Linguistic Inquiry and Word Count (LIWC) Analysis: CNN versus Fox News
Updated: Mar 31
This was written for CAS 839: Media Analytics Communication in the Michigan State University Strategic Communications MA program. The assignment was based on an LIWC analysis of these articles:
CNN LIWC Results
Fox News LIWC Results
Reviewing the LIWC results for each article, I saw results there parallel the both the mindset of Fox News (FN) and CNN and the current U.S. political divide. There are some similarities in number, but a deeper look explains the qualitative context. For example, self-references (or I-Words), were nearly the same (FN-0.3, CNN-0.5). This struck me as a neutral statement that readers on both sides were insecure, nervous, and potentially depressed about how the election turned out (Beaty, 2021a)
If social words indicate an outgoing, socially connected posture, then CNN has a clear lead. Perhaps this difference stems from FN staffers’ beliefs being bolstered by a sense of righteousness (“Trump was chosen by God”) and attitudes formed by the FN news programs. CNN leadership might feel more outgoing because they feel they have to change minds; subsequently, the fact must be presented in a way to illuminate the facts and to create an argument to dissuade. FN might not have felt this pressure because of their devout, dedicated base. Positive emotions were close also (FN-1.0, CNN-1.6). This makes sense because again, each side may have been equally stalwart in the truth and validity of their beliefs. Thus, both sides felt optimistic that “their side” would triumph. So even though we are talking about two different ideologies, positive emotions are the same. The slight increase on the CNN side could have come from hope for change in a new administration. This is the case also for negative emotions. Again, we have a close distinction (FN-1.1, CNN-1.6) which shows anxiety or pessimism on both sides. The slight increase in CNN here, though, might show a fear that the politics or potential bad actors in a recount process would give credence to the stolen election framing. Cognitive processes were lower on the FN side (7.7) than on the CNN side (8.5). My guess is that the FN viewers were so confident that the election would be overturned, they didn’t have to think actively because they “had it in the bag” and because they had prominent members of congress “on their side.” A higher cognitive process at CNN may indicate, again, that news must be presented to uncover nefarious alignments and to help people see the truth of what was behind the electoral college recount process. Both articles scored closely in the analytics category (FN-96.3, CNN-91.5), which suggest both news organizations employed formal, logical, and hierarchical thought patterns (LIWC, n.d.). Clout, a measure of social status and confidence in leadership, marked a large difference (FN-58.6, CNN-70.5). LIWC (n.d.). suggests that confidence is different than power; thus, a confident leader may still harbor no interest in others’ social standing. Considering this sentiment, one could argue that CNN was more confident in their reporting and in the fact that their content contributed to the social betterment of the country. Authenticity scores (FN-10.7, CNN-8.9) would indicate that FN leadership reveal themselves honestly and they are more personal, humble, and vulnerable (LIWC, n.d.). With emotional tone, higher numbers indicate more positive tones—although anything under 50 trends negatively. FN scored 40.4, and CNN scored 25.8. This could represent a feeling of dread, worry, or fear on the CNN side, hence the lower number (and more negative tone).
I definitely think the story was framed differently. Right from the gate, the framing is clear. In the FN article, Biden “narrowly edged Trump” (Steinhauser & Donner, 2020, p. 2). Many states demonstrated a failure “to follow their own election laws” (Steinhauser & Donner, 2020, p. 3). Describing Sen. Josh Hawley, FN uses the verb “charged” to describe Hawley’s refusal to certify. This creates a heroic tone and positions Hawley as a champion of American ideals. In the CNN article, Herb, Mattingly, and Fox (2020) come right out and say any objectives “will not change the outcome of the election (p. 3). They write that Hawley’s objections may put his Republican colleagues in the political hot seat as they must essentially go on record whether they support former President Trump or if they are open to the will of American voters. Here, is positioned as a dissenter and possibly a chaos agent. Partisan Politics
These differences carry clear partisan connotations. To watch FN is to watch the GOP’s greatest hits (messaging, posture, talking points, and camaraderie with FN hosts). Sean Hannity on paper is a journalist, but when he speaks with Hawley, Lindsey Graham, Tom Cotton, Mo Brooks, Rudy Giuliani or other frequent guests, one doesn’t get the idea that the guests are being “grilled” or asked “the tough questions.” FN hosts position and prop the GOP guests in a favorable light. Some experts have accused FN of being a Trumpy “propaganda arm” (Klein, 2020). Others go further, stating that FN has become a “language” in which an alternative reality is presented in an “echo chamber”/”information silo” (Garber, 2020, p. 1). I don’t disagree with either of these sentiments.
In general, it's interesting to mention the names referred to and cited (and how they change as the story progresses). This is an indicator of shifting allegiances, ties, and election results. For example, the FN article mentioned Kelly Loeffler, but she is no longer a member of congress.These election-related stories are dynamic; the focal people on any given day speaks volumes. Based on this week's lectures, the FN and CNN articles could have been analyzed further deductive methods with a pre-defined codebook using categories like "GOP," "Democrats," "Electoral College," "Trump," and "Biden." The Qualitative analysis could reveal themes based on the text which researchers could use to examine text to extrapolate intention and interpretations (Beaty, 2021b; Medelyan, n.d.). Machine-aided text analysis could prove effective also. Rule-based study, with an emphasis on date stamps (to show changing sentiment) and text properties (to identify words before "Trump" and "Biden") could researchers view this story through a wider lens (Beaty, 2021a). Dictionaries could help pinpoint emotion or sentiment by calculating ratios of positive and negatively toned words (Beaty, 2021a). Finally, document clustering could help one identify category structure within a collection of documents—counting appearance of terms to indicate similarities and differences between documents (Beaty, 2021a). The addition of machine learning can help journalists and researchers incorporate qualitative, interactive, clean, and accurate algorithms—a "supervised" classification— into their reporting (Stray, n.d.; Brown, 2016, p. 11).
Beaty, J. (2021a, February 21). LIWC Program. [Video]. Retrieved from the Michigan State University, Media Analytics Com. Desire 2 Learn: https://mediaspace.msu.edu/media/LIWC%20Program%206m%2026s/1_e3vl3ao5
Beaty, J. (2021b, February 21). Machine Driven Text Analysis. [Video]. Retrieved from the Michigan State University, Media Analytics Com. Desire 2 Learn: https://mediaspace.msu.edu/media/Machine%20Driven%20Text%20Analysis%20(10m%2014s)/1_sudz3npc
Brown, S. (2016, January 26). Tips for computational text analysis. University of California Berkeley.
Garber, M. (2020, September 16). Do you speak Fox? How Donald Trump’s favorite news source became a language. The Atlantic. https://www.theatlantic.com/culture/archive/2020/09/fox-news-trump-language-stelter-hoax/616309/
Herb, J., Mattingly, P., & Fox, L. (2020, December 30). GOP senator to delay affirming Biden victory by forcing votes on Electoral College results. CNN. https://www.cnn.com/2020/12/30/politics/josh-hawley-force-votes-electoral-college-results/index.html
Klein, G. (2020, September 8). Fox News has become Trump’s propaganda arm, CNN’s media reporter says. National Press Club. https://www.press.org/newsroom/fox-news-has-become-trumps-propaganda-arm-cnns-media-reporter-says
LIWC. (n.d.). Interpreting LIWC output. http://liwc.wpengine.com/interpreting-liwc-output/
Medelyan, A. (2020). Codinq qualitative data: How to code qualitative research. Thematic. https://getthematic.com/insights/coding-qualitative-data/
Steinhauser, P., & Donner, J. (2020, December 30). Hawley says he'll object to Electoral College certification of Biden Victory on Jan. 6. Fox News. https://www.foxnews.com/politics/josh-hawley-to-object-electoral-college-certification-jan-6
Stray, J. (n.d.). What do journalists do with documents? Field notes for natural language processing researchers. Columbia Journalism School. https://journalism.stanford.edu/cj2016/files/What%20do%20journalists%20do%20with%20documents.pdf