Phobia: a corpus study of political diagnostics

The paper situates the study amidst the 2020 Black Lives Matter protests and the concurrent COVID-19 pandemic to illustrate how scientific and political discourses intertwine (e.g., the metaphor of racism as a pandemic). It argues that politics and science share a vocabulary of illness and focuses on the lexical item phobia as a productive element within this shared vocabulary. The introduction reviews the historical and lexical development of -phobia terms via the OED, from medical senses (hydrophobia, photophobia; agoraphobia/claustrophobia) to socially evaluative senses (xenophobia, homophobia), noting the expansion to terms like transphobia and Islamophobia. The research aims are to investigate who uses -phobia terms in online alternative media, how they are used rhetorically (collocation and patterning), and for what purposes, thereby illuminating the entanglement of political and medical vocabulary.

Background sources include lexicographic accounts from the OED on -phobia terms and their semantic shifts; corpus linguistics frameworks on collocation, colligation, semantic preference, and semantic prosody (Sinclair; Stubbs; Partington; Stewart; Louw; Whitsitt); and sociopolitical theory on identity politics and intersectionality (Crenshaw; Yuval-Davis; Laclau & Mouffe). The paper also references psychiatric classification debates (APA/DSM, Drescher) and public health/policy discourse surrounding bias and its training. These works provide context for understanding the rhetorical and political deployment of -phobia and its alignment with broader discursive struggles.

Data derive from the Genealogies of Knowledge (GoK) English Internet corpus (over 4 million tokens from >2900 texts across 35 outlets, mostly 2010–2020), representing Anglophone alternative media at the science–politics interface with a left-leaning skew. Texts are accessed via the GoK concordance browser, which provides keyword-in-context (KWIC) lines, an extract function, and visualization tools. The study employs both quantitative and qualitative methods characteristic of rhetorical corpus analysis. Quantitatively, the corpus is queried for all tokens matching the regular expression ".*phobi[ac]" to capture nouns ending in -phobia and adjectives ending in -phobic (e.g., Islamophobia, xenophobic). Collocations are examined within standard windows (typically 4L–4R; with focus also on N−2 to N+2) and assessed using measures available in the GoK Mosaic tool (including log-log, mutual information, z-score). The Metafacet tool is used to visualize distribution across outlets. Frequency lists provide broader corpus context (>80,000 word types). Qualitatively, patterns are interpreted in terms of colligation, semantic preference, and semantic prosody to identify extended units of meaning (e.g., formulaic lists of prejudicial terms), rhetorical functions (accusation/condemnation), and ideological positioning (who accuses whom, and from which outlets).

• The search returned 345 concordance lines: 197 ending in -phobia and 148 in -phobic. • Top items by frequency (Table 1): xenophobia (73) / xenophobic (73); Islamophobia (72) / Islamophobic (35); homophobia (21) / homophobic (19); transphobia (11) / transphobic (6); chemophobia (4) / chemophobic (6). • Strong collocational associations link xenophobia/xenophobic with racism/racist across multiple positions; for xenophobic, collocates to the left are often adjectives (e.g., racist, nationalist) and to the right nouns indicating mental states or behaviors (e.g., prejudice, attitudes, impulses, reaction, violence, attacks). • Nouns of prejudice (e.g., homophobia) typically occur in formulaic, open-ended lists with other prejudices (racism, sexism, transphobia, xenophobia, ableism/disablism). The order often follows implicit norms (e.g., racism before sexism; homophobia after sexism). • The semantic prosody of -phobia/-phobic usage is predominantly negative and judgmental; lists function rhetorically to accuse/condemn and signal prohibited attitudes. • Outlet distribution (Metafacet) shows concentration in left-wing/radical-left outlets. Salvage.zone has 93 occurrences across >30 of 73 articles (28 instances of Islamophobia). Conversely, far-right Amerika.org has only one quoted instance (no original use) across 99 articles. • Of 35 outlets, 13 have no -phobia/-phobic occurrences (e.g., Bad Science, Desmog, Answers in Genesis), reflecting topic focus rather than general internet language use. • Right-leaning/science-contrarian outlets show selective, polemical uses: The Risk-Monger uses chemophobia (e.g., anti-GMO discourse) to critique opponents; Creation Ministries International quotes left-leaning lists (e.g., queer-, trans-, intersex-phobia) to argue against politicization of “science.” • Evidence of co-existence between social and clinical senses is minimal but present (e.g., rare mention of claustrophobic alongside xenophobic). • Overall, -phobia/-phobic terms cluster with other isms/phobias in list structures, aligning with intersectional discourse and functioning as markers of ideological positioning and accusation.

Findings indicate that in Anglophone alternative online media, -phobia/-phobic terms are predominantly mobilized in left/progressive outlets as elements in formulaic lists of condemned prejudices (racism, sexism, homophobia, transphobia, xenophobia, ableism, etc.). This pattern supports the research aim by showing how these terms act as rhetorical devices of accusation and alignment: they identify both targets (accused) and the ideological stance of accusers. The strong collocational ties and list structures reflect intersectional framing, where multiple forms of oppression are articulated together without fixed hierarchical ordering. The study situates this within broader cultural and political dynamics: the medical metaphor (prejudice as disease), the historical reversal of medicalization regarding homosexuality/trans issues, and ongoing attempts to “treat” bias (e.g., unconscious bias training) despite limited evidence of behavioral change. Polarization shapes adoption: left-leaning spaces normalize -phobia lexicon as a performative call to action, while right-leaning spaces either avoid it, deploy competing labels (e.g., chemophobia) to criticize opponents, or quote it to contest the fusion of scientific and political discourse. The paper interprets -phobia as tending toward an empty signifier that enables coalition-building through the logics of equivalence/difference, functioning as political identification and mobilization.

The study shows that -phobia/-phobic in online alternative media exhibits a marked collocational profile centered on condemnatory, often intersectional lists, especially in radical-left outlets. It argues that this usage reflects and reinforces the contemporary entanglement of scientific and political discourse, with phobia terms operating as tools of antagonism and political identification. Future work should examine more balanced and diverse corpora across the political spectrum, track diachronic changes, compare languages, and further theorize how medical metaphors and empty signifiers structure public debates on science, identity, and policy.

The GoK Internet corpus is thematically curated and skewed toward left-leaning Anglophone alternative media; it is not representative of general internet usage. Outlet coverage is uneven (35 outlets; 13 with no -phobia/-phobic occurrences). Conclusions about ideological distribution are therefore tentative. The analysis emphasizes qualitative interpretation of collocational patterns; results may be sensitive to window sizes, measures, and corpus updates. Copyright limits access to full texts, potentially constraining contextual analysis.