CHAPTER 4

MESSAGE-MEANING ANALYSES

W. JOHN SMITH

SEMIOTIC AND ZOOSEMIOTIC ASPECTS

In 1946 Morris presented a division of semiotic (compare Chapter 1) into three different sorts of relationships which has proved of considerable value in directing research. All three relationships need study in animal communication, and, although the results must be combined to yield a complete picture of any species’ communication pattern, it is necessary to treat each sort of relationship in terms of its own particular features. When the scheme used by Morris (1946) is applied in a zoosemiotic study, it is necessary that all of the hypotheses employed be consistent with the principles of zoology, and in particular with the central theory of biology—evolution through natural selection [discussed in detail by Mayr (1963)].

Syntactic Level

By observing, listening to, or otherwise recording an animal’s specializations for signaling, it is possible to distinguish (1) discrete, or largely discrete, units and (2) “grading” (Konishi, 1963)—types of essentially continuous variation which modify the forms of these units —from discontinuous variations in signaling activity. This step in research yields a list of signals and of different ways of modifying signals; signals and the relationships between them are the subject matter of the syntactic level. Further work at this level can deal with comparing the relative frequencies of usage of the members of the signal list, studying ways in which they are normally distorted by the vicissitudes of their environments, etc.

The first problem of research at this level is to identify the signals, and this is made more difficult because a zoologist often wishes to study primarily a special class of signals which he calls displays. A display is an output of an organism which has been selected and modified in the course of evolution to function efficiently as a social signal [for discussions of “ritualization,” the evolutionary process involved, see Tinbergen (1959, p. 44); Blest (1961)]. Displays, however, are not the only animal features which carry information and are involved in communication. In fact, the term “communication” has exceedingly broad application—often it is used to include all events in which information is transferred (see, e.g., Ayer, 1955, or Haldane, in the same volume). In living systems alone such events range from sensation of an inanimate external environment to interindividual signaling employing displays and include intraindividual hormonal and nervous activity. Yet, within all this, there is a class of potentially important information which is often not easily available unless displays are employed.

One of the basic functions of a display (see below) is to make the behavior of the communicator more predictable to a recipient by making available some information about the internal state of the communicator. Other aspects of the recipient’s external environment, say, rocks, plants, or such maintenance behavior of other animals as foraging, have not been structured by evolution to have signal function, even though the recipient does obtain information through perceiving them. Some features of an individual’s hormonal and nervous activity are more comparable with displays, however, since they have been evolutionarily structured for a signaling function. But such signals are private and, unlike displays, are not designed specifically to impart information to another individual. Fringe cases exist (such as plant-toanimal communication, exemplified by the guidelines with which some flowers attract some bees to their nectar stores) which make it clear that the limits of what I have in mind should not be drawn too sharply. Yet in this chapter I shall be concerned only with that part of the whole of communication in which a communicator produces displays or things very like them and which therefore is distinguished by a history of evolution for social signal function.

Because evolution is a continuing process, we must sometimes, or often, encounter instances in which it is hard to know whether we are dealing with a display or with something that is designed for a nonsignal function and is only incidentally functional for signaling. Thus Brockway (1964) was not able to detect much ritualization in most of the agonistic movements of the budgerigar, which she found to have both signal and other functions. Evolution of postural and movement displays, such as those with which she was working, must usually be based upon previously existing activities: on intention movements (Daanje, 1951), autonomic responses (Morris, 1956), or other comfort movements (e.g., McKinney, 1965). The unritualized activities always occur in contexts and in some of these contexts provide socially relevant information; if it is relevant enough, and if other functions do not oppose ritualization, then evolution should occur. But evolution cannot occur in the absence of a positive net value of favoring selection, and a nice example of this can be seen in the case of anatid wing-flapping. McKinney, in his detailed study of comfort movements of the Anatidae (1965) lists seven situations in which wing-flaps occur: after bathing, after rising from sitting on the ground or sleeping, after alighting, after a disturbance (as when an airplane flies over), before flying, in aggressive encounters, and, in an unpredictable way, during feeding. In some of these situations, such as before flying, wing-flapping probably has a signal function, being one of the movements by which, as McKinney puts it, “one bird transmits the flying mood to others, and ensures synchronous flight in the group.” On the other hand, this wing-flapping is not different in appearance from the flapping which removes water from the wings after bathing. In relevant situations, it might function better if it were obviously specialized as a display, but perhaps anatids already have sufficient redundancy in the other activities which also occur in these situations and very little or no selection pressure exists to modify it.

Semantic and Pragmatic Levels

To annotate the syntactic lists into something resembling a lexicon, we need to study the users of the signals. In the event of any communication [discussed more fully in Smith (1965)] there must be at least two users who share a code about the signals: one, the communicator, sends the signal, and the other user is its recipient. Two different sets of hypotheses are required, one to deal with that which the communicator encodes in the signal (its “message”) and the other to deal with any “meaning” the event might have for the recipient.

As a first hypothesis at the semantic level, I suggest that messages be considered to represent some aspect(s) of the state (or central nervous state) of the communicator. Additional hypotheses about messages are necessary if we are to be able to recognize them, so I suggest that messages can be recognized (1) in part by taking the common denominator, where one can be found, of the communicator’s other activities over a number of samples in which the display is given and (2) in part by determining to what extent the signal could serve to identify the communicator in any way.

A display can evolve by means of natural selection only in so far as the message it carries is relevant in the social relations of the signaling animals. An irrelevant message not only has nothing working for it in the mechanism of evolution, but in addition, will almost always carry a negative advantage for the signaler. In most cases production of a display is costly in one or more senses—it requires equipment and the maintenance of that equipment, energy for its production, and it may expose the signaler to more than one recipient—including such undesirable recipients as predators. Hence another postulate of zoosemiotics should be that ritualization will not act to produce functionless displays.

As a corollary, if we can recognize that something is a display, then we can expect to find it serving at least one function in the social relationships typical of the species. In actually employing the corollary, however, we might suppose that we can recognize as displays all events which are organized but which have little or no function other than signaling. This would involve us in two assumptions. One is basic to evolutionary theory: Only disorganization will result from random genetic happenings; organization requires that there is an organizing process, and since in this case it is natural selection, we are able to assume that some function accrues. The second assumption, however, is dangerous: We can scan for types of function and establish their presence or absence. This is not easy in a great many cases [as pointed out by Tinbergen (1965)], and we may sometimes mislead ourselves into the conviction that something has a display function because we cannot detect any other function.

At least three very broad and somewhat overlapping sorts of social function must exist: displays identify the communicator in one or more ways, from indicating the communicators position in space to giving the communicators species, and displays make subsequent behavior of the communicator more predictable and/or permit monitoring of current activities, within different time spans for different displays. While a signal may enable prediction, we can say only that a recipient behaves as if it predicts the subsequent behavior of the communicator. The recipient’s responses to the signals could be, and probably are, usually, to a considerable degree, genetically programmed. We should be most cautious about assuming the operation of rational behavior, and I do not mean to imply at any point in this chapter that I believe a recipient to be necessarily rational.

Even a “purely” identifying display enables prediction of behavior. A bird song, for instance, cannot just identify the sex and species of the singing individual without also indicating that he would court with, or defend territory primarily, from birds of his own species and would not so interact with grossly different species. At the other extreme, there are probably very few displays which do not in some way provide identifying clues, if only so that they will be clearly labeled “pertinent to read” for the right species. The hypothesis put forward above, that a message should represent some aspects of the state (or central nervous state) of the communicator, is consistent with this view, although at times perhaps in an almost trivial sense. Animal species differ from one another in their behavioral repertoires, for instance, and a display which identifies a species also specifies that the behavior of the communicator will probably be chosen from the repertoire of his species.

By way of example, any worker honey bee, receiving the dance and accompanying sounds of a returned forager, is obtaining identifying information; either the signals fit an “expected” code or they do not. The dancers message is sometimes described as being about a particular location, but this is not necessarily the case. It is equally likely [or even more probable, considering the possible evolution of this dance (Blest, 1961, pp. 120-122)] that the message is reporting only on the activities of the communicator, or on the probability of her repeating a particular set of activities in a particular way. Such a message is relevant, and hence can be selected for, if a recipient worker can obtain the meaning (see below) that she should repeat those activities herself. This repetition can be for one or more ends (foodor water-gathering or hive-site selection), some of which are apparently not specified by any known feature of the displays.

At the pragmatic level, to determine what the meanings of a signal are we can employ Cherry’s (1957) operation of using the recipient’s responses—if, indeed, we can actually observe the responses. Often, though, uncomplicated responses can be nearly inaccessible, and there is need for an hypothesis enabling us to predict the likely responses so that we may devise ways of checking for them. This could be that the responses in any particular circumstance should usually be appropriate to the role the individual should assume within the social framework characteristic of its species.

Employing the hypotheses from the semantic (message) and pragmatic (meaning) levels of semiotic entails that we prepare for the lists constructed at the syntactic (signal) level not one lexicon, but two. This further requires, then, that we study the transformations by which recipients obtain meanings from messages. There are too few signals available to provide a different signal for each meaning, so recipients must use other information in addition to that which is actually encoded in the messages of the signals. Thus another hypothesis about meanings should hold that we consider them to be derived not simply from messages, but from messages in contexts (Smith, 1965). The transformations of messages are governed by the total relevant context, including among other things the code about the signals which recipient and communicator share and the ways in which the recipient is programmed, by heredity and experience, to select environmental features as context to modify the perception of a signal.

Within the repertoire of a species, should the message of any one display, or of any one type of grading of displays, always be essentially the same under different circumstances? Suppose it were not. Since displays, received in different circumstances, can lead to different meanings for a recipient, there might be a variety of messages to correspond to the variety of meanings. But all of these messages would still be encoded by the same single display, and the recipient could recognize them only by knowing the contexts. However, if a recipient must know the context to use the message effectively, it is not necessary that the message vary as well as the context and the meanings. A constant message yielding different meanings in different contexts is a simpler assumption, and is the assumption implicit in my suggestion that a message be (in part) identified as the common denominator in an individual’s activities over a series of samples while using that particular display. Further, it is difficult, although perhaps not impossible, to envisage the evolution of a display which encoded messages with nothing in common. How would the unlike messages both come to coincide with the same display output?

The relationship between meaning to a recipient and the function of a display is close, but the two notions are by no means identical. For one thing, functions may in many cases be different for the communicator and the recipient (s), as illustrated in the following example. If one of the meanings of a display is “seek cover,” its function in this usage, from the standpoint of recipients, will be to alert at least some conspecifics to a dangerous situation they have not yet detected for themselves. The functions of such a potentially dangerous display for the communicator are harder to see, but probably inelude the following. (1) The protection of associating individuals who may act similarly to protect the communicator in the future, i.e., individuals whose continued association is valuable to him. (2) The protection of descendants of the communicator carrying some portion of his genotype (see Hamilton, 1963; Maynard Smith, 1965)—it is his success in leaving descendants that is the crux of natural selection.

Natural selection can act in the evolution of a display only if that display has a function, or functions, of value. Hence the postulate about the transformation of messages to meanings by recipients is important and must be biologically plausible. The postulate requires that an animal be able to receive simultaneously, or nearly simultaneously, a number of sensory inputs, sort them, and somehow integrate the relevant ones with each other and with historical context while discarding or perhaps temporarily storing the remainder [see Broadbent (1961, 1965) and Marler (1961) for discussion and evidence of perceptual selecting abilities]. For some simple animals this may be asking too much, and in their communication we might expect the few displays they have to lead a recipient to one meaning each. But higher animals clearly do deal with a highly complex sensory environment. Imagine a bird, flying at about thirty miles per hour among the twigs and branches of a forest, integrating a complex array of visual inputs, achieving a complex control of adjustments in its flight, navigating, alert to yet other visual clues (food, conspecifics, predators, etc.) and to auditory clues. Small wonder the racing pigeons tested by Herrnstein and Loveland (1964) for ability to recognize a complex and variable shape performed about as well as the experimenters. In the light of these abilities the postulated transformation of messages into meanings surely is biologically plausible.

It is often the case that a particular meaning is not dependent upon only one particular message, and there is likely a considerable overlap among messages in the meanings to which they can lead in different contexts. Thus a message of either intense fear or intense aggression may, in different contexts, lead to a meaning which might be paraphrased as “I should flee/’ It may be that natural selection needs considerable freedom in structuring some parts of the communication pattern of a species. It may happen, for instance, that precursors for displays which could encode some categories of message are unavailable and that opportunistic use must be made of those messages which can be communicated. This would be successful if the available messages and the right aspects of context could combine to give a recipient a usable meaning. Yet certain meanings will be sufficiently advantageous within particular kinds of social organizations that, whatever combinations of messages and contexts can be exploited by a particular species, these meanings must be obtained. If the set of messages which could lead to these meanings is small, then the perceptual choice of contextual features may become relatively important and the amount of freedom for evolutionary choice may be quite limited.

On the other hand, there may be considerable evolutionary freedom in the remaining aspect of the pattern, the physical characteristics of the displays. Certain criteria, some of which are reviewed by Tinbergen (1959) have to be met—the displays must be recognizably distinct from other displays within the same repertoire, they must be distinct among the displays of sympatric species, etc. The known criteria (there may be more) still leave a great deal of latitude, however, and the forms of displays with the same message may be quite different in some aspects from species to species, even where such differences are not required [see, for example, the various and distinct ways in which different parts of intention movements of leaping have been ritualized into displays, reviewed by Daanje (1951)].

APPLICATION OF THE HYPOTHESES ABOUT MESSAGE AND MEANING

The approach described above has been employed in an analysis of the communication patterns of the species of New World flycatchers in the genus Tyrannus (Aves, Tyrannidae). This analysis (Smith, 1966) deals primarily with displays and messages, since observation has been concentrated thus far on following a displaying individual in order to study its actions before, during, and after displaying. The set of messages common to most of the species appears to encompass the following general message categories: (1) inhibition of a tendency to locomote, (2) anxiety over potentially disruptive situations, (3) aggressiveness, usually with an indication of defensiveness or offensiveness, (4) fear possibly, and (5) identification of various sorts. Approximately seven or eight different vocal displays are found in most of the species (the existence of a variable amount of continuous grading of one signal into another makes it difficult to determine an exact figure), and these encode, with some overlap, the various messages. Approximately eleven nonvocal displays (mostly movements and postures) have been found which also encode the messages, providing both a good deal of redundancy, which is usually designed for other sensory receptors, and some nèw features such as directed transmission (effectively by pointing) instead of broadcast transmission [terms from Hockett (I960)].

All but three (which were used only in attack) of the roughly eighteen or nineteen different displays were used in more than one significant context, and some occurred in many different contexts. Judging from observed behavior and the social relationships typical of members of these species, these displays probably mediate seven social functions: (1) attraction and repulsion of conspecifics, (2) individual recognition, (3) appeasement and reassurance, (4) threat, (5) providing environmental monitoring information to conspecifics, (6) alerting, and (7) the associating of adults and dependent offspring. Taken alone, most of the available messages do not relate directly to these functions, but when placed in their contexts they should lead to a very large set of meanings which are quite relevant. A number of these meanings have been verified by observation; still others are known to be probable from the extensive ethological literature on the displays of other species; many, though, will not be verified without further research. The whole set of possible meanings is too great to reproduce here, but one particularly interesting vocalization can be used as it occurs in the repertoire of T. tyrannus, the eastern kingbird, as an example for both message and meaning analyses.

The Locomotory Hesitance Vocalization (LHV) is a series of repeated units in which each unit is complex, usually consisting of two or three parts and sounding somewhat like “kt-ter,” or “kt-ter-ter.” A good deal of grading variation is possible within a series, but little of this has yet been correlated with the different situations in which the display is used; for the most part the display can be considered to be very similar in its different occurrences. Its many uses include the following: (1) A male, patrolling the vaguely defined limits of an area he defends in the spring, gives an LHV every time he ceases flying, and even at times when he approaches a potential perch and veers off without completing a landing; this behavior is more common before he has obtained a mate than after. (2) Within a pair, particularly a newly formed pair, when one bird approaches its mate, it may use the LHV and receive a similar reply. (3) Within the family, a few days after the young have fledged, the offspring may use the LHV when persistently following their parents from whom they still beg— although now with a greater likelihood of eliciting an aggressive response from too close an approach. (4) A male eastern kingbird who has just chased a hawk from the region of the nest may call the LHV as he slows his flight and begins to turn back. (5) A female may call as she flies from the nest in the absence of her mate, who usually guards the nest in her absences.

Surely the motivational states of birds using LHVs in different circumstances are often very different, but they seem to share one main feature—the displaying eastern kingbird is experiencing some internal conflict with a tendency to locomote. This I have taken to be the message of the display.

A list of all the possible meanings of the LHV requires considerably more space. Let me give one example, phrased first as if it were an assessment by the recipient bird (an unobservable meaning, and one which would very probably not take the syntactical form of expression I must use): “I am not likely to be attacked by the approaching, vocalizing, conspecific individual.” Depending upon the identity of the recipient, this hypothetical meaning might lead to various observable responses which we could interpret directly as meanings. For instance, an adult conspecific opposite in sex to the communicator, if in breeding state and established as the communicators mate, would probably repeat the LHV without fleeing or attacking, although there are measurable but much lower probabilities of other responses in variations of this context. Suppose, however, that an adult recipient was being approached, not by its mate, but by one of its offspring; the response then might be to avoid, or to ignore, or even to attack the approaching bird. It would take an extremely thorough set of observations in different contexts to enable one to state the probabilities of each of these different meanings.

In another category of contexts, where the communicator is a patrolling male moving rapidly from point to point within a restricted area, some of the hypothetical, unobservable meanings might be: (1) “I may approach and attempt to initiate pair-formation activities” (recipient an unpaired female), (2) “I must not intrude unless ready to fight” (recipient a territorially inclined male), or (3) “I must not display” (recipient a nonaggressive individual which is migrating through or a foraging neighbor or even a member of the local floating population). Again, in another context the LHV is given by a female flying from her nest to feed when her mate is elsewhere and the meaning to him might be “I should go to the nest vicinity and guard”; observational evidence for such a meaning is suggestive, but not yet conclusive. Still other meanings are possible in the many other contexts, but the above provide a sufficient example to demonstrate the workings of the approach [a more detailed presentation may be found in Smith (1966)].

Since first using this approach in studying the communication patterns of the various Tyrannus species, I have extended it to many other sorts of birds and recently to some mammals while exploring the possibilities of generalizing about message sets, their employment, and the displays which encode particular kinds of them. In some other species it has been much easier than it was in the flycatchers to study certain sorts of meanings. One such species is the herring gull, and I should like to give one example from its vocal repertoire to illustrate just how different the meanings of a single display can be. One of the herring gull’s displays is a simple, wheezing call which the dependent young use when begging for food or when mildly uncomfortable for reasons other than hunger. At this stage in their lives, it can lead adults to feed them [see Tinbergen (1953); my own experience with birds of this age is almost entirely with a hand-reared group, and I have not watched the behavior of adults of their own species in response to the chicks’ calls]. Later, my wife and I have watched young gulls in their first winter of life beg from adults without getting fed, but have found that the call still, at times, enables the young to obtain food by a very different method. Some adult herring gulls defend winter feeding areas along the shoreline (Drury and Smith, in press), and exclude many other gulls from them. Young gulls cruise about; when one finds an adult with a conspicuous food resource, it attempts to intrude into the defended area, all the while uttering the same wheezing call. Its actions attract other herring gulls of all ages which also intrude, or attempt to intrude, until soon there are too many individuals present for the owner to cope with, and the food becomes available to all. The inferred meaning in this instance is not “feed the communicator’ but “the calling youngster has discovered a food source, and I should try to get some of it.” Note that only the meaning and the relevant recipients have changed; the message is the same, and the function, at least with respect to the communicator, is also the same. Tinbergen (1953, p. 179) indicates that this call occurs in yet other contexts, e.g., both adults use it as part of their precopulatory behavior—so that apparently it has still other meanings.

Much more work is necessary before we can know just how extensively distributed certain sorts of messages are among animals or how similar are the sets of messages possessed by different species. Certainly the studies of Lorenz, Tinbergen, Moynihan, and many others have shown that messages pertaining to agonistic and sexual tendencies are very widespread, but few other sorts of messages have yet been extensively investigated. The locomotory hesitance message of the LHV in T. tyrannus seems to be widespread among tyrannids of other genera (personal observations), and displays with apparently similar messages have been reported in other passerine birds (e.g., Marler, 1956; Andrew, 1957, 1961; Snow, 1958; among others) and even in mammals [e.g., night monkeys, see Moynihan (1964)]. Message sets like that of T. tyrannus may also be common, and clearly similar sets occur in many other tyrannids. Species which have different social arrangements, however, will probably be shown to emphasize different sorts of messages. Eastern kingbirds, for instance, do not form integrated flocks, and their closest knit social organization is the pair. They are relatively aggressive birds and communicate mostly about approach, inhibited aggression, and anxiety. Some other nongregarious species [e.g., the green-backed sparrow, studied by Moynihan (1963) ] also express considerable inhibited aggression. Such flockforming birds as chaffinches and buntings, on the other hand, are relatively nonaggressive and have more messages expressing fear. Since flocks are probably efficient predator-locating devices (see Moynihan, 1960), displays which indicate different levels of fear probably are very valuable within the social framework of these species. Both Marler’s (1956) presentation of the chaffinch repertoire and Andrew’s (1957) discussion of the calls of buntings show this higher number of fear messages among the non-song vocal displays. Detailed message analyses could surely be made of these and other species, although the published accounts have been prepared for other ends and do not usually present data that can easily be turned to this purpose.

One feature of the repertoire of many passerine birds which might make detailed comparisons with tyrannids difficult is the loosely defined sort of signal commonly called bird “song.” There are probably at least two or more general kinds of song (see Armstrong, 1963), one of which is characterized by its rigidly constant form and rapid rate of delivery by a bird which is usually perched alone. Often this sort of song is uttered principally in the very early morning, around dawn, and many tyrannids do have a vocalization (which I have elsewhere termed the Regularly Repeated Vocalization) of this sort. Most tyrannids lack, however, the somewhat more variable, less rapidly repeated songs which so many other birds use through the day. The members of the genus Tyrannus, for instance, do not sing during the day—but, on the other hand, they are far from silent and vocalize readily from a repertoire of less complex calls under a wide variety of circumstances.

The functions of bird song have been reviewed in detail by Marier (I960), Thorpe (1961), and Armstrong (1963). The best understood functions can be summarized briefly as territorial proclamation (acting to repulse males and to attract females for pair formation), species and individual identification, the maintenance of the pair-bond, and physiological synchronization of the pair. Armstrong lists various usages of songs (rather broadly defined) which in some cases suggest other functions, such as stimulating nestlings to beg for food.

The complexity of temporal variety in the song of many species seems to be much in excess of what might be required for the better known of the above functions. In fact, experiments by Busnel and Bremond (1961, 1962), Weeden and Falls (1959), and Falls (1963) indicate that some of the elements of this complexity can be removed and males will still come toward the sound and/or sing in response to it. Some of this additional complexity may be to make the song distinct against a complex sonic environment (Marler, 1960), or may be other sorts of necessary redundancy, and some might be explained by Hartshorne’s “monotony” principle (1956). But all of these explanations together may still leave a good deal of complexity unaccounted for. To suggest that this residue of complexity is nonfunctional might not be consistent with one of my earlier hypotheses, although it does not necessarily follow from that hypothesis that all minor aspects of a display must be functional. Significantly, however, experimenters have as yet tested only the reactions of males to playbacks of their altered songs. The mates of these birds are also potential recipients of the songs, of course, and it might be that a variable song enables a listening mate to monitor some of the activities of an unseen male.

I have tested this hypothesis on one of the few tyrannid flycatcher species which does “sing” during the day, the eastern wood pewee (Contopus vir ens) (Smith, ms.). The daytime song of this species comprises a nearly continuous repetition of two units: “pee-a-wee” (designated RV1) and “pee-oh” (RV2). Under most circumstances an interval of from about six to ten seconds separates each unit. The RV1 unit may be repeated from one to sixteen or more times before the string is terminated by an RV2, but the latter call is only very rarely repeated in sequence. Thus the singing of a male eastern wood pewee can be analyzed as a succession of strings of varying numbers of RV1 units, each string terminating with an RV2. In a certain proportion of cases an RV2 is correlated with the cessation of flight, but no positive correlations between activities and the RV1 have been discovered. Neither unit is uttered in flight.

One of the findings of this study has been that for strings of different lengths there are different probabilities that the singing bird will fly during the string. It also appears that different levels of flight tendency and its inhibition also associate with different lengths of strings. Thus, without watching a singing male, it is possible to form some notion of the amount and kind of his locomotory activity. My next working hypothesis, to be tested shortly, is based on the fact that males of this and many other tyrannid species often guard the nest while the female is away foraging; thus a female could select her times of leaving the nest to forage at least partly on the basis of such information. There is one particular range of string lengths in which I suspect the singing male is both ready to guard and yet does not perceive any danger in the area, and this is the range in which the female should be most likely to quit the nest. Other species are known in which a male indicates to an incubating female that he is ready to guard [often, as in the song sparrow (Nice, 1937), by coming and singing near the nest] and then the female leaves to forage. So far as I am aware, however, no attempt has been made to show that a change in the relationships of units in a song provides this information.

A male eastern wood pewee apparently can modify the messages encoded in his singing by changing the quantitative relationships between the two discrete units of this song. Whether or not such simple grammars will be found in other bird song remains to be seen. If correlations between the activities of singing birds and variations in their songs are found to be widespread, however, then the same methods of analysis for message and meanings that apply to the study of the repertoires of birds which do not sing through the day will also work for the analysis of the vocalizations of singing birds.

APOLOGIA

Is the message-meaning distinction productive; are there any circumstances in which an approach based on this distinction has any advantages over other approaches? These questions can best be asked in terms of the ends of the investigator. If one is interested only in function, then this approach is cumbersome and one may deal just with the meanings recipients take from communication events and the fit of these meanings to a functional pattern. It often happens, however, that functional labels are assigned to displays, labels like “contact call,” “alarm note,” “flocking call,” “social call,” and “courtship dance.” This is a perfectly useful procedure, of course, although it must be kept in mind that a display may have more than one function. But cross-level interpretation may follow when there is not recognition of the different sorts of relationships that signals have with each other, their referents, and their meanings, or of the difference between meaning and function. It is not necessarily correct to assume that the message of a contact call is specific to a flock-seeking tendency if the call has other uses, or that the message of a courtship display is necessarily related to sexual tendencies if the display also occurs in encounters between members of the same sex. Such error is less likely in a scheme in which it is recognized that messages are only very indirectly related to functions and that the meanings taken by recipients of displays are related both to messages and to contexts. Functions depend upon these meanings, but the functional viewpoint is more limited than the communication viewpoint, and it seeks to learn, at a variety of levels, how the ecological and social needs of the organism are met.

On the other hand, if one is interested also in the process of communicating by displays, then the message-meaning distinction becomes more useful. The value of the proposed scheme which employs it depends upon the recognition of Morris’s three levels of analysis, and in many other respects this scheme is not markedly different from various others long in use. (Different investigators often look for different sorts of messages, but such preferences are not inherent in this approach.) Perhaps its most important feature is that it recognizes the role of context in the generation of meaning and thus implies that evolution does not have to produce a different signal for every meaning or for every function. Another feature is that it places the behavior of the communicator within a scheme in which, by employing additional information about specific social relationships and ecology, an estimate of probable meanings can be arrived at indirectly. It is easier, often very much easier, to study the relation between the signal and the activities and tendencies of the communicator before trying to find that between the signal and the reaction of a recipient. Granted that the latter relationship must be investigated at some stage, it is at present often difficult to know to what extent the signal itself, and to what extent different features of the immediate and historical contexts of the event, interacted to determine the reaction—if, indeed, the event was followed by an observable action which could be assigned to it and not just by some ephemeral change of mood. For instance, it can be very difficult to measure the extent to which a recipient is filtering its various inputs. In contrast, the communicators coincident and/or subsequent behavior and the display are necessarily related to at least some aspect(s) of a particular mood, as both are products of one and the same individual.

REFERENCES

Andrew, R. J., 1957. A comparative study of the calls of Emberiza spp. (buntings). Ibis, 99:27-42.

——, 1961. The motivational organization controlling the mobbing calls of the blackbird (Turdus merula). IV. A general discussion of the calls of the blackbird and certain other passerines. Behaviour, 18:161-176.

Armstrong, E. A., 1963. A Study of Bird Song. London, Oxford Univ. Press.

Ayer, A, J., 1955. What Is communication? In: Studies in Communication. London, Martin, Seeker and Warburg.

Blest, D., 1961. The concept of ritualization. In: W. H. Thorpe and O. L. Zangwill (Eds.), Current Problems in Animal Behaviour. New York, Cambridge Univ. Press.

Broadbent, D. E., 1961. Human perception and animal learning. In: W. H. Thorpe and O. L. Zangwill (Eds.), Current Problems in Animal Behaviour. New York, Cambridge Univ. Press.

——, 1965. Information processing in the nervous system. Science, 150:457-462.

Brockway, B. F., 1964. Ethological studies of the budgerigar (Melopsittacus undulatus): Non-reproductive behavior. Behaviour, 22:193-222.

Busnel, R.-G., and J. C. Bremond, 1961. Étude préliminaire du décodage des informations contenues dans le signal acoustique territorial du Rouge-gorge (Erithacus rubecula L.). C. R. Acad. Sei. Paris, 252:608-610.

——, and J. C. Bremond, 1962. Recherche du support de l’information dans le signal acoustique de defense territorial du Rouge-gorge (Erithacus rubecula L.). C. R. Acad. Sei. Paris, 254:2236-2238.

Cherry, C., 1957. On Human Communication. New York, Wiley.

Daanje, A., 1951. On the locomotory movements in birds and the intention movements derived from them. Behaviour, 3:48-98.

Drury, W. H., Jr., and W. J. Smith, In Press. Defense of feeding areas by adult Herring Gulls and intrusion by young. Evol.

Falls, J. B., 1963. Properties of bird song eliciting responses from territorial males, Proc. 13th Intern. Ornithol. Congr., pp. 259-271.

Hamilton, W. D., 1963. The evolution of altruistic behaviour. Am. Naturalist, 97:354-356.

Hartshorne, C., 1956. The monotony-threshold in singing birds. Auk, 73:176-192.

Herrnstein, R. J., and D. H. Loveland, 1964. Complex visual concept in the pigeon. Science, 146:549-551.

Hockett, C. F., 1960. Logical considerations in the study of animal communication. Am. Inst. Biol. Sei., 7:392-430.

Konishi, M., 1963. The role of auditory feedback in the vocal behavior of the domestic fowl. Z. Tierpsychol., 20:349-367.

McKinney, F., 1965. The comfort movements of Anatidae. Behaviour, 25:120-220.

Marler, P., 1956. The voice of the chaffinch and its function as a language. Ibis, 98: 231-261.

——, 1960. Bird songs and mate selection. Am. Inst. Biol. Sei., 7:348-367.

——, 1961. The filtering of external stimuli during instinctive behaviour. In:

W. H. Thorpe and O. L. Zangwill (Eds.), Current Problems in Animal Behaviour, New York, Cambridge Univ. Press.

Maynard Smith, J., 1965. The evolution of alarm calls. Am. Naturalist, 99:59-63.

Mayr, E., 1963. Animal Species and Evolution. Cambridge, Mass., Harvard Univ. Press.

Morris, C. W., 1946. Signs, Language and Behavior. Englewood Cliffs, N.J., Prentice-Hall.

Morris, D. J., 1956. The feather postures of birds and the problem of the origin of social signals. Behaviour, 6:271-322.

Moynihan, M. H., 1960. Some adaptations which help to promote gregariousness. Proc. 12th Intern. Ornithol. Congr., pp. 523-541.

——, 1963. Display patterns of tropical American “nine-primaried” songbirds. III. The green-backed sparrow. Auk, 80:116-144.

——, 1964. Some behavior patterns of platyrrhine monkeys. I. The Night Monkey (Aotus trivirgatus). Smithsonian Inst. Misc. Collections, 146(5):1-84.

Nice, M. M., 1937. Studies in the life history of the song sparrow. I. Trans. Linnaean Soc. New York, 4:1-247.

Smith, W. J., 1965. Message, meaning, and context in ethology. Am. Naturalist, 99:405-409.

——, 1966. Communication and relationships in the genus Tyrannus. Cambridge, Mass., Publications of the Nuttall Ornithological Club, No. 6.

——, Ms. Bird “song” as communication. I. Preliminary studies of Contopus virens.

Snow, D. W., 1958. A Study of Blackbirds. London, Allen and Unwin.

Thorpe, W. H., 1961. Bird-Song. New York, Cambridge Univ. Press.

Tinbergen, N., 1953. The Herring Gull’s World. London, Collins.

——, 1959. Comparative studies of the behavior of gulls (Laridae): A progress report. Behaviour, 15:1-70.

——, 1965. Behavior and natural selection. In: J. A. Moore (Ed.), Ideas in Modern Biology, Troc. XVI Intern. Zool. Congr., Washington, 1963, Vol. 6, New York, pp. 521-542.

Weeden, J. S., and J. B. Falls, 1959. Differential responses of male ovenbirds to recorded songs of neighboring and more distant individuals. Auk, 76: 343-351.