Many forms of alternative communication utilise non-speech sounds to convey information. From the simple but effective ship's fog-horn, to the intricate drum and whistle languages of Africa and beyond, alternative communication appears in many forms.

All these methods of passing on information share one important property; namely, the use of sound to communicate. By looking at the different approaches taken by these alternative forms of communication, it is hoped that a useful insight might be gained into how information can be transmitted using non-speech audio.

Calls Requiring Horns

The simplest call which makes use of a horn is the popular ship's foghorn. Blasts on such horns are made on ships to inform other traffic of the presence of the ship when the usual visual cues are not available. As a communicative system, it is extremely primitive as the occurence of the call broadcast only two pieces of information; the presence of the ship and its direction.

More elaborate systems using horns have been developed, most notably for use by the military. Battle calls dating from the sixteenth century stand alongside the fanfares and bugle calls scattered across military history (for example the American Civil War Bugle Calls, Garbeck 1999). The purpose of the later bugle calls was to signal commands to a potentially large body of men either on camp or in the field. Many common traits occur within a large number of these calls. Examples include rising arpeggio patterns and rapidly repeating initial notes. In some ways, this is hardly surprising as the number of unique pitches used by the calls is usually restricted to four. Such a restriction is bound to result in recurring patterns. However, there are some repeated similarities across calls where the messages share component parts.

There is considerable variation in the duration of each call. Some last a mere four to eight bars, other for thirty-two/sixty-four, etc. It seems that calls used to assemble a large body of men urgently or for some important occasion are more prolonged than those used for a more trivial annoucement. This highlights the fact that the calls' melodies are chosen to firstly draw attention to the call (reflected in the length of the call). The actual meaning of the call is then determined from learnt associations aided by context. Therefore, most calls are made as distinct as possible, with many of the similarities purely being coincidental.

Calls have also been put to use by huntsmen since the Middle Ages (Carr 1976). As the men were often separated during the hunt, a series of signals were employed to maintain contact through dense woodland. The original type of horns used to blow the calls were capable of producing only one pitch of note, so different rhythms have been used over the centuries to signal the presence of a fox in a field or to call the hunting party together, etc. Once more, a limited signal system shows repetition across calls without an underlying shared structure.

Calls Requiring Whistling

The Boatswain's Call is a whistle that is still used by the Royal Navy to issue orders to a ship's crew and to welcome a visiting dignitry onboard. It is a high pitched call which is ideally suited to piercing through the noise of a dangerous sea-storm. By using three different tones and two main pitches (with noticeable glides between), a number of orders such as hoist sails, stand still or away boat, can be made.

The format of the calls is similar to those of the bugle calls discussed above. Nine commands are represented by nine calls that last from between two and thirty seconds. The pitch and tone of each call either mimics the essential nature of the command (e.g. stand still is a single held pitch), or is a fairly arbitrary choice. Therefore, a simple mapping is defined that must be learnt by rote. There is no scope for new commands or variations on old commands unless a new call is designed and added to the list of current calls; thus, the call system is a closed communication system.

Caving is a pastime where communicating with a whistle is sometimes necessary when voices cannot be heard (for example near waterfalls) (Australian Speleological Federation Inc, 1990). The recommended signals are extremely basic and involve a mere five messages:

One blastStop
Two blastsUp (I want to come up)
Three blastsDown
Four blastsOK (Affirmative and safe)
One long blastHelp

As can be seen, the associations between the signal and the meaning are completely arbitrary. Consequently, a fair amount of confusion could arise if the caller and/or listener were unsure of the number and type of blasts required.

Sheepdogs are trained to recognise about six basic commands when rounding up sheep (Ryder 1983). These are normally spoken, but can be whistled over long distances. The common commands are stop (steady or that'll do), lie down, walk on (where the dog moves away from the shepherd), go left (usually known as come away), go right (often come by meaning go past the sheep) and come on (to bring the sheep towards the shepherd).

The exact commands used depend on the shepherd and the number of dogs in use. On large hillsides, it is common to see two dogs working as a pair, where the commands are specific to each dog.

The form of the whistled calls is unknown. It appears that most dog training has been passed on by word of mouth and therefore very little literature exists describing how to whistle each command.

The Use of Bells

Bells have been used to transmit messages for centuries. Even in the time of the Romans, bells were rung to inform locals that the bath house had opened (Coleman 1971; Camp1988). This idea of signalling events continued with church bells which not only called parishioners to different services, but also marked different parts of the service. Different messages were generally inferred by the context in which the bell was heard. For example, a peal of bells at midnight on New Year's Eve would be recognised as signalling the death of the old year.

A simple coded message would often be hidden in the bells ringing. When the death knell was rung, the number of rings would indicate whether a man, woman or child had died. One set of three rings for a child, two sets of three for a woman and three threes for a man. Alternatively, a large bell may be rung for adults, a small one for children. Three rings for a male, two for a female.

The normal ringing of church bells would be a descending scale down the set (ring) of bells. When the bells were rung backward, it signified an alarm (such as a fire or an uprising).

The use of bells is not limited to church belfries. Bells are often used at sea to mark dangerous rocks or on board ships to issue commands and warnings to hands on deck. Bells are also found in large households where particular servants were (and sometimes still are) called to certain rooms to perform some service.

Drum Languages

Music signalling is used in Africa, Middle/South America, Europe, South East Asia and Oceania and often involves drums (Adler 1979b; Sebeok 1976). The traditional drumming found in Africa is actually of three different types. Firstly, a rhythm can represent an idea (or signal). Secondly it can repeat the profile of a spoken utterance or thirdly it can simply be subject to musical laws.

Drum communication methods (along with whistle languages discussed below) are not languages in their own right, they are often merely based on actual natural languages (CHAINS 1997). The sounds produced are idiophones. The messages are normally very stereotyped and context-dependent. They lack the ability to form new combinations and expressions.

In central and east Africa, drum patterns represent the stresses, syllable lengths and relative pitch and tone of the particular African language (Adler 1979b). In tone languages where syllables are associated with a certain tone, some words are only distinguished by their tonal pattern. Therefore, syllable drum languages can often communicate a message using the tonal phonemes alone.

In certain languages, the pitch of each syllable is uniquely determined in relation to each adjacent syllable. In these cases, messages can be transmitted as rapid beats at the same speed as speech as the rhythm and melody both match the equivalent spoken utterance.

Misinterpretations can occur due to the highly ambiguous nature of the communication. This is reduced by context effects and the use of stock phrases. For example in Jabo, most stems are monosyllabic. By using a proverb or honorary title to create expanded versions of an animal, person's name or object, the corresponding single beat can be replaced with a rhythmic and melodic motif representing the subject.

Some tribes such as the Melanesians extend this idea further by freely inventing signs to make up their drum signals. This is in sharp contrast to the Efik tribe of Nigeria who use notes which exactly correspond to the tones of their morphemes. Different still is the Ewe language found in Togo, where only full sentences and their combinations are translated into the drum language. No smaller units are used; a sound picture represents a whole thought. This is similar to the Tangu tribe of New Guinea, where signals represent phrases, the mnemonics of which are parts of song melodies, quasi-poetic rhythms or purely personal rhythms.

As can be seen, when a drum is used in speech mode, it is culturally defined and depends on the linguistic/cultural boundaries. Therefore, communication suffers from translation problems as in vocal communication. There is no international drum language.

Whistle Languages

In modern Western society, whistling communicates emotions but not information. Informational communication through whistling can and does occur however, (for example, between people in deep valleys surrounded by mountains) and is found in the Americas, Africa, Asia and New Guinea (Adler 1979c).

Most of these several hundred whistle languages are based on tonal languages. Only the tonal contour of speech is conserved in the whistle (along with duration and stress). All other sound variations of speech such as articulation and phonation, are eliminated. These are replaced by prosodic features, e.g. rhythmical and stress variations.

Other whistle languages include articulation so that consonants interrupt the flow of the whistle. The people from Aas in the Pyrenees speak a Spanish-derived dialect which used to be adapted into a whistle language in a similar way to the Silbo Gomera whistle language used by shepherds high up in the mountains in the Canary Island of La Gomera (Classe 1957). Both languages share common traits: the Spanish linguistic framework, the method of whistling, the similar form of signals and the functional purpose of the signalling.

The Silbo works by whistling whilst articulating as normally as possible as in speech. Due to the nature of the Spanish dialect found on La Gomera, the resulting pitch contour is sufficiently recognisable for non-native Spanish speakers to pick up the Silbo within a matter of months. As the whistling on La Gomera preserves the structure of the spoken dialect, it is powerful enough to allow whistlers to converse about any subject.

Similarly in Turkey in the village of Kusköy, people use a whistle-version of Turkish to communicate across mountains. Many other communities appear to have such communication methods; Indian tribes in Mexico, tribes in New Guinea, Nepal, Burma, West Africa, etc.

Whistle languages have evolved in situations where speech is impractical. As they rely on the nature of the language they emulate, they are not suitable for all languages. However, they are superior to drum languages as they function at the same rate as speech.

Implications of Communicative Systems

All of the alternative forms of communication discussed above fall into three classes of communicative system. The first class comprises calls which transmit a message simply by the presence of the call, whilst communication involving a small finite number of pre-defined messages constitutes the second class. The most expressive systems are those of the third type; a potentially infinite number of different messages can be produced. Moreover, novel messages that conform to the system should be understandable by anyone who understands the system. The first two types of system are closed communicative systems; the third is open.

It is impossible to form an open communicative system without an underlying structural approach to message formation. Simply assigning arbitrary mappings between acoustic signals and messages will result in a closed system. Any new messages will have to be learnt to be understood. This is reflected in many of the call systems.

There is a practical limit to the number of distinct meaningful elements a communicative system can have. The more meaningful elements there are, the more similar they will become. Consequently, the less discriminable they will be. Hockett argues that for a highly complicated system duality of patterning is a pre-requisite. In other words, the only way to support many more meanings is to allow smaller units to combine in different ways, thereby representing different meanings. This can be taken to extremes by the structuralist view, which would argue that meaning can be discussed solely in terms of structure. Words are understood by their place within a system of other words that oppose them. The actual words themselves are completely arbitrary. Meaning arises from a positional value; the sense of a word is defined by its place in a system of relationships that link it with other words (Holtzman 1996).

Language itself began as a closed system. It is only through blending that language became an open system (Hockett). When two distinct calls of a closed system are merged by accident, a new call results from the combination of parts of the original calls (e.g. a cross between enormous and gigantic results in ginormous); this is known as blending. One of the side-effects of blending is that smaller components of the two blended sounds take on an independent structural role which can combine with components of other words. This can be seen most clearly with the many prefixes and suffixes found in English.

To gain the expressive power of an open system, a methodology for combining units is required. Once the method of combination has been defined, the set of possible structures which can be generated by such a system can be determined. Generative grammars are a useful way of capturing structural constraints given a set of basic units and form the topic of the following section.

These pages originally belonged to John Hankinson, but are now maintained by Alistair Edwards (

21st February 2002