Perceptions of Social Change

Perceptions of Social Change:  100 Years of Front Page Content

in The New York Times and The Los Angeles Times

By

Wayne A. Danielson

and

Dominic L. Lasorsa

July 28, 1994

Abstract

Perceptions of Social Change:  100 Years of Front Page Content

in The New York Times and The Los Angeles Times

This chapter deals with thematic content analysis over historical time periods -- in this case 100 years of front page news in The New York Times and The Los Angeles Times.  The computerized content analysis method chosen for the study reveals some of the major social and political changes in American society.  Trends that might go unnoticed over short time spans are clearly observable when longer periods are studied.

The chapter attempts to summarize and integrate the findings theoretically, and the predictive value of content analysis as an indicator of social change is assessed.  Specific themes addressed include the rise and fall of communism as a social threat; the perception of change in American society; the increasing importance of the executive branch of government; the ascendancy of “experts” and “professionals” in American society;  the rise of “quantification” in the social perception of reality; the general decline of the individual and the rise of the group; and the change from  agriculture to manufacturing/ information as the economic base of society.

Perceptions of Social Change:  100 Years of Front Page Content

in The New York Times and The Los Angeles Times

By

Wayne A. Danielson

and

Dominic L. Lasorsa

This paper has deep roots.  In many ways it relates most strongly to the research of Lasswell, Lerner and de Sola Pool (1952) undertaken at Stanford University in the early 1950s.  The Lasswell group was then immersed in one of the first comprehensive and integrated series of content studies of mass communication.  They were using a technique of quantitative semantics, which they labeled “content analysis.”

Lasswell and his colleagues saw human beings as living mainly in a “symbolic environment.”    Our physical environment as humans is basic, of course, but the environment that counts the most to us is the environment of words and images, the environment of meanings that surround us from our earliest moments of existence.

Lasswell and his associates believed that these symbols, although they can be frozen and studied in the short term, are most meaningful if studied over long time periods, so that trends or changes can be observed.  Lasswell, Lerner and de Sola Pool (p. 78) were interested in creating “models of symbolic behavior which will enable us to formulate and validate propositions about the mechanisms underlying the flow of symbols.”

To study the flow of symbols, Lasswell and his colleagues developed the technique they called content analysis.  “When it is desired to survey politically significant communication for any historical period on a global scale,” they wrote (p. 16), “the most practicable method is that of counting the occurrence of key symbols and cliches.  Only in this way can the overwhelming mass of material be reliably and briefly summarized.  By the charting the distributions in space and time, it is possible to show the principal contours of ... political history.”

Lasswell, Lerner and de Sola Pool made a concerted effort to map symbolic environments and to study changes in these environments.  As director of research for the Hoover Institute at Stanford, Lasswell oversaw a series of studies under the auspices of a major institute research program entitled Revolution and the Development of International Relations (the RADIR project).  The methods they used to study the flow of symbols are not unlike those we use today.  They sampled text over time.  They established the reliability and validity of their coding techniques.  They counted the occurrence of key symbols in political documents and in the press.  They counted what they called themes (symbolic condensations of textual units).  They applied statistical techniques of histogram and contingency  table and correlation coefficient to help describe their findings.

Lasswell and his RADIR associates came to believe that newspapers were an especially apt source of data about the symbolic environment of most people.  They (p.17) wrote:  “Newspapers appear regularly and frequently, in uniform formats.  Also, they have a more or less explicit point of view.  The press is mainly an information medium rather than an entertainment medium; and the most significant category on which the press regularly presents news and views is the political, including the ideological.  As compared with such verbal flows as after-dinner speeches, golf-club stories, psychiatric interviews -- all of which provide data symbols of great value -- the press is both accessible and rich in the vocabulary of political ideology current among the elite of any given time.  For these reasons the RADIR symbol studies concentrated on the press.”  Noting that “politically significant symbols are usually concentrated in the front page or  editorial page,”  Lasswell (1942, p. 14) suggested that analysts studying such symbols might limit their attention to such pages, depending on their particular objectives.

Lastly, Lasswell noted that the coder who describes a given symbol needs to consider what amount of context is necessary for a symbol to be properly understood, and it is important to note that bigger is not necessarily better.  Meaningful symbols often are made up of a single word or a simple phrase consisting of two or three words.  An idea, which often is made up merely of the connection between two or more symbols, most often is embodied in a single sentence.  Lasswell recognized that a map of the symbolic environment requires a judicious sampling of symbols, and he realized that one normally should need no more context than a sentence to understand the meaning of its words.  “In general,” (p. 18) he noted, “ the analyst is directed to limit his attention to the sentence, coding it separately without reference to preceding or succeeding sentences.”  It should be kept in mind that Lasswell was suggesting criteria for the study of a prominent and enduring symbolic environment -- the front page of the daily newspaper.  While some other content analyses may require larger (or smaller) contexts to maintain what Weber (1990) calls semantic coherence, one should consider carefully the costs and benefits associated with this important choice.  Choosing a smaller context than necessary is likely to result in symbols improperly coded because of ambiguity or other semantic problems.  The solution, however, is often not to err in the other direction.  Casting a larger semantic net may help the coder do a better job of capturing the meaning of symbols but choosing a larger context than necessary is likely to result  in cuts being made elsewhere in the study, such as in the size of the sample or in the complexity of the coding scheme.  These tradeoffs should be considered carefully before a decision is made regarding the size of the context needed to code symbols accurately.

Using this approach for analyzing large sets of symbols, the Lasswell group was able to study an array of interesting topics from xenophobia in five major countries over the years 1890-1950 (Pool, 1951) to appeals to the masses in The New York Times from 1900-1949 (Pool, 1952).  The work was difficult and wasdone almost entirely by hand.  Only at the end of their complex and meticulous operations where the symbols transformed into numerical data on punched cards that could be run through a counter-sorter.  Reading the Lasswell group’s work today, one cannot help but be impressed not only by what they invented but by the audacity and energy with which they approached their task.  To test even a simple hypothesis often took weeks or even months of labor.  In one study, 105,004 symbols were examined.  As Lasswell, Lerner and de Sola Pool (p. 53) noted, “It took over a year to analyze the data.”  In some studies, they (p. 63) said, “Without machine tabulation, analysis would have been impossible.”

A decade after the Stanford group described their method for analyzing symbolic environments and other large text sets, a group of scholars at Harvard University began trying to use electronic helpers not just in the latter statistical analysis stages of the process, but in the earlier coding stages as well.  These scholars were able to employ  the newly invented computer to analyze linguistic information found in texts.  Thus, the symbols themselves -- as electronically coded -- had become data, and content analysts found that throughout the entire content analysis process they could enlist the aid of a new, fast and tireless coder that never made mistakes. (Stone, Dunphy, Smith and Ogilvie, 1966).

Since the pioneering work at Harvard, further developments in computer s and computer programs have helped automate the text analysis process to such an extant that today it is not just possible but practical to analyze massive texts.  The authors of the studies reported in this chapter used a computer software package written by Danielson (1988) that searches electronically stored text files for the occurrence of words and phrases contained in electronic dictionaries. These dictionaries are similar in construction to those made by Lasswell, Lerner and de Sola Pool in the 1950s.  The results obtained are similar as well, as are the forms of statistical analysis and display empl|oyed.  What is different is that the studies that took the Stanford group months to perform can now be completed in days.   And an error that might have meant weeks or even months of reanalysis can now be corrected easily and the study rerun in hours or even minutes.

In the studies reported in this chapter, the authors have attempted to document social changes in the United States over a century by examining prominent and long term flows of symbols.  By sampling the symbolic environment judiciously, they hoped to discover particular patterns of change.  The front pages of two prestige American newspapers, The New York Times and The Los Angeles Times, were examined over 100 years.  These papers were selected for their size, longevity and influence.  The New  York Times was founded in 1851, and The Los Angeles Times was founded in 1882.  By the 1980s, the daily circulations of both papers surpassed 1 million, and the papers were highly regarded as “newspapers of record” in the United States.

A computer sampling program was used to sample text in the two papers for the period 1890 to 1989.  Specifically, the program selected days at random during the year and 10 sentences at random from the front page of that day’s edition of each paper.  Sentences were typed into the computer by hand. The resulting data set was a multi-stage cluster sample of 300 sentences per decade per newspaper, or 3,000 sentences for each paper, 6,000 sentences in all.  With sentences averaging 25 words in length, the text includes about 150,000 words.

FINDINGS

Initially, we decided to  look at a study of the occurrence of a simple set of words in the text sampled.  These words were “Communist,”  “Communists,” and “Communism.”  If we plot the occurrence of these words over the last 100 years, we get a picture of the rise and fall of attention to Com—munism in the symbolic environment of the  United States as reflected on the front pages of two leading U.S. newspapers.  As Figure 1 shows, the symbolic flow of these terms rises and falls across the decades.  The peak occurs in the 1950s, a time of great concern in the United States over domestic and international Communism.   Notice how interest ebbs toward the end of the century, however.  Even during the Vietnam War, the symbolic presence of Communism on the front pages was on the decline.  Communism as a strong symbolic factor was already dying at a time when some American politicians were attempting to arouse concerns about a “domino” effect in Southeast Asia.

*****FIGURE 1 ABOUT HERE *****

For those readers who lived through significant parts of the “Communist” era, the graph probably has a kind of face validity.  Many will look at it an say, “Yes, that’s the way it was,” or “That’s the way I remember it.”  The similarity of symbol flows in sentences taken on different dates in two widely separated newspapers also argues for the validity of the technique.  Other interpretations are possible, of course, but the close tracking of symbol frequencies in the two newspapers supports the notion that both were observing and reporting the same social phenomenon.

*****FIGURE 2 ABOUT HERE *****

Figure 2 deals with attention given in the newspapers studied to terms related to agriculture.  In addition to showing the gradual decline in the frequency of agricultural terms in The New York Times and The Los Angeles Times, the graph also shows national figures for the percent of the work force engaged in agriculture.  The overall tracking of the external occupational measure with the symbolic measure offers additional evidence that symbol flows in the press can and do index some of the social changes taking place in U.S. society over broad stretches of time.  Both symbolically and in fact,  the United States in the time period studied was changing from an agricultural society to an industrial and information-based society.

These initial attempts to assess the validity of the method used in this chapter were encouraging.  Beginning with Figure 3, however, our studies began to  turns toward symbolic representations of social change that cannot be checked so directly from memory or from government statistics.

*****FIGURE 3 ABOUT HERE  *****

Figure 3 tracks an interesting phenomenon  involving  the disappearance of people from the front page.  This trend was first noted by White (1989) in her study of male and female references in The New York Times and The Los Angeles Times.  White  wanted to document the well-established notion that more men than women appear in the news, and she wanted to see whether references to women were on the increase in recent decades.  Although the expected dominance of male references appeared as expected, the predicted gradual increase in female  references did not seem to occur.  Instead, both male and female references appeared to decline with the passage of time.  Ghanem (1990) repeated the White study with a somewhat more inclusive dictionary of what she called “human element” terms.  The symbols in this dictionary referred both to men and women and various relational  words such as “husband,”  “wife,” “son,” and “daughter-in-law.”  Confirming White’s findings, she found that such terms declined both in The New York Times and in The Los Angeles Times over most of the decades studied.

***** FIGURE 4 ABOUT HERE *****

The disappearance of references to individuals from the flow of front-page symbols may be partly due to a growing tendency of the media  to observe society  not from the point of view of single persons, but from the point of view of groups or  “publics” or “masses”  or other collective entities such as “consumers” or “viewers” or “voters.”  “Don’t treat me as a statistic,” we often hear people saying, but that may be precisely what is happening in front page depictions of American society.   Ghanem took the notion of depersonalization a step further.  She hypothesized that the decline in references to people as individuals over time would be paralleled by an increase in references to numbers generally.  As Figure 4 shows, her hypothesis seems to be supported.  The frequency of occurrence of what she called “quantification words” appears to be an increasing function both in The New York Times and in The Los Angeles Times.  Do the contrasting patterns shown in Figures 3 and 4 reflect a social world in which people as individuals are becoming less important and numbers more important?

***** FIGURE 5 AND 6 ABOUT HERE *****

While the definitive answer to that question remains to be discovered,  Figure 5 seems to show that the people who are referred to on the front pages of the papers studied are increasingly people of power and influence.  Figure 5 shows an increase in the use of terms that refer to experts and professionals, such as accountants and lawyers, nurses and doctors.  Figure 6 shows an increase in the use of references to officials, such as elected office holders and government appointees.  Barker-Plummer (1988) found an almost linear increase over 100 years in verbs of attribution such as  “said” (by far the most common), “stated” and “reported.”  She thought this indicated an increasing tendency of newspapers to rely for news on officials, particularly government officials, and to quote them as to the meaning of what was happening.

***** FIGURE 7 ABOUT HERE *****

Closely related to the increase of references to experts and professionals, is an increase in references to the president and the presidency over the century.  In a similar study,  O’Donnell (1989) found an increase in regulatory terms related to the federal administrative bureaucracy, and Willats (1990) found an increase in terms related to the national government in general, while terms related to local government remained stationary or declined slightly.  The sociopolitical world portrayed symbolically in all these studies is one in which attention increasingly seems to center on central government, particularly the administrative b ranch, and particularly the occupant of a distant, almost mythical White House.  Depictions of practical, day-to-day political activities on the local level are increasingly rare.

***** FIGURE 8 ABOUT HERE *****

Johnson (1989) studied the perception of change itself.  Figure 8 depicts the relative frequency over time of words such as “new”, “different,” and “innovative.”  The figure shows change words are on the increase in both newspapers.  Thus, the common perception that we live in a “speeded up” world is supported in the symbolic world of the press.  The comforting, stable world of the past is gone.  Increasingly, at least on the front pages of the two prestige papers studied, we can expect the unexpected.

***** FIGURE 9 ABOUT HERE *****

Finally, Figure 9 shows changes in the two papers regarding the distribution of words and phrases referring to religion and religious rituals.  These terms include words such as “Christmas,” “funeral,” “marriage,” and “synagogue.”  As the figure shows, a steady decline in such references has occurred over the century.  The world depicted in these front pages is one increasingly devoid of people, places and things religious.  Rites, feasts, and other public displays once held to be sacred and central to the focus of the community have declined steadily, some transformed into secular rituals, others all but disappearing from the symbolic environment.

DISCUSSION AND CONCLUSIONS

As Lasswell, Lerner and de Sola Pool would have expected, the flow of symbols through time does seem to reflect real social change.  The flow of front-page symbols depicts a decrease in emphasis on the individual and an increase in emphasis upon the group.  It shows an increasing reliance upon numbers to describe social reality.  It shows power shifts from local to central government, particularly to the chief executive and to the regulatory bureaucracy.  It increasingly relies on  “experts” and “officials” (often anonymous) to tell us what is happening.  It sometimes seems indeed that nothing happens until someone in authority “says” that it does.  It depicts a world in which change, often bewildering change, has become a permanent feature.   And finally it depicts a world in which deeply held values, such as those associated with religious beliefs, are on the decline.  It suggestions a world, in short that seems to be moving from what Tonnies (1988) called gemeinschaft, a community pf persons closely knit by strong sentiments based on kinship and spirit, to gesellschaft, a secular society loosely bound by impersonal interactions based on formal contracts and social functions.

The gemeinschaft-gesellschaft formulation is theoretically rich, and it has provided one of the most enduring interpretations of the direction and character of social change in the modern world.  If our symbolic studies track, as we believe they do, some of the major social changes of the last century, Tonnies’s century-old theory seems to us to provide one of the most descriptive and integrative interpretations of the data we have observed.

We have assumed here that news appearing on the front pages of two major U.S. newspapers represents an important aspect of the symbolic environment of Americans living at the time of publication, and that the symbols used may serve as an index of public attention.  For most of the 19th and 20th centuries, the daily newspaper was the premier method for distributing news, but in the last few years that has gradually changed.  Today, the mass communication channel with even greater penetration and influence is network television news.  If Tonnies were alive now, he probably would have recognized television, too, as providing an important representation of the symbolic environment.  Were we not interested here in the symbolic environment  for many years before the introduction of television, that medium might have been another excellent source of information about the flow of symbols in society. 

It is also important to note that even if we accept the daily newspaper as a convenient repository of socially relevant symbols,  newspapers have themselves changed over the last 100 years.  Some changes in front-page news, therefore, may not be due to events directly related to the symbolic environment.  For example, newspapers have moved toward a more sectionalized or departmentalized approach to the news.  Thus, “medical sections” or “science sections” are popular today, and a story that might have appeared on the front page in times past today appears on the medical pages.  An observed decline in front-page medical terms, may not represent a decline in public attention to medicine; it may in fact mean the opposite: medical news is now so important that it gets its own section.

Similarly, the decline in religious ritual words in The New York Times and The Los Angeles Times noted in this chapter may be due in part to a more general and subtle change in the attitudes of editors and reporters.  Over the years, both papers have become major players in American politics. Occasionally, their owners and employees have seemed to act not as observers of government but almost as if they considered themselves to be part of government.  Thus the difference in reporting about religion may in a way be attributable to papers imitating government and deliberately trying to avoid an undesirable mixing of church and state.   As long as we keep such limitations in mind and recognize that some content changes may be due to developments unrelated to public attention, the symbols appearing on the front pages of these newspapers appear to serve us quite well in describing the symbolic environment Americans experience in given eras.

Many hypotheses remain to be tested.  Studies of changes in the “key symbols and cliches” of other societies would allow patterns of social change to be compared and contrasted around the globe.  For example, it might be possible to determine whether the United States has become, as Tonnies (p. 221) predicted nearly a century ago, “the most modern and gesellschaft-like state.”  In addition, studies of the symbolic environment using contemporary versions of Lasswell’s venerable method, may give us additional insights into such grand sociological theories as those of Marx, Durkheim, Parsons and their intellectual descendants.

With the completion of more than a dozen computer-aided textual studies so far, we are encouraged to think that we may be coming ever closer to the goal envisaged by Lasswell and his colleagues of creating “models of symbolic behavior which will enable us to formulate and validate propositions about the mechanisms underlying the flow of symbols.”  As more general and user-friendly  content analysis programs emerge, and  as more texts are stored electronically and made available, researchers undoubtedly will continue to find ingenious and ambitious uses for such programs and data.  The studies reported in this chapter have tried to show that just as scientific tools have helped us describe and explain the physical world we inhabit, they now as well can help us describe and explain the symbolic world in which we live.

REFERENCES

Bernadette Barker-Plummer, “Look Who’s Talking:  Trends in Attribution and Sources in Two Elite Newspapers,”  Southwestern Mass Communication Journal 4:55-62 (1988).

Wayne Danielson, User Manual, General Content Analyzer,  Version 1.1 (Austin, TX: Wayne Danielson, 1988).

Salma Ghanem, “The Change in the Reporting Style of Newspapers,” University of Texas, Department of Journalism Seminar Paper, 1990.

Doug Johnson, “Examination of Change-Related Words Over Time, “ University of Texas, Department of Journalism Seminar Paper, 1989.

Harold D. Lasswell, “The Politically Significant Content of the Press: Coding Procedur˙es, “  Journalism Quarterly 19(1): 12-23 (March 1942).

Harold D. Lasswell, Daniel Lerner and Ithiel de Sola Pool, The Comparative Study of Symbols (Stanford: Stanford University Press, 1952).

Jayne O’Donnell, “Public Administration Values as Portrayed in The New York Times and The Los Angeles Times, 1886-1986,”  University of Texas, Department of Journalism Seminar Paper, 1989.

Ithiel de Sola Pool, Symbols of Internationalism (Stanford: Stanford University Press, 1951).

Ithiel de Sola Pool, Symbols of Democracy (Stanford: Stanford University Press, 1952).

Philip J. Stone, D.C. Dunphy, M.S. Smith and D.M. Ogilvie, The General Inquirer: A  Computer Approach to Content Analysis (Cambridge: MIT Press, 1966).

Ferdinand Tonnies, Gemeinschaft und Gesellschaft (Community and Society), translated by Charles P. Loomis (New Brunswick: Transaction Books, 1988).

Robert Philip Weber, Basic Content Analysis, second edition. (Newbury Park, CA: Sage, 1990).

Alice White, “Female and Male References on Front Page Samples of The Los Angeles Times and The New York Times, 1900-1987.”  University of Texas, Department of Journalism Seminar Paper, 1989.

Bryan D. Willats, “The Growth of National Government: A  Content Analysis,”  University of Texas, Department of  Journalism Seminar Paper, 1990.

The first author is a professor and the second is an associate professor in the Department of Journalism, University of Texas, Austin, TX 78712.  They wish to acknowledge the help of the students in their undergraduate theory classes and their graduate seminars who helped collect and analyze data in this chapter.

For more information about the computer software package or the data used in the studies reported in this chapter, please contact the authors.

Interestingly, Tonnies would have agreed with the Lasswell group that newspapers are an especially apt source of data about the symbolic environment.  Information about the nature of human relationships, Tonnies (p. 221) said, “is prepared and offered to our generation in the most perfect manner by the newspapers, which make possible the quickest production, multiplication and distribution of facts and thoughts, just as the hotel kitchen provides food and drink in every conceivable form and quantity.