[PPT] - John J. Ohala Department of Linguistics University of California, PowerPoint Presentation

SLIDE 1

John J. Ohala Department of Linguistics University of California, Berkeley

TAL Nanjing 2012

1

SLIDE 2

S tructure of the Talk

1. Introduction: Why study physiology of tone

and a brief history of discoveries.

The principal mechanism:

contraction of the crico-thyroid *and adj acent muscles of larynx)

The influence of Ps (subglottal pressure)
The role of larynx height
The effect of the pharyngeal muscles (speculative)
2. Ethological determinants of speech F0

TAL Nanjing 2012

2

SLIDE 3

TAL Nanjing 2012

3

Why study the physiology of F0 variation?

It has been demonstrated many times (by Rousselot, Passy, S ievers, Chiba & Kaj iyama, etc.) that an understanding of how speech is produced and perceived can yield explanations

f sound patterns: speech sound inventories,

sound changes, phonotactics, etc. This should apply to tones, tonal sound patterns, and intonation in genera.,

SLIDE 4

S peech may be considered as a sound (voice) modulated by articulations of the upper vocal tract. The latter were known from ancient times (e.g., by the S anskrit grammarians, notably Panini --, by the Arab and Persian grammarians, and the phoneticians

f the Renaissance and Enlightenment, notably

Amman, Wilkins, etc.) But there was some confusion and difference of opinion about what we now refer to as “ suprasegmentals” , including tone.

TAL Nanjing 2012

4

Brief History of understanding of F0 variation.,

SLIDE 5

From early times voice was understood as similar to certain musical instruments, notably, wind instruments, esp. flute, horn, etc. In the mid 18th century, the idea grew that voice was generated by a strnged instrument. This was precipitated especially by the research of Ferrein 1741, hence his term: “ les cordes vocales” . This was an advance but: In fact, a musical instrument that is a closer analogue to voice is the sheng, a musical instrument used in East and S

utheast Asia which

uses a ‘ free reed’ , a mechanism that produces a pulse-like sound rich in harmonics over a wide frequency range.

TAL Nanjing 2012

5

SLIDE 6

The sheng: an ancient musical instrument (from East and S

utheast Asia). It employs a free reed (as do some

modern Western instruments such as the accordion and the harmonica).

TAL Nanjing 2012

6

SLIDE 7

I can’ t resist the impulse to inj ect parenthetically that the free reed was used in the first mechanical speech synthesizer by, Christian Gottlieb Kratzenstein, in 1781. Kratzenstein was a German physician who practiced in Russia and Denmark. He is credited with introducing the free reed in the West.

TAL Nanjing 2012

7

SLIDE 8

Now a brief review of laryngeal anatomy:

TAL Nanjing 2012

8

My apologies: this duplicates to some extent elements in the presentation of Hiroya Fuj isaki.

SLIDE 9

TAL Nanjing 2012

9

SLIDE 10

S

me of the most extensive and quantitative

research on the action of the larynx was conducted by the pioneering German physiologist, Johannes Müller. The next slide shows one of his experimental set-ups.

TAL Nanjing 2012

10

SLIDE 11

Excised larynx

TAL Nanjing 2012

11

SLIDE 12

Müller demonstrated that tilting the thyroid cartilage with respect to the cricoid cartilage was sufficient (and plausible) to account for the increase of F0 over the range found in humans. Moreover he identified the cricothyroid muscles as being located and structured in a way to accomplish this. More modern studies using EMG (incl. some by Hirano, Vennard, and Ohala) suggest that other laryngeal muscles, notably the lateral crico- arytenoid, the vocalis, and the interarytenoid are also active during F0 increases. In some cases such activity may be necessary to counteract an abductory tendency due to contraction of the cricothyroid.

TAL Nanjing 2012

12

SLIDE 13

TAL Nanjing 2012

13

SLIDE 14

The result of Müller’ s quantification of the effect

f subglottal pressure (Ps) on F0 yielded values

such as 4.5 Hz/ cm H2O.

TAL Nanjing 2012

14

Müller quantified the effect of the weights pulling on the cartilages to stretch and thus tense the vocal cords but such quantification was difficult to translate into in vivo muscle contractions. But the results on subglottal pressure were more easily understood in real term:

SLIDE 15

TAL Nanjing 2012

15

SLIDE 16

TAL Nanjing 2012

16

SLIDE 17

This is decidedly a second-order effect and in any case it is not clear that speakers have much control over Ps in a way to purposely vary F0 . Ps is affected by glottal resistance which in turn is affected by glottal opening and vocal cord tension.

TAL Nanjing 2012

17

SLIDE 18

What Müller did was in vitro studies in that he used excised larynges but his results have been supported by modern in vivo studies on intact speakers. These require the skill of an otolaryngologist. I had the privilege of working with Dr., Minoru Hirano from Kurume University.. He pioneered methods of inserting electrodes into all the intrinsic and some of the extrinsic muscles of the larynx via palpation of the landmarks on the larynx and developing “ proof” criteria for verifying correct placement.

TAL Nanjing 2012

18

SLIDE 19

From this we found general patterns of laryngeal muscle activity for various functions including increasing pitch (F0).

TAL Nanjing 2012

19

SLIDE 20

This, then, is the primary mechanism for varying F0, which is the primary (though not the

nly) phonetic correlate of tone.

But that still leaves some other aspects of the mechanism of tone to be discovered and explained. One of these is how F0 is lowered. One might think that it is simple relaxation of the muscles raising Fo that should suffice. I thought that but it turns out that isn’ t the case.

TAL Nanjing 2012

20

SLIDE 21

TAL Nanjing 2012

21

SLIDE 22

TAL Nanjing 2012

22

SLIDE 23

TAL Nanjing 2012

23

SLIDE 24

From this we found general patterns of laryngeal muscle activity for various functions including increasing and decreasing pitch (F0).

TAL Nanjing 2012

24

I don’ t fully understand this

SLIDE 25

Why should the sternohyoid be involved in lowering F0? Let’ s look at the anatomy;

TAL Nanjing 2012

25

SLIDE 26

The sternohyoid is not connected to the larynx but it is connected to the hyoid bone which is connected to the larynx.

TAL Nanjing 2012

26

As it happens there is a good correlation between larynx height (in the neck) and F0. (This was known qualitatively to Greek physicians and was thought to be the principal mechanism for F0 control. It’ s not, of course, but we need to try to understand it.)

SLIDE 27

TAL Nanjing 2012

27

SLIDE 28

TAL Nanjing 2012

28

SLIDE 29

TAL Nanjing 2012

29

SLIDE 30

TAL Nanjing 2012

30

SLIDE 31

TAL Nanjing 2012

31

This is a tracing of the larynx of a person who has paralysis of the crico-thyroid muscle. He is able to change F0 by changing the vertical position of the larynx.

SLIDE 32

TAL Nanjing 2012

32

Now, why should altering the vertical position of the larynx affect F0? The best answer I can give is that it changes the

vertical tension of the vocal cords.

That is, along with anterior-posterior tension of the vocal cords – controlled by the crico-thyroid, the vertical tension, controlled by the so-called strap muscles ion the neck, can also influence F0.

SLIDE 33

TAL Nanjing 2012

33

It is difficult to get direct evidence for this – at least in an intact, living, speaker, however some coronal x-rays taken by Arnold 1961 show the laryngeal ventricle increasing with increasing F0. I believe this means increasing vertical tension, too.

SLIDE 34

TAL Nanjing 2012

34

S umming up up so far: we have posited two maj or mechanisms for varying F0: contraction

f the crico-thyroid and variations in larynx
height. In addition we have identified one

minor mechanism: changes in the transglottal air pressure. Although it is speculative, I would like to suggest that we consider a 4th mechansim: contraction of the pharyngeal muscles.

SLIDE 35

TAL Nanjing 2012

35

It is well known that F0 is slightly different after voiced and voiceless obstruents: higher after voiceless and lower after voiced. The different is not that great: 10 ~ 15 Hz in a male speaker. But this is large enough to be detectable by a listeners. It is, moreover, the best option for the explanation of why tonal distinctions high vs. low develop after voiceless and voiced obstruents, respectively.

SLIDE 36

TAL Nanjing 2012

36

It was thought to be due to aerodynamic factors: a higher pressure drop across the vocal cords in the case of voiceless obstruent vis-à-vis voiced ones. But there is some evidence that undercuts this explanation

SLIDE 37

TAL Nanjing 2012

37

This figure show subglottal pressure for an American English voiceless aspirated stop on the left and a simple voiced stop on the right., The Ps is lower on the vowel immediately after the period of aspiration in comparison to the vowel after the voiced stop. Yet the different in F0 is has been uniformly found for American English stops.

SLIDE 38

TAL Nanjing 2012

38

The reason for suspecting the pharyngeal muscle – specifically the inferior constrictor – of being involved is that it has been found by some (including Joe Perkell in his 1969 dissertation) that the pharynx is more expanded for voiced stops than for voiceless stops., Another way

f stating this is that the voiced stops have more relaxed

walls so they can take advantage of the compliance of the walls to “ give” to the impinging pressure that builds up during a stop, thus keeping the pressure drop across the glottis – necessary for vocal cord vibration – low

enough. On the other hand it would seem necessary

that the pharyngeal walls should be tense (and less compliant) in the case of the voiceless stops – to ensure a pressure drop that guarantees voicelesssness.

SLIDE 39

TAL Nanjing 2012

39

Another topic: F0 can be lowered faster than it can be raised. I have the impression that falling tones are executed faster than rising tones. I’ m not sure why but I think his pattern has relevance for tonology.

SLIDE 40

TAL Nanjing 2012

40

Finally, I’ d like to move to a completely different domain where I think there is an influence on the function of tone, not its shape, namely, ethology. Ethology is the study of comparative behavior,

esp. behavior that imparts survival or “ fitness”

to the behaver. That is, it is behavior viewed from a Darwinian perspective.

SLIDE 41

TAL Nanjing 2012

41

It has been argued l(by Morton and by me) that certain face-to-face vocal signals are shaped to convey to the other their relative size and that they do this – even deceitfully -- in part by modifying their vocal F0: the confident aggressor produces as low an F0 as possible and the submissive and appeasing individual produces as high an F0 as possible. I have argued that this pattern – which I call ‘ the frequency code’ -- influences humans’ vocal F0: low and high F0 to designate large and small obj ects, respectively and utterances expressing self- sufficiency or confidence vs. requests fop help or information, respectively.

SLIDE 42

TAL Nanjing 2012

42

I believe this accounts for the common cross- language pattern where questions –

when they are

not signaled by certain words or morphemes or by syntactic means -- use high F0 (and low F0 for statements).

SLIDE 43

TAL Nanjing 2012

43

I have reviewed some anatomical and physiological and ethological factors – that is factors outside language as such – which help to explain patterns involving tones. Certainly these do not exhaust the tonal patterns that might be so explained.

SLIDE 44

TAL Nanjing 2012

44

John J. Ohala Department of Linguistics University of California, Berkeley

S tructure of the Talk

and a brief history of discoveries.

contraction of the crico-thyroid *and adj acent muscles of larynx)

Why study the physiology of F0 variation?

It has been demonstrated many times (by Rousselot, Passy, S ievers, Chiba & Kaj iyama, etc.) that an understanding of how speech is produced and perceived can yield explanations

sound changes, phonotactics, etc. This should apply to tones, tonal sound patterns, and intonation in genera.,

S peech may be considered as a sound (voice) modulated by articulations of the upper vocal tract. The latter were known from ancient times (e.g., by the S anskrit grammarians, notably Panini --, by the Arab and Persian grammarians, and the phoneticians

Amman, Wilkins, etc.) But there was some confusion and difference of opinion about what we now refer to as “ suprasegmentals” , including tone.

Brief History of understanding of F0 variation.,

uses a ‘ free reed’ , a mechanism that produces a pulse-like sound rich in harmonics over a wide frequency range.

The sheng: an ancient musical instrument (from East and S

modern Western instruments such as the accordion and the harmonica).

Now a brief review of laryngeal anatomy:

My apologies: this duplicates to some extent elements in the presentation of Hiroya Fuj isaki.

S

research on the action of the larynx was conducted by the pioneering German physiologist, Johannes Müller. The next slide shows one of his experimental set-ups.

Excised larynx

The result of Müller’ s quantification of the effect

such as 4.5 Hz/ cm H2O.

Müller quantified the effect of the weights pulling on the cartilages to stretch and thus tense the vocal cords but such quantification was difficult to translate into in vivo muscle contractions. But the results on subglottal pressure were more easily understood in real term:

This is decidedly a second-order effect and in any case it is not clear that speakers have much control over Ps in a way to purposely vary F0 . Ps is affected by glottal resistance which in turn is affected by glottal opening and vocal cord tension.

From this we found general patterns of laryngeal muscle activity for various functions including increasing pitch (F0).

This, then, is the primary mechanism for varying F0, which is the primary (though not the

But that still leaves some other aspects of the mechanism of tone to be discovered and explained. One of these is how F0 is lowered. One might think that it is simple relaxation of the muscles raising Fo that should suffice. I thought that but it turns out that isn’ t the case.

From this we found general patterns of laryngeal muscle activity for various functions including increasing and decreasing pitch (F0).

I don’ t fully understand this

Why should the sternohyoid be involved in lowering F0? Let’ s look at the anatomy;

The sternohyoid is not connected to the larynx but it is connected to the hyoid bone which is connected to the larynx.

As it happens there is a good correlation between larynx height (in the neck) and F0. (This was known qualitatively to Greek physicians and was thought to be the principal mechanism for F0 control. It’ s not, of course, but we need to try to understand it.)

This is a tracing of the larynx of a person who has paralysis of the crico-thyroid muscle. He is able to change F0 by changing the vertical position of the larynx.

Now, why should altering the vertical position of the larynx affect F0? The best answer I can give is that it changes the

vertical tension of the vocal cords.

That is, along with anterior-posterior tension of the vocal cords – controlled by the crico-thyroid, the vertical tension, controlled by the so-called strap muscles ion the neck, can also influence F0.

It is difficult to get direct evidence for this – at least in an intact, living, speaker, however some coronal x-rays taken by Arnold 1961 show the laryngeal ventricle increasing with increasing F0. I believe this means increasing vertical tension, too.

S umming up up so far: we have posited two maj or mechanisms for varying F0: contraction

minor mechanism: changes in the transglottal air pressure. Although it is speculative, I would like to suggest that we consider a 4th mechansim: contraction of the pharyngeal muscles.

It was thought to be due to aerodynamic factors: a higher pressure drop across the vocal cords in the case of voiceless obstruent vis-à-vis voiced ones. But there is some evidence that undercuts this explanation

The reason for suspecting the pharyngeal muscle – specifically the inferior constrictor – of being involved is that it has been found by some (including Joe Perkell in his 1969 dissertation) that the pharynx is more expanded for voiced stops than for voiceless stops., Another way

walls so they can take advantage of the compliance of the walls to “ give” to the impinging pressure that builds up during a stop, thus keeping the pressure drop across the glottis – necessary for vocal cord vibration – low

that the pharyngeal walls should be tense (and less compliant) in the case of the voiceless stops – to ensure a pressure drop that guarantees voicelesssness.

Another topic: F0 can be lowered faster than it can be raised. I have the impression that falling tones are executed faster than rising tones. I’ m not sure why but I think his pattern has relevance for tonology.

Finally, I’ d like to move to a completely different domain where I think there is an influence on the function of tone, not its shape, namely, ethology. Ethology is the study of comparative behavior,

to the behaver. That is, it is behavior viewed from a Darwinian perspective.

I believe this accounts for the common cross- language pattern where questions –

not signaled by certain words or morphemes or by syntactic means -- use high F0 (and low F0 for statements).

I have reviewed some anatomical and physiological and ethological factors – that is factors outside language as such – which help to explain patterns involving tones. Certainly these do not exhaust the tonal patterns that might be so explained.

Thank you!