Anais da Academia Brasileira de Ciências (2004) 76(2): 405-412
(Annals of the Brazilian Academy of Sciences)
Bioacoustics of human whistled languages: an alternative approach
to the cognitive processes of language
Laboratoire de Dynamique du Langage (DDL)-CNRS, Institut des Sciences de l’Homme (ISH)
14 Avenue Berthelot, 69363 Lyon Cedex 07, France
Manuscript received on January 15, 2004; accepted for publication on February 5, 2004.
Whistled languages are a valuable heritage of human culture. This paper gives a ﬁrst survey about a new
multidisciplinary approach to these languages. Previous studies on whistled equivalents of languages have
already documented that they can provide signiﬁcant information about the role of rhythm and melody in
language. To substantiate this, most whistles are represented by modulations of frequency, centered around
2000 Hz (±1000 Hz) and often reach a loudness of about 130 dB (measured at 1m from the source). Their
transmission range can reach up to 10 km (as veriﬁed in La Gomera, Canary Island), and the messages can
remain understandable, even if the signal is deteriorated. In some cultures the use of whistled language
is associated with some ‘‘talking musical instruments’’ (e.g. ﬂutes, guitars, harps, gongs, drums, khens).
Finally, whistles as a means of conveying information have some analogues in the animal kingdom (e.g. some
birds, cetaceans, primates), providing opportunities to compare the acoustic characteristics of the respective
signals. With such properties as a reference, the project reported here has two major tasks: to further
elucidate the many facets of whistled language and, above all, help to immediately stop the process of its
Key words: human whistled languages, whistle communication, cognitive processes, cultural universals.
Besides spoken language, some populations in dif-
ferent parts of the world use a complementary sys-
tem of vocal communication, which is based on
modulated whistles and thus called ‘whistled lan-
guage’. Whistled languages can be regarded as a
transposition of a given local language into a reper-
toire of whistles. Almost any language (non-tonal,
tonal or accent-pitch) could be whistled and nearly
anything can be expressed this way. Such adapta-
tions have been used by various cultures, mainly in
response to a speciﬁc ecological situation: a cer-
tain isolation of individuals in their everyday activ-
ities. Therefore they are mostly used in mountains
or dense forests.
Reports about whistled languages were docu-
mented since the treaty of the Tao in Asia (6
tury B.C.) and since the 17
century in the island
of La Gomera (Canary Islands). First studies con-
cerned mainly anthropological aspects(Quedenfeldt
1887, Lajard 1891, Labouret 1923, Eboué 1935),
whereas later investigations also included linguis-
tic (Cowan 1948, Classe 1956) and acoustical is-
sues (Busnel 1966, 1970b). Today, twelve whistled
languages have been partially described and stud-
ied linguistically or bioacoustically (see Sebeok and
An Acad Bras Cienc (2004) 76 (2)
406 JULIEN MEYER
Umiker-Sebeok 1976). In addition, as many as sixty
other languages are suspected to still have a whistled
equivalent, but these have not been studied yet.
GEOGRAPHICAL AND CULTURAL DISTRIBUTION
The spatial distribution of whistled languages is as-
sociated with special conditions of human commu-
nication, particularly (a) long distance between peo-
ple living in places where rugged topography sepa-
rates them in terms of travel times, even when they
may be in visual contact (which results in a cer-
tain isolation of individuals), or (b) local secrecy in
speaking about the environment (e.g. for hunting
or ﬁshing) or about others (e.g. in terms of love,
religious, political or social matters).
The twelve whistled languages that currently
have been described and studied in linguistic or
bioacoustic approaches can be listed in Table I.
Many other examples have been described in
articles either by travelers, conquistadores, anthro-
pologists or ethnomusicologists. Southern China,
Papua New Guinea, the Amazon forest, Sub-
Saharan Africa and Mexico encompass most of
these languages. They are generally of everyday
use but are often endangered.
The whistled languages have a linguistic extent that
reﬂects totally the basic knowledge of the population
in the vocalspoken language. Thelinguistic range of
these languages is therefore not physically limited,
but only culturally, just like for spoken languages.
Through the signal of the whistled languages, dif-
ferent dialects from a village to another can often be
recognized. For example, in the Mazateco culture
of Oaxaca, Mexico, 14 different variations of the
Mazateco language exist in the highlands, where we
were able to record four of them in whistled versions
(Tenango, Eloxochitlan de Flores Magon, Huautla
and San Mateo). They reﬂect the differences among
vocabularies resulting from various contacts with
other languages and the differences of evolution in
The whistles mediate many different messages: the
gender of the whistler, his/her age (as we could ver-
ify it by a perception test in the Mazateco language),
his/her mood (to a certain extent) and, in most of the
cases, the identity of the whistler, i.e. the signature
of his/her voice.
Link with Talking Musical Instruments (TMI)
In locations where whistled languages are still very
developed like in South China, they occur often
tightly linked to talking musical instruments [ﬂutes,
local guitars, harps, drums (Fig. 1), gongs, khen
(small and tall)]. The whistles as well as the in-
struments are used indifferently for speaking or for
playing music in bands. They have a key position
in every popular ceremony because the messages
they convey represent the living memory of their
oral culture. For example when somebody has died
or is born, they are used to tell the life of the dead
or of the ancestors of the newborn to other villages.
Therefore they are one of the main means of cultural
transmission (Xian-Ming 2002).
PHYSICAL CHARACTERISTICS OF THE SIGNAL
The whistle is most of the time either practiced with
one ﬁnger in order to reach high levels of power for
long distance speaking (Fig. 2), or directly with the
lips for the short distance discussions (Fig. 3) and
sometimes with a leaf. The whistles are represented
by modulations of frequency, centered around
2000± 1000 Hz (Fig. 4). At a distance of one meter
from the mouth of the whistler, they can reach an
amplitude of 130 dB. Variations of amplitude levels
follow approximately those of frequency. This indi-
cates that the whistler has to increase the air pressure
to raise the pitch of the whistle.
In general, there is a concentration of the infor-
mation within a relatively short band of frequency
compared to spoken language. This explains both
An Acad Bras Cienc (2004) 76 (2)
HUMAN WHISTLED LANGUAGES 407
Whistled languages Authors
Silbo, on the small island of Classe (1956, 1957, 1963), Busnel and Classe (1976),
La Gomera (Canary Islands)
Busnel and Siegfried (1990), Trujillo (1978),
Rialland and Gautheron (2000)
Aas, in the French Pyrénées Busnel et al. (1962a, b)
Kusköy, in Turkey Busnel (1970a, b), Lenneberg (1970), Leroy (1970),
with M. Moles and B. Vallancien
Evia, in Greece Charalambakis (1993)
Abu’-Wam, in Papua New Guinea Nekitel (1992)
Tepehua, in eastern Mexico Cowan (1952, 1972, 1976, 1981)
Mazateco, in Mexico Cowan (1948, 1952), Busnel (1981)
the H’mongs, in Vietnam and Guyana Busnel ( pers. comm. 2003)
the Bai, in Yunnan, China Xian-Ming (2002)
the Chepang of Nepal Caughley (1976)
Diola language of Casamance, Sénégal Moreau (1997)
Fig. 1 – Manuel Ruis Mibeco, a Bora shaman playing on the Manguarey (double slit drum).
An Acad Bras Cienc (2004)
408 JULIEN MEYER
Fig. 2 – Maestro de Silbo of La Gomera showing to a class of
children how to use the ﬁnger to whistle.
its use in conditions of noise or isolation and the ef-
ﬁciency of the whistles at long distance. Such prop-
erties increase the signal’s resistance against rever-
beration. In the mountains of La Gomera whistles
can travel up to 10 km, yet the messages can still
remain understandable, even if, at this distance, the
signal deteriorates (Busnel and Classe 1976).
In summary, the whistled forms of language
seem to encode the essential part of human language
by an apparently simple signal that is constituted by
one main band of frequency, which survives propa-
gation over a long distance.
The whistled forms of languages are far more than
a code or conventional signalization system with
stereotyped sentences as they sometimes have been
described. They work exactly like vocal speech,
with the vocabulary, the grammar and, in many
cases, the phonology of the local language, espe-
cially in the domain of prosody. As far as the tonal
languages are concerned, the pitch level character-
izes the composition of the syllable and has a phono-
logical signiﬁcation. In this situation, the whistles
transmit two components together in one band of
frequency: the characteristics of the melodic
line of the spoken words and the characteristics of
their articulation. Moreover, it has been shown that
whistled forms of non-tonal languages reproduce
the phonemes of spoken language by using speciﬁc
levels of whistle frequencies for the different vow-
els, whereas speciﬁc modulations of the whistle fre-
quency (and sometimes its harmonics) are used for
the different consonants in the non-tonal languages
(Busnel and Classe 1976) and even in some tonal
languages when the tones are not sufﬁcient to get
rid of the confusion (Moore, pers. comm. 2003).
Theoretically, the vocabulary of the whistled
languages appears inﬁnite. But, as it concerns rural
populations, mainly with an oral culture, far from the
centers of information, it is not surprising that the
collection of words may seem limited to researchers
who investigate vocabularies used in the cities. It
has been observed that there is not less creation of
vocabulary in the whistled speech, but many modern
words are not adaptable to local life and therefore
are not whistled. Such tendencies are also noticed
for the domain of spoken language.
It has been stated that every individual who has been
trained with the whistled language of La Gomera for
about 3 years obviously reached a high global intel-
ligibility, despite the simplicity of the signal (Luis
Maestro, pers. comm. 2003). After sufﬁcient train-
ing, for example, 95% of messages encoded by the
‘Silbo’ of La Gomera (Busnel et al. 1962b) and
80% of messages given by the ‘Bai’ tonal language
of China were successfully understood (Xian-Ming
2002). A ﬁrst study in neuroscience has shown that
the BrocaArea as well as theWernickeArea were ac-
tivated in well-trained listeners, but not in untrained
ones, when they were presented with sentences com-
posed of Silbo (Carreiras 2003).
Some structural properties of whistled language
suggest comparing it to a few other forms of com-
munication. These are given, for instance, by the
use of so-called talking musical instruments (TMI)
which are found inAfrican cultures and China, or by
An Acad Bras Cienc (2004) 76 (2)
HUMAN WHISTLED LANGUAGES 409
Fig. 3 – Labio-dental technique practiced for Mazateco and Spanish whistled languages
Fig. 4 – Stereo distribution of the frequencies of a Mixteco whistled sentence.
The signal that conveys the information at long distance is centered on 2000Hz.
the application of whistle-like vocalizations which
frequently occur even in nonverbal organisms, such
as birds, several primates and marine mammals.
The sound of TMIs and similarly whistled lan-
guages can be regarded as an abridgment of speech.
Thus, comparisons between instrumental sound pat-
terns and whistles could positively inﬂuence further
studies of language. Cross-comparisons of human
and animal whistles, on the other hand, are interest-
ing from a different perspective. For example, both
classes of whistles serve for long-distance commu-
nication mostly (Marler 1955, Busnel 1966; see also
Brumm 2004, Oliveira and Ades 2004), but they dif-
fer in many other respects, e.g. the way they are pro-
duced. Thus, contrasting them could indeed help to
sharpen the idiosyncrasy of our own whistles. Such
a task, however, requires a formal deﬁnition of their
An Acad Bras Cienc (2004) 76 (2)
410 JULIEN MEYER
In human beings, a whistle consists of a nar-
rowfrequency band (typically located between 1000
and 3000 Hz), which lasts for a certain amount of
time and the relevant information it contains resists
degradation due to reverberation. This deﬁnition
could be the same for animal species with particu-
lar variations in the frequency level or the degree of
Human Whistles and Talking
Musical Instruments (TMI)
Many cultures play their native language on instru-
ments, and whistle it too. Each instrument takes
advantage of speciﬁc aspects of the musical and
linguistic transmission. The TMIs have been stud-
ied mainly in Africa, but also in Melanesia and in
China. It is interesting to consider that these means
of communication also convey messages with a lin-
guistic attitude. The technique used for the abridg-
ment of the speech depends on the instrument and
on the structure of the local language. Even differ-
ent strategies have been observed between cultures
using similar instruments on similar structures of
languages (Stern 1957). From the whistles to the
drummed forms of the languages, a continuum of
abridgments of speech can be noticed. Therefore it
is important to study the acoustical comparison of
these different ways of conveying information.
The extinction of whistled languages in La Gomera,
in Aas and currently in Antia in Greece is the di-
rect drawback of the aging rural populations and
their progressive weakening through, for example,
the disappearance of traditional activities like shep-
herding. When a culture is threatened, its whistled
or played form of speech is one of the ﬁrst parts of
the native language to disappear.
Today, every investigator who studies these
forms of communication should contact the Depart-
ment of Intangible Patrimony of Humanity in the
UNESCO in order to be part of the international
collaboration between these cultures which ﬁghts to
enable the next generations to learn and study their
own oral richness, but also to learn the endangered
status of these played and whistled languages.
The various people using whistle speech consti-
tute a minority culture in the country where they live.
They lack ofﬁcial recognition. They have been the
victims of different waves of domination. Therefore
the native forms of the languages are in danger of
disappearing and the cultural knowledge contained
in the oral traditions is less and less perpetuated by
the new generations.
To conclude, whistled languages are products of
human intelligence and not just curiosities or
‘‘surrogates’’ in the pejorative sense. They show a
widespread distribution across cultures and have ob-
viously developedquite independently of each other,
but mostly related to a particular local environment.
They are quite clearly deﬁned and represent an orig-
inal adaptation of the spoken language, like a ‘lo-
cal cellular phone’ for the needs of isolated human
groups. They document a cultural heritage and are
still fully in use among millions of Mexicans and
Chinese, as well as thousands of Turks, Africans
and people of Papua New Guinea.
The study of whistled and played languages
(e.g. TMI) might be a means by which to better
understand the role and the status of prosody in the
language faculty and some elements of its evolution
linked to musical aspects. They are also particularly
convenient for research on the intelligibility of lan-
guages and the recognition of an individual signature
in speech. Moreover their principles may inspire an
alternative development either of systems that con-
vey coded signals or of applications of languages for
The current situation of endangered languages
underlines that the cultures they represent are dying.
But all of these forms of communication hold
secrets about communication in general and there-
fore when they are in danger, it is also our culture
that is in danger.
An Acad Bras Cienc (2004) 76 (2)
HUMAN WHISTLED LANGUAGES 411
I would like to thank the Engineer Laure Dentel for
her constant support, the persons of the Mazateco,
Mixteco, Gomero and Bora cultures for their col-
laboration, René-Guy Busnel, Colette Grinevald and
Denny Moore for their advice, Jacques Vielliard and
Dietmar Todt for their professionalism and kindness
during the preparation of this manuscript.
Línguas assobiadas são um patrimônio valioso da cultura
humana. Este artigo fornece um panorama preliminar da
nova abordagem multidisciplinar sobre essas línguas. Es-
tudos anteriores dos equivalentes assobiados de idiomas
já mostraram que podem fornecer informações signiﬁca-
tivas sobre o papel do ritmo e da melodia na linguagem.
Para concretizar este papel, a maioria dos assobios con-
siste em modulações de freqüência em torno de 2000 Hz
(±1000 Hz) e freqüentemente atingem uma intensidade
de cerca de 130 dB (medidoa1mdafonte). O al-
cance pode atingir até 10 km (como veriﬁcamos em La
Gomera, Ilhas Canárias) e a mensagem pode permanecer
inteligível, mesmo que o sinal seja deteriorado. Em certas
culturas o uso de linguagem assobiada é associado a cer-
tos ‘‘instrumentos musicais falantes’’ (por exemplo, ﬂau-
tas, violões, harpas, gongos, tambores, ‘‘khens’’). Final-
mente, como meio de transmitir informação, os assobios
têm algumas analogias no reino animal (como certas aves,
cetáceos, primatas), oferecendo oportunidades para com-
parar as características acústicas dos respectivos sinais.
Com essas propriedades como referência, o projeto re-
latado aqui tem duas tarefas principais: elucidar melhor
os diversosaspectos da linguagem assobiada e, sobre tudo,
ajudar a interromper imediatamente o processo de seu de-
Palavras-chave: línguas humanas assobiadas, comuni-
cação por assobio, processos cognitivos, constantes cul-
Brumm H. 2004. Causes and consequences of song am-
plitude adjustment in a territorial bird: a case study
in nightingales. An Acad Bras Cienc 76: 289-295.
Busnel R-G. 1966. Information in the human whistled
language and sea mammal whistling. In: Norris KS.
(Ed), Whales, Dolphins and Porpoises. Berkeley and
Los Angeles: University of California Press.
Busnel R-G. 1970a. Historical brieﬁng of the whistling
language of Kusköy. Rev Phonétique Appliquée
Busnel R-G. 1970b. Recherches expérimentales sur la
langue sifﬂée de Kusköy. Rev PhonétiqueAppliquée
Busnel R-G. 1981. Bioacustica de la lengua silbada
mazateca. Ciencia y desarrollo 39: 178-184.
Busnel R-G and Classe A. 1976. Whistled languages.
Berlin: Springer Verlag.
Busnel R-G and Siegfried JR. 1990. Paroles, lan-
gages et langues sifﬂées. Paris: Service du Film de
Busnel R-G, Moles A and Vallancien B. 1962a. Un
cas de langue sifﬂée dans les Pyrénées françaises.
Logos 5: 76-91.
Busnel R-G, Moles A and Vallancien B. 1962b.
Sur l’aspect phonétique d’une langue sifﬂée dans les
Pyrénées françaises. Proc Int Congr Phonetic Sci,
Helsinki. The Hage: Mouton, p. 533-546.
Carreiras M. 2003. Activación cerebral durante el
procesamiento del silbo gomero. ICongr Int Lengua-
jes Silbados. La Gomera: Casa cultural de Valley
Caughley R. 1976. Chepang Whistle Talk. In: Se-
beok TA and Umiker J. (Eds), Speech Surrogates:
Drum and Whistle Systems. The Hague: Mouton, p.
Charalambakis C. 1993. A Case of Whistled Speech
from Greece. In: Philippaki-Warburton I, Nico-
laidis K. and Siﬁanou M. (Eds), Themes in Greek
linguistics: papers from the ﬁrst international confer-
ence on Greek linguistics. Amsterdam: John Ben-
jamins Publ Co, p. 389-396.
Classe A. 1956. Phonetics of the Silbo Gomero.
Archivum linguisticum 9: 44-61.
Classe A. 1957. The whistle language of La Gomera.
Sci Amer 196: 111-124.
Classe A. 1963. Les langues sifﬂées, squelettes infor-
matifs du langage. In: Moles AA and Vallan-
cien B. (Eds), Communication et Langage. Paris:
An Acad Bras Cienc (2004)
412 JULIEN MEYER
Cowan G. 1948. Mazateco whistle speech. Language
Cowan G. 1952. El idioma silbado entre los Mazate-
cos de Oxaca y las Tepehuas de Hildago. Mexico:
Tlatoani I, 31-33.
Cowan G. 1972. Segmental features of Tepehua whis-
tle speech. In: Rigault A and Charbonneau R.
(Eds), Proc Int Congr Phonetic Sci, Montréal. Paris:
Mouton, p. 695-698.
Cowan G. 1976. Whistled Tepehua. In: Sebeok T and
Umiker-Sebeok DJ. (Eds), Speech surrogates: drum
and whistle systems. Paris: Mouton, p. 1400-1409.
Cowan G. 1981. Whistled communication. N Linguis-
tics 20: 22-24.
Eboué F. 1935. La clé musicale des langues tambou-
rinées et sifﬂées. Bull Comité Afrique occid franç
Labouret H. 1923. Langage tambouriné et sifﬂé. Bull
Comité Afrique occid franç 6: 120-158.
Lajard M. 1891. Le langage sifﬂé des Canaries. Bull
Soc Anthropol Paris 4: 469-483.
Lenneberg EH. 1970. An acoustic analysis of the Turk-
ish whistling language of Kusköy. Rev Phonétique
Appliquée 14/15: 25-39.
Leroy C. 1970. Étude de phonétique comparative de
la langue turque sifﬂée et parlée. Rev Phonétique
Appliquée 14/15: 119-161.
Marler P. 1955. Characteristics of some animal calls.
Nature 176: 6-8.
Moreau ML. 1997. La communication sifﬂée chez les
Diola (Casamance, Sénégal). In DiversCite Langues.
Vol. II. http://www.uquebec.ca/diverscite.
Nekitel O. 1992. A perceptual analysis of Abu’-Wam
whistled speech. In: Dutton T, Tryon D and Ross
M. (Eds), Language Game: papers in memory of Don
C. Laycock. Canberra: Australian National Univer-
sity, p. 299-311.
Oliveira DAG and Ades C. 2004. Long-distance calls
in Neotropical primates. An Acad Bras Cienc 76:
Quedenfeldt HM. 1887. Pfeifsprache auf der Insel
Gomera. Zeitschr Ethnol 19: 731-741.
Rialland A and Gautheron B. 2000. Phonologie et
phonétique d’un langage sifﬂé: le silbo gomero. 2nd
Round Table in Phonology of the GDR (Groupe De
Recherche) 1954. Bordeaux: C.N.R.S.
Sebeok T and Umiker-Sebeok DJ. 1976. Speech surro-
gates: drum and whistle systems. La Haye and Paris:
Stern T. 1957. Drum and whistle languages: an analysis
of speech surrogates. Amer Anthropol 59: 487-506.
Trujillo R. 1978. Análisis lingüístico del silbo gomero.
Tenerife: Universidad La Laguna.
Xian-Ming Y. 2002. The talking musical instruments of
the province of Yunnan. Ethnomusicology research
seminar. London: Goldsmith University.
An Acad Bras Cienc (2004)