Sound Symbolism: An Explanatory Theory from Ethology

1
Sound Symbolism An Explanatory Theory from
Ethology
John J. Ohala Department of Linguistics Universi
ty of California, Berkeley Ohala_at_socrates.berkele
y.edu
2
Outline
0. Introduction
1. Brief overview of common cross-language
manifestations of sound-meaning correlations.
2. A theory initially applied to non-human
vocalizations
3. Elaborations of the theory application to
human speech
4. Further implications of the theory to
a. The origin of the smile
b. The reason for sexual dimorphism in vocal
anatomy in humans
c. Speculations on laryngeal morphology for the
origin of speech
3
0. Introduction
Regarding non-arbitrary or systematic
correlations between sound and meaning
(onomatopoeia or sound imitation) has been
recognized for centuries, even millennia. There
is little controversy about this. It was
acknowledged by the Classical Grammarians and
treated as a case where the usual sound laws did
not apply.
Examples the bark of a dog
English bAU wAU Hindi b??) b??) Swe
dish vov vov
Thai h?N ? h?N ?
4
Even as regards non-arbitrary sound-meaning
correlations where the thing designated is not
inherently acoustic, there has been speculation
for centuries, arguably for millennia. But in
this case there has been controversy. The
evidence for it often did not live up to the
otherwise high standards established by the
Classical Grammarians. Saussure felt safe in
declaring the sign is arbitrary.
Example (from a novelty song of the 1960s)
Itsy Bitsy Teenie Weenie Yellow Polka Dot Bikini
5
But beginning in the 1920s psycholinguistic
evidence was amassed (by, among others, the
highly respected Edward Sapir) and in the
following decades statistical evidence was also
presented based on analyses of large textual
corpora and of searches through extensive
lexicons.
It seemed inescapable that for some vocabulary
and for some few semantic dimensions there was a
provable link between sound and meaning.
Still, with some exceptions this concept, now
generally called sound symbolism has been
regarded with suspicion by mainstream linguists.
Several UC Berkeley linguists, Emeneau, Haas,
Malkiel, Hinton, Matisoff, Aoki, Nichols, Berlin,
Rhodes, Ohala have written about it.
6
Is this due to the great influence of Saussure or
due to some other factor?
At least one cause of skepticism is that there is
no convincing theory to account for sound
symbolism. (Among the theories ignored, not
without reason, are those by des Brosses 1765,
Court de Gebelin 1776, Paget 1930 and others.)
My purpose in this presentation is to present
what I think is a viable theory accounting for at
least some forms of sound symbolism. This theory
is ethological in nature.
(Ethology is the comparative study of behavior.)
7
1. Brief overview of common cross-language
manifestations of sound-meaning correlations.
As discussed and documented by Hermann 1942 and
Bolinger 1978, there is an uncanny similarity in
the use of intonation, i.e., voice F0, in
signaling question vs. statement (declaration)
(in those cases where questions are not signaled
by special words like who, what or by
syntactic means, such as auxiliary inversion in
English.
Namely declarations have low F0 and questions
high F0.
8
The exact use of high F0 in questions may vary

• Like English, the high F0 may appear as a
  terminal rise

• Or the contours for question and statement may be
  similar but the F0 for questions in overall higher

• Or questions may simply have a higher F0
  somewhere in the utterance, not necessarily at
  the end.

9
Exceptions to this pattern may exist in some tone
languages or when other pragmatic factors
override, e.g., a terminal rise on statements
when politeness or deference is being conveyed.
10
This kind of cross-language similarity in
sound-meaning correlation, even in languages
genetically unrelated, is important.
It is unheard of with other meanings. E.g., cf.
English cup, Hindi pjala, French tasse,
Japanese honpon.
Lieberman 1967 suggested that since questions are
marked syntactically (often marked by special
lexical items, etc.), they are also marked
phonetically and have extra tension in the vocal
cords. By this argument, any marked grammatical
or semantic class, e.g., plural or past tense
should also be characterized by extra tension in
some part of the vocal anatomy. (Anyway, as it
happens, it takes extra tension in certain
laryngeal muscles to lower F0 for statements,
too.)
11
Among other meanings showing similar sounds
across languages is SIZE (and related dimensio
ns such as distance, familiarity, status)
Relevant data (next slide)
12
(No Transcript)
13
And some data on words for large
14
 
15
Words for small tend to have vowels (and
consonants, and tone, where relevant) that have
high acoustic frequency, e.g., in the case of
vowels i I e.
Words for large tend to have vowels (and
consonants and tones) that have low acoustic
frequencies, e.g., in the case of vowels A ? o.
This can be seen more easily on the acoustic
vowel space (F1 vs. F2)
16
(No Transcript)
17
This tendency has psychological validity as
demonstrated in psycholinguistic tests by Sapir,
Roger Brown, Eli Fischer-Jørgensen, among others.

The results are also supported by statistical
studies of vocabulary by Thorndike 1945,
Chastaing 1958, Ultan 1978, among others.
But, why and how?
18
2. A theory initially applied to non-human
vocalizations
The hint of a possible answer was suggested by
Eugene Morton 1977, although he studied the
agonistic vocalizations of
28 avian species and 28 mammalian species
Agonistic vocalizations are those sounds made
in face-to-face threatening situations. (Does
not necessarily include territorial calls,
food-finding calls, mother-infant exchanges,
courtship vocalizations.)
19
His findings an uncanny cross-species
similarity
Threatening vocalizations were low in F0 (for the
given animal), and often somewhat rough and
aperiodic.
lion
Submissive and affinitive vocalizations were high
in F0 and usually more tone-like (i.e., periodic).
kitten
20
His explanation
The threatening individual tries to appear as
large as possible (since size, possibly
correlated with age and maturity, usually decides
the outcome of any competition), and employs both
visual and acoustic signals to accomplish this.
Low F0 is mandated since the natural frequency of
structures like vocal cords or syringeal
membranes is inversely proportional to the
overall body size of the vocalizer.
A submissive individual tries to appear as small
and non-threatening as possible. It may engage
in infant mimicry. Both visual and acoustic
signals are employed. High F0 is mandated.
21
3. Elaborations of the theory application to
human speech
My elaboration on Mortons account
1. The same principles apply to humans use of
voice F0. The person asking a question can be
viewed as requiring the cooperation of the person
to whom the question is addressed. Therefore a
supplicating intonation is appropriate. A
declarative statement, on the other hand, signals
the speakers self-confidence and control of the
information conveyed.
2. More importantly, I suggest that not only F0
but also the resonances of the vocal tract can
also indirectly convey an impression of size of
the vocalizer. As with F0 the acoustic
frequencies of the resonances are inversely
proportional to the overall body size of the
vocalizer.
22
Acoustically high resonances should convey an
impression of a small vocalizer and low
resonances, of a larger vocalizer.
How does this explain the association of the
phonetic classes of sounds noted above with words
meaning small and large? On the surface it
does not because the speaker uttering small
words like the English teeny ÈtHini, does not
himself wish to appear small rather he wants to
convey the smallness of the thing which that word
describes.
But since as a word such teeny is supposed to
evoke in the listener an impression of something
small, the particular sounds which the word
consists of may enhance that evocation.
23
I have labeled the cross-language and
cross-species use of acoustic frequency whether
in the sound source or the sound resonances -- to
convey impressions of size (and related concepts)
the   frequency code.
24
4. Further implications of the theory
25
a. The origin of the smile
The smile manifested widely across different
cultures and across different species (at least
those with a movable facial cover such as canids
and primates)  
A puzzle why should a non-threatening facial
display show so many teeth, which are potential
weapons and thus more appropriate for a threat
display?
http//www.newequations.com/type1.htm
26
The smile
http//humanityquest.com/topic/Photos/
27
Also, in many species (I think humans included)
there is another facial display which is part of
threat signals which is opposite to the smile in
that it involves protrusion of the lips and thus
greater concealment of the teeth (I call this the
o-face).
http//www.emotioneric.com/anger.jpg
28
(No Transcript)
29
I propose instead that the smile -- and its
opposite, the o-face -- originally served as a
component of the acoustic element of these
displays.
In the smile the mouth corners are drawn back to
effectively shorten the vocal tract and to give
rise to higher resonant frequencies.
(This shortening is much more effective in
species that have a snout, i.e., a mouth that
protrudes from the plane in which the eyes are
set. Most primates and certainly canids have
snouts the human species used to have a snout
and it is a reasonable guess that these facial
expressions existed then and have been retained
even after homo sapiens lost its snout.)
The o-face, on the other hand, through mouth
constriction and/or lip protrusion serves to
lower the resonant frequencies.
30
(No Transcript)
31
Possible objection humans and some primates
smile with mouth closed or without an
accompanying vocalization -- thus undercutting my
claim that these mouth shapes serve an acoustic
end.
However, a gesture serving a given purpose can
become ritualized and thus performed in the
absence of its original purpose or even to be
performed for a completely different purpose.
Today the smile may serve in addition as an
independent visual signal.
Bauer (1987) found a correlation in the predicted
direction between mouth opening and F0 of
vocalization in chimpanzee threat and non-threat
displays. There is no physiological necessity
for such a correlation -- the muscles serving
facial expressions do not interact with those
modulating vocal cord tension -- so if it exists,
it must be for the behavioral reasons adduced.
32
Weak support for my hypothesis is that smiles may
also be toothless! -)
33
b. The reason for sexual dimorphism in vocal
anatomy in humans
Adult male and female humans have different
average dimensions of parts of their vocal
anatomy.

• Males have
• a vocal tract that is about 15 to 20 longer than
  females
  
• vocal cords that are about 50 longer (and
  considerably more massive).

These two differences account for the lower
resonant frequencies of males vowels and the
nearly 1 octave difference in typical male and
female F0s.
34
(No Transcript)
35
I propose that these differences exist in order
to enhance the acoustic component of the adult
males threat display.
Perhaps these anatomical features increase a
males chances of success in garnering scarce
resources mates or food and territory.
36
There may also be an element of sexual selection
in the evolutionary development of these
anatomical features since it is known that
females prefer as mates those with more masculine
traits such as a low-pitched voice.
Even in species where the young are nurtured
exclusively by the female, there would still be
an advantage to the female to choose mates with
these traits because they would help to guarantee
the same traits in her offspring and thus the
propagation of her genes.
The sexually dimorphic aspects of the vocal
anatomy take their place alongside similar
dimorphic aspects that serve in visual displays,
e.g., beards. Beards in humans serve to make the
head appear larger than it really is and thus the
beard owner more potent.
37
c. Speculations on laryngeal morphology for the
origin of speech
The following a paper given at the 2000 Paris
Mtg on Origin of Language Speech
38
The irrelevance of the lowered larynx position in
modern man for the origin and development of
speech
39
Why has man developed a lower larynx position --
and thus larger pharynx -- than his anthropoid
cousins?
There are a number of hypotheses

• It is a response to upright posture, i.e.,
  bi-pedal locomotion.

• It is a response to man having less of a snout
  than his ancestors.

• It was an essential element in the origin and
  evolution of the capacity for speech.

40
Why has man developed a lower larynx position --
and thus larger pharynx -- than his anthropoid
cousins?
There are a number of hypotheses

• It is a response to upright posture, i.e.,
  bi-pedal locomotion.

• It is a response to man having less of a snout
  than his ancestors.

• It was an essential element in the origin and
  evolution of the capacity for speech.

41
Lieberman suggested that mans larger and longer
pharynx endows him with the capacity to make a
full range of vowels and that this is an
essential element for articulate speech (as we
know it). Other species (Neanderthal, apes) as
well as human newborns, lack this ability because
of their lack of a large pharynx.
42
I will offer a critique of the last proposal and
will propose an entirely different explanation
for the lowered larynx and larger pharynx found
in man.
43
My main points of criticism of the proposal that
the large pharynx in man is an adaptation for
speech are these

• who has the large pharynx

• when, in the human life cycle, the pharynx
  enlargement occurs

• what other developmental events occur when
  pharynx enlargement occurs

• similar enhancements of the vocal apparatus are
  found in other species that dont have speech

44
Who Has the Large Pharynx?
Up to now I have deliberately exploited an
ambiguity in the phrase the large pharynx in
man man homo sapiens or man adult
male?
In fact, it is adult male humans that have a
large pharynx adult female humans do not have
such a remarkably large pharynx.
Carré Lindblom argue that females still have a
larger pharynx than our ape cousins and could
still qualify as an adaptation for speech.
45
Okay, but if males have a larger pharynx than
females (undeniable fact) then it would follow
that males should have more language and speech
capability than females.
But this is contradicted by much evidence

• Females perform better on verbal tests in all
  cases where society gives both sexes equal
  educational opportunity and encouragement.

• Males outnumber females -- in some cases by 4 or
  5 to 1 -- in several forms of speech disorders,
  including stuttering and autism, and delayed
  acquisition of language.

Background assumption when there is sexual
dimorphism, that sex that has a greater need for
a structure has it or has more of it, e.g.,
nuptial pads on the feet of male frogs (used in
copulation).
46
Conclusion the larger pharynx cannot be an
adaptation for speech and language because, if
so, it is not working.
47
When, in the Human Life Cycle, the Pharynx
Enlargement Occurs
It occurs at puberty.
Background assumption when age dimorphism
occurs, the given anatomical feature appears when
the organism can use it (and disappears when it
no longer needs it), e.g., the egg tooth that
appears on some birds and reptiles just before
hatching from the shell.
The descent of the larynx (thus enlarging the
pharynx) occurs in males long after they have
begun to speak. The timing would seem to be very
bad if the large pharynx were an adaptation for
speech.
48
(No Transcript)
49
Conclusion the large pharynx must not be an
adaptation for speech but it might be an
adaptation for some other activity connected with
the sex role of males.
A challenge to those who believe the lower larynx
in humans is an adaptation for speech please
make your case with respect to the vocal tract of
a 3 or 4 year old child, i.e., the approximate
age by which a sufficient phonology exists to
support reasonably complex vocal communication.
My assumption such a case cannot be supported
by the evidence. Very young children who
nevertheless speak very well have a very small
pharynx, relative to the rest of the vocal tract.
50
What Other Developmental Events Occur When
Pharynx Enlargement Occurs?
There is another even more remarkable change in
the males vocal apparatus that happens at
puberty his vocal cords increase in length and
mass.
In fact, although the pharynx in males is up to
about 15 longer than that in females, the vocal
cords are fully 50 longer and much more massive.
Another change at puberty is the growth of facial
hair.
51
Can the pharynx enlargement, vocal cord
lengthening, and growth of facial hair all be
related?
I suggest that they are. They all enhance the
males ability to produce a more convincing
threat display.
Facial hair -- crucially around the perimeter of
the face -- makes the person seem larger since
the head subtends a larger angle in the visual
field of a viewer. It serves the same function,
then, as the mane of lions, the forehead hair of
male gorillas, and similar hair growth on the
heads of other primate species.
52
Adult Female Gorilla
Adult Male Gorilla
53
If facial hair functions in a visual component of
a threat display, then the enhancements to the
vocal apparatus functions in the auditory
component of threat displays.
The larger pharynx and longer vocal cords both
give the males voice lower frequencies (both the
sound source or fundamental frequency and the
vocal tract resonances) consistent with a
larger and thus more threatening vocalizer.
Background there is abundant evidence, both
from experimental psychology and everyday life,
that low-pitched voices sound more authoritative,
more confident, and more threatening.
54
This is consistent with the generally accepted
point that males are, as a rule, more aggressive
than females.
Males may engage in competition more since they
have to compete for females. The male requests
the female grants. This is consistent with the
general ethological principle that females are
more valuable than males they possess the more
valuable gamete ova are more metabolically
costly than sperm, and there are orders of
magnitude fewer of them.
55
Similar Enhancements of the Vocal Apparatus Are
Found in Other Species (That Obviously Dont Have
Speech)
Many other (non-human, non-speaking) species show
enhancements of their vocal system and some of
these have sexual dimorphism in this

• Some geese, virtually all cranes (notably the
  whooping crane)

• A bird of paradise, phonygammus (in French oiseau
  trompette) has a trachea longer than the bird
  itself -- the extra length coiled like a rope
  between the sternum and the outer skin.

56
(No Transcript)
57
The length and position of the trachea are
critical features of crane anatomy, and shape the
distinctive voices of the various cranes
(Niemeier 1983). With the exception of the
Siberian Crane, the trachea of all Grus species
coils on the vertical plane within the sternum.
In the Brolga, Sandhill, and Sarus Cranes, the
coiled trachea fills most of the anterior half of
the sternum, while in the White-naped, Eurasian,
Whooping, Hooded, Black-necked, and Red-crowned
Cranes the trachea penetrates the entire sternum.
The bony rings of the trachea fuse with the
sternum to create thin plates. When cranes
vocalize, the plates vibrate. This amplifies the
cranes calls, which can carry several kilometers
(Gaunt et al. 1987).
58

• Orangutans have laryngeal ventricles that inflate
  during vocalizations and probably act as added
  resonators during vocalizations.

• The howler monkey (Alouatta caraya) (the male
  only!) has an enlarged and hollow hyoid bone that
  probably serves as an added resonance chamber for
  its vocalizations -- which can be heard almost
  than 5 km. in dense jungle.

• Elephant seals males have a proboscis that
  probably functions like the hand of the French
  horn player it is inserted into the seals
  gaping mouth during bellowing vocalizations.
  This probably serves to lower the resonances.

59
Howler monkeys
60
Adult female and juvenile
Two adult males fighting
Adult male vocalizing note that proboscis is
covering the mouth
http//cal-parks.ca.gov/default.asp?page_id1116
61
There is nothing unique about enhancements of the
vocal apparatus many species have them. They
serve to shape vocalizations to make them louder
(and possibly contain lower frequencies -- but
this needs further research).
They apparently serve in threat displays and
often supplement visual components of threats.
62
In various publications I have woven these ideas
into an ethological tapestry that attempts to
give a unified account of a wide range of human
vocalizations and agonistic displays ranging from
the acoustic character of sound symbolic
vocabulary, affective prosody, how questions vs.
declarations may be signaled using fundamental
frequency, the origin of the smile, and sexual
dimorphism of the human vocal tract.
I have only discussed a fraction of this but I
hope I have presented sufficient evidence that
the lowered larynx and enlarged pharynx of humans
(males) is not a plausible adaptation for speech.
It probably is an adaptation for threatening
vocal displays but this ability surely pre-dated
the acquisition of speech in human evolution.
63
Thank you!