Within and About Communication:

A Time Series Study of Communication Monographs

Davis Foulger
Temple University
Fall, 1979
Department of Speech Communication

(more papers by Davis Foulger)


Secondary Analysis of data describing the publishing history of Communication Monographs, obtained from Nation and Nation (1975), was performed in order to trace substantive trends in Communica tion research. Clusters of periodicities were found, including one in the range of 6 to 11 years and another of roughly 20 years. An attempt was made to assess the extent to which journal editors may have acted as "gatekeepers", favoring certain areas of research over others. Weak evidence was found, but it cannot be confidently concluded that gatekeeping bias occured.


As a scholarly discipline, communication (speech and mass, where such divisions are made) finds its roots in the intersection of many fields, including sociology, rhetoric, psychology, anthropology, english, and others, as well as a number of emergent technologies, including new and refined mediums of mass communication and methods for recording and dissecting both sight and sound. This diversity of roots has brought communication a vast range of ideas and interests which are reflected in the existence of a variety of sub-disciplines. Although there is often overlap between the interests of these sub disciplines, each includes members with little or no interest in the doings of any of the other sub-disciplines. Thus, although all share a basic interest in how people communicate, there is competition for both membership and forums to present ideas.

There are a variety of ways to split a field into subdisciplines. The National Center for Educational Statistics splits Speech Communication in ten parts (McBath and Jeffrey, 1978) and classifys Mass Communication under an entirely different heading. The International Communication Association only makes eight such divisions--Information Systems, Interpersonal Communication, Mass Communication, Organizational Communication, Intercultural Communication, Political Communication, Instructional Communication, and Health Communication--but, although including Mass Communication under its banner, ignores the more traditional divisions of Speech. The older and more traditional Speech Commnunication Association makes nine such parsings at Forensics, Instructional Development, Interpersonal and Small Group Interaction, Interpretation, Mass Communication, Public Address, Rhetorical and Communication Theory, Speech and Language Science, and Theatre. Although the International Communication Associations parsings are probably the more sensitive to current directions in Communication research, the Speech Communication Associations classifications are are broader and more sensitive to the history of the discipline.

The strength of any scholarly discipline is its communication system. This system, in the form of books, journals, conventions, training, and interpersonal contacts, enables ideas to circulate and develop, not only among those who share ones immediate interests, but among those in other areas of a field whose peculiar insights give one's ideas their stiffest test. Without methods for the exchange of ideas, there cannot be a scholarly discipline, for if scholarship is nothing else, it is the sharing and refining of ideas.

With a variety of conventions sponsered by competing associa tions, a growing range of titles, and several high quality journals, the field of communication has a reasonably strong communication system. The diversity of interests within the field easily challenge this system, for although at least sixteen journals publish almost exclusively within the field, few publish work representing the bredth of the field. Competition among divisions for panels at conventions is often intense, moreover.

One of the more exceptional communication journals from the standpoint of representing the bredth of interests within the field is Communication Monographs. Following Matlon and Matlon (1975) who classify journal articles within the field using the parsings of The Speech Communication Association, Communication Monographs contains articles published representing eight of the nine divisions. The one sparsely represented division, Interpretation, is, moreover, much more a performance than a research division. Other journals unquestionably better represent particular sets of interests within the field (Quarterly Journal of Speech for rhetoric; Journalism Quarterly for Mass Communication; Human Communication Research for interpersonal communication and information systems; Journal of Communication for the sociology/anthropology of communication), Communication Monographs probably makes the best combination of prestige and breadth of interests represented.

It should be expected then, that an examination of Communication Monographs might shed light on where the field of communication has been and where it is going. Just such and examination is the focus of this study.


The choices one makes in conducting a study set boundaries on what can be learned about the object of study. In conducting this study I made the following choices:

  1. Communication Monographs was selected as a representative sample for studying the development of Communication as a field.
  2. The interests of the various editors of Communication Monographs were assessed (to see if these editors served as gatekeepers, regulating the content, as well as the quality, of the journal).
  3. Instead of conducting my own content analysis, I relied upon an existing content analysis, that of Matlon and Matlon (1975)
  4. The existance of patterns in the data was assessed statistically.

Each of these choices involved subchoices and had both positive and negative implications which will be assessed below. By restricting the study to Communication Monographs, I kept the scope of the study within manageable limits. It was doable. A larger study, encompassing more journals would unquestionably have been more representative, but such a study would also introduce problems of equivalence that would have made the study considerably more difficult. These problems involve quality, length of manuscript, number of articles per issue, number of issues per year, date of first publication and other problems that would have made the question "What constitutes an article" more difficult to assess. The choice of Communication Monographs as most representative may also be percieved as a limitation. Other scholars might have made different choices. Narrower interests would have made other choices inevitable. It is my personal opinion, however, that no other journal matches Communication Monograph's combination of scope and prestige. Thus, for me, it is the best choice.

Communication Monographs brings problems in its choice, however. First, it began publishing in 1934, limiting the present investigation to the time period 1934 to present. As, however, the development of communication as a discipline was barely beginning at that time and the development of several sub-disciplines was not to start for several years, this limitation should not be considered too stringent. Second, although there has been a steady increase in the number of articles published within the journal over the years, this growth has not kept pace with the overall growth of the field. With more scholars competing for less relitive publishing space, the quality of articles should be expected to have increased over the years while the extent to which the journal is representative of all scholarship in the field has undoubtably decreased. This bias may or may not be reflected in the extent to which divisions are represented. A third limit occurs in changes in the shape of the field which have occured since Communication Monographs was first published. Speech and Hearing Science and Mass Communication were once much more active in the affairs of the Speech Communication Association than they are currently. The development of competing departments, journals, and professional associations have decreased the extent to which they can be adaquately represented by studying Speech Monographs. Indeed, current developments in the field make it likely that the same thing will happen for scholars of interper sonal communication and information science.

A fourth and potentially most important boundary imposed by the choice of Communication Monographs is found in the methods by which articles are selected. The scholar who chooses a journal as an archival source is not obtaining anything even remotely resembling a random sample. What is obtained is a selection of articles chosen through the critical judgment of readers and the editorial policies of the editor, usually on the basis of percieved quality of scholar ship, percieved quality of writing, and the percieved relevance of the article to the field of communication. Any of these factors can be biased, within limits, by the editor's choice of readers, explicit and implicit standards for quality, and perception of what constitutes Communication research to favor one division over another, one set of interests over another.

The possibility of such 'gatekeeping' led to the second choice of the study, that of assessing the editor's interests (before they became editor) on the basis of their previously published research. Such gatekeeping would not only bias the contents of the current data archive, but have long range consequences in shaping the inter ests of reader/researchers. This choice opens the possibility of overcoming the fourth limitation posed by the choice of journal and adds an interesting extra dimension to the present study. The recent deposing of an editor of a major journal in psychology due, it is said, to excessively stringent editorial requirements adds relevance to this question. Do editor's act as gatekeeper's? Let's find out.

Deciding to rely on a pre-existing content analysis opens several potential problems. The benefit is the saving in time and resources accrued. There is no need to select a set of content categories, no need to check those categories for their reliability, no need to set up criterion for discriminating among categories, and no need to spend time doing the actual classification. One inevitably loses, through this choice, much of the flavor of the data, however. Ethnographic and semiotic treatment become impossible. Such depth, however, was not the aim of the study. A second boundary is imposed by the quality of the content analysis chosen. This study can only be as good as the content analysis selected. My choice, Matlon and Matlon (1975) seems secure in its classification scheme. It is that of the Speech Communication Association. The taxonomy has been criticized in several quarters, however, on the grounds of having misclassified articles. Such criticism is probably inevitable, however. I doubt the critics could agree on a single scheme or set of criteria for setting up such a scheme. What is important is the consistancy of classification which should result from having such a content analysis done by one set of scholars. The publishing date of the Matlon and Matlon index, 1975, sets a third boundary, 1974, the last year for which articles are classified within the index. Continuing the index beyond that date would eliminate the boundary but introduce new inconsistancies to the data set. Inevitably, my classifications would be inconsistant with those made by Matlon and Matlon, in part because I don't have access to the criteria on which their choices were made; in part because even given those criteria, my perceptions of how articles should be classified would differ from theirs.

A fourth limitation imposed by my choice of the Matlon and Matlon content analysis is imposed by interaction with my choice to evaluate patterns in the data statistically. They do not treat their category as being mutually exclusive. As such the use of finite stochastic modeling techniques is made impossible without the violation of assumptions. The decision to treat the data statistically invokes limitations in its own right, as well. Statistical treatments are generally drier reading than thick descriptions. Because they are more precise, they are more open to argument and disconfirmation. In defence, however, it should be noted that a statistical description of pattern is a description. Such descriptions are not perfect, but few descriptions ever are. Indeed, the criterion of perfection is open to debate. It should be noted, moreover, that statistical treatment of data does not restrict one to exploring linearity. Although the general linear model will be used thoughout in testing the existance of pattern, I will be much more interested in finding periodic patterns in the data than I will be in finding linear patterns.


Data was collected from the Index to Journals in Communication Studies through 1974 (Matlon and Matlon, 1975). The index is a cumulative taxonomy in which all articles appearing in a large number of communication journals are classified using, as major divisions, the parsings of the Speech Communication Association and two additional levels of sub-classifications. Although data was collected for various categories at each of the three levels of classification, only the main headings were examined for this study. The data set used here consists of thirty-two variables, eight representing the eight subdivisions of the Speech Communication Association parsings which were sufficiently represented in the journal to allow analysis, seven representing the various interest areas of the journal's editors, fifteen representing sub-classes of the various divisions, one representing the total number of articles published in any given year, and one which recounted the year.

The total number of article variable, the eight divisions, and the fifteen subclasses were recorded as frequency variables, the total number of articles attributed to each area by the journal in any given year. Editor interests, assessed by identifying the appropriate classification of their articles in the field published prior to their editorship, were dummy coded with 1 representing interest and 0 recording lack of published interest. Although not sensitive to changes in interest, depth of interest, or unpublished interest, the variable seems both sufficient to the task and available. Finally the years for which measurements were taken (1934-1974, inclusive) were represented in the last variable using the last two digits of each year.

The data analysis involved three methods: an 0-style principle component analysis, spectral analysis, and multiple regression analysis. The 0-style analysis (single subject, across variables, between occasions)was performed using the twenty-three content variables (eight divisions and fifteen subclasses) to see if patterns existed in the data such that sets of years clustered together. The obliquely rotated factor patterns were then appended to the data set and treated as the dependent variables in a multiple regression analysis. In the multiple regression, both the content and editor interest variables served as regressors, providing a first level at which to examine the proposed gatekeeper effect.

The spectral analysis, which assessed the existance of periodic tendencies in the data, was performed on the eight principle division variables and the seven editor interest variables. Spectral Analysis (Jenkins and Watts, 1968; Chatfield, 1975; Watt, 1979) is a time series method which enables the researcher to examine a data set for the existance of cyclic movements in the data. Human existance is filled with cycles including eating cycles, sleep cycles, news cycles, and the seasons. It is felt in some quarters (Rosnow, 1978) that themes within the social sciences are also cyclic, that interest in various research topics ebbs and flows in a semi-regular rythem of interest and disinterest, expansion and retrenchment. It was hoped that such cycles might be found in the data. The dominant spectral method used was that of frequency optimization (Watt, 1979). Those cycles identified in the spectral analysis were then entered into a regression analysis along with the year (a linear component), the total number of articles published in each year (a non-linear growth component) and lagged editor interest and area variables as regressors and unlagged editor interest and area variables as dependent variables. All regressions were performed stepwise with the criterion for entry to or removal from the equation set at an F-ratio of 4.1.

Results I: The 0-style analysis

With the criterion for accepting components set at eigenvalues greater than or equal to one, five components where found whichaccounted for 89 percent of the variance between years. The oblique rotation of the factor pattern converged after thirty-six iterations at the pattern shown in table one and the interfactor correlations shown in table two. Three periods are clearly discernable in the pattern. The first, which extends from 1934 to 1946 represents Communication Monographs early years. This period is marked by a high level of consistancy in the distribution of articles among divisions and can be seen as Factor two of tables one and two. The second period extends from 1943 to 1962 and is best visualized as a fight between competing interests within the field. For the most part this fight is waged on Factors three and four, but is joined by the eventually victorious Factor one. The third period, represented on Factor one, extends from 1958 to 1974 and can be seen as paralleling the emergence of the field of communication as we know it today.


Table four shows the regression of Factor two on the thirty area and editor frequency variables. The seven variable stepwise regression which results accounts for ninety one percent of the variance in the factor. The dominant variables in the regression are Speech and Language Science, Rhetorical and Communication Theory, and Instructional Development. Speech and Language Science is notable for its presence during the period. It so dominates the journal during this period that one imagines that the journal was started largely as a service to the interests of that division. Instructional Development and Rhetorical and Communication Theory are more notable for their absence during the early history of the journal. The above pattern is further re-enforced in the regressions minor components. Voice Science accompanies Speech and Language Science in its presence during the period. Modern Rhetorical and Communication Theory is even less apparent than is the division to which it owes allegience. Not only is public address less important than it will later be, but the editor's during this period aren't interested in the area either. Overall, this period is best described as the era of Speech Science. In later years members of this division will start their own journals and profes sional associations, but during this period they emerge as a major interest area.


In the regression of Factor three, seen in Table five, four variables account for ninety-two percent of the variance in the factor. The five variables in Factor four account for eighty percent of its variance. The former is most clearly an Instructional Development component. The latter is dominated by Public Address. As noted above, this period seems to be a fight. On the one side sit forces interested in using the journal as a forum for exploring pedagogy and curriculum options for an emerging discipline. On the other side forces more interested in exploring the theory of rhetoric and public address. Basically, this can be viewed as a conflict between the forces of theory and the forces of application, of those interested gaining knowledge and those interested in sharing it.

Ultimately this conflict is resolved in 1952 with the establishment of a new journal, Communication Education, intended to service the interests of the Instructional Development division. The conflict between Instructional Development and Rhetoric and Public Address should be viewed, however, more as a symptom of deeper unrest within the field than as the sum of the unrest. This period sees a range of new ideas, themes, and interests take root among scholars of communication. These include the emergence of readability formulas, the Shannon and Weaver model, systems theory, information theory, propaganda research, and the Human Relations school of organizational functioning. Thus, just as instructional development was taking itself out of the competition to control Communication Monographs, whole new fields of interest began to enter the fray.



The regression of Factor one, in table three, accounts for ninety- five percent of the variance in the factor. It is clearly dominated by the Interpersonal and Small Group Division and, more dominantly, the Rhetoric and Communication Theory Division. Indeed, they are not unlike the dominant interests of scholars of Speech Communication today. Although the data doesn't extend to the present, the first component appears to offer a sort of modern consensus on the shape of the field of communication. Indeed, the latter part of this period displays so much consistancy that it can be inferred to represent a period of retrenchment.

Factor Five

Table seven shows the regression of the fifth component on the area and editor interest variables. The fifth component is obviously a minor component. It lacks both the consistancy of location necessary to its association with a period and the consistancy of effect necessary to concluding it is a dynamic influence. The strong loadings of the factor in table one are all secondary loadings to loadings on factor one. Yet the factor seems, despite all this, to offer a useful additional perspective on the development of the field which is not available through the other four components. Here we see a continuation of the fields interest in Speech and Hearing Science, this time divorced from voice science. Although some of the interests of the second component are continued in this component, is should not be interpreted as a continuation so much as an attempt, apparently unsuccessful, to recessatate some of the interesting issues of the past.

Results II: The Spectral Time Series Analysis

The spectral analysis was performed on each variable individually. The results will be presented in the same way. Because the equation of spectral analysis differs markedly from that of linear regression, tables eight through twenty-two may seem odd to the reader. Instead of the one component (alpha or beta) involved in each section of a linear or polynomial equation, each cyclic component requires three. These are w (the length of the cycle divided into one and multiplied by 6.28), A, and B (A and B are similar in purpose to regressions beta) expressed in the equation: Y=A*cosine(w*t)+B*sine(w*t), where t is a point in time. Feeding successive points in time through the above equation creates a wave form not unlike the ripples in a pond formed when one drops a stone in the water.

The present analysis attempts to obtain an a posteriori view of the extent to which such ripples exist in the data, but is not limited to cyclic description. The identification of cycles was, however, the first step of the analysis. In this the procedures of Watt (1979, p. 15-17) were followed using Watt's computer program. Once periodicities in the data were identified, they were built into variables and the second stage, which regressed the area and editor interest variables on both the cycles and the year and total articles per year variables. The regression was stepwise, entering only those variables which added significantly to the prediction. The residuals were then entered into a bivariate spectral analysis with the data (area residuals with editor interest data; editor interest residuals with area data). The interest here was first in the possibility that there were additional periodicities that could be removed from the data (several were found and added to the analysis) and second that there might be significant lag covariances between area and editor interest (none were found). The stepwise regressions were then repeated with the addition of those items found in the Spectral procedure.

The results, when viewed as a whole, are as interesting for what is missing as for what is present. The lack of cycles of lengths between twenty-four and thirty-six years is potentially a trivial result of the nature of spectral analysis, but not finding a single periodicity in the fifteen analyses in the range from twelve to sixteen years should be viewed as an important finding. Issues in Communication do not seem to cycle within that range.

Longer periods, in the range of twenty years (17 to 23 years with what appears to be a valid normal distribution) are common, however, particularly for the periods of editor interest. Five of the seven editor interest variables are characterized by a cycle that falls within this range, editor interests in Rhetoric and Communication Theory (table nine and figure one), Public Address (table eleven and figure three), Speech and Hearing Science (table thirteen and figure five), Instructional Development (table fifteen, figure seven), and Small Group and Interpersonal Communication (table nineteen, figure eleven). Six, moreover, of the eight area variables contain, either directly or indirectly, a periodicity within this range. Three, Public Address (table ten, figure four), Interpersonal/Small Groups (table eighteen, figure twelve), and Theatre (table twenty-two, figure fifteen) chose this period directly and show it in their tables. Another three are associated with the period through their selection of the total number of articles published in any given year as a predictor. Eighty seven percent of the variance (r=.93, F=128.5 @ 2/38) in the total number of articles published each year is explained with appeal to only two components, one linear (year) and the other a 19 year cycle (see table twenty three). Thus through their association with 'total', Rhetoric and Communication theory (table eight, figure two), Speech and Hearing Science (table twelve, figure six),and Instructional Development (table fourteen, figure eight) indirectly take on this nineteen year cycle as well.

There is some great consistancy here. Only Mass Communicationand Forensics do not have an approximately one generation (twenty year) cycle associated with them. All of the other areas contain this periodicity for both the cycling of subject matter (area) and editors (editor interest). It would appear that this is a basic periodicity for interests within the field, ten years of interest, ten years of disinterest; a twenty year period for the recycling of ideas.

A second dominant set of cycles within the data is found at approximately six years (6 or 7 years). Again, four of the seven editor interest variables are represented, Public Address, Instruc tional Development, Forensics (table sixteen, figure nine),and Interpersonal/Small Groups. Generally, this cycle is broadly interpretable as saying that every other editor (editors of Communication Monographs serve three year terms) of the journal had an interest in these areas. Although the interpretation is only directly witnessable for Forensics and Instructional Development (see their respective tables), it makes sense for all four. (For those ready to make the objection, editor interests were not treated as mutually exclusive categories.) Two area variables, Public Address and Forensics (table fifteen, figure ten) also contain six year periodicities. As this result may be evidence for gatekeeping, we will return to it later.

Another set of periodicities which will be similiarly interpreted are found for three of the editor interest variables in the range of ten to eleven years (It should be noted that none of the variables contained eight or nine year cycles) indicating less frequent less frequent changes in the interests of the editors. They are Editor Interest in Rhetorical and Communication Theory, Speech and Hearing Sciences, and Interpersonal/Small Groups. The interpre tation is most valid for the former, less valid for the latter, and invalid for Speech and Hearing Science, where the cycle, on visual inspection, is clearly performing a smoothing function for an area that dominated the interests of early editors, but has held little interest to editors selected since. This kind of smoothing function is probably basic to the six year cycle of Editor Interest in Interpersonal and Small group Communication.

The last major cluster of periodicities is at roughly two years. Short cycles of this sort are difficult to interpret, but two interpretations should be considered seriously, depending upon the type of the variable. First, such cycling can be due to negative autocorrelation within the data. If this interpretation applies, and it should apply to its occurance with any Editor Interest Variable (there are three) the effect of the cycle is largely one of smoothing the edges in the data (one should recall that editor interest is dummy coded). The alternitive interpretation might be called a time lag hypothesis. Basically, this states that it takes several years for the publication an idea to multiply into new publications. If a researcher gets an idea from reading an article in one year, it may take anywhere from one to three years (and even longer) to refine, test, write up, and submit for publication ones results. Once this is done, moreover, even assuming acceptance for publication as is, it will most of a year or longer before the article is actually published.

Thus it should take at least two years for one one publication to beget another. A third interpretation, that publication in an area is sufficiently seldom that articles are only likely to appear every other year may validly explain theatre, but publication time lag seems the more likely explanation for the 2 year periods found in Rhetoric and Communication Theory and Mass Communication. Indeed, the interpretation is likely to also apply to the three and four year periodicities associated with some area variables.

Long periodicities (of 36 years or longer in this study) should not be regarde as cycles at all. The sample (41 years) is two short. None of the long periodicities found are adaquately explained by linear or simple curvilinear variables, however, so the trend lines were retained for their explanatory power. They trace a major line of interest. Indeed, over a larger sample, some might even be valid cycles.




It is possible to test for gatekeeping effects at three different points in the data. First, if a particular area of interest proved predictive of one the the 0-style factor patterns, there could be evidence for gatekeeping during the period represented. One editor interest variable does prove predictive of the period represented by the second component (table one), Editor interest in Public Address. However, the variable is negatively related to the factor (table four) and visual inspection of the data (figure two) shows that editors did not have a published interest in public address during this period. It must be concluded that gatekeeping is not evidenced in this element of the data.

The second test involves time lagged editor interest variables predicting area at distances of anywhere from zero to four or five years. The possibility of such effects was examined, it should be recalled in the bivariate spectral analysis. Although some significant cross-covariances were found (both from area to interest and from interest to area), none survived the final stepwise regression. Thus it must be concluded that no evidence for gatekeeping can be found at this level either.

The third test involves examining overlapping area and editor interest cycles. This test can provide evidence of possible causality, but cannot extablish causality because of the possibility of a third variable, the general interest of members of the field in given areas, causing both. The certainty of this correlative hypothesis is relitive, however. If the overlapping cycles are exactly the same (if they are in phase, if they rise and fall at the same time) prior causality must be assumed. But if Editor Interest precedes area by only a year or two, it can be infered that there may be a gatekeeping effect. Conversely, if area precedes Editor Interest, it can be infered that the interests of the field play a role in the selection of editors. There are nine overlapping sets of cycles within the data. Most are out of phase by enough to make the hypothesis irrelevant, but two, the six year cycles of Public Address and Forensics, are extremely close.

Editor Interest leads area by one year for both. Again, causality cannot be concluded on the basis of this result. But there is a consistancy between the results in public address and Forensics that speaks for attention to the gatekeeper hypothesis. Simply, the cycles in Public address and Forensics exactly parallel each other. Where the cycle for editor interest in Forensics rises, the cycle for editor interest in Public Address rises. One year later, the six year cycles of the areas of Public Address and Forensics both fall. Visual inspection of the editor interest data for Public Address and Forensics reveals one basis for the parallel. In no case is an editor interested in Forensics where there is not also an interest in Public Address. This easily explains the parallel in the editor interest six year cycles. One is left to wonder, however, why the two area variables exactly parallel each other. Partial explanation may be given in overlap. Articles in Forensics might also have been classified under public address. But it may be more reasonable to view Public Address and Forensics as allies who must fight other areas of the field for publishing space. When they profit, they profit together. When they suffer, they suffer together. This lends credence and some reliability to a gatekeeper hypothesis. If accepted, however, one still cannot identify the gatekeepers. Are editors with other interests keeping the Forensics and Public Address literature down or are editors with interests in Forensics and Public Address promoting their own interests. No answer can be given here and it may well be that both are true.

In any case, this much can be said. There is some evidence to support gatekeeping in Communication Monographs, at least with respect to Forensics and Public Address. Such gatekeeping can not, however, be credited with much effect. Indeed, the evidence is not strong enough for the conclusion that there is any gatekeeper effect in Communication Monographs to be warrented. There is the possibility, none-the-less, of there being some effect.

Discussion and Conclusion

I began this study in a search for pattern in the interests of the various sub-disciplines of communication. A great deal of pattern has been found. Changes in the published interests of editors seems to ebb and flow in cycles of six to eleven years. as editors change. Changes in the fields general interests seem to ebb and flow over the course of about twenty years. Interest increases. Interest declines. The old becomes new and the new is forgotten, eventually lost to five and ten year literature searches that inevitably rediscover old interests and make them current.

No judgement should be made here. The new becomes boring. The overused becomes reified, trite, and/or meaningless. After a period of disinterest, scholarship can often return to old problems with new vigor, new perspectives, and new ideas that are not rediscoverys. We are allowed, when we return to an old idea, to advance it, clarify it, build it better.

As a journal, Communication Monographs has had three distinct periods in which to observe these patterns. The early period of the journal accented Speech Science, an area of interest that has very nearly left the domain of Speech Communication to become a distinct area of study. It has not, however, left either communication or Communication Monographs. The journals second period was characterized by a fight for control, one eventually resolved in the establishment of another journal. The third period sees the ascendance of modern communication in the interest areas of Communication Theory and Interpersonal and Small Group Communication. There is pattern in this too. For although the mid-life of the journal was characterized by conflict, the conflict was among nearly kindred spirits when compared to what surrounds them on either side. Speech and Hearing Science aspires to Science. So does todays communication theory. By comparison, the middle period is quite different than either side. Theatre, public address, instructional development, and the rhetorical half of rhetoric and communication theory all aspire more strongly to art than to science.

Here we see the possibilities of yet unrecorded (or at any rate unconfirmed)cycles of forty or fifty years in which the interests of art and science ebb and flow though the field of communication. Such ebb and flow can certainly be seen in the past.

I think I have been successful, within the limits of the data, in describing where Communication Monographs, and perhaps the field with it, has been in the past. I have made no attempt to be complete in my verbal description of the data. The figures and tables at the back of the paper are more elequent in that description than I can be with this data set. The reader is encouraged, if curious, to spend time there. Nonetheless, much that is interesting, at least to me (I hope for you) has been described. Communication is unquestionably a diverse discipline, but patterns of interest do exist within that diversity. Such patterns may well exist in all areas of scholarly endeavor.

Where is communication going? If Communication Monographs three periods comprise a cycle, I suspect communication will start moving back toward art and application and away from science. But there are other signs of this movement within the field and it may well already be happening. But I also suspect that in returning to art, the field will niether forget science nor move backward. If old ideas are recycled, they will be given new and better rubber. In the end, we will have learned more about what constitutes communication.