Findings & Conclusions

{ENRON} and {TSUNAMI}
{ENRON} and {TSUNAMI} resulted in longitudinal keyword frequencies in line with the expectation for non-mythical catastrophes. At their time, each was a disaster worthy of history books: the collapse of Enron was the largest bankruptcy in American history and the “Boxing Day Tsunaimi” was the most deadly natural disaster in modern history. The New York Times gave each substantial attention during the Catastrophe phase and some attention during the Aftermath, but neither had an appreciable Latency beyond the first Memorial.

Enron had been high-profile, media-savvy, politically connected corporation prior to its catastrophic bankruptcy. Prior to revelations of its economic difficulties in November 2002, {ENRON} had a mean 16wf. This rose to 1019wf during the four-month Catastrophe, falling to an average 178wf in the one-year Aftermath, and to 34wf and 22wf in the following years. {ENRON}’s word frequencies rose again (twice past 100wf) between December 2005 and May 2006, during the criminal trial of founder Ken Lay and CEO Jeff Skilling.

{TSUNAMI} received 906 hits December 2005-January 2006 and had a slightly higher latency. The 12-month Background had 11 of 12 months with fewer than 3 hits and the Latency period witnessed 8 of 12 months with 30 hits or fewer. A more detailed content analysis is required to identify the source of this increase, but a haphazard sample of articles in July 2006 (the last spike) reveals news of a smaller tsunami, a reference to the tsunami relief efforts of former Presidents Bush and Clinton within the context of solving Israel/Palestine conflicts, and criticism of the reconstruction efforts of aid agencies leveled by President Clinton.
{HURRICANE} and {KATRINA}
In contrast to the first two examples, the drowning of New Orleans in 2006 illustrated some of the features expected of a catastrophic myth. Because a Sign was pre-assigned by authorities (like Enron) prior to the apex of the Catastrophe, there was never any doubt as to what to call Hurricane Katrina (unlike the tsunami). Unsurprisingly, there was a high degree of concordance between {HURRICANE} and {KATRINA} during the Hurricane Katrina catastrophe and in subsequent newspaper articles. It was also clear, however, that the background pattern of use is disrupted after September 2005.

In the {HURRICANE} graph we saw five distinct spikes in September of 2003, 2004, 2005, and 2007 and in August of 2007. Each of these spikes in word frequency corresponded to the American Atlantic “hurricane season”. In 2003, Hurricane Isabel made news more for hitting the Washington DC metropolitan area than for her destructive power, while 2004 witnessed a convergence of Hurricane Ivan “the Terrible” and US presidential candidates campaigning in a storm-ravaged swing-state (Florida). The catastrophes in 2003 and 2004 had short Aftermaths and word frequencies quickly returned to Background levels, a pattern that was clearly broken in 2005 with Hurricane Katrina.

In both the {HURRICANE} and {KATRINA} graphs, we saw a long Aftermath and a long latency above Background levels. The 10-month Background for {HURRICANE}, between Ivan and Katrina, had a mean 57wf, which rose to 232 in the 12-month Aftermath, and 65wf and 56wf for each of the following years (excluding spikes for 2007 and 2008 hurricane seasons). The 12-month Background for {KATRINA} had a mean 0.67wf versus 130wf for the 12-month Aftermath, and a mean 56wf and 44wf for each of the following two years.

The frequency of “hurricane” remained elevated from two per day to over seven per day until the first anniversary of Hurricane Katrina, after which it returned to Background levels. The frequency of “Katrina”, on the other-hand, remained elevated from once every-other month to four per day for the Aftermath, and remained elevated at a level equal to “hurricane” for the following two years. Haphazard samples suggest that every hurricane was compared to Katrina, that Katrina was often politically invoked as a failure or challenge independently from “hurricane” (Krugman 2007). The September 2008 spike in “Katrina” was almost certainly the result of Hurricane Gustav (an otherwise unremarkable storm) coinciding with the Republican National Convention.
Terrorist Attacks of September 11th 2001
Each of the four catastrophes so far discussed varied in the details of their profiles, but collectively hold to the expected form. The initial query for the terrorist attacks of September 11th 2001 – most commonly termed “9/11” – revealed an unexpected deviation from this form.

According to the {9/11} graph, there was a long build-up to the first peak4 which did not coincide with the actual attacks. The six peaks occur in September 2002, 2003, 2005, 2006 (the anniversaries of the attacks), and between March and October 2004 (the 2004 US Presidential campaign season). While the occurrence of anniversaries and the strong Latency evident in {9/11} fits with the expectations, the slow build-up to the first peak suggests other keywords needed to be investigated.

Further queries were run for {OSAMA}, {BIN LADEN}, {AL QAEDA}, {TERRORIST ATTACK} and {WTC}, all signs commonly associated with the catastrophe. The results of these queries all fit with the general expectation of catastrophic events. The latter three terms fit with elevated Latency expected of myth-level catastrophes while the former two probably do not. {OSAMA} and {BIN LADEN} had higher background levels for (15wf and 43wf) compared to the other terms (9wf, 7wf, 5wf) and a longer study period may reveal that the first two terms were already mythic, owing perhaps to the truck bomb attack on the World Trade Center in 1993.

As with {9/11}, {OSAMA} and {BIN LADEN} saw anniversaries in 2002, 2004 and 2006. There is, however, a conspicuous absence of substantial anniversaries in 2003 and 2005. The sustained the increase in frequency during the 2004 US Presidential campaign season and peaks during anniversaries which coincide with mid-term elections suggests the terms {OSAMA} and {BIN LADEN} are related to election activity.

{WTC} and {TERRORIST ATTACK} showed expected catastrophic peaks, Aftermaths, and substantial Latencies. Recognizing that {9/11} was synonymous with the {TERRORIST ATTACK} on the {WTC}, it was worth returning to Barthe’s Mythologies: load latter two signs into the first and construct the myth {9/11}. This was graphed with a “stacked area graph” which displays the sum of the word frequencies while keeping the individual terms visually distinct.

In this graph we saw Background levels as negligible, and the moment of the catastrophe in the first peak of {WTC} and {TERRORIST ATTACK}. With the first anniversary we witnessed a peak of all three terms and the beginning of the rebranding of the event into {9/11}. By the 2004 US Presidential campaign {9/11} substantially outnumbered both of the other two terms, was clearly the favored term for the catastrophe, and suggested that the event has passed into myth.

Conclusion

The findings support the hypothesis that the explanatory power of an event would be found not in the word frequency of the catastrophe, but in the “latency” of the word frequency over long periods. The example of “9/11” demonstrates the myth-creation process described by Barthes, but warns of the misleading results that can be arrived at with inadequate attention to keyword selection. The co-incidence of periods of high-word frequency with periods of political importance (like presidential and mid-term election seasons) suggests that studies of concordance may find significance, and that similar patterns may be observable for other myths. The graph for {IRAQ} in Appendix B, for example, more closely resembles {9/11} than any other query, suggesting that multiple variables were driving word frequency. While the familiar 2004 US presidential election-season spike appeared to be present, the earlier 2002-2003 period resembles “False Statements by Month” graph (Appendix C) produced by the Center for Public Integrity (Lewis & Reading-Smith, 2008). If further studies consistently link myth-like word frequencies with presidential communication they may provide quantitative support for The Rhetorical Presidency and its thesis of a “Newsmaker-in-Chief” (Friedman, 2007).

Notes

1. Queries for “Jumped the shark” or “jump the shark” resulted in over 91,000 hits on Google Blog Search and 900 hits on a LexisNexis query of major US and world publications and newswires.
2. Queries for “epic fail” resulted in over 114,000 hits on Google Blog Search and 53 hits on a LexisNexis query of major US and world publications and newswires.
3. This exercise revealed that Hurricane Katrina is still mentioned an average of once a day even three years after the event, that 10% of those occurrences overtly reference failure. While such a concordance is likely significant, a robust statistical method would be required confidently express significance with weaker keyword pairs.
4. The term “catastrophe” is avoided here because it denotes a sudden calamity, and suddenness is not represented in the {9/11} graph.