The Bee Waggle Dance: A Linguistic Analysis of Insect Language

William Lewis

Writer’s comment: My knowledge of insects was minimal, so when I took Professor Diane Ullman’s Entomology 111 class—Insects and Human Affairs—I was not sure what to expect. What I found was a fascinating, well-organized course. When it came to choosing a term paper topic, however, I was at a loss. I wanted to write a paper that tied in with my major field of interest, linguistics, but what relationship does linguistics have with entomology? Very little, I assumed, until the third week of the course, when we studied eusocial insects. Professor Ullman’s lectures on the subject intrigued me because they concerned the ability of eusocial insects to communicate with one another. My paper topic was born! I presented the idea of doing a linguistic analysis of an insect “language” to Professor Ullman and one of her TAs, both of whom were quite receptive. They suggested the bee waggle dance. Although the paper took careful research, I had fun doing it. I never expected, though, that my offbeat idea would turn into an award-winning essay.
—Will Lewis

Instructor’s comment: Insects and Human Affairs (ENT 111) is a General Education course designed to teach basic concepts in biology and entomology to nonbiology majors. The course goals are to provide students with an appreciation of insect natural history and the important role insects play in the daily lives of humans, e.g., how insects both benefit and harm humans and the position of insects in human beliefs, art, literature, and music. The purpose of the term paper was to give students an opportunity to choose, develop, and creatively explore a topic revolving around any interaction between insects and human affairs. The following essay by Will Lewis is beautifully written and particularly creative as he applies the fundamentals of human linguistics to the communication systems of a group of highly socialized insects, the honey bees.
—Diane E. Ullman, Entomology Department

The issue of language use among animals other than humans has been a source of great controversy in the behavioral sciences for much of this century. Most of this debate has centered on language use in the apes, with a concentration on training apes in human language (Patterson 1981:2-6). But what of naturally occurring language use among other animals, such as the waggle dance “language” used by the honeybee, Apis mellifera? Is the ability of honeybees to communicate locations of food sources to each other just a very sophisticated form of communication, or can it be described by the lofty term “language”?
         The difficulty here is what is meant by language. Webster’s Seventh New Collegiate Dictionary (1965) describes language as “a systematic means of communicating ideas or feelings by the use of conventionalized signs, gestures, marks.” Honeybees use a combination of sounds and gestures to convey information about the location of food to other members of a hive. Based on this definition, the honeybee’s form of communication is language. However, the Thorndike Barnhart Dictionary (1981) defines language as “human speech, written or spoken,” which obviously rules out the possibility of calling the honeybee waggle dance a “language” (as well as any form of human sign language).
         I suggest a more systematic approach. First, I suggest devising a set of criteria which define what language is. These criteria should be widely accepted and should be applicable to any form of human language. Second, I suggest applying these criteria to the waggle dance of the bees and determining how closely it meets these criteria.


         The criteria that I suggest are those that would be used by a linguist. Although there is no consistent set of criteria used by linguists for what constitutes language, I suggest that the following criteria would be acceptable to most linguists. If applied to any human language, these criteria would give positive results:1

1. The purpose of language is to communicate. Language, then, must allow communication to occur between at least two entities.
2. Language can be broken into smaller units. The smaller units of a language like English would be sentences, words, morphemes, phonemes, etc.
3. Language has, and must follow, certain pre-defined rules. These rules quantify and describe the various generalities and regularities of a given language, and do not necessarily indicate actual cognitive processes.
4. Language possesses grammar. Grammar is the cement that links words together into a sentence, creating a framework for a string of words to provide a meaning beyond their combined meanings. It also changes the shapes of the words. For instance, the string of words
         She give ball I
does not constitute language because it possesses no grammar. On the other hand, the string of words
         She gave the ball to me
does constitute language because it contains grammatical information beyond the semantics of the content words. The tense of give, the declension of I, and the grammatical function of the words the and to convey grammatical information.
5. Language is unbounded in scope. Thus, there is no limit on what language can be used to “talk” about.
6. Language is a learned phenomenon. Although humans possess innate abilities for language (there is some controversy on this point), language is always learned.2


         The purpose of language is to communicate, and it is generally accepted that honeybees are able to communicate and that their language meets this criterion: A honeybee returning from a successful foraging trip can communicate to her hivemates the precise location of the food source. The information contained in her message describes both distance and direction, and she uses samples to help recruit other hivemates (Frisch 1967:57-60, Weiss 1989:283). Although the mode of communication is a dance (rather than spoken words or gestural signs), bees do communicate successfully.3

         The honeybee waggle dance does not consist of one gesture, sound, or sign, but represents a combination of them. I would suggest that the honeybee waggle dance includes the following elements, which can be thought of as the units, or “words,” of the honeybee’s language: pattern, number of iterations, liveliness, angle, request for sample, giving of sample, sound.4 Each of these units conveys some meaning or performs some necessary function in the dance. In particular:
         • The pattern of the dance is instrumental in conveying distance information. Distances less than 10 meters are communicated by a “round” dance, so called because the general shape of the dance is a circle (although the bee may not complete the entire circle with each iteration) (Frisch 1967:29). The bees communicate distances greater than 100 meters by what Frisch calls the “tail-wagging dance,” whose pattern consists of two semi-circles, performed side-by-side (Frisch 1967:57). Between 10 and 100 meters, bees use a hybrid of the two patterns.
         • The number of iterations of the dance that a bee performs apparently conveys distance information as well. Frisch (1967:64) indicates an inverse relationship between the number of iterations and the distance of a food source: 100 meters = 9 to 10 iterations, 500 meters = 6 iterations, and 1500 meters = 4 iterations.
         • The liveliness of the dance conveys information about the quality of the food source. The more excited the bee appears, the better the quality of the food (Frisch 1967:34).
         • The angle of the dance conveys the direction of the food source (Frisch 1967:129-37).
         • The request for a food sample is not submitted by the dancing bee, but by an observing bee. A bee that is actively observing the dance (and, in fact, can be engaged in the dance itself by following the lead bee) will at times emit a short squeak by thrumming the honeycomb in order to request a sample of food. The dancing bee will stop her dance and give out a food sample at the other bee’s request (Kirchner 1994:76). Although this “word” is actually not used by the dancing bee, it is still part of the discourse between bees and is thus a relevant unit of the waggle dance.
         • The doling out of samples by the dancing bee is important in the recruitment process, as shown by experiments done with robot bees to decipher the components of the waggle dance (Kirchner 1994, Moffett 1990, Weiss 1989). In these studies, scientists developed robot bees controlled by a computer that mimicked the waggle dance. Because a robot bee could not detect the “thrumming” request for food from the observing bees, it could not dole out samples at the appropriate times (Moffett 1990:137). Observing bees, frustrated by this apparent violation of one of the “rules” of their language, would often appear annoyed and leave without having been recruited (Weiss 1989:283, Moffett 1990:140).
         • The sound component of the waggle dance eluded scientists until fairly recently. In fact, until the mid-1980s, there was no proof that bees could even hear (Kirchner 1994:74). At that point, it was determined that bees could respond to sound (Kirchner 1994:79). Subsequent research indicated that bees hear only at short distances (Sanford 1993). It was not until the development of the robot bees, however, that the sound component of the waggle dance became apparent. Earlier studies using robots proved unfruitful because they omitted sound (Weiss 1989:283). In fact, the robots in these other studies were either ignored by the bees in the hive or attacked as intruders. The Michelsen robot described by Kirchner (1994), Weiss (1989), and Moffett (1990) was unique because it used sound. As a result, it was successful in recruiting bees.
         The sound emitted by a bee during a dance apparently serves two purposes: to get the attention of bees in the immediate vicinity, and to keep their attention. Since bees cannot see the dance (they are in dark, enclosed hives), they evidently also need the sound to perceive the dance.5 Because the dance language breaks down into distinct units, it meets this criterion of language.

         The combination of units in the honeybee’s waggle dance must be structured in certain ways and cannot omit elements, or the communication event does not occur (Kirchner 1994). These rules govern the waggle dance:
         • The dancing bee, in order to successfully recruit, must respond to requests from other bees by doling out samples. If she fails to do so (as evidenced by the robot bees), she will be less successful in recruiting.
         • The dancing bee must use the appropriate dance patterns for the appropriate distance. If the bee uses an inappropriate pattern (no matter the number of iterations), the attending bees will not find the food source.
         • The dancing bee must also use the appropriate angle to convey the direction of the food source.
         • The dancing bee must use sound to get and keep the attention of other bees. Failure to do so causes the dancing bee to be ignored, or, worse, attacked.6
         Although these rules are primarily pragmatic ones, they still constitute identifiable rules of the waggle dance. It might be possible to determine other rules for the waggle dance by further isolating its components, or by further defining the interaction between the components. This list, however, provides sufficient evidence to indicate that the waggle dance does contain “rules.”

         Isolating the grammatical components of the honeybee’s waggle dance, if they existed, proved to be a significant challenge. What was necessary to prove the existence of grammar in the waggle dance was a unit of the dance that did not provide content information, but rather linked or changed the units in some way (such as tense or plurality).
         When examining a sentence in English—even a simple sentence—it is possible to identify its grammatical units quickly. For example, in the sentence
         She gave the ball to me
grammar includes the past tense in gave, the direction of the act of giving in to, and the uniqueness of the ball with the. The tense, the objective preposition, and the modifier do not have easily identifiable meanings when isolated (what is the definition of the?) or they cannot be isolated (how can past tense be isolated?).
         A waggle dance discourse, if translated to English words, might look something like this:
1. Hey guys! The sound component gets the attention of local bees.
2. 1000 meters away. The type of dance and the number of iterations indicate the distance of the food source.
3. 45-degree angle. The food source is at a 45-degree angle from the hive.
4. Keep listening! The sound compo- nent keeps the attention of the attending bees.
5. Really good! The excitement level of the bee indicates the quality of food.
6. Give me! An attending bee requests food.
7. Here is some. The dancing bee doles out food upon request.
         Units 2 through 5 would overlap. The attention-getting sound precedes the dance, as well as participating in it. (It is thus indistinguishable from unit 4.) The information imparted does not appear in sentences, as we would expect in English, but as a structured list of facts. English sentences for this exchange might be:
         “Hey guys! I found a really good food source, 1000 meters away at a 45-degree angle.”
         “May I have some?”
         “Here’s some.”
         Here grammar, though minimal, does exist. More importantly, it could not be eliminated without adversely affecting or altering the meaning or the intent of the communication. If a human attempted to communicate by stringing sets of facts together, as described in the waggle dance example, they would likely be misunderstood (unless the context dictated such a listing of facts). Bees, on the other hand, appear to have no problem accepting a list of facts, and do so without exception.

         Because the bee language conveys a limited set of information, it also fails the criterion of unboundedness. Human languages allow us to express virtually anything we can think or feel. We are also quite capable of describing topics and abstractions that we do not or cannot completely understand (such as infinity, for instance).
         It may be argued that the reality of the bee becomes bounded because it consists primarily of food-related tasks and that bees thus cannot communicate beyond this reality. The problem with this argument is that bees also make honeycombs, lay eggs, cool off the hive (by fanning), sting, and engage in a number of other non-food-related activities (Frisch 1967). No evidence could be found that communication as sophisticated as the waggle dance exists to “talk” about these other activities.

         Bee language is not a learned phenomenon, but appears genetically predetermined. As Rosin (1992:83) aptly points out, “[H]oneybee dances, which are neither learned from experienced dancers nor discovered in a crude manner and perfected through practice, must be instinctive (i.e., genetically predetermined).” In other words, bees do not go through a period of language acquisition as humans do, but are able to use their language as soon as they can forage.


         It is obvious that the waggle dance of the honeybee constitutes a sophisticated form of communication. The waggle dance, however, does not meet several key criteria of human languages: grammar, unboundedness, and learnability. It cannot, therefore, be called a “language” in the sense of a human language, but can only be considered a sophisticated form of communication.
         This conclusion in no way belittles the evolutionary achievement of the honeybee waggle dance, nor the uniqueness of this achievement. It is profound that the honeybee, with a mere 800,000 neurons (Weiss 1989), compared with our 2 trillion, has evolved such a highly sophisticated form of communication. It is also humbling that bees come remarkably close to possessing something that many would describe as a uniquely human capacity, the ability to use language.


1 These criteria were derived in part from my own course of study (and thus constitute common knowledge in the field of linguists), and also from Akmajian, et al. (1990:5-9).
2 Caplan (1987:43-133) presents good arguments in favor of localization of language function in humans and the innateness of language ability arising therefrom.
3 R. Rosin is a major opponent of the dance “language” school, favoring instead the odor-search paradigm. This paradigm states that bees primarily rely on odor stuck to their bodies for recruitment to food sources (Wenner 1990:73-89). Other sources of R. Rosin’s arguments can be seen in Rosin (1992) and Wenner (1990:240-50).
4 Frisch (1967) was used as the source for determining these units. His is still the most definitive source on the structure of the waggle dance.
5 The related species Apis florea nests outside. Apparently because it can use visual cues, it does not use sound in its dialect of the waggle dance (Kirchner 1994:80).
6 The Michelsen bee actually was tested in situations that violated some of these rules (Kirchner 1994:80).


Akmajian, A., Demers, R. A., Farmer, A. K., and Harnish, R. M. (1990). An introduction to language and communication. Cambridge, Mass.: MIT Press.

Barnhart, C. L. (1981). The Thorndike Barnhart Dictionary. New York: Doubleday.

Caplan, D. (1987). Neurolinguistics and linguistic aphasiology. Cambridge, Mass.: Cambridge University Press.

Frisch, K. von (1967). The dance language and orientation of bees. [Trans. Leigh E. Chadwick.] Cambridge, Mass.: Harvard University Press.

Gove, P. B. [Ed. in Chief] (1965). Webster’s Seventh New Collegiate Dictionary. Cambridge, Mass.: G. & C. Merriam.

Kirchner, W. H., and Towne, W. F. (1994). The Sensory Basis of the Honeybee’s Dance Language. Scientific American, 270, 74-80.

Moffett, M. W. (1990). Dance of the electronic bee. National Geographic 177, 135-40.

Patterson, F. and Linden, E. (1981). The education of Koko. New York: Holt, Rinehart and Winston.

Rosin, R. (1992). Do honeybees dance? BioScience 42, 83-85.

Sanford, M. T. (1993). Can bees hear? Online. University of Florida Newsletters Database. Internet. 31 Jan. 1993. Available: Gopher://

Weiss, R. (1989). New dancer in the hive: an insect imposter helps scientists decipher honeybee lingo. Science News 136, 282-83.