acoustic and perceptual effects of air sacs bart de boer university of amsterdam
TRANSCRIPT
Acoustic and perceptual effects of air sacs
Bart de Boer
University of Amsterdam
The air sac
Siamang
Brown Howler monkey
Orang Utan
Humans
Air sac prevalence
• All apes have air sacs
• But humans don’t
• Why?– Speech?
The anatomy of air sacs
“siamang” “howler monkey” (subhyoid)
The function of air sacs?
• Speculation
• Acoustics:– Resonators (e. g. Avril, 1963; Schön, 1971) – Impedance matchers (e.g. Fitch & Hauser, 1995)– Suppressing resonances (Haimoff, 1983)
• Non-acoustic– Accidental byproduct (Brandes, 1932) – Rebreathing air (Negus, 1949) – Preventing hyperventilation (Hewitt et al., 2002)
Why is this even important?
• For the evolution of speech
• Correlation air sacs – hyoid bone
– Bones fossilize!
Shape correlation
• When an air sac is present (at least in primates) the hyoid has a cup-shaped front (the bulla)
Chimpanzee(Avril 1962)
Human
(http://www.anatomyatlases.org/atlasofanatomy/
plate01/08hyoidbone.shtml)
Brown howler monkey
Fossil evidence
• Four fossil hyoid bones are known– Neanderthal (60 KyA)– 2x H. heidelbergensis (530 KyA)– Australopithecus afarensis (3.3 MyA)
Dikika baby, A. afarensis(Alamseged et al. 2006)
Neanderthal (Arensburg et al 1989)
H. Heidelbergensis (Martínez et al. 2008)
Fossil conclusion
• The latest common ancestor of Neanderthals and modern humans did not have air sacs
• Australopithecus afarensis did
• (+ possibility to find evidence for intermediate ancestors)
Implications for speech?
• Did air sacs disappear because of speech?
• But we need to be sure about their acoustic effect!
An acoustic model
• Important elements:– The neck– The cavity– The wall– Radiation
• Can be analyzedas an electricalcircuit
Pressure
wall loss
radiation
input
LwLN CwRwRN
Rc Cc
Rr Lr
20 50 100 200 500 1000 2000 400010
4
105
106
107
frequency (Hz)
impe
danc
e (k
gm
-4s-1
)
softcartilagebone
The acoustic effect
• A new peak at ~the resonance of the air sac• Original formants get shifted up and closer together
[a]
[ə]
[y]
Perception issues
100 500 1000 1500 2000 2500 3000 3500 4000
10
-5
10
0
10
5
Frequency (Hz)
Rel
. Pow
er
[y]
Fs1
F1 Fs2F2 F3 Fs3
F4
• We could measure distances in formant space– But which formants do we take?– Does the air sac resonance take over the role of F1?– And how about F2 – F2’?
Perception experiment
• Expose subjects to stimuli, and let them classify– [a]/[ə], [a]/[y], [ə]/[y], [a]/[ə]/[y]– (Two or three - unforced choice)– With and without air sac
• Iteratively find noise level at which performance halfway between perfect and chance
Example trial
2 4 6 8 10 12 14 16 18 20-30
-20
-10
0
10
20
trial
Sig
nal/N
oise
(dB
)
Without air sac
ae
ayey
aey
2 4 6 8 10 12 14 16 18 20-30
-20
-10
0
10
20
trial
Sig
nal/N
oise
(dB
)
With air sac
ae
ayey
aey
Perception experimental results
**
Conclusion
• Apes have air sacs, humans don’t– Neanderthals don’t– Australopithecines do– Lost in evolution– Why?
• Change spectrum of speech– Can hear the difference less well
• Lost because of speech
• Hypothesis:– Neanderthals could speak, Australopithecines not– (proto-) speech is at least 500 000 years old
Analysis