pith. machine review for the scientific record. sign in

arxiv: 2604.20454 · v1 · submitted 2026-04-22 · 💻 cs.CL

Recognition: unknown

Not all ANIMALs are equal: metaphorical framing through source domains and semantic frames

Yulia Otmakhova , Matteo Guida , Lea Frermann
Authors on Pith no claims yet

Pith reviewed 2026-05-10 00:00 UTC · model grok-4.3

classification 💻 cs.CL
keywords metaphorical framingsource domainssemantic framesdiscourse analysiscomputational linguisticsclimate changeimmigrationpolitical ideology
0
0 comments X

The pith

Metaphors derive their framing power from the specific pairing of source domains with semantic frames, not domains alone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Metaphors shape understanding of complex issues through more than just their source domains. The paper argues that the semantic frame activated inside a given domain further steers interpretation, and it supplies a computational framework to extract and compare these paired elements from large text collections. When applied to climate change news, the method surfaces both familiar domains and finer frame-level distinctions in how the topic is portrayed. In immigration coverage, it finds that liberals and conservatives systematically choose different frames even within the same source domains, with conservatives favoring uncontrollability and liberals selecting neutral or victim-centered options. This yields the first NLP pipeline for discovering discourse metaphors at that level of granularity.

Core claim

The interplay between source domains and semantic frames determines how metaphors shape understanding of complex issues. A computational framework derives salient discourse metaphors by identifying both elements jointly, revealing nuanced associations in climate change reporting and ideological differences in immigration discourse where conservatives prefer uncontrollability frames and liberals prefer neutral or victimizing ones.

What carries the argument

A computational pipeline that jointly extracts source domains and semantic frames from metaphorical expressions in text to enable fine-grained discourse analysis.

If this is right

  • Discourse studies can quantify ideological differences in metaphor use at the frame level rather than the domain level alone.
  • Media portrayals of issues like climate change can be compared for subtle frame associations within shared source domains.
  • Automated discovery of metaphors becomes feasible without exhaustive manual coding of every expression.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same framework could track shifts in metaphorical framing across time or media outlets on any contested topic.
  • Metaphor detection tools in NLP would gain accuracy by incorporating frame information alongside domain labels.
  • Experimental tests could check whether the documented frame choices within domains measurably affect reader attitudes toward immigration or climate policy.

Load-bearing premise

The automatic detection of source domains and semantic frames accurately reflects the distinctions that shape human interpretation of metaphors.

What would settle it

A controlled study in which participants rate the framing impact of the same source-domain metaphors when paired with the different semantic frames identified by the method, finding no measurable difference in perceived meaning or attitude.

Figures

Figures reproduced from arXiv: 2604.20454 by Lea Frermann, Matteo Guida, Yulia Otmakhova.

Figure 1
Figure 1. Figure 1: Interaction between semantic frames and do [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Framework overview with semantic frames and source domains for the input metaphors (bold). semantic frames as a proxy for conceptual map￾pings, in order to detect (Li et al., 2023), or gen￾erate (Stowe et al., 2021b) metaphors. We are the first to leverage the interaction between the two frameworks. According to the constructionist view of metaphors, their conceptual understanding is bound by linguistic co… view at source ↗
Figure 3
Figure 3. Figure 3: Relative frequency of source domains in cli [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The five semantic frames with highest saliency [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance comparison on test set for classes [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Distribution of mentions of “climate change” [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Screenshots of the annotation instruction interface. Annotators were guided through an interactive [PITH_FULL_IMAGE:figures/full_fig_p022_7.png] view at source ↗
read the original abstract

Metaphors are powerful framing devices, yet their source domains alone do not fully explain the specific associations they evoke. We argue that the interplay between source domains and semantic frames determines how metaphors shape understanding of complex issues, and present a computational framework that allows to derive salient discourse metaphors through their source domains and semantic frames. Applying this framework to climate change news, we uncover not only well-known source domains but also reveal nuanced frame-level associations that distinguish how the issue is portrayed. In analyzing immigration discourse across political ideologies, we demonstrate that liberals and conservatives systematically employ different semantic frames within the same source domains, with conservatives favoring frames emphasizing uncontrollability and liberals choosing neutral or more ``victimizing'' semantic frames. Our work bridges conceptual metaphor theory and linguistics, providing the first NLP approach for discovery of discourse metaphors and fine-grained analysis of differences in metaphorical framing. Code, data and statistical scripts are available at https://github.com/julia-nixie/ConceptFrameMet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript introduces a computational framework combining source domains from conceptual metaphor theory with semantic frames to identify salient discourse metaphors and analyze their framing effects. It applies the framework first to climate change news to uncover both established source domains and nuanced frame-level associations, then to immigration discourse across liberal and conservative sources, reporting that conservatives preferentially use frames emphasizing uncontrollability while liberals select neutral or victimizing frames within the same domains. The work positions itself as the first NLP approach for discovery of discourse metaphors at this granularity and releases code, data, and statistical scripts.

Significance. If the framework's identification of domains and frames proves robust, the paper offers a novel, reproducible method for fine-grained analysis of metaphorical framing in public discourse on contested issues. The dual applications demonstrate utility for distinguishing ideological patterns beyond source domains alone. Explicit credit is due for the open release of code, data, and scripts, which supports reproducibility and extension by others. This bridges conceptual metaphor theory with computational methods in a way that could enable systematic, large-scale studies of framing effects.

major comments (1)
  1. [§5] §5 (immigration discourse analysis): The central claim that liberals and conservatives systematically differ in semantic frame choice within the same source domains (conservatives favoring uncontrollability frames, liberals neutral or victimizing ones) requires that the two ideological corpora be comparable on subtopic coverage. The manuscript provides no description of topic balancing, keyword stratification, event matching, or sub-corpus controls (e.g., border security vs. asylum cases), so observed frame differences could arise from differing topic distributions rather than framing preferences.
minor comments (2)
  1. [Abstract / §1] The abstract and introduction assert this is 'the first NLP approach' for discourse metaphor discovery via domains and frames; a short related-work paragraph contrasting with existing metaphor detection pipelines would clarify the precise novelty.
  2. [§3] Operational details on how source domains and semantic frames are automatically or semi-automatically extracted (lexicons, models, annotation guidelines) are referenced but could be expanded with a brief example or pseudocode for reproducibility.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback, particularly the emphasis on ensuring subtopic comparability to strengthen causal claims about ideological framing differences. We address this point directly below and will incorporate revisions to enhance the robustness of the immigration analysis.

read point-by-point responses

  1. Referee: §5 (immigration discourse analysis): The central claim that liberals and conservatives systematically differ in semantic frame choice within the same source domains (conservatives favoring uncontrollability frames, liberals neutral or victimizing ones) requires that the two ideological corpora be comparable on subtopic coverage. The manuscript provides no description of topic balancing, keyword stratification, event matching, or sub-corpus controls (e.g., border security vs. asylum cases), so observed frame differences could arise from differing topic distributions rather than framing preferences.

    Authors: We agree that subtopic comparability is necessary to isolate framing preferences from potential topic distribution confounds. The current manuscript selects broad immigration coverage from liberal and conservative outlets without explicit stratification or matching on subtopics such as border security, asylum processing, or deportation events. In the revised version, we will add a dedicated subsection describing the topic distributions (via keyword categorization and LDA topic modeling) across the two corpora. Where imbalances are detected, we will apply event-matched subsampling or include subtopic indicators as covariates in the frame association models, then re-report the conservative-liberal frame differences to confirm they hold within comparable subtopics. This directly addresses the concern while preserving the core finding that frame choices differ within shared source domains. revision: yes

Circularity Check

0 steps flagged

Empirical framework on external corpora exhibits no circular derivation

full rationale

The paper introduces a computational framework for identifying discourse metaphors via source domains and semantic frames, then applies it to independent news corpora on climate change and immigration. No equations, first-principles derivations, or predictions are presented that reduce by construction to fitted inputs, self-definitions, or self-citation chains. All central claims rest on external data processing and standard linguistic resources rather than internal closure, satisfying the condition for a self-contained empirical study.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no technical details on implementation are provided, so no free parameters, axioms, or invented entities can be extracted. The framework is described at a conceptual level without equations or procedural specifications.

pith-pipeline@v0.9.0 · 5469 in / 1215 out tokens · 41409 ms · 2026-05-10T00:00:38.326844+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 4 canonical work pages · 1 internal anchor

  1. [1]

    concepts: A meta-analysis.Lan- guage and Cognition, 11(1):41–65

    Metaphorical framing in political discourse through words vs. concepts: A meta-analysis.Lan- guage and Cognition, 11(1):41–65. Tuhin Chakrabarty, Xurui Zhang, Smaranda Muresan, and Nanyun Peng. 2021. MERMAID: Metaphor gen- eration with symbolism and discriminative decoding. InProceedings of the 2021 Conference of the North American Chapter of the Associat...

    2021

  2. [2]

    InProceedings of the First Computing Social Responsibility Workshop within the 13th Language Resources and Evaluation Con- ference, pages 24–34, Marseille, France

    Framing legitimacy in CSR: A corpus of Chi- nese and American petroleum company CSR reports and preliminary analysis. InProceedings of the First Computing Social Responsibility Workshop within the 13th Language Resources and Evaluation Con- ference, pages 24–34, Marseille, France. European Language Resources Association. Minjin Choi, Sunkyung Lee, Eunseon...

  3. [3]

    do it all

    MelBERT: Metaphor detection via contextual- ized late interaction using metaphorical identification theories. InProceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Tech- nologies, pages 1763–1773. 10 Jacob Devasier, Yogesh Gurjar, and Chengkai Li. 2024. Robust frame-semantic mod...

    2021

  4. [4]

    George Lakoff and Mark Johnson

    Master metaphor list (technical report).Cog- nitive Linguistics Group University of California, Berkeley. George Lakoff and Mark Johnson. 2008.Metaphors we live by. University of Chicago press. Mark J Landau, Daniel Sullivan, and Jeff Greenberg

    2008

  5. [5]

    RoBERTa: A Robustly Optimized BERT Pretraining Approach

    Evidence that self-relevant motives and metaphoric framing interact to influence political and social attitudes.Psychological science, 20(11):1421– 1427. Chee Wee (Ben) Leong, Beata Beigman Klebanov, and Ekaterina Shutova. 2018. A report on the 2018 VUA metaphor detection shared task. InProceedings of the Workshop on Figurative Language Processing, pages ...

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  6. [6]

    Haohan Meng, Xiaoyu Li, and Jinhua Sun

    A framework for the computational linguistic analysis of dehumanization.Frontiers in artificial intelligence, 3:55. Haohan Meng, Xiaoyu Li, and Jinhua Sun. 2025. Large language models prompt engineering as a method for embodied cognitive linguistic representation: a case study of political metaphors in Trump’s discourse. Frontiers in Psychology, 16:159140...

    2025

  7. [7]

    InPro- ceedings of the 62nd Annual Meeting of the Associa- tion for Computational Linguistics (Volume 1: Long Papers), pages 15407–15428, Bangkok, Thailand

    Media framing: A typology and survey of computational approaches across disciplines. InPro- ceedings of the 62nd Annual Meeting of the Associa- tion for Computational Linguistics (Volume 1: Long Papers), pages 15407–15428, Bangkok, Thailand. Association for Computational Linguistics. Vinodkumar Prabhakaran, Marek Rei, and Ekaterina Shutova. 2021. How meta...

    work page arXiv 2021

  8. [8]

    InFindings of the Association for Computational Linguistics: NAACL 2025, pages 3953–3967, Albuquerque, New Mexico

    ImaRA: An imaginative frame augmented method for low-resource multimodal metaphor detec- tion and explanation. InFindings of the Association for Computational Linguistics: NAACL 2025, pages 3953–3967, Albuquerque, New Mexico. Association for Computational Linguistics. Yuan Tian, Nan Xu, and Wenji Mao. 2024. A theory guided scaffolding instruction framewor...

    2025

  9. [9]

    Cli- matebert: A pretrained language model for climate-related text,

    Drum up SUPPORT: Systematic analysis of image-schematic conceptual metaphors. InProceed- ings of the 3rd Workshop on Figurative Language Processing (FLP), pages 44–53, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computa- tional Linguistics. Dian Wang, Yang Li, Suge Wang, Xin Chen, Jian Liao, Deyu Li, and Xiaoli Li. 2025. CKEMI: Con- cept kno...

    work page arXiv 2025

  10. [10]

    Read the entire sentence to establish a general understanding of the meaning

  11. [11]

    Take into account what comes before and after the lexical unit

    (a) Establish the word’s meaning in context, that is, how it applies to an entity, relation, or attribute in the situation evoked by the text (contextual meaning). Take into account what comes before and after the lexical unit. (b) Determine if the target word has a more basic contemporary meaning in other contexts than the one in the given context. For o...

  12. [12]

    climate change

    If yes, the word is metaphorical. Otherwise it is literal. Please return a json object which consists of the following field: metaphors: a list of extracted metaphor spans. Do not output any explanations or reasoning steps. 15 Figure 5: Performance comparison on test set for classes with different frequency in the training set. To evaluate the extracted s...

    work page arXiv 1986

  13. [13]

    First comprehension check failure: partic- ipants were allowed to retry after reviewing the instructions

  14. [14]

    Second comprehension check failure: par- ticipants were permanently rejected and in- structed to return their submission on Prolific

  15. [15]

    vision” could not be established, the labels for “visions

    Incomplete submissions: participants who did not complete all 50 annotations in their assigned batch were automatically instructed to return their submission. Participants were informed during the consent process that failing comprehension checks would result in submission return or rejection, and that repeated failures would lead to ineligibility for fu-...

  16. [16]

    treating immigrants like animals,

    **Abusing** - Liberals frequently use animal metaphors to describe how immigrants are being mistreated, with phrases like "treating immigrants like animals," "caging children like animals," and "treating them worse than dogs"

  17. [17]

    cages,"

    **Containing** - Extensive use of metaphors describing detention facilities as "cages," "kennels," "dog pounds," and "pens" where immigrants are held

  18. [18]

    hunting down immigrants,

    **Hunting** - Metaphors describing ICE and border patrol activities as "hunting down immigrants," "stalking," and "preying on" vulnerable populations

  19. [19]

    animals,

    **Dehumanization** - References to how Trump and officials call immigrants "animals," "snakes," "cockroaches," and other dehumanizing terms

  20. [20]

    **Protecting** - Calls to protect immigrants from being treated like animals, with references to animal welfare standards being better than immigrant treatment **Conservative Semantic Frames:**

  21. [21]

    **Categorization** - Frequent use of "animals" to categorize MS-13 gang members and violent criminals, distinguishing them from law-abiding citizens

  22. [22]

    illegal aliens

    **Being_in_category** - Classifying certain immigrants as "illegal aliens" versus legal immigrants, often using animal terms for the former group

  23. [23]

    swarms,"

    **Swarming** - Describing immigration as animal-like mass movement with terms like "swarms," "herds," "flocks," and "stampedes" of immigrants

  24. [24]

    hunting down

    **Hunting** - References to "hunting down" and "rounding up" illegal immigrants for deportation 20

  25. [25]

    breeding like rabbits

    **Proliferating_in_number** - Using breeding metaphors to describe immigrant population growth, with terms like "breeding like rabbits" and concerns about demographic changes A.12.2 WATER source domain - based on tweets that contain WATER metaphors **Liberal Side - Most Salient Semantic Frames:**

  26. [26]

    flow of migrants,

    **Fluidic_motion** - This is the most prominent frame, with liberals using water metaphors to describe the movement and flow of immigrants in neutral or sympathetic terms. Examples include "flow of migrants," "stream of immigrants," "waves of immigration," and "tide of migrants." These metaphors often emphasize the natural, ongoing nature of migration

  27. [27]

    pouring in,

    **Motion** - Closely related to fluidic motion, this frame captures the directional movement of people, often described as "pouring in," "flooding," or "streaming" across borders, but typically used to critique anti-immigration policies rather than the immigrants themselves

  28. [28]

    drowning

    **Catastrophe** - Used to describe the humanitarian crisis and dangerous conditions migrants face, such as "drowning" in the Rio Grande, being "swept away," or facing "floods" of persecution in their home countries. **Conservative Side - Most Salient Semantic Frames:**

  29. [29]

    flood of illegals,

    **Fluidic_motion** - Also the most prominent frame for conservatives, but used with negative connotations to describe an overwhelming, uncontrolled influx. Examples include "flood of illegals," "pouring across the border," "stream of illegal aliens," and "tide of illegal immigration." The emphasis is on the volume and lack of control

  30. [30]

    flooded,

    **Catastrophe** - Used to describe the perceived negative impact of immigration on American society, such as the country being "flooded," "swamped," or "drowning" in illegal immigrants, suggesting an overwhelming disaster

  31. [31]

    draining

    **Emptying** - This frame appears in contexts of "draining" resources, the economy, or social services due to immigration, suggesting that immigrants are depleting American resources like water draining from a container. A.12.3 WATER source domain - based on tweets that do NOT contain WATER metaphors **Liberal Tweets - Most Salient Water Metaphor Semantic...

  32. [32]

    immigrants...pour in and infest our country

    **Fluidic_motion** - Examples: "immigrants...pour in and infest our country", "flooding our country with immigrants", "immigrants are swarming to the border", "migrants fleeing violence in El Salvador still plan for the U.S."

  33. [33]

    more hopeless. I feel like we are about to get buried by a tsunami...Iran/Roe/Immigration

    **Catastrophe** - Examples: "more hopeless. I feel like we are about to get buried by a tsunami...Iran/Roe/Immigration", "We’re not being invaded from any direction, we’re not under attack by immigrants"

  34. [34]

    Trump refers to immigrants as an infestation 46 times

    **Abounding_with** - Examples: "Trump refers to immigrants as an infestation 46 times", "immigrants...pour in and infest our country", "calling migrants invaders and an infestation" **Conservative Tweets - Most Salient Water Metaphor Semantic Frames:**

  35. [35]

    illegals are flooding states

    **Fluidic_motion** - Examples: "illegals are flooding states", "Democrats divert 137B tax$ ANNUALLY, to support & defend illegals", "illegals flooding the country", "flood our country with immigrants", "illegals pouring in"

  36. [36]

    NATIONAL EMERGENCY: 2,000+ Migrants Quarantined After Bringing Disease

    **Catastrophe** - Examples: "NATIONAL EMERGENCY: 2,000+ Migrants Quarantined After Bringing Disease", "This is an invasion", "illegals flooding your border overwhelming your system"

  37. [37]

    illegals cost America over 300 Billion

    **Scarcity** - Examples: "illegals cost America over 300 Billion", "illegals are costing us billions", "Each illegal immigrant costs taxpayers $70k per year", "illegals cost US well over 300 Billion a yr" 21 Figure 7: Screenshots of the annotation instruction interface. Annotators were guided through an interactive tutorial explaining (top) the concept of...