Can journalists be opinionated on Twitter?

To conduct this experiment I analysed tweets from Hamid Mir, Raza Rumi, Javed Chaudhry, Moeed Pirzada and many others.

Usman Shahid October 27, 2014
The events taking place around us affect our feelings, which as a result, affect our conversations throughout our normal lives. Since August, the prolonged Azadi march has been affecting the feelings of average Pakistanis on the road.

Being associated to the data-mining and text-mining field, I carried out a little experiment to explore the sentiments of Pakistani journalists who are reporting current events in Pakistan. I took journalists as my test subjects as ordinary people generally take their (journalists’) opinions seriously and even adopt them as their own. But according to the journalism objectivity principle, these journalists should remain impartial and convey only facts without including their feelings, emotions, opinions and political views along with the news. The result was rather interesting.

In academic literature, extracting the writer’s opinion or feeling from the text using the machine learning computational technique is termed as sentiment analysis or opinion-mining. Generally speaking, by using sentiment analysis, researchers try to extract the opinion or feeling of a writer, leaving behind only factual text. The attitude of the writer normally involves his or her judgment, emotional state and emotional effects that writers normally intend to convey to their readers. The results of this analysis appear as positive, negative, or neutral for every piece of “news” conveyed by writers or journalists.

Positive results indicate the writer’s happy mood and excitement while writing the text. Negative refers to the presence of sad emotions or negative opinion. When equal polarity of positive and negative sentiments exists in the text, it becomes difficult for the computational techniques to pass their judgment, and therefore refer to them as neutral. For example, if the writer is talking about somebody’s death, but doing so without expressing or including his opinion or emotions on the demise.

Twitter has become a famous tool to extract people’s personal and political views. Researchers in the US used Twitter data to understand people’s emotions during the presidential elections. It is due to the same reason that John Morton, who was a former newspaper reporter and is now the president of a consulting firm that analyses newspapers and other media properties, wrote an article on how social networking sites like Twitter and Facebook have changed the way journalists interact with the public. He says,
“Journalists should not reveal their political views, Twitter or no Twitter”

For a good journalist, journalism objectivity must be ensured. Journalism objectivity is the basic principle of journalistic professionalism and refers to the neutrality of the journalist. It ensures that he/she remains neutral without choosing to be on either side of the argument. That is when only the facts are reported for what they are.

In Pakistan, almost every journalist claims to be neutral. In this experiment, I put this claim to test via sentiment analysis using tweets of famous journalists in Pakistan so as to show how neutral their reporting is under the context of political activity regarding Pakistan Tehreek-e-Insaf (PTI)’s Azadi march.

I used R Software’s Twitter library (Twitter) for mining or extracting tweets from August and Datumbox API, a text analysis service, to rate the sentiment of each tweet as positive, negative or neutral. I took into account tweets of various journalists such as Cyril AlmeidaFahd HusainFereeha IdreesHamid MirIftikhar AhmadJasmeen ManzoorJaved ChaudhryKashif AbbasiMoeed PirzadaMushtaq MinhasRauf KlasraRaza RumiShahzeb Khanzada, and Talat Hussain. Due to limitation of time, I wasn’t able to conduct the experiment on other renowned journalists.

Initially, I obtained the sentiments from some of the tweets individually.

Moeed Prizada tweets:

Pirzada’s mood and feelings were detected as positive.

The disappointment in Hussain’s tweet was concluded as negative. The sentiment analyser considered Hussain to be feeling rather sad whilst tweeting this text.

However, the sentiment analyser was unable to extract a positive or negative analysis from Mir’s tweets.

Clearly, it was difficult to conclude whether he was excited or sad. He conveyed the message but did not include any emotions or opinions.

Every tweet was carrying different sentiments, and so for the next stage, I collected tweets of journalists, who tweeted in August, and calculated the twitter profile of each journalist to understand what type of aggregate sentiment they are spreading amongst their followers. And the results of the experiment turned out to be quite interesting. The results of the sentiment analysis are shown in Table 1 and Figure 1 below.

Figure 1

Table 1

Pirzada and Husain’s sentiments are obtained as most positive amongst all of their peers. Minhas appeared as the one who mostly displays negative sentiments in his tweets. For neutral sentiments, Mir tops the list. Not only did he top the list, but there were barely any positive or negative sentiments portrayed in any of his tweets, which proved that he is better at hiding his emotions and opinions as compared to his peers. Ahmad, Klasra, and Khanzada had similar results to Mir.

Positive sentiments remain higher than negative for Husain, Mir, Chaudhry, Pirzada and Rumi. Whereas Almeida, Idrees, Manzoor and Abbasi try to balance their sentiments in their tweets, thus, their neutral sentiments remain lower than their positive and negative sentiments.

As per the results, it is obvious that most of the Pakistani journalists are tweeting with emotions and not taking into consideration the existence of journalism objectivity. And this is surely affecting their followers.

A journalist’s job is to be impartial and let the people decide for themselves which politician is correct or not. Our journalists should only convey the facts without their own judgments about the political events, parties or people.

If I manage to get time in the future, I will conduct an experiment to analyse the sentiments of the people responding to these journalists to understand how much their tweets affect the people reading them. Till then, I leave you with the results of my current experiment to ponder upon journalism objectivity.
Usman Shahid A PhD candidate at North Dakota State University (USA) and a lecturer at COMSATS, Abbottabad, Pakistan.
The views expressed by the writer and the reader comments do not necassarily reflect the views and policies of the Express Tribune.


Haider | 8 years ago | Reply Why in the world did the blogger prefer to use the sentiment analysis function (which claims only to measure emotion) when the Datumbox API he has used already provides a Subjectivity Analysis function to measure objectivity? How can sentiment measure objectivity? Natural language processing applications like these measure sentiment by assigning negative/positive scores to words. For example, the statement "Pakistan win by 200 runs" may be categorised as positive because of the word "win", while "Australia lose by 200 runs" may be negative because of the word "lose". What does that say about whether the statement was objective or based on personal opinion? Absolutely nothing! And in this particular case, Datumbox does a bad job of even gauging emotion. I see no reason why that Moeed Peerzada tweet should be categorised as positive. Did the blogger evaluate the results? Even the author of the API says on his blog that analyses may be prone to errors depending on the dataset. 'Neutrality' here is basically the program's inability to gauge emotion, NOT neutrality in the journalistic sense. The only opinion in this blog is the writer's assumption that Pakistani journalists are opinionated. This 'research' is a complete waste of an otherwise promising piece of software. A better use of this API could be to mine for tweets containing "Nawaz Sharif" and see what kind of sentiment prevails on Twitter regarding the PM.
an indian | 8 years ago | Reply true. I try to stick to Ndtv ,but I find that some reports are unique to their channels & not reported by all four (Times,Ht & CnnIbn) I end up watching all of them.
Replying to X

Comments are moderated and generally will be posted if they are on-topic and not abusive.

For more information, please see our Comments FAQ