The design-feature approach to animal communication was developed in a series of papers written separately, though not independently, by the coauthors of this note (Hockett, 1959, 1960a,b; Altmann, 1962; Hockett, 1963; Hockett and Ascher, 1964; Altmann, 1967). The latest of this series incorporates the content of its predecessors, and is so recent that a new summary at this time would be inappropriate. Our purpose here is to describe a change of emphasis that we believe desirable.
The notion of a design feature is, at bottom, quite simple-minded. We begin with a rather loose conception of what sort of behavior would constitute an instance of “communication.” We note that the communicative behavior in some species, or perhaps only in some local community of a species, has a certain describable property. For example, human speech involves the production and detection of sound; the courtship ritual of sticklebacks involves physiologically controlled seasonal shapes and colorations of the participants. At this point we do not know—or we pretend we do not know—whether the property we have noted is a universal characteristic of all communicative behavior or is found in some forms and not in others. Subsequent empirical research is aimed at discovering which is the case.
We may find that the property recurs in all communication because it is characteristic of all organic behavior of any sort, or even of the behavior of all matter, living or inert. However important such a property may be biologically or physically, it has no relevance for our investigation because it distinguishes neither between communicative and noncommunicative behavior nor between different kinds of communication.
In some cases we may find that the property cannot even be looked for in some systems. Thus we might note, of written English, that some words begin with the letter “A” and others do not. Do gibbon calls have this property? An affirmative answer seems to be false, a negative answer misleading rather than true. A system cannot either have or lack this property unless it has such things as words, and such things as letters, and unless one of the latter is the letter “A”. The question is meaningless because neither answer is verifiable.
Or we may find that the property is present in some communication systems, lacking in others. In this case, it is a design feature, and communication systems can be classed into those that have it and those that do not. The two examples first given above (from human speech and stickleback courtship) are both of this sort.
Sometimes we find that a property is very widespread and that we are uncertain as to its universality only because of the vagueness of our definition of “communication.” At such junctures a common first impulse is to say something like “Perhaps systems that lack this property aren’t really communicative.” One may then argue about what communication “really is.” Eventually, however, we realize that all this is beside the point. Any discernible property affords a criterion more exact than our original loose conception of “communication.” So we take the obvious step: we devise new terms with which the facts can be stated simply and unargumentatively. Suppose, for example, that we have discovered two properties, a and ß, both very widespread but each missing in a few cases. We merely say that any system with property a, however marginal as judged by other criteria, is indeed a system with property a; and similarly for ß. If it is true that most systems have both, we can say so. If it turns out that no system has ß unless it also has α, we can say so, and try to find out why. The original term “communication” is retained in its original vague sense, because productive discussion is awkward unless one has broadside epithets as well as pinpointing labels.
There is nothing novel in this strategy. It recurs throughout science in the tentative and constantly changing classification of chemical elements, of organic compounds, of stars, of languages, of organisms or species—indeed, of anything. Thus any novelty in the design-feature approach lies in the application of a time-honored method to a new subject-matter.
The first of the papers cited earlier (Hockett, 1959) listed seven design features, all shared by all human spoken languages but each apparently lacking in one or another animal communication system. In Hockett (1960a) the number was increased to thirteen, and in Hockett (1963) to sixteen, still with human language as the major point of departure. In Altmann’s papers the emphasis has been shifted so that language is no longer the only point of reference or even the major one. There is also another change: although all sixteen of the design features from Hockett (1963) are clearly subsumed by Altmann (1967), the newer presentation is such that one is not tempted to summarize by saying that it presents exactly so many design features. One has an enlargement of the frame of reference that is not reducible to mere numerical terms. This second shift is a modification, we believe an improvement, of the original strategy, and it is this that we wish to explain and carry further here.
For this purpose, it will be convenient to cite the sixteen design features of Hockett (1963); the relation of what we have to say to the more elaborate system of Altmann (1967) should be obvious:
DF1. Vocal-Audi tory Channel.
DF2. Broadcast Transmission and Directional Reception.
DF3. Rapid Fading. (The sound of speech does not hover in the air.)
DF4. Interchangeability. (Adult members of any speech community are interchangeably transmitters and receivers of linguistic signals.)
DF5. Complete Feedback. (The speaker hears everything relevant of what he says.)
DF6. Specialization. (The direct-energetic consequences of linguistic signals are biologically unimportant; only the triggering consequences are important.)
DF7. Semanticity. (Linguistic signals function to correlate and organize the life of a community because there are associative ties between signal elements and features in the world; in short, some linguistic forms have denotations.)
DF8. Arbitrariness. (The relation between a meaningful element in a language and its denotation is independent of any physical or geometrical resemblance between the two.)
DF9. Discreteness. (The possible messages in any language constitute a discrete repertoire rather than a continuous one.)
DF10. Displacement. (We can talk about things that are remote in time, space, or both from the site of the communicative transaction.)
DF11. Openness. (New linguistic messages are coined freely and easily, and, in context, are usually understood.)
DF12. Tradition. (The conventions of any one human language are passed down by teaching and learning, not through the germ plasm.)
DF13. Duality of Patterning. (Every language has a patterning in terms of arbitrary but stable meaningless signal-elements and also a patterning in terms of minimum meaningful arrangements of those elements.)
DF14. Prevarication. (We can say things that are false or meaningless.)
DF15. Reflexiveness. (In a language, we can communicate about the very system in which we are communicating.)
DF16. Learnability. (A speaker of a language can learn another language.)
Of course, these features are related in various ways, some of which are discussed in Hockett (1960a, 1963) and Altmann (1967). But even if the reader examines these earlier papers, he will probably be struck by two things.
The first is that each feature seems to be set forth in an all-or-none manner, although upon closer scrutiny some of them are surely matters of degree. Displacement will serve as an example. Just how far away from the site of a communicative transaction must the topic of the message be before we will speak of displacement? And suppose the transmitter and receiver of the message are themselves some distance apart: for determining displacement, or its degree, do we measure the distance of the topic from the transmitter or from the receiver? The correction of this defect began in Altmann (1967), and is carried further below, but only in an implicit way—the point will not be mentioned again explicitly.
The other questionable point is that the sixteen features tie into communicative behavior in divers ways: some have to do with the channel, some with the repertoire of messages, some with the mechanisms by which the system is passed from generation to generation, and so on. One feels that there should be some superordinate grouping, and that appropriate grouping might suggest further things to be looked for in the communicative behavior of any one species or community.
When an investigator observes behavior in the field, he cannot physically isolate a dyad of animals, or a stimulus, or a channel, for separate study. He observes an ongoing community with complex networks of interactions and interrelations; he sees as much as he can of what is going on, and records as much as he can of that. Although there obviously are points that can be settled only by laboratory experimentation, it is clearly this behavior in the natural setting to which the laboratory results must eventually be related, since it is behavior in the natural setting that is adaptively significant for the animals themselves. Communication takes place in context, not in a vacuum. It is from this that we take our cue in seeking to eliminate from the design-feature approach the two defects described above.
If, in a specific social and ecological setting, a particular animal transmits a message to one or more others (or to himself), use is necessarily being made of a channel, or of several channels at once. This means that it makes sense to inquire about the following:
FRAMEWORK A. Features relating to the channel or channels (if several, the following points for each) :
A1. The emitting apparatus and its capacities.
A2. The receiving apparatus and its capacities.
A3. The material medium for the signal, and the type or types of energy (or matter) involved.
A4. The source of energy (in vocal-auditory systems, the transmitter; in “sonar,” the receiver; in visual communication, except for bioluminescence, external to both).
A5. The rate of travel of signals.
A6. The rate of decay or fading (damping out) of signals.
A7. The sources and sorts of channel noise.
A8. Orientation of transmission; e.g., broadcast or tight-beam; straight-line or crooked-line:
A8a. as imposed by physical factors;
A8b. as modified or narrowed by address cues (see Framework B, below).
A9. Orientation of reception: to what degree directional; adjustment of receiver’s posture to receive signal optimally. Factor of distance in determining the response (a territorial cry from a long way off may be audible but ignored).
These factors will vary within a community from one specific communicative transaction to another, but they will cluster around one or more norms. If there are several norms—say, two different types of energy for signals, or two different mechanisms for the generation of signals—then we may speak of two or more distinct communication “systems” (we shall say more below on the use of the term “system”). Thus, in a literate human community, one finds both speech and writing, perhaps also drum-signaling or whistle-speech or telegraphy. These different systems are not used simultaneously for a single message; thus, the situation seems to be different from that in which a single system makes use of two or more channels at once. Deeper investigation may show that what have been interpreted as two distinct communication systems show a significant connection: for example, there may be some kinds of information that can be transmitted with either of two systems, the choice of system depending on such circumstances as distance and noise. But a deeper connection of this sort is not likely to appear merely from an examination in terms of Framework A.
Framework A might be thought of as a generalization from design features DF1, DF2, DF3, and perhaps DF5. Surely, if we observed human language with the points of Framework A in mind, those features would be noted. But the framework implies research questions that are better because they are more open-ended. For example, instead of asking of a particular animal communication system, “Is it vocal-auditory?” we ask, “What is its channel?”
Now, if we focus not on channel but on the social setting of the communicative transaction, we get a somewhat different array of points to look for:
FRAMEWORK B. Features of the social setting:
B1. Number and location of transmitters (usually one at a time in human speech, many at a time in avian mobbing responses; sometimes one at a time but with rapid interchange of roles during a “single” transaction).
B2. Number and location of addressees. This is closely related to A8b, and one may at least distinguish directed messages (oneto-one, one-to-many, many-to-one, many-to-many), to-whomit-may-concern messages, and soliloquy.
B3. Mechanism(s) by which directness of message is achieved, e.g.
B3a. tight-beam transmission;
B3b. low volume or its equivalent;
B3c. use of an encoding system unknown to unwanted eavesdroppers or of a channel for which such eavesdroppers have no receiving apparatus;
B3d. indication of addressee(s) within the message (vocatives);
B3e. mechanisms for the denial of a possible addressee (as by facing away), for directedness at no one, and for directedness at self.
B4. Number and location of receivers, whether or not they are addressees (thus allowing for eavesdropping).
B5. Relevant differentiations of participants as to age, sex, and even species; and as to community membership.
B6. Nature and role of feedback to transmitter(s) from receivers (perhaps especially from addressees): signs of attentiveness, compliance or noncompliance with requests, reply in same channel or in some other. (We turn our head and eyes towards one who is speaking to us.)
B7. Number, nature, and location of topics of the message (what the message is about), relative to the locations of the various participants.
In some of the above, particularly B3 and B6, we have to do with what has been called “metacommunication”; i.e., communication that is some sort of commentary on other communication (Bateson, 1955). In human speech we can have “primary” communication and metacommunication in the same system, as when we interrupt what we are saying with “You’re not listening!” or with “I guess I expressed that badly.” But we also carry on, in paralinguistic and kinesic form, a virtually uninterrupted running commentary on what we are saying in words. Something much like this seems to be the case with many other animals: virtually all social messages are accompanied by contextual or “framing” cues that affect the interpretation or response. Human communication, centering on language, is so very complicated in these meta relationships and in the devices that mark them that “code noise” is common, leading, on the one hand, to the various antinomies discussed by logicians and, on the other, perhaps to certain kinds of emotional illness (Bateson et al., 1956). In the laboratory, comparable bizarrenesses of behavior have been induced in other animals; the extent to which other animals are subject to such communicative pathology in their natural environments is not clear.
For B7 we must realize, first, that a message may be about more than one thing at a time; second, that it may not be “about” anything at all as we ordinarily use that expression; and, third, that there is a difference between what an actual message is about (if anything) and the conventions by which the message, or its parts, can be about something. On the first point, note that a young boy and girl who are apparently (in the sense of “displacement”) talking about a popular singer, or Shakespeare, or their geometry teacher, are likely also to be communicating about their own interpersonal relationship; this is, in fact, generally the case of all face-to-face human conversations. On the second point, note that the seasonal coloration of a male stickleback is really not “about” anything—it has no denotatum whatsoever—although it is important in the courtship ritual. On the third point: the denotation of a word (say, “sugar”) must clearly be distinguished from the topic (or “content” or “import”) of a particular uttering of a sentence that contains that word (e.g., “Please pass the sugar,” uttered at a lunch counter with the speaker, the sugar, and other people and objects distributed in a particular way).
Framework B stems from design features DF4, DF5, DF15, in part (as to metacommunication and similar matters) D14, and in part (especially B7) from DF10.
But DF10, together with DF6, DF7, and DF8, and aspects of DF11, can be generalized to yield a third framework:
FRAMEWORK C. Features related to the behavioral antecedents and consequences of communicative acts:
C1. To what extent is the communicative act (the signal) effective in itself, via direct energetic consequences, and to what extent only via triggering?
C2. To what extent is the trigger effect of a signal due to an associative tie between the signal and other familiar things or situations in the life of the community? (For example, is the reaction to the danger call like the reaction to actual danger?)
C3. If there are such ties, to what extent do they rest on perceptual similarities between the signal and what it stands for? (Thus, there is no such similarity in the case of the English word “sugar” and the substance denoted by that word.)
C4. If there are such ties, does the signal merely call an instance of its denotation to the attention of the receiver, or does it function effectively when no instance of its denotation can be directly perceived by the receiver? (A hungry gibbon will move in the direction of the food call whether or not he can perceive the food.)
C5. If the signal functions in the absence of any instance of its denotation, is it ever transmitted falsely or “deceptively”?
Next we turn to the implications of DF12 and DF16. Here we must tread carefully to avoid pseudo questions; and we must remember that the significant questions may be extremely hard to answer empirically.
The actual course of moment-to-moment, day-to-day behavior of a single animal is the life history of that animal. We know that the life history of any animal shows changes, often gradual but occasionally abrupt, in typical capacities, roles, and habits, as well as physiological-anatomical growth and change. If we think of this as a genetically monitored unfolding conditioned by environmental circumstances, we call it maturation. If, on the other hand, we think of it as the acquisition and alteration of habits through experience, conditioned by genetics, we call it learning. In a sense, we are talking about the same thing whether we take the one or the other of these two views, and our vocabulary depends as much on the view as on the facts. Both views are justified; neither is complete.
What we should always like to know (though it may be enormously difficult to find out) is implied by the phrasing in the next framework:
FRAMEWORK D. Continuity and change in communication systems. Relative to each of the following, what is the exact synergy of genes, of tradition (social conditioning), and of the physicogeographical environment?
D1. The variations from group to group, community to community, in the communication systems of a species.
D2. The variations with age and sex in the participation of an individual in the communication systems of his community.
D3. The variations through successive generations in the communication systems of a community.
In the foregoing we have spoken rather loosely of communication “systems,” but without much discussion of what constitutes a “system.” The best approach to this seems to be in terms of the receivers situation. In any momentary situation, a receivers “state” is defined in terms of all his previous experience (insofar as it has registered) and in terms of everything he perceives at the moment. If, in a given state, the receiver can predict exactly what some other animal is going to do, then that other animal’s actual next act conveys no information to the receiver, since information is defined precisely in terms of reduction of indeterminacy or unpredictability. To count as communicative, therefore, a particular act of an animal in a setting must be, from the point of view of some receiver, one of a set of two or more acts any one of which might occur in just that perceived setting, but each of which precludes all the others. A honeybee whose dance means “northwest by north” cannot at the same instant be indicating some other direction. At a lunch counter, one can ask for both sugar and cream, but one cannot utter the two words “sugar” and “cream” at exactly the same moment. Any such set of mutually exclusive alternative communicative acts, or alternative pieces of larger communicative acts, defines a system. We do not say that it constitutes a system because the system surely includes much more—in the end, all the sorts of things that are being discussed in this paper.
Since a communicative act may be complex, composed of smaller acts that can be performed more or less simultaneously, a system as thus defined may consist of smaller subsystems—as when a human being may speak and gesture at the same time, with partly independent ranges of choice for the speech and for the gesture. On the other hand, the array of alternatives for a particular situation may overlap, or form a proper part of, the array for some other situation; thus what is revealed by certain observations as a system may turn out, through wider observations, to be only a subsystem of some more inclusive system. Here we need flexibility: we must be able to speak of a particular case with the bare term “system” without implying anything necessarily as to its position in a hierarchy of more or less inclusive superand sub-systems.
As a limiting case of this size-relativity of systems, we must recognize that a particular act that sometimes conveys information to some receiver may also occur in a setting in which, for a particular receiver, it is wholly predictable. We are occasionally sure—or almost sure— how someone is going to end a sentence before he actually finishes it. Such a case must obviously involve only a tiny subsystem of some larger system, in which we are more interested; if events of this special kind were too frequent the behavior involved would cease to have any adaptive importance.
For any system, then, large or small, there is a total repertoire of possible communicative acts (messages). Our last framework, stemming from design features DF9 and DF13, comprehends the structure of messages and the structures of repertoires:
FRAMEWORK E. Features of repertoire and of messages for a single system.
E1. Is the repertoire open or closed? Continuous or discrete? Are new messages sometimes transmitted, and received and interpreted appropriately? If so, how often, and what is the mechanism of coinage?
E2. Do the messages of the repertoire differ from one another as total gestalts, or are they built out of recurrent partials in some patterned way? If both, with what separation of these two mechanisms?
The last part of E2 is to allow for situations such as this: Among gibbons, the danger call and the food call are holistically different from each other, and these two seem to belong to a rather small finite set of calls from which the individual gibbon, at any moment, can choose exactly one or else remain silent. Yet the so-called “danger call” itself proves, on examination, to be a continuous range of possible calls, varying within certain rough bounds as to intensity, duration, and number of repetitions. To some extent, such actual physical variations from one instance of the danger call to another are “nondistinctive,” i.e., do not serve to convey different sorts or amounts of information, just as a speaker of English will vary somewhat his actual pronunciation of the word “sugar” without in any way altering the communicative import of the signal—it is still the same word. But this is not wholly true. To some extent, greater danger triggers a louder, longer, or more often repeated danger call, lesser danger a softer or less prolonged one. Thus, we must say that the single gibbon call (as usually described) is in fact a continuous open repertoire of possibilities within which any actual utterance occupies a systematic position; in this respect the “single” gibbon call is like the entire repertoire of bee dances. A human language is open, but discrete rather than continuous, so that the mechanism of openness (of free coinage of new messages) is necessarily quite different from that of bee dancing. The duality of patterning of human spoken languages (DF13) turns up in the detailed answers we get to the question of the way in which the messages of the repertoire differ, as stated in E2.
The frameworks we have set forth above are tentative; surely they need refining. We believe that they represent a net improvement over the old item-by-item design-feature technique first described. No loss is entailed by the framework method, since one is still free to ask the older type of atomistic questions whenever one is impelled to do so, but can then fit the answers into more meaningful frames of reference.
Altmann, S. A., 1962. Social behavior of anthropoid primates: analysis of recent concepts. In: Eugene L. Bliss (Ed.), Root of Behavior. New York, Hoeber-Harper.
——, 1967. The structure of primate social communication. In: S. A. Altmann (Ed.), Social Communication among Primates, Chicago, Univ. Chicago Press, pp. 325-362.
Bateson, G., 1955. A theory of play and fantasy. Psychiat. Res. Rept., 2:39-51.
——, et al., 1956. Toward a theory of schizophrenia. Behavioral Sci., 1:251-264.
Hockett, C. F., 1959. Animal “languages” and human language. In: J. N. Spuhler (Ed.), The Evolution of Man’s Capacity for Culture. Detroit, Wayne State Univ. Press, pp. 32-39.
——, 1960a. Logical considerations in the study of animal communication. In: W. E. Lanyon and W. N. Tavolga (Eds.), Animal Sounds and Communication. Washington, D.C., Am. Inst. Biol. Sei., pp. 392-430.
——, 1960b. The origin of speech. Sei. Am. 203:89-96.
——, 1963. The problem of universals in language. In: Joseph H. Greenberg (Ed.), Universab of Language. Cambridge, Mass., M.I.T. Press, pp. 1-29.
——, and R. Ascher, 1964. The human revolution. Current Anthropol., 5:135-68.