Back to home page
Back to "Occasional Papers"
Back to Contents page
In many of my writings I have argued that poetic images
have no fixed predetermined meanings. In my 1992 book
What Makes Sound Patterns Expressive? -- The Poetic
Mode of Speech Perception (originally published in
1987) I propounded the view that speech sounds do not
have fixed predetermined symbolic values either. Poetic
images as well as speech sounds are clusters of features,
each of which may serve as ground for some combinational
potential. The resulting combinations of images and
speech sounds give rise to figurative meanings and
sound symbolism. Unforeseen contexts may actualise
unforeseen potentials of images and speech sounds.
Language users may shift attention from one potential
to another in the same speech sound or poetic image,
and realise new figurative meanings and sound-symbolic
qualities. Thus, the handling of figurative language
and sound symbolism in poetry is governed by a set
of homogeneous principles. The acquisition and use
of language require considerable creativity. This creativity
is heightened and turned to an aesthetic end in the
writing and understanding of poetry. In my writings
I have explored the sources of these potentials, and
how human intuition handles them in generating poetic
In this way, a sophisticated interplay between sound and meaning is generated. Relevant features can be multiplied indefinitely, and one may discover unexpected phonetic or phonological features. In my 2001 paper "Onomatopoeia: Cuckoo-Language and Tick-Tocking: The Constraints of Semiotic Systems" I consider a minimal pair that can illustrate this. In Hebrew, metaktek means "ticktocking"; we attend to the repeated voiceless plosives and perceive the word as onomatopoeic. metaktak, by contrast, means "sweetish", derived from matok (sweet). In Hebrew, the repetition of the last syllable is lexicalized, suggesting "somewhat (sweet)". A wide range of such "moderate" adjectives can be derived in this way from "main-entry" adjectives: , adamdam (reddish) from adom (red), yerakrak (greenish) from yarok (green), and so forth. Hebrew slang even derives gevarbar ("somewhat man") from (man). The meaning directs our attention to this redoubling of the syllable, and we attend away from the acoustic features of the specific consonants. Benjamin Hrushovski (1968; 1980) pointed out that the sibilants have different (even opposite) effects in "When to the sessions of sweet silent thought / I summon up remembrance of things past" and in "And the silken, sad uncertain rustling of each purple curtain". In my book I explore the different aspects of the sibilants that may generate such conflicting effects. In the former quote, meaning components related to "silent" activate one set of aspects; in the latter, meaning components related to "rustle" activate another set. This is what Wittgenstein (1976) called "aspect-switching".
I wish to point out an additional issue, crucial for an understanding of how sound symbolism works. The Haskins Laboratories researchers distinguish between a speech mode and a nonspeech mode of aural perception, which follow different paths in the neural system. In the nonspeech mode we listen to a stream of auditory information in which the shape of what is perceived is similar to the shape of the acoustic signal; in the speech mode we "attend away" from the acoustic signal to the combination of muscular acts that seem to have produced it; and from these elementary movements away to their joint purpose, the abstract phoneme sequence. In this mode, all the rich precategorial sensory information is shut out from awareness (listen in my 2001 article to the "ba-da-ga" series and the series of glides and whistles after Figure 2). In verbal communication it is the abstract phoneme that counts, not the precategorial sound stream or the articulatory gestures that led to the abstract category. There is, however, experimental literature that gives evidence that some of the rich precategorial sensory information is subliminally perceived. I have claimed that there is also a "poetic mode" of speech perception, in which emotional and perceptual qualities are generated when the precategorial auditory information is available for combination with meaning components. In the Hebrew word for "ticktocking" the meaning directs attention to the sensory information underlying the voiceless plosives; in the Hebrew word for "sweetish" it directs attention away, to an abstract lexical model.
The present paper was prompted by three chapters, by John Ohala, Eugene Morton, and Gérard Diffloth, in the mind-expanding collection of essays Sound Symbolism (Hinton, Nichols and Ohala eds., 1994). In the light of these essays I will recapitulate two issues from the chapter "Some Spatial and Tactile Metaphors for Sounds" of my above-mentioned book: the relation between sound frequency and the size of the body that produced it; and the relation between "high" and "low" vowels and the suggested size of their referents. The former may account for the rise of certain crucial potentials active in the latter.
Sound Symbolism and Source's Size
Sounds can be located along dimensions whose extremes are marked by spatial notions as low~high, thick~thin or space-related notions as heavy~light, and the like. These dimensions seem to be correlated in certain meaningful ways. There is plenty of anecdotal as well as carefully controlled experimental evidence that intuitions concerning the "spatial" as well as the "tactile qualities" of sound are fairly consistent from observer to observer, and sometimes even from culture to culture. Some such experiments have been reported by Roger Brown in his classic of psycholinguistics (Brown, 1968: 110-154). The whole chapter testifies to Brown's usually brilliant insights and subtle ways of analysis. Here, however, I am going to quote only two passages with which I disagree.
A concept like "boulder" is referred to rocks and stone and, in comparison, judged to be "heavy", "large", "thick", and "wide". These terms are directly applicable to boulders. However, boulders have no voices. Where, then, does the concept belong on the "bass-treble" or "loud-soft" scales? We cannot doubt the answer. If Disney were to give a boulder a voice it would be "bass" and "loud" in contrast to the piping of a pebble. This could be a mediated association: a boulder must have a bass voice because creatures that do have bass voices are usually heavy and boulders are heavy. It is not necessary to assume that there is any subtle inter-sensory quality found in boulders and bass voices.The cognitive approach to Man, of which Brown is one of the most outstanding exponents, tends to regard such explanations as "mediated associations" as the last resort of the scientist, where all structural explanations fail. Now, what seems to be wrong with the "mediated associations" theory is that it reverts to a rather strong version of associationist theory, assuming that people in various cultures have been uniformly conditioned by external conditions. It seems to be all too easy to invent some mediating story that appears to be pretty convincing, until one becomes aware of not less convincing counter-examples. Thus, for instance, red colour is felt to be "warm", whereas blue is felt to be "cold"; this feeling is not culture-dependent, and thus cannot be explained by cultural conditioning. Now there is a rather widely accepted explanation, that fire is red in all cultures whereas the blue sea is relatively cold in all cultures. However, the blue sky on a tropical (or even European) summer-noon is not exactly associated with cold. The sun, on the other hand, at its hottest, would be associated with gold rather than red, whereas red would be associated with the setting rather than with the shining sun.
Subjects in the study of Brown et al., felt that "thick" and "thin" simply do not apply to voices. However, "loud" and "resonant" do. Now thick people and animals and violin strings are usually loud and resonant. So, if the subject is required to guess, he will call the loud and resonant voice "thick". This need not be because the voice shares some inter-sensory quality with the visual or tactile apprehension of thickness. It could be because the voice is loud and creatures who have loud voices are usually thick, a mediated association (152-153).
Animals in competition for some resource attempt to intimidate their opponent by, among other things, trying to appear as large as possible (because the larger individuals would have an advantage if, as a last resort, the matter had to be settled by actual combat). Size (or apparent size) is primarily conveyed by visual means, e.g. erecting the hair or feathers and other appendages (ears, tail feathers, wings), so that the signaler subtends a larger angle in the receiver's visual field. There are many familiar examples of this: threatening dogs erect the hair on their backs and raise their ears and tails, cats arch their backs, birds extend their wings and fan out their tail feathers. [...] As Morton (1977) points out, however, the F0 of voice can also indirectly convey an impression of the size of the signaler, since F0, other things being equal, is inversely related to the mass of the vibrating membrane (vocal cords in mammals, syrinx in birds), which, in turn, is correlated with overall body mass. Also, the more massive the vibrating membrane, the more likely it is that secondary vibrations could arise, thus giving rise to an irregular or "rough" voice quality. To give the impression of being large and dangerous, then, an antagonist should produce a vocalization as rough and as low in F0 as possible. On the other hand, to seem small and non-threatening a vocalization which is tone-like and high in F0 is called for. [...]. Morton's (1977) analysis, then, has the advantage that it provides the same motivational basis for the form of these vocalizations as had previously been given to elements of visual displays, i.e. that they convey an impression of the size of the signaler. I will henceforth call this cross-species F0-function correlation "the frequency code" (Ohala, 1994: 330).
Voice frequency gives, then, information not about the mass of the body, but about the mass of the vibrating membrane which, in turn, may or may not be correlated with the mass of the body. A bass singer may be slim, it is his vocal chords that must be of a substantial size.
In another paper in the same book, Eugene Morton explores avian and mammalian sounds used in hostile or "friendly," appeasing contexts. He provides two tables in which sounds given by aggressive and appeasing birds and mammals are listed. "Aggressive animals utter low-pitched often harsh sounds, whose most general function is to increase the distance between sender and receiver. Appeasing animals use high-pitched, often tonal sounds, whose most general function is to decrease the distance or maintain close contact by reducing the fear or aggression in the receiver" (Morton, 1994: 350-353). Subsequently (353-356) he expounds a conception of sound--size symbolism in animals similar to the one quoted above from Ohala.
This conception may have far-reaching implications, beyond what is conspicuously suggested by Ohala and Morton. At the end of an important theoretical statement of research done at the Haskins Laboratories, Liberman (1970: 321) says: "One can reasonably expect to discover whether, in developing linguistic behavior, Nature has invented new physiological devices, or simply turned old ones to new ends". I will suggest that in some cases at least old cognitive and physiological devices are turned to linguistic, even aesthetic, ends. This seems to reflect Natures parsimony.
What is the relationship between being dangerous and having an irregular or "rough" voice quality; or between seeming non-threatening and a vocalization which is tone-like? To answer this question, one must realise that "noises" are irregular sounds, "tones" are regular, periodic sounds. Ohala and Morton mention this issue merely as a corollary of "deep" and "high" voices. But this aspect of nonhuman vocalisation may throw an interesting light on certain widespread intuitions in the poetic mode of speech perception, namely, that periodic consonants (e.g., [m], [n]) are perceived as soft, mellow, and appeasing, whereas aperiodic continuants (e.g., [s], [z]) as harsh, strident, turbulent, and the like. In fact, what I wrote about the poetic effects of periodic and aperiodic speech sounds may apply, mutatis mutandis, to this echological problem as well:
Periodic sounds have been described (May and Repp, 1982: 145) as "the recurrence of signal portions with similar structure", whereas aperiodic stimuli as having "randomly changing waveform", that "may have more idiosyncratic features to be remembered". The recurring signal portions with similar structures may arouse in the perceiver a relatively relaxed kind of attentiveness (there will be no surprises, one may expect the same waveform to recur). Thus, periodic sounds are experienced as smoothly flowing. The randomly changing waveforms of aperiodic sounds, with their "idiosyncratic features", are experienced as disorder, as a disruption of the "relaxed kind of attentiveness" (Tsur, 1992: 44).
In some circumstances unpredictability is a dangerous
thing. Sound gives information about physical changes
in one's environment. Randomly changing sounds give
information about unpredictable changes. So they force
one to be constantly on the alert. The survival purpose
of such alertness is conspicuous. Even in animal communication, however, an irregular or "rough" voice quality is sometimes "symbolic"; it constitutes no danger in itself, but has a common ingredient with dangerous circumstances: unpredictability. In the poetic mode
of speech perception, response to regular or randomly
changing waveforms is turned to an aesthetic end: it
assumes "purposiveness without purpose".
The foregoing conception may illuminate the motor theory of speech perception too, from an unexpected angle. This theory assumes that in the production as well as in the perception of speech we attend from the acoustic signal to the combination of muscular movements that produce it (even in the case of hand-painted spectrograms); and from these elementary movements to their joint purpose, the phoneme. The best approximation to the invariance of phonemes seems to be, according to Liberman et al. (1967: 43, and passim), by going back in the chain of articulatory events, beyond the shapes that underlie the locus of production, to the commands that produce the shapes. "There is typically a lack of correspondence between acoustic cue and perceived phoneme, and in all these cases it appears that perception mirrors articulation more closely than sound. [...] This supports the assumption that the listener uses the inconstant sound as a basis for finding his way back to the articulatory gesture that produced it and thence, as it were, to the speaker's intent" (Liberman et al., 1967: 453). If Ohala and Morton are right, this mechanism underlying speech perception is a less recent invention of evolution than might be thought. The lion's roar, for instance, follows a similar course. The F0 of voice can convey an impression of the size of the mass of the vibrating membrane and, indirectly, of the size of the signaler; in other words, the listener uses the inconstant sound as a basis for finding his way back to the articulatory organs and gestures that produced it and thence, as it were, to the roarer's intent.
Sound Symbolism and Referent's Size
The foregoing discussion has established a causal relationship as well as structural resemblance between the frequency and perceived size of sounds on the one hand, and the physical size of their vibrating source on the other. The association of small size with high frequency and of large size with low frequency becomes a "meaning potential" of sounds, which may be actualized in sound--referent relations too. In the chapter "Some Spatial and Tactile Metaphors
for Sounds" of my 1992 book I also discussed vowel
symbolism for size and distance. Among others, I quoted
Ultan (1978) who, by examining a total of 136 languages,
tested the hypothesis that diminutive sound symbolism
is associated with marked phonological features (high
and/or front vowels and palatal or fronted consonants).
He found that diminutive is most often symbolized by
high or high front vowels, high tone, or various kinds
of consonantal ablaut. Proximal distance is symbolized
overwhelmingly by front or high vowels. To take a language
not included in Ultan's sample, my native Hungarian,
itt means "here", ott means "there",
ez means "this", az means "that".
Így means "in this fashion", úgy
means "in that fashion"; ilyen means "of
this kind", olyan "of that kind", and
so forth. "Since high front vowels reflect proportionately
higher second formant frequencies, and the higher the
tone the higher the natural frequency, there appears
a correspondence between a feature of high frequency
(= short wavelength in physical terms) and the category
of small size" (Ultan, 1978: 545). Likewise, for
the same reasons, the received view is that in Western
languages /i/ is small and /a/ is big.2
In a mind-expanding paper on the word class of "expressives"3 in Bahnar, a Mon-Khmer language of Vietnam, Gérard Diffloth claims that in this word class /i/ signifies "big", and /a/ "small". This throws my foregoing argument into an exciting perspective. At first sight the paper provides outright refutation of one of my pet beliefs; but in the final resort it lends massive support to my wider conceptions, that speech events (speech sounds and articulatory gestures) do have certain (sometimes conflicting) combinational potentials, which may be activated, after the event, by certain meaning components. Diffloth points out the following relationships between referent size and vowel height in Bahnar:4
Examples ("D. red." = "Descriptive reduplication"):
In both the two-way and three-way division "the iconic values of the vowels are, roughly speaking: High = Big and Low = Small, exactly opposite to the English /i/ = Small and /a/ = Big, claimed to be universal. There is nothing peculiar about this Bahnar system, and one can easily find an iconic basis for it. In the articulation of high vowels, the tongue occupies a much larger volume in the mouth than it does for low vowels. The proprioceptive sensation due to this, reinforced by the amount of contact between the sides of the tongue and the upper molars, is available to all speakers and is probably necessary to achieve a precise articulatory gesture. [...] In this perspective, two different languages may easily use the same phonetic variable (vowel height) to convey the same range of sensations (size), and come up with exactly opposite solutions, both being equally iconic; all they need to do is focus upon different parts of the rich sensation package provided by articulatory gestures, in our case the volume of the tongue instead of the size of the air passage between it and the palate". Now consider such pairs of English synonyms as big and large, or small and little one member of which contains a high vowel, the other a low one. One may account for their coexistence in one language in one of two ways: either by assuming that the relationship between sound and meaning is arbitrary, or by assuming that speakers and listeners intuitively focus upon different parts of "the rich sensation package" provided by either the articulatory gestures or the speech signal in pronouncing these words. Shifting attention from one part of "the rich sensation package" to another is what Wittgenstein called "aspect-switching", prompted by the meanings of the words.
There are two conspicuous common features in Diffloth's corpus and my foregoing examples from Hungarian. First, the sound--meaning relationship, if present, does not take the shape of a statistical tendency in a huge aggregate of isolated words; it is displayed by minimal pairs of straightforward antonyms. Second, phonetically, these pairs are opposed in only one pair of vowels; semantically, too, they are contrasted in one feature. All the rest is really equal. In other words, size--sound symbolism is formally lexicalised. This lexical feature reflects creative phonetic intuitions in the distant past which have fossilised by now; the present-day language-user may attend away from the sound symbolism of "high" and "low". So, these pairs of words are structurally different from such clusters of synonyms and antonyms as big and large, or small and little. The two systems, however, are opposed in one interesting feature. In Hungarian there is vowel harmony. Consider the pair ilyen and olyan The size-symbolic contrast is carried by the /i~o/ opposition; but this affects the relative height of the second vowel too. In Diffloth's examples from Bahnar, by contrast, the other vowels may vary independently.
I have a vested theoretical interest that Diffloth's explanation should be valid. It would reinforce my conception according to which sound symbolism is part of a complex event, comprising meanings, articulatory gestures, sound waves, etc. Each one of these components has an indefinite number of features, which give rise to a multiplicity of sometimes conflicting combinational potentials. Strong intuitions concerning sound symbolism are generated by selecting a subset of available features on the semantic, acoustic, and articulatory levels. When conflicting intuitions are reported, attention is shifted from one subset to another.
When, however, I tried to pronounce the speech sounds
which Diffloth designates "high", I noticed
that his description suits [i] extremely well;
but not [u].5 In view of the examples he provides, whatever
explanation suits [i] should suit [u] too.
Figure 1 The acoustic and articulatory location of the synthetic vowels,
plotted according to the frequency positions of the first and second formants.
Back to home page
Back to "Occasional Papers"
Back to Contents page
This page was created using TextToHTML. TextToHTML is a free software for Macintosh and is (c) 1995,1996 by Kris Coppieters