Phonosemantics is a portmanteau word which suggests the meaning that comes from sounds. The idea that individual vocal sounds have meaning is not a new idea. You can find precedents in the Upaniṣads and in Plato. However it is not a popular idea amongst linguists who deny the possibility of such a relationship.
Try this. Get a dictionary of English and chose a consonant - but be aware that this experiment works on sounds so the hard /c/ and /k/ are the same sound. To eliminate the confusion I will refer to a phoneme. Phonemes are indicated by being placed between slashes. So c can be /c/ or /k/; and j or g can be either /j/ or /g/; but k is always /k/. Take each word which does not have a prefix or suffix, and put it into as many categories of meaning as you need to cover the basic senses. Words typically have more than one meaning. Work through all the words that start with that phoneme. At the end of the process most of the words will have fitted into a small number of reasonably well defined categories. Those that do not fit a category are typically concrete nouns - they are names for specific things. In /d/ daffodil is just a daffodil for instance, and has no other referent. This should take 60-90 miuntes. Repeat for another sound and compare.
I can summarise the results you will find because I have done this experiment as suggested by Margo Magnus the person who first performed it as part of her doctoral research at Trondhiem University, Norway.
- Firstly a very high percentage of words beginning with any phoneme, approaching 100%, which are not concrete nouns will fit into one of about a 10-20 broad categories of meaning. We call this effect clustering.
- Secondly comparing two different initial phonemes you will find only limited cross over of categories, and words beginning with one phoneme will not fit well into the categories associated with another phoneme.
- Thirdly more subtle patterns emerge when comparing similar phonemes such as /p/ an /b/ which are respectively the unvoiced and voiced bilabial stops. Taking a catergory - such as impacts - words beginning with different phonemes seem to emphasise different aspects of the referant. Impacts with /b/ are broad and blunt, stay on the surface; where /p/ words indicate precision, pin-points, and the impact with puncture the surface. This effect of the phoneme on the symbolism underlying a word we call iconism.
If you work through the whole dictionary you will find the that, at least for the consonants, that this pattern holds true for all the phonemes in English. Vowels appear to behave in a different manner.
Why should a pattern like this emerge. I know of three explanations.
- 1. Bounded Chance
- 2. verbal roots
- 3. Sound symbolism
Chance is a slight misnomer. Contemporary linguists hold that the relationship between sounds and what they signify is arbitrary. If this were so it would predict a random distribution of meanings and sounds. As the simple experiment above shows there is not a random distribution. Linguists solve this problem by invoking social convention. Words that mean the same are given similar sounds because this has become conventional. The relationship is still fundamentally arbitrary because the sound itself carries no universal meaning, but it does have a relative value in each language. However this is a weak answer to the pervasive phenomenon of sound symbolism.
One way to disprove this hypothesis would be to demonstrate that clustering goes across languages. So far as I know a conclusive demonstration of this is wanting, but there are some results which are suggestive. For instance people have a better than random chance of guessing the meaning of a unfamiliar word in a language they do not speak. Also some phonemes, such as the compound /str/ do appear to cluster in a similar way across languages from different language families - the clustering in one language matches the clustering in another unrelated language for /str/.
As far as I know the scholarly literature has not considered the idea that notional verbal roots underpin clustering. It occured to me when studying Pāli and becoming familiar with Sanskrit words that verbal roots give rise to a range of words all starting (on the whole) with the same phoneme. These variations fill up dictionaries. However verbal roots are notional, which is to say that they were arrived at by considering similarities between words and working back to theoretical root entities - verbal roots do not exist in the wild.
It may be that verbal roots in fact offer a confirmation of a relationship which clustering is a manifestation of.
The theory says that there is relationship between phonemes and meaning. That relationship is non-arbitrary, but also non-linear. We say that there is a degree of "motivation" behind the choice of phonemes in making a new word. A strong form of this theory would say that speakers have little freedom in choosing; while a weak version would say that speakers have much choice. But any theory of sound symbolism suggests that says that the sounds that make up words, the sounds themselves, carry information. One very simple experiment, now known from it's result as "the bouba/kiki effect" shows that sounds can and do act as metaphors for forms in normal people.
The Bouba/Kiki effect
If you look at the image on the right you will see two letters from the "Martian" alphabet. One of the letters is called "bouba", and one "kiki". Which one is which?
Irrespective of the language they speak around 95% - 98% of people will say that kiki is on the left (the spikey shape), and bouba is on the right (the rounded shape). This suggests that some sounds do indeed have an inherent symbolic "meaning". That meaning is metaphoric and fuzzy rather than literal and linear. The effect was first noted in 1929, but more recent studies have confirmed it. Interestingly children of 2.5 years, too young to read, also make this distinction.
Conventional linguistic theories cannot account for this result. If the relationship between sound and meaning is arbitrary then there is only a 50% chance of any two disparate people choosing the same label for each shape. However experimental results show that across language groups it is extremely unlikely that individuals will make a different choice.
The mechanism of sound symbolism is as yet unknown. However one suggestion is that verbal gesture - the use of the lips and tongue - is an aspect of it. Margo Magnus for instance suggests that a large portion of words beginning with /b/ come under the headings "barriers, bulges, and bursting" because of the action of lips when creating the /b/ sound: viz they come together to form a barrier to the flow of air, resulting in a build up of pressure (and a bulge), and finally a burst of air and sound. Plato suggested that /r/ connoted movement and activity.
Another promising area is research into synaesthesia. Someone with synaesthesia interprets sensory experience in one mode via the language of another: they see music sounds as colours for instance, or experience words as taste sensations. Vilayanur S. Ramachandran has shown that not only are synaesthetes not faking it, but that there is a possible neurologicla explanation. His work is promising because it demonstrates how sounds might function as metaphorical images for us. Ramachandran thinks that cross activation between different areas of the brain explain not only synaesthesia, but the whole development of metaphorical thought, and language itself. In particular areas of the brain which process aural information are next door to those which process visual information. Cross linking between them gives rise, accoridng to Ramachandran, sight-sound synaesthesia. Some cross linking is normal and it is this that allows us to perceive for instance sounds as dull or bright; or colours as loud or muted.
Problems with the Phonosemantic Hypothesis
The many languages/many words problem
The key objection from the mainstream is this: that different languages use different words for the same thing, therefore the relationship must be arbitrary. However this is a weak argument that involves a number of assumptions which are not legitimate. Listing the assumptions is useful.
The first assmuption is that a single word refers to a single 'thing', and that there is a one to one correspondance between referants between languages. So for instance a "dog" is a single thing, and the word for that thing in French, say, chien, is a word for that same thing. This is an extremely naive assumption as anyone who has ever tried to translate from one language to another will know. Many words do not translate easily, and even when they do they may have both subtle and gross differences. Language is more frequently marked by ambiguity or multiplicity of reference (polysemy), than by singularity.
We might also ask whether a dog really is a simple phenomena? Well, obviously it is not. A dog has a general body plan and physical characteristics, and a set of behaviours which mark it out as a dog. The single word dog can take in chihuahua and great-dane; poodle and bloodhound. We know it is a dog because we are able to interact with it through our senses in a way which conforms to our previous experience of other dogs. Some dogs are closer to our mental picture of a dog, and some are much further away. Given that the dog is complex, and our relationship with it, would it then surprise us to find that there are many English words for a dog? Not really. So why would it surprise us to find that other languages, which often involve quite different patterns of thinking, use different words. If Swedes say "hund", might they not relating to the dog in a similar way to us when we choose to call a dog a hound?
Experience is not simple and linear. Experience is complex and non-linear. It is not just processing facts and churning out actions. Linguistics has a behaviouralist model of the relationship of people to their world as an unstated assumption. That assumption demonstrably is flawed and misleading.
The foundations of a new theory of the relationship may well be found in the work of George Lakoff and others working in the area of "linguistic categorization". Lakoff acknowledges the complexity of experience, the possibility of polysemy, and the importance of relationship in definition and the making of meaning. Lakoff emphasises the role of metaphor and metonym (where a part stands in for the whole) in creating meaning.
Comparing languages with radically different sound pallets
There is an unsolved problem with the phonosemantic hypothesis which is what happens when you compare two languages with very different sound pallets: Māori has 10 consonants, while Sanskrit has 34. Māori has at least one sound, /f/, that Sanskrit lacks. So far as I know there has been no research on this aspect of sound symbolism.
The vowels do not have the same kind of symbolic associations. The initial phoneme /o/ shows the clustering, but the various sounds associated with the English vowel 'a' do not seem to. Research on the symbolism of vowels in words has shown a fairly consistent effect in regard to size - stressed high vowels (e, i) sound smaller, that rounded back vowels (o, u). Think minimum, medium, maximum (i,e,a). Also in sequences back vowels are later than front vowels: sang, sung. However no one has researched this in the comprehensive way that Magnus has done for the consonant sounds.
Applications to Mantra
A thorough application of the principles of phonosemantics to mantra has yet to be performed. But we can at least look at some general applications.
One of the features of aural experience is that we often take in and process sounds unconsciously. It is well known, for instance, that music evokes emotions without us making a conscious effort to "understand" the music. There is a good evolutionary bilogical explanation. Reacting to sounds has a positive survival value: we hear in 3 dimensions, and can typically pin-point the direction of a sound to about 1° of arc. This we hear some threat approaching and are launhced into action often before we can think. Our aural systems is fine tuned for this survival: we have emotional responses to sounds. More than this however we are a social species and many of our interactions are conducted in the medium of vocal sounds. Anecdotally we say that only a very small percentage of the information in a conversation is down to the words we use. The bulk of what we are communicating is non-verbal. Included in this are factors such as intonation, and phrasing: features of language known collectively as prosody. Prosody gives us clues as to the attitudes and emotions of another person. It is prosody that allows us to completely change the meaning of a phrase through inflection (technically known as conversation implicature). I could say "I like your new hair cut" and convey many different messages by varying prosodic and gestural elements, even making it mean the complete opposite.
As a social species this kind of non-verbal information is vital to the healthy functioning of our group, and therefore once again we are highly attuned to it. We can tell simply from tone of voice for instance, if someone is angry with us, or genuinely pleased to see us. So on this level also we are atuned to sounds.
If we accept that vocal sounds are also gestures (in the broadest sense), or at least can symbolise a gesture, then we accept that a verbal sound may communicate information from other modes. This in a way is obvious, but I am drawing attention to it in the light of sound symbolism. If sounds are no arbitrary, then our choice of sounds conveys information. What I mean is that we convey information with a word, but that the choice of that word in particular, with those sounds, is a layer of non-verbal information.
Sound symbolism posits a symbolic link between sounds and meaning. The link is metaphoric or poetic even. Upon hearing a vocal sound we do not immediately leap to a concept. We may associate it with an image, or series of images; or we may simply respond emotionally. Linguistics on the whole tries to set aside this aspect of language, and to focus on trying to refine the meaning of words, or the structure of grammar and syntax. Anything which has no precise meaning, or which does not convey a concept is often dismissed out of hand. Linguists working within this paradigm come to conclusions which are a variation on "mantras are meaningless". The exemplar of this is Fritz Staal who uses the tools of Straucturalism and Semantics to try to understand ritual and mantra. Ritual according to Staal is a set of rules with no meaning, and mantra is not a form of language but a relic of a time before language, a persistance of pre-human animal sounds into the human arena. Neither of these theories are very attractive, nor do they offer much in the way of insight into the effects of mantra, or why they are such potent symbols. Staal is a more sophisticated version of the Victorian scholars of Buddhism who frequently concluded that if they did not understand something, then it was incomprehensible to all.
However linguists working within the paradigm of Pragmatics often find more meaning in mantras. They see for instance poetic devices such as repetition, alliteration, and rhyming in some mantras. Despite the lack of proper words, a mantra can be using the effects of language to create an image.
Phonosemantics goes even further. It says that vocal sounds, even single syllables, have meaning. That meaning is imprecise and metaphoric. If we accept the phonosemantic hypothesis then we need not conclude that the sounds of mantras have no meaning. Mantras can be meaningful in the sense that phonemes communicate in images and feelings. Compare this idea with research that shows that the inarticulate noises we make during conversations - hmm, uh, uhuh, mmm, etc - are packed with meaning. (see for instance: http://www.nigelward.com/egrunts/
My understanding of mantra in Tantric Buddhism is that its purpose is to give us an experience. That experience should help us to transform ourselves in line with the Buddhist teachings. We are highly attuned to Sounds, and vocal sounds in particular, and they are able to produce a physical/emotional response in us without involving conscious cognition. Sounds can indeed evoke an involuntarily response in us, that is, they can give us an experience. What is communicated is non-verbal and non-linear - it is metaphorical and symbolic. What's more by choosing specific syllables, that have built up and possibily even intrinsic associations, we may be able to direct that experience. Anecdotally practitioners do often report that different mantras have a different "feel" to them.
A lot of what I am saying here is speculative, and based on an admittedly limited understanding of linguistics, but I think it is plausible. In this essay I have not appealed to any mystical entity or process. This is not metaphysics, although a metaphysical explanantion may be more satisfying to some people. Personally I want to understand the process to better apply it. If I know how the average person might respond to a given combination of sounds, for instance, then it will be possible to create new mantras designed to evoke a particular kind of experience. If however mantras are a kind of devine revelation that we mustn't tamper with or investigate then we are stuck. Perhaps Sanskrit mantras do not work so well for Westerners? What if one or other of the sounds in English which are not in Sanskrit (/f/ for instance) might work better for me? What if, having grown up with no retroflex consonants, they merely confuse my aural processing facility?
A lot of work remains to be done, and there are few people to do it. If you want to follow up this thread then the place to start is Margaret Magnus's website: The Magic Letter Page. There is a lot of info here, and links to other material.
I also recommend:
- Lakoff, G. and Johnson, M. 1981 Metaphors we live by Chicago University Press. (a very interesting introduction to the pervasiveness of metaphor in communication and language).
- The 2003 BBC Rieth Lectures by V.S. Ramachandran included a talk, Purple numbers and sharp cheese, on his ideas about how synaesthesia might provide an account of metaphor more generally, but also insights into the development of language.
A growing number of resources are appearing on the web...