kɛ̂ɲ PRON:P-DEM PAST-take:2SG PREP place:S.DEM ‘Somebody has taken them gradually in this place.’

Fall gìn-ánɪ́ á-lɛ̂́ŋ kɪ̀ kɛ̂ɲ PRON:S-DEM PAST-beat PREP place:S.DEM ‘Somebody has beaten it in this place.’


Alignment in a falling contour tone (Shilluk example)

Figure 1: Schematic representations of Low, High and Falling tone patterns on a CV syllable, preceded by a High target and followed by a Low target. 11

Alignment in contour tones (schematic representation) Fu

ndam

enta

l fre

quen

cy (F

0)

“[I]t might be that in some languages pitch changes are timed relatively early in the syllable, and in other languages they are timed relatively late. Such control would only be phonetic, never phonological.” [Odden 1995:450]

“[T]here is no possible opposition between two HL or two LH contours where the two tones are synchronized differently within the syllable.” [Hyman 1988:51]

“If the contours are composed of levels, the existence of two falls implies at least three levels, and then in fact we might expect up to three contours of the same shape.” [Yip 2002:29]

12

Null hypothesis: tonal alignment not contrastive in contour tones

This configuration is assumed not to occur:

13

Figure. Schematic representations of early- vs. late-aligned falling contours within the syllable domain.

Null hypothesis: tonal alignment not contrastive in contour tones

14

Evidence for contrastive tonal alignment in contour tones: Yoloxóchitl Mixtec

Figure: From DiCanio, Amith & Castillo García (2014): time-normalised f0 traces, for early- vs. late-aligned rises, on monosyllabic words in citation form.

15

Early-aligned Fall (allophonic variant of Low toneme)

Late-aligned Fall (realisation of Fall toneme)

ràaan ā-lel person:S AG.S-isolate:3S ‘You isolate a person.’

ràaan ā-lel person:S AG.S-isolate:PASS ‘A person is being isolated.’

ràaan ā-leel person:S AG.S-isolate:3S ‘He isolates a person.’

ràaan ā-leel person:S AG.S-provoke:PASS ‘A person is being provoked.’

Dinka

•  Example:

Evidence for contrastive tonal alignment in contour tones: Dinka

Figure. Averaged f0 traces on normalised time axis (4 sets, 13 speakers). From Remijsen (2013).

ER

B

Evidence for contrastive tonal alignment in contour tones: Dinka

•  Confronted with a constellation like the one below, representing the contours as level tones is not an option.

17 Figure. Schematic representations of four tone categories distinguished by tonal alignment.

A maximally complex configuration of tonal alignment

18

Low gìk-ánɪ́ á-lɛ̀ŋ kɪ̀ kɛ̂ɲ PRON:P-DEM PAST-take:2SG PREP place:S.DEM ‘Somebody has taken them gradually here.’

Early-aligned

Fall

gìn-ánɪ́ á-lɛ̂́ŋ kɪ̀ kɛ̂ɲ PRON:S-DEM PAST-beat PREP place:S.DEM ‘Somebody has beaten the it here.’

Late-aligned

Fall

gìn-ánɪ́ á-lɛ́ŋ̀ pâac PRON:S-DEM PAST-beat:FUG village:S ‘Somebody went away to the village to beat it.’


Evidence for contrastive tonal alignment in contour tones: Shilluk

19

Figure. Averaged f0 traces of the target word on a normalised time axis (whole dataset).

F0 (H

z)


20

Figure. Means and standard deviations for tonal alignment of the high turning point by Tone.

nucleus coda


21

Figure. Means and standard deviations for the F0 height of the high turning point, by Tone. Separate panels for raw f0, in hertz (left), and after conversion to ERB scale and z-transformation per speaker (right).

F0 height (ERB, z-trans) F0 height (Hz)


Bird, Steven & Haejoong Lee (2014). Computational support for early elicitation and classification of tone. In Steven Bird & Larry Hyman (eds.) How to study a tone language? Language Documentation and Conservation 8, 453-461. Bird & Larry Hyman (eds.) (2014). How to study a tone language? Language Documentation and Conservation 8. DiCanio, Christian, Jonathan D. Amith & Rey Castillo García (2014). The phonetics of moraic alignment in Yoloxóchitl Mixtec. In Carlos Gussenhoven, Yiya Chen & Dan Dediu (eds.) The 4th International Symposium on Tonal Aspects of Languages. 203–210. Available (August 2014) at http://www.isca-speech.org/archive/tal_2014. Greenberg, Steven & Eric Zee (1979). On the perception of contour tones. UCLA Working Papers in Phonetics 45, 150-164. Himmelmann, Nikolaus (1998). Documentary and descriptive linguistics. Linguistics 36, 161-195. House, David (1990). Tonal perception in speech. Travaux de l’institut de linguistique de Lund 24. Lund: Lund University Press. Hyman, Larry M. (1988). Syllable structure constraints on tonal contours. Linguistique Africaine 1, 49-60. Hyman, Larry M. (2011). Tone: is it different? In John A. Goldsmith, Elizabeth Hume, & Leo Wetzels John Goldsmith, Jason Riggle, Alan C. L. Yu (eds.) The handbook of phonological theory (2nd editition). Berlin: De Gruyter Mouton, 197-239. Hyman, Larry (2014). How to study a tone language. In Steven Bird & Larry Hyman (eds.) How to study a tone language? Language Documentation and Conservation 8, 525-562. Kuang, Jianjing (2013). The tonal space of contrastive five level tones. Phonetica 70, 1-23. Odden, David (1995). Tone: African languages. In John A. Goldsmith (ed.) The Handbook of Phonological Theory. Blackwell, 444-475. Pike, Kenneth L. (1948). Tone Languages, A Technique for Determining the Number and Type of Pitch Contrasts in a Language, with Studies in Phonemic Substitution and Fusion. Ann Arbor: University of Michigan Press. Remijsen, Bert (2013). Tonal alignment is contrastive in falling contours in Dinka. Language 89(2), 297-327. Remijsen, Bert (2014). Investigating tone in languages with a quantity contrast. In Steven Bird & Larry Hyman (eds.) How to study a tone language? Language Documentation and Conservation 8, 672-689. Rossi, M. (1978). La perception de glissandos descendants dans les contours prosodiques. Phonetica 35, 11-40. Smalley, William A. (1963). Manual of articulatory phonetics. New York: Practical Anthropology. [The sound-based exercises have been digitized and are publicly available, by Piet Mertens [http://bach.arts.kuleuven.be/MOAP/]] Snider, Keith (2014). On establishing underlying tone contrast. In Steven Bird & Larry Hyman (eds.) How to study a tone language? Language Documentation and Conservation 8, 707-737. Stevens, Kenneth N. (1989). On the quantal nature of speech. Journal of Phonetics 17, 3-45. Stevens, Kenneth N. & Samuel J. Keyser (2010). Quantal theory, enhancement and overlap. Journal of Phonetics 38, 10-19. ‘t Hart, Johan, René Collier & Antonie Cohen (1990). A perceptual study of intonation: an experimental-phonetic approach to speech melody. Cambridge University Press. Wedekind, Klaus (1983). A six-tone language in Ethiopia – tonal analysis of Benč4non4 (Gimira). Journal of Ethiopian Studies 16, 129-156. Woodbury, Tony (2003). Defining documentary linguistics. Language Documentation and Description 1, 35-51. Xu, Yi (1999). Effects of tone and focus on the formation and alignment of F0 contours. Journal of Phonetics 27: 55-105. Xu, Yi & Xuejing Sun (2002). Maximum speed of pitch change and how it may relate to speech. Journal of the Acoustical Society of America 111, 1399-1413. Yip, Moira (2002). Tone. Cambridge University Press. Yu, Kristine M. (2014). The experimental state of mind in elicitation: illustrations from tonal fieldwork. In Steven Bird & Larry Hyman (eds.) How to study a tone language? Language Documentation and Conservation 8, 738-777. Zhang, Jie (2001). The effects of duration and sonority on contour tone distribution – Typological survey and formal analysis. UCLA PhD dissertation.

Bibliography

23

The following slides have bits and pieces that might come in handy in response to questions

24

Methods / Appendix

•  Number of utterance tokens: 293

•  Aggregated over repetitions, ahead of quantitative analyses: 154 types. So there are 8 missing values, given a logical total of 162 (18 items x 9 speakers).

•  ANOVAs are based on balanced dataset, with 105 types. So there are 3 missing values, given a logical total of 108 (12 items [3 for each level of Tone] x 9 speakers).

Shilluk •  A Western Nilotic language,

within the Nilo-Saharan language family

•  Spoken in South Sudan

•  At least 200,000 speakers

Figure. Map from Storch (2005), showing the Western Nilotic languages

25

Background on Shilluk and Dinka

House (1990)

•  Evidence from perception experiments, in which tonal alignment is varied across the syllable

•  Perception of falling f0 patterns according to House (1990:133ff):

26

But there is some counterevidence

V C V C V Segmental sequence

F0 contour

27

House (1990)

“For the movement contour features Falling [...] and Rising to be optimally perceived, three conditions must be fulfilled”:

1.  The fall takes place through a zone of relative spectral stability;

2.  The fall sets in 30-50 milliseconds into the vowel;

3.  The vowel is at least 100 milliseconds long.


28

House (1990)

•  “Another area altogether is that of tonal distinctions: what if anything is quantal about them? Candidates for quantal characterizations of tone do not leap immediately to mind.”

[Stevens & Keyser 2010:13]

•  House’s hypothesis is in effect a proposal for such a quantal threshold.


Noise in f0 traces: voicing in plosive consonants

Figure. F0 traces of 2 question-word question in Kuot (Papuan, New Ireland). Data from Eva Lindström.

Noise in f0 traces: spikes between vowels and (nasals, [l])

Figure: F0 trace of a question-word question in Kuot (New Ireland). Data from Eva Lindström.

elicitation and documentation of tone and documentation of tone what is the place of quantitative...

Documents