Joseph 1
Filsy Joseph
Dr. Mithilesh Kumar
Digital Humanities MEL349A
09 August,2025
Silencing Resistance: A Computational Analysis of Social Media Censorship on
Palestine-Related Content
1. Research Question(s)
● How is resistance discourse on Palestine represented, filtered, or censored across major
social media platforms?
● Which linguistic patterns, thematic elements, or narrative framings are most likely to be
flagged, removed, or algorithmically deprioritized in Palestine-related digital content?
● What patterns both quantitative and qualitative can Python-based text mining reveal
about the silencing or amplification of certain voices, and how do these patterns intersect
with broader questions of media power and digital rights?
2. Background and Significance
In the 21st century, social media platforms such as YouTube, Twitter (now X), Facebook,
and Instagram are not just tools for personal communication, they function as global stages for
political discourse, cultural storytelling, and activist mobilization. The Israel–Palestine conflict is
one of the most visible and contentious political issues on these platforms. While such spaces
ostensibly promote free expression, mounting evidence suggests that algorithms and content
moderation policies systematically suppress certain political narratives, particularly those aligned
with Palestinian resistance movements.
Censorship in this context may take many forms: removal of posts, shadow banning,
downranking in feeds, flagging content as “sensitive,” or altering the visibility of politically
charged hashtags. These practices raise urgent questions about who controls the global digital
Joseph 2
public sphere, whose voices are amplified or erased, and how narratives of resistance survive
under algorithmic scrutiny.
From a digital humanities perspective, this study sits at the intersection of computational
analysis and critical cultural inquiry. Digital humanities is uniquely positioned to interrogate how
technological systems mediate political discourse, because it values both data-driven methods
(text mining, pattern detection) and interpretive critique rooted in history, politics, and ethics. By
examining the structural silencing of Palestinian narratives, the research addresses key concerns
in communication studies namely, the interplay between media technologies and the politics of
representation.
The project’s significance lies in its ability to:
● Map macro-level patterns(frequency, deletion rates, flagging patterns).
● Identify micro-level discourse markers (specific words, phrases, and rhetorical
strategies targeted by moderation systems).
● Reveal the power dynamics embedded in content moderation algorithms, thus
contributing to ongoing debates about transparency, bias, and digital colonialism.
3. Proposed Data Sources
● Primary: YouTube Transcripts: Transcripts from approximately 20 curated videos
discussing Palestine. These will represent a diversity of perspectives—grassroots
activism, journalism, governmental speeches, and on-the-ground reports—to avoid
overrepresentation of any single narrative style.
● Optional Expansion:Social Media Posts: Publicly available Palestine-related posts from
Twitter/X, Facebook, and Instagram that have been flagged, labeled, or removed.
● Secondary Sources: Reports and studies from organizations such as 7amleh – The Arab
Center for the Advancement of Social Media, Human Rights Watch, and Access Now
that document cases of political content suppression.
Joseph 3
4. Python Techniques/Tools to Be Used
Data Preprocessing:
- Cleaning and normalization of text (removing stopwords, punctuation, special characters,
and non-textual elements).
- Handling multilingual or code-switched data (Arabic–English) if present, with
language-specific tokenization.
Text Mining & NLP
- Tokenization and frequency analysis to identify high-frequency terms and recurring
semantic fields.
- Extraction of bigrams and trigrams to uncover common collocations (e.g., “Free
Palestine,” “Israeli occupation”) and to detect expressions disproportionately associated
with censorship.
- Sentiment analysis (optional extension) to explore whether negative or positive emotional
valence affects moderation likelihood.
Clustering & Comparative Analysis
- Grouping transcripts/posts into categories such as “flagged” vs. “non-flagged” or
“removed” vs. “retained,” and comparing linguistic patterns between these categories.
Visualization & Interpretation
- Word clouds for thematic overviews.
- Bar charts for frequency comparisons.
- Network graphs to show relationships between censored terms and broader thematic
clusters.
Technical Environment
- All analysis performed in Python using libraries such as NLTK, spaCy, pandas, and
matplotlib, executed in Visual Studio Code for integrated development.
Joseph 4
5. Expected Outcomes
Quantitative Findings
- Ranked lists of the most frequent words, bigrams, and trigrams in Palestine-related
discourse.
- Comparative statistics showing which terms or themes appear disproportionately in
censored vs. uncensored content.
Qualitative Insights
- Identification of recurring narrative frames in censored content (e.g., calls to action,
critiques of state violence, solidarity messages).
- Examination of how resistance discourse adapts to circumvent censorship (e.g., using
euphemisms, alternative spellings, visual memes instead of text).
Contribution to Digital Humanities
- A methodological blueprint for integrating computational text analysis with political and
cultural critique.
- Evidence of how Python can be a tool for “distant reading” political texts, while
remaining anchored in close, contextual reading.
Broader Social Impact
- Strategic recommendations for activists, journalists, and educators on navigating
algorithmic censorship.
- Policy suggestions for greater transparency in the moderation process.
Ultimately, this research seeks not only to map patterns of censorship in Palestine-related
discourse, but also to situate these patterns within broader historical, cultural, and political
struggles over voice, representation, and power. By bridging computational analysis with critical
interpretation, the study reaffirms the role of digital humanities as both a rigorous scholarly
practice and a socially engaged discipline one capable of interrogating the infrastructures that
shape public narratives and of amplifying silenced perspectives in the digital sphere.
