[One of only two CDA papers I ever wrote. This is the first one, it's about 3 years old. I cut all the graphs and most of the technical stuff from it and reworked some parts. I hope it's readable. I kept the "(graph x)" as a reference, but there is IMO no need to include the graphs themselves as their content is revealed in the text, I think. I will post two of the appendices as well as the bibliography as "comments" to this post, so they will not clog up blog space. You don't need to read them. But by all means, read the essay. It's kinda interesting, I think, as far as Linguistic essays go.]
In the second half of the 20th century, Halliday (1978) and Fowler et al. (1979) uncovered the ideologies hidden in newspapers that supposedly strove for objectivity. Correcting the misapprehension of journalistic objectivity has been the subject of Critical Discourse Analysis (henceforth CDA) research, which showed that news is a social construct and that different representations of the same issue through distinctive expressions carry different ideologies (cf. Fowler 1991: 10).
Lately, new focus has been placed on the process of framing, whereby “journalists get to pick and choose from the frames offered by the political elites” (Callaghan/Schnell 2005: 11), meaning that the selection of facts by the jounalists is restricted by politicians and their staff. Reading about highly controversial news items, one has to wonder, though, whether the newspapers are really that powerless and whether they do not “generate their own frames” (11), a question that becomes all the more urgent considering the possibility “that most Americans, on most issues, do not really possess opinions” (Kinder/Nelson 2005: 117).
This would give newspapers enormous power: shaping the public’s opinion to their liking. This invites questions: how do newspapers treat controversial news? Do they have their own agenda? What hand do they play? Do they try to apply their influence? Do newspapers differ in the frames and ideologies that they spread? In review the reasearchers found that the particular subject of different yet similar frames was never properly treated in CDA research and consequently went for an analysis in that specific area of research.
The particular news topic that is used as a vehicle for this paper’s analysis is the case of the late Terri Schiavo, a brain-damaged woman from Florida (cf. Appendix 1). On March 18, 2005, her husband had her hospice withdraw her feeding tube, which right to do so he won after years of legal disputes with her parents. On March 31, she died from dehydration. In the two weeks between the withdrawal of the feeding tube and her death, a conflict was waged in American politics and American courts, as her parents, religious leaders, most of the Republican Party, Florida governor Jeb Bush, but also a select few Democrats, tried to have the tube reinserted but ultimately failed because of the insistence of both state and federal courts (cf. Appendix 2).
In the struggle for opinion leadership all analysed newspapers, as this paper will demonstrate, pursued their own agendas. In order to show this the researchers opted for a comparison of four of the largest daily American newspapers, the Washington Times, the Washington Post, the New York Times and the Boston Globe. This analysis is centered around actor analysis as explicated by Van Leeuwen (1996) and analysis of lexical items as undertaken by van Dijk (1988a) and Fowler and Kress (1979). Also, in the discussion of the results we will be drawing heavily from the discussion about media bias in American Jounalism as expounded in the works of Kuypers, Alterman and Brock.
The basis for this analysis is obtained in van Dijk’s notion about the relation of discourse structures, such as lexical items and grammatical structures, to societal structures, which, van Dijk argues, is to be found in social actors and their minds (cf. Garrett and Bell 1998: 6ff.). News, van Dijk says, is “a particular form of social [...] practice” (Van Dijk 1988a: 176).
The researchers have decided on treating ‘news’ not as a homogenous force, not as one social actor, but as a set of actors who compete with each other for opinion leadership. Instead, it was agreed on treating the readers as a homogenous mass, which might be influenced either way, especially since even the most clearly opinionated paper in our selection, The Washington Times, has repeatedly been classed a “relevant read” (Freedman 1995), read by both sides of the political spectrum.
Three of the newspapers in this paper’s selection, however, have been severely criticized for being biased and suppressing conservative opinions, a criticism that has been so pervasive in American discussions, that “[t]oday people pronounce ‘media bias’ as if it were a single word” (Nunberg 2003). Especially conservative media reviewers such as L. Brent Bozell claim that the American press “sets itself up as an independend advocate for particular political ends” (Kuypers 2002: 12), “presents a consistent left-wing point of view among news stories [and] shuts out information that contradicts its point of view, thus acting to limit information” (235f.) This alleged bias is commonly called ‘liberal bias’.
Others, most prominently Eric Alterman, say that these claims present a “myth [that] empowers conservatives to control debate in the United States” (Alterman 2003: 3). It is further asserted that “the most important sectors of the political media [...] are powerful propaganda organs of the Republican Party.” (Brock 2004: 2)
Proof, either way, has been scarce. Media bias research (where it is not just anecdotal, as, for instance the work of Alterman (2003) and Conason (2003) is) as undertaken for example by Groseclose and Milyo (2003) has mainly been trying to use mathematical formulas to prove assertions of bias and has thoroughly been criticized for both method and execution (cf. Nunberg 2004). Linguistic studies are conspicuously absent.
In assembling the corpus for this analysis the researchers decided on four important daily nationwide American newspapers: the Washington Times, the Washington Post, the New York Times and the Boston Globe. It was taken care to use newspapers that were all located on the east coast, close to each other, so that they could be said to compete for opinion leadership locally as well as nationwide. Thus, of the major nationwide daily newspapers, only the Christian Science Monitor and the Los Angeles Times are missing.
In each newspaper the one article was taken wherein track is kept of new developments in the Schiavo case, most of which articles wherelocated on the first two pages. All newspapers also carried stories about certain actors such as the Schiavo family or about certain topics such as euthanasia but those were ignored, to keep the corpus focused upon the news.
Finally, the frame of time the articles were picked from begins with March 18, the day that Terri Schiavo’s feeding tube was withdrawn and ends with March 31, the day that she died. These dates were selected because that was the time when the struggle for Terri Schiavo’s life was the most heated. On the first day, her death seemed imminent and on the last day, she died.
It was tried to adjust the amount of articles so that the total amount of words per newspaper was comparable. This explains certain gaps between the dates, which was inevitable, as the quantity of coverage of the events was unevenly spread. This of course presents a weakness of this paper’s analysis, but the researchers are confident that the analysis was performed systematically enough to make up and still be convincing.
As most of the studies pertaining to this paper’s topic examine a very small corpus, such as one (Kitis and Milapides 1997) or two (Choi 2002) articles or just the headlines (Schrabback 2000) of the articles, the researchers could not do an analysis as thorough as theirs on each article.
Instead it was deemed preferable to focus on a few choice aspects, such as the treatment of the person of Terri Schiavo with the linguistic strategies of nomination/categorization, collocation and transitivity, the removal of the feeding tube with strategies of nominalization, transitivity and lexical choice and, as a final aspect, the rhetoric structure of each newspaper.
Each newspaper reflected the debate rhetorically by quoting and citing experts, politicians and various parts of the involves families, the Schindlers and the Schiavos. In doing so, an argumentative pattern noticeably evolved in each newspaper, marked by the words ‘but’ and ‘yet’, which had the advantage of being countable. The occurences of ‘but’ and ‘yet’ were split in three groups. The first presented arguments and circumstances for Michael Schiavo’s and/or the position of the Democratic Party, such as
(1) House Speaker J. Dennis Hastert (R-Ill.) and House Majority Leader Tom DeLay (R-Tex.) said in a statement: “It’s unconscionable that Senate Democrats . . . would not allow a vote to move forward on critical legislation the House passed last night to save and protect Terri Schiavo’s life.” But key GOP senators [...] defended their democratic collegues [...] (Wash. Post, 18-03-05)
the second presented arguments and circumstances for the Schindler’s and/or the position of the Republican Party, such as
(2) Rep. Barney Frank, Massachusetts Democrat, said the caption for the House floor debate should be “We’re not doctors. We just play them on C-SPAN.” He said that “ideology is driving this” legislation. But Mr. Bush came back to Washington last night from his ranch in Texas expressly to sign the measure into law immediately upon the House giving its approval. (Wash. Times, 21-03-05)
the third presented neutral occurrences, such as
(3) But House members adjourned before the bills could be reconciled. (Wash. Post, 18-03-05)
Categorization means the representation of social actors “in terms of identities or functions they share with others” (Van Leeuwen 1996: 52), whereas nomination simply refers to the naming of social actors. Terri Schiavo is not always referred to by both her names, first and last. She is referred to as ‘Terri’, ‘Terri Schiavo’, ‘Ms. Schiavo’ and ‘Mrs. Schiavo’. It should be noted that each newspaper uses either ‘Ms. Schiavo’ or ‘Mrs. Schiavo’. There is no overlapping. Also, it should be said that all seven mentions of ‘Mrs. Schiavo’ in the Boston Globe occur within quotations. In all of the analysed newspapers, Ms. Terri Schiavo, apart from being referred to by her first and/or last name, is talked about also as ‘wife’ (of Michael Schiavo ) or as ‘daughter’ (of Robert and Mary Schindler). It is worth mentioning that nomination and categorisation may occur together, the two categories of referring to Terri Schiavo are not mutually exclusive, but the combination of the two categories provided no useful patterns whatsoever. Also, not all references to Terri Schiavo were collected thus, as she is being talked about in still more ways, which, however, were found to be of little interest.
On surveying (table 1) there are two distinctive numbers that are fascinating: the number of times the NY Times and the Washington Times referred to Terri Schiavo by ‘Ms. Schiavo’ and ‘Mrs. Schiavo’ respectively. The exclusiveness of use of one or the other form may be attributable to house style, which means that might be customary in the NY Times to use ‘Ms.’ instead of ‘Mrs.’ at all times, a form that, interestingly, “can be used when you do not want to state whether she is married or not” (Wehmeier 2000).
The Washington Times, on the other hand, seems to be particularly appealed by the married status of Terri Schiavo, using such a unique form in such disproportionate amounts. Where the NY Time’s interest lies, is hard to ascertain from just that one form. Certainly, the way it uses Terri’s first name far less than all the other newspapers and the formal adress far mor often , seems to indicate a kind of distancing.
Not distancing at all seems to be the Globe’s attitude, as he uses Terri’s first name by far the most times, just as he heads the list in (table 2) when it comes to referring to her as the Schinder’s daughter, a category which the Washington Times mentioned the fewest times.
This is another attention-grabbing table as far as the Washington Times is concerned, for whereas the Times holds back with the category of ‘daughter’, it can boast the most occurrances by far of the category of ‘wife’. This is consistent with (table 1), which already suggested an effort by the Washington Times to link Terri Schiavo to her marriage. How striking this attitude is becomes clearer once it is noted in how balanced an quantity the two categories are mentioned in the other three papers, which divide Terri Schiavo evenly between those two roles.
Collocation may be defined as “the habitual co-occurrence of individual lexical items” (Abdel-Hafiz 110) such as, in this case: “the brain-damaged woman”. Careful study of the selected newspapers revealed that certain collocations were occurring persistently in the vicinity of references to Terri Schiavo, the most remarkable ones being the words ‘brain-damaged’, ‘helpless’ and ‘incapacitated’, as in the following sentences:
(4) Terri Schiavo, who has been incapacitated since suffering a temporary heart
stoppage in 1990, has had her feeding tube removed twice previously, including
for a six-day stretch in 2003. (Boston Globe, 22-03-05)
(5) ”This decision is at odds with both the clear intent of Congress and the
constitutional rights of a helpless young woman,” DeLay said. (Boston Globe,
(6) The Supreme Court rejected an appeal, and a Florida judge ruled that the
governor lacks the authority to order the state to take custody of the severely
brain-damaged woman. (Boston Globe, 25-03-05)
Having said that Terri Schiavo in the analyzed newspapers is divided evenly between being ‘daughter’ and being ‘wife’, the data in (table 3) insinuates that through collocation, i.e. linking of Terri Schiavo with words emphasizing her passivity, she is to be given a subordinate position, she is to remain in that pattern, being somebody’s wife, somebody’s daughter. The most appealing fact about this data is the consistency, as this time the rifts between the newspapers disappear or become negligeable, it seems that the newspapers are all of one mind in this respect.
Transitivity is one of “the deeper semantic features of a text” (Fowler and Kress 198), highlighting actions (predicates) and both those who act (agents) and those who are being acted upon (patients). Analysis of transitivity enables the analyst to take into account syntactical transformations such as passivation and nominalization, the latter of which refers to the conversion of verbs to nouns.
As is to be expected of a woman in a persistent vegetative state, in most constructions where Terri Schiavo is involved, she occupies the role of patient. Still, there are a few sentences, wherein she is activated. Appendix 4 contains a list which includes all essential variations; it does not, however, include multiple mentions of the same or a similar construction, even if they happen to be in different newspapers; these multiple mentions are not included because it is of no importance which phrase appeared in which newspaper how often, as no pattern emerged.
This is not the only aspect where the newspapers seem to have a similar agenda. The sentences, of which examples are collected in Appendix 4, are equally distributed over all four newspapers. They share one characteristic: except for sentence (8), the activation is unidirectionally taking place in phrases that detail Terri Schiavo’s will to live. Examples of sentence (8) occur once in each newspaper, which obviously puts the sentences detailing Terri’s possible will to die in the minority, not even proponents of euthanasia were cited anywhere, thus neatly fitting the pattern established by (table 2) and (table 3), a pattern which could be called “selektive Diskursrepräsentation” (Pollak 2001: 151).
This next part of the analysis ressembles the last part closely in that both are centered around Terri Schiavo’s death, although, this time around, Terri Schiavo is no direct participant as all the sentences and phrases analysed for this part focus on the circumstances of the removal of Terri Schiavo’s feeding tube. The sentences that are analyzed are all concerned with the removal of the feeding tube. Agents are Michael Schiavo, on whose behalf it was removed, the judiciary, who granted the removal and refused the appeals and finally the doctors, who manually removed the tube. All mentions of the judiciary in the agent position, such as ‘Judge Greer’, ‘Judge Whittemore’, ‘the court’ etc., were merged into one item for clarifying purposes. Nonetheless it is not without significance, how exactly the judiciary was referred to.
At first glance, (table 4) does not appear to be fraught with interesting data, as most of the differences seem slight enough to be attributable to either the choice of articles or the variations in the quantity of analyzed material. The only three assertions that can safely be made from this data are, for one thing, the complete omission of the doctors both in the NY Times and in the Washington Times.
For another thing, the Washington Times mentions more often agents than the other newspapers, a difference that is of course partly originates in the higher number of sentences or phrases concerned with the removal of the tube. Another reason for this difference will be explained in the next section.
The third point to be made about the data in (table 4) is that both the Boston Globe and the Washington Post seem to be trying to balance the agents responsible for removing Terri Schiavo’s feeding tube fairly, a point which will be elaborated, too, in later paragraphs.
Meanwhile, the higher amount of agents in the Washington Times and, thus, of blame that can be apportioned, is striking, all the more so when (graph 1) is taken into account. Whereas the other newspapers do not prefer one judge over the other, the Washington Times’ preferences are clear. If blame is to be apportioned, (graph 1) shows, to whom it is due, at least to which judge.
After having done a transitivity analysis on the sentences in question, it was noticed that the strategy of nominalization recurred frequently, with the verb ‘(to) remove’ as object of the transformation.
Not the apportioning of blame but instead the averting of said blame is a possible effect of nominalization, according to Fowler and Kress, a result of “the rendering of a process as an object” (1979: 208). In the case of ‘removal’, the data is conclusive, as the differences between the occurrences are considerable.
It is hardly surprising that the Washington Times, who has been made out as wanting to lay blame on Michael Schiavo and the Judiciary’s doorstep, does not employ nominalization, as this would obviously be contradictory. Neither is it unexpected that the NY Times, who has a tendency to be formal, as (table 1) showed as well as a tendency to be balanced, as both (table 2) and (table 4) indicated, would sidestep the temptation to blame someone straightaway when the possiblity to be more formal and balanced via nominalization exists.
It is worthwhile, however, to discuss the data concerning the Globe. There is a consistency, too, even if it is not that apparent. It should not be startling that the newspaper who called Terri Schiavo most often by her first name, something that suggests a more intimate relationship, would not shy away from treating the process that would lead to her death, like a regular process, with agents and a patient, all the while staying balanced as far as the agents are concerned; this (table 4) and (Graph 1) demonstrated convincingly.
The depiction of the removal of the feeding tube did not only yield insights into stategies such as nominalization and agent analysis, it also demonstrated significant differences in the choice of lexical items to describe the process and its consequences. Of special interest proved to be the verb ‘(to) starve’ and the noun ‘starvation’.
With most of the newspapers settled into their characteristic profiles, the data collected in (table 6) does not shake things up, but instead fits into the data that has already been collected. Especially the Washington Times’ profile keeps sharpening. Not only is the use of the verb ‘(to) starve’ in collocations such as “starve her to death” in sentences (18), (19), (20) and (23), severely misleading, as Terri Schiavo, as physicians kept saying all along, died not of starvation but of dehydration. Also, sentences as (18) not only blame Michael Schiavo, but it could almost be said that these sentences attack him or his integrity. Writing that a judge enabled someone to starve his wife to death is an argument in itself against both the judge and the person doing the starving without even taking the pain to actually argue rhetorically, a technique, to which the words “enhanced persuasion” (Van Dijk 1988b: 16) fit perfectly.
Without a doubt, arguments do not need to be formally argument-like, as the Washington Time’s use of ‘(to) starve’ formidably demonstrated. Nonetheless, argumentation is part of any news text and it works on many levels. On the most basic level, through the use of markers such as ‘but’ and ‘yet’. Having counted and divided them into three groups as described in section 3.2, the findings, collected in (table 7), were astonishing.
Not only the fact that the proportions appear to be nearly identical, but also the clarity of the data astonished the researchers. Having worked hard to piece together the four newspaper’s agendas, the final bit came quite easily.
Yet this data, that indicates a strong tendency of the Washington Times to the cause of the Schindlers and the Republican Party and a strong tendency of the remaining newspapers to side with Michael Schiavo and those Democrats who supported his cause, does not preclude the present analysis.
It should be admitted that dividing the whole struggle in two opposed parties, as the researchers did, could be called pushing the envelope slightly. None of the Democrats who because of this division were said to ‘side’ with Michael Schiavo, did actually issue a statement to that effect in the news material that was analyzed. The struggle between Democrats and Republicans was a dispute over the merits of the seperation of powers.
It was shown, that each newspaper has a characteristic profile, and as far as the Washington Times and the Boston Globe go, they also have a characteristic agenda. Both have generated a frame of their own: the Globe’s agenda could be said to be centered around defending Michael Schiavo and the Judiciary not by neutralizing the anguish caused by Terri Schiavo’s situation (as the NY Times attempted to do) but by being intimate and balanced at the same time, an agenda of conservation, whereas the Washington Times’s agenda promotes change, as the statement by Robert Schindler, Terri’s father, suggests: “We can change laws, but we cannot change them today” (Washington Times, 03-29-05)
It has been attempted in the present paper to show how newspaper coverage of a controversial event such as the Terri Schiavo case not only exhibits certain patterns, but that newspapers take equal part in that struggle and develop an agenda of their own. Even newspapers as similar to each other as the New York Times and the Boston Globe, who is even owned by the New York Times, have distinct agendas and promote different frames, even if, globally speaking, they are similar.
So why go do all that work if the results, globally, are not different from the results of other studies, that did not split ‘news’ into more than two newspapers? If the researchers’ claim in section 5.4. that the newspaper’s struggle closely mirrors the political struggle which takes place simultaneously can be accepted, the consequences are obvious.
Early on in the discussion a certain unidirectionality was noted, that did not seriously take into account that Terri Schiavo would maybe actively have wanted to die. A whole branch of possible debate was not included at all, i.e. proponents of euthanasia.
As newspaper agendas mirror political, this analysis allows for a critique of the political habits, of the exclusion that is perpetrated by political frames.
The limitations of the present study are obvious, the most weighty being the big scope of what it attempts to show and the small-scale demonstrations that it succeeded in mounting. Also, as nearly everything was chosen out of a pool with many alternatives: the newspapers, the articles, the linguistic instruments, the passages with which to use these instruments, it cannot be denied, that possibly personal bias distorted this study.
A comprehensive study of different newspapers reporting on controversial events remains to be done, it might possibly be hoped that this paper proved the need for such a study.