Emphasis¶

Version: v2 · 2026-08-30 · History
Prime #: 226
Origin domain: Rhetoric
Also from: Art & Aesthetics, Communication & Media Studies, Information Theory
Aliases: Salience Engineering, Signal Amplification, Foregrounding
Related primes: Contrast, Hierarchy, Emphasis (Focal Point), Attention, Figure-Ground, Measurement Uncertainty and Observational Noise

Core Idea¶

Emphasis is the rhetorical and linguistic mechanism for foregrounding selected information against a background of less-prominent information. The concept rests on four interdependent components: (a) the foregrounded element — the specific piece of content (a word, phrase, visual element, claim, or computational token) targeted for heightened salience; (b) the contrast background — the field of surrounding or competing elements against which the foregrounded element stands out; © the emphasis vehicle — the technique or apparatus that produces the foregrounding effect, which may be prosodic (stress, intonation, duration), typographic (bold, italics, color, size, whitespace), positional (opening/closing placement, central visual location), syntactic (marked word order, isolated clause), rhythmic (repetition, meter, cadence), or computational (attention weight); and (d) the communicative function — the intended downstream effect on audience attention, comprehension, belief formation, decision-making, or action. Emphasis is recognized across disciplines with a unified structural signature: rhetorical tradition from Aristotle onward; modern phonological and prosodic linguistics (Bolinger 1972, Halliday 1967, Selkirk 1995, Pierrehumbert 1980); graphic and typographic design (Bringhurst 2004, Tufte 2001); information-structure linguistics (Lambrecht 1994, Krifka 2008, Chafe 1976, Prince 1981, Vallduví 1992, Büring 2007); contemporary visual-hierarchy design (Müller 2018); and machine-learning computational formalization (Vaswani et al. 2017, Harris 2020, Cresswell 1973).^[1]

How would you explain it like I'm…

Making One Part Stand Out

When you really want someone to hear ONE word, you say it louder. When you want a word to pop on paper, you can write it in BIG bold letters. Emphasis is any trick that makes one part stand out from the rest so people notice it first.

Highlighting What Matters

Emphasis is when you make one part stand out against everything around it. You can do it with your voice by saying a word LOUDER or sloo-ow-er. You can do it on a page using **bold**, italics, color, or big letters. You can do it in a picture by putting one thing in the bright spot or by leaving space around it. In every case there's a foreground (the thing you want noticed) and a background (everything else), and a technique that pulls the eye or ear toward the foreground.

Foregrounding Selected Information

Emphasis is the mechanism for making selected information stand out against a background of less-prominent information. It works through four interdependent pieces: (a) the *foregrounded element* — the word, image, claim, or token you want noticed; (b) the *contrast background* — everything else against which it stands out; (c) the *emphasis vehicle* — the technique that produces the contrast (loudness, bold type, color, position, isolation, repetition, or attention weight in a machine-learning model); and (d) the *communicative function* — the downstream effect you want, such as drawing attention, shaping belief, or prompting action. The structural signature shows up across spoken language, typography, visual design, information layout, and even neural networks.

Emphasis is the rhetorical and linguistic mechanism for foregrounding selected information against a background of less-prominent information. It rests on four interdependent components: (a) the *foregrounded element* — the specific content (a word, phrase, visual element, claim, or computational token) targeted for heightened salience; (b) the *contrast background* — the surrounding field against which it stands out; (c) the *emphasis vehicle* — the technique that produces the foregrounding, which may be prosodic (stress, intonation, duration), typographic (bold, italics, color, size, whitespace), positional (opening or closing placement), syntactic (marked word order, isolated clause), rhythmic (repetition, meter), or computational (attention weight); and (d) the *communicative function* — the intended downstream effect on attention, comprehension, belief, or action. The same structural signature is recognized across rhetorical tradition (Aristotle onward), prosodic linguistics, information-structure theory (Lambrecht, Krifka, Vallduví), typography and visual design, and transformer attention mechanisms in machine learning.

Structural Signature¶

The abstraction is defined by six italicized role-phrases that capture its recurring structural pattern across domains: the foregrounded element, the contrast background, the emphasis vehicle, the focus-information structure, the prosodic-vs-typographic mode, and the communicative function. In any instance of emphasis, these six roles are present: an element is selected for heightened salience from a field of potential elements; techniques that create differential salience are applied (stress, formatting, position, rhythm, or computational weighting); the techniques function against a baseline or contrast set; the resulting structure guides information flow according to information-structure principles; the implementation may privilege prosodic channeling (speech, oral delivery) or typographic channeling (visual design, written text) or both; and the entire operation exists to shape downstream cognition or action toward the emphasized subset. This role-set remains invariant even as the media, domains, and specific techniques change.^[2]

What It Is Not¶

Emphasis is not mere bold typography — it is a broader concept encompassing vocal stress, spatial position, repetition, syntactic marking, and computational attention weighting. Typography is one vehicle among many; the abstraction is the salience-engineering principle itself, not its implementation in any single medium. Emphasis is not all rhetoric or persuasion — rhetoric encompasses argument structure, emotional appeal, credibility-building, and many other techniques; emphasis is the specific operation of foregrounding selected content for heightened attention. It is not equivalent to stress in phonology alone — stress is one prosodic manifestation of emphasis, but emphasis manifests across multiple channels (visual, structural, computational) and stress alone does not exhaust the concept. Emphasis is not mere redundancy or repetition — while repetition can be an emphasis vehicle, not all repetition is emphasis (wallpaper patterns repeat without emphasizing any element), and emphasis can occur without repetition (a single bold word emphasizes via contrast, not through repetitive structure). Emphasis is not synonymous with all information-structuring operations — information structure is broader, encompassing topic-focus articulation, givenness-newness distinctions, and predication patterns; emphasis specifically targets salience-differential creation. Emphasis is not equivalent to all attention-direction mechanisms — while attention is closely related, attention encompasses pre-attentive perceptual processing, orienting reflexes, and many autonomous allocation processes; emphasis is the deliberate engineering of salience structure to bias or guide that allocation.^[3]

Broad Use¶

In rhetoric and oratory, emphasis operates through vocal stress (pitch variation, volume modulation, duration elongation), structural placement (key claims at speech openings and closings, climactic positioning), and figure-of-speech deployment (anaphora's repetitive emphasis on claims, epistrophe's closing-word emphasis, antithesis's structural emphasis through parallelism). King's "I Have a Dream" speech exemplifies multi-layered emphasis: vocal stress on dream, freedom, today, justice; anaphoric structural emphasis on "I have a dream that…"; climactic placement of the dream-sequence; allusive emphasis importing resonance from biblical and patriotic traditions. In written typography and graphic design, emphasis manifests through formatting (bold, italics, underline, capitalization, color contrast), structural devices (chapter breaks, pull quotes, margin notes, visual hierarchy), and compositional placement (rule of thirds, focal-point positioning, spatial isolation). Bringhurst's typographic principles and Tufte's data-visualization standards codify emphasis as a foundational design discipline. In information visualization and UX design, emphasis directs attention through visual hierarchy (size, color, contrast), spatial positioning (primary actions centered and enlarged, secondary actions peripheralized), and interactive affordances (button prominence, warning colors). In advertising and marketing, emphasis aggressively prioritizes brand and call-to-action (logos emphasized over disclaimers, sales signage over regular pricing, headline prominence over body copy), exploiting scarcity of attention to drive consumer behavior. In speech-language pathology and clinical linguistics, emphasis modeling is used to help patients with prosodic disorders develop or recover stress and intonation control. In AI text generation, emphasis modeling is critical to producing coherent, persuasive, and appropriate outputs — computational models must learn which tokens to weight heavily (emphasized) and which to de-weight (de-emphasized) to produce fluent, task-appropriate language.^[4]

Tensions and Failure Modes¶

T1 — Prosodic versus typographic emphasis. Speech relies primarily on prosodic channels (stress, intonation, duration) to create emphasis, while written text and visual media rely on typographic channels (formatting, size, color, spatial position). The tension emerges when translating emphasis across media: vocal stress in speech must be converted to typographic devices in print (capitalization, bold, italics, variable sizing); visual emphasis in graphic design must be converted to rhythmic or vocal devices in audio (pacing, dynamic range, pitch variation). Mismatched translation (applying speech-like repetitive emphasis to static visual design, or assuming silent readers will infer vocal stress from text alone) produces emphasis mismatch. The corrective is deliberate medium-aware translation, with different techniques optimized for different channels.^[5]

T2 — Universal emphasis mechanisms versus language-specific and culture-specific variation. Certain emphasis principles appear to be universal (contrast creates salience differentials across all sensory and linguistic systems), but the specific techniques, thresholds, and interpretations vary substantially by language, culture, and historical period. Stress-accent systems differ across languages (English places stress on the first syllable of many words; French distributes stress more evenly; tonal languages use pitch for lexical distinction, requiring phonological emphasis to operate on pitch contours not available in non-tonal languages). Typography conventions vary culturally (bold and capitalization carry different weight in German, English, French, and other writing traditions; right-to-left scripts require different compositional positioning than left-to-right scripts). Emphasis effectiveness depends on audience cultural repertoire: experts and novices respond to different emphasis signals; different eras valorize different stylistic conventions. The failure mode is assuming universal emphasis effectiveness; the corrective is audience-aware and culture-aware emphasis calibration.^[6]

T3 — Emphasis inflation and the collapse to no-emphasis. When emphasis is applied too broadly, its differential effect erodes and the field returns to flatness. If every sentence is bolded, nothing stands out as bold; if every organizational goal is labeled top-priority, no goal is genuinely prioritized; if a neural-network attention mechanism distributes weight uniformly across all inputs, the model has no way to emphasize task-relevant positions. The failure mode arises from well-intentioned over-emphasis: communicators believe everything is important, so they emphasize everything, producing a flat field with no differential salience. The corrective is disciplined selectivity — explicit constraints on how many elements can receive emphasis, with the principle that effective emphasis requires de-emphasis (implicit or explicit) of the surrounding field. Attention mechanisms in machine learning exhibit this pathology as "attention collapse," requiring regularization and normalization to prevent uniform or degenerate distributions.^[7]

T4 — Emphasis in sign languages and visuo-spatial modulation. Spoken emphasis relies on prosodic modulation along a temporal axis (stress, pitch, duration variations over time). Sign languages operate in visuo-spatial channels: handshape, location, movement, and non-manual markers (facial expression, body position, head tilt). Emphasis in sign language manifests through modulation of these visuo-spatial parameters: increased movement amplitude, repeated movements, specific hand configurations, facial intensity, or body shift. The tension is that the parametric structure differs fundamentally from spoken prosody, yet the communicative function (foregrounding selected content) remains similar. The corrective is recognition that emphasis principles transfer across modalities, but the implementation must respect the affordances of each modality.^[8]

T5 — Emphasis modeling in AI text generation and the computational emphasis alignment problem. Modern language models and generative AI systems must learn to generate appropriately emphasized outputs: key claims emphasized, supporting details de-emphasized, warnings and safety-critical information emphasized, filler de-emphasized. The tension is that emphasis patterns in training data are often implicit (conveyed through wording choice, structural placement, and subtle cues) rather than explicit signals. Models must infer emphasis distribution from text alone, without access to prosodic markers or visual formatting during training. The failure mode is misaligned emphasis: models generate technically correct but pragmatically inappropriate outputs (equal emphasis on routine and critical information, insufficient emphasis on task-relevant content). The corrective is explicit training on emphasis alignment, incorporating information-structure annotations and task-specific emphasis patterns.^[9]

T6 — Cultural and historical variation in emphasis conventions. Emphasis conventions are not universal across cultures or time periods. Some cultures favor direct, explicit emphasis (using superlatives, exclamation, repetition); others favor indirect, subtle emphasis (relying on positioning and implication). Victorian writing emphasized through elaborate syntactic construction and formal diction; contemporary writing often emphasizes through brevity and directness. Academic writing emphasizes claims through hedging and qualification; marketing emphasizes through assertion and superlatives. Audience interpretation of emphasis markers varies: a raised voice signals urgency in some contexts, anger or loss of control in others; bold text signals importance in contemporary design, but would have been unusual or read as error in 19^th-century typography. The failure mode is applying emphasis conventions from one cultural or historical context to an audience from another, producing misalignment and misinterpretation. The corrective is audience-aware emphasis, with explicit recognition that effective emphasis requires understanding the audience's interpretive norms.^[10]

Examples with Mapped-Back Closers¶

Formal/Abstract Example: Bolinger 1972 Prosodic Stress Placement. Dwight Bolinger's foundational work on accent and intonation demonstrates how prosodic emphasis forgrounds different information from identical semantic content. The sentence "John ate the apple" can be uttered with prosodic emphasis (marked by stress, intonation peak, duration elongation) on different words: "John ATE the apple" emphasizes the action (foregrounded element: the action of eating; contrast background: other possible actions; emphasis vehicle: prosodic stress and intonation peak on ate; communicative function: directing audience attention to what John did, not whether he did it). Alternatively, "JOHN ate the apple" emphasizes the subject (foregrounded element: John as the agent; contrast background: other possible agents; emphasis vehicle: prosodic stress on John; communicative function: directing attention to who ate, perhaps in answer to "Who ate the apple?"). The identical proposition is uttered with radically different emphasis structure, producing different discourse-contextual appropriateness and different audience interpretation of what information is being foregrounded. This example maps back to the core abstraction: (a) the foregrounded element differs (action vs. agent), (b) the contrast background shifts (other actions vs. other agents), © the emphasis vehicle is prosodic stress and intonation, (d) the communicative function is discourse-contextual appropriateness and information salience direction. Bolinger's work is foundational because it demonstrates that emphasis is not ornamental but structural — it carries information-allocation work that is not reducible to semantic content alone.^[1]

Applied/Industry Example: Tufte Data-Visualization Emphasis. Edward Tufte's principles of effective data visualization codify emphasis as a design discipline. In a chart comparing quarterly revenue across product lines, effective emphasis uses: bold lines for the primary product line, lighter or grayed lines for secondary lines; color contrast (saturated color for the story-line, desaturated for context); size hierarchy (large axis labels, small reference gridlines); and compositional positioning (the key story-line positioned to be read first, secondary comparisons supporting it). A poorly-designed chart emphasizes uniformly (all lines the same color and weight), forcing the audience to parse emphasis from raw data; an effective chart engineers differential salience (the important story-line emphasized through color, weight, position, and visual isolation) so the audience sees the point immediately. Modern UX design extends this principle: call-to-action buttons are emphasized (size, color, position); secondary actions are de-emphasized (neutral color, smaller size, peripheral position); danger actions receive special emphasis (warning colors) precisely to prevent accidental clicks by making them salient and distinctive. This example maps back: (a) the foregrounded element is the key insight or action (revenue trend, primary call-to-action), (b) the contrast background is supporting data or secondary actions, © the emphasis vehicle is visual hierarchy (color, size, position, line weight), (d) the communicative function is efficient information intake and action guidance. Tufte's work is influential because it demonstrates that emphasis is not decorative but functional — well-engineered emphasis reduces cognitive load and error, while poor emphasis produces confusion and misinterpretation.^[11]

Substrate Independence¶

Emergent Formalization (Language) is a moderately substrate-independent prime — composite 3 / 5 on the substrate-independence scale. The process — informal usage conventionalizing into formal structure through frequency-driven crystallization — is structurally suggestive and could in principle describe cultural evolution or the hardening of institutional norms. But it is grounded in linguistics and language change, and the proposed extensions to other domains read as metaphorical rather than structural reuse. So while the underlying logic hints at wider applicability, in practice the prime stays a language-change phenomenon, which holds it to the middle of the scale.

Composite substrate independence — 3 / 5
Domain breadth — 3 / 5
Structural abstraction — 4 / 5
Transfer evidence — 2 / 5

Relationships to Other Abstractions¶

Current abstraction Emphasis Prime

Parents (2) — more general patterns this builds on

Emphasis presupposes Attention Prime

Emphasis presupposes attention because foregrounding selected information against a background only matters as a mechanism for directing scarce attentional resources.
Emphasis is a decomposition of Figure-Ground Prime

Emphasis is the specific shape figure-ground takes in rhetorical and linguistic foregrounding of selected content.

Children (5) — more specific cases that build on this

Flashing arrow Domain-specific is a kind of Emphasis

The proposed strict upward parent is prime:emphasis.
Intensive Pronoun Domain-specific is a kind of Emphasis

Proposed parent: prime:emphasis. Every intensive-pronoun occurrence foregrounds its associate through a specialized grammatical vehicle and makes that referent prominent relative to alternatives.
Emphasis (Focal Point) Prime is a kind of Emphasis

A focal point is a specialization of emphasis in which the foregrounded element is positioned to register as the primary site of attention in a visual field.

▸ Show 2 more

Hierarchy paths (2) — routes to 2 parentless roots

Emphasis → Attention

Show alternative path (1)

Neighborhood in Abstraction Space¶

Emphasis sits in a sparse region of abstraction space (74^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.

Family — Unclustered & Miscellaneous (424 primes)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-10

Not to Be Confused With¶

Emergent Formalization must be distinguished from Formalization itself, though the two are closely related. Formalization is the process of taking informal, tacit, or intuitive knowledge and expressing it in explicit, structured, often mathematical form—a set of axioms, definitions, rules, or algorithms that capture the essential logic of the domain. Emergent Formalization specifically describes formalization that arises not by initial deliberate design but through iterative discovery and synthesis from concrete practice, patterns, examples, and tacit understanding that practitioners accumulate. Formalization is the act and outcome of expressing knowledge explicitly; Emergent Formalization names the temporal and epistemic process by which formal systems gradually crystallize from informal grounding. Both involve the creation of explicit structure, but formalization can be top-down (designing axioms and rules from first principles and then applying them to practice) or bottom-up (observing practice, extracting patterns, gradually articulating rules). Emergent Formalization emphasizes the bottom-up, discovery-based aspect. The distinction matters because it clarifies why some formal systems (mathematics developed from axioms) feel elegant but disconnected from practice, while others (statistics developed from practical problems of insurance and agriculture) feel grounded and transferable. Both pathways are valuable, but they produce different epistemological stances and practical consequences.

Emergent Formalization is also distinct from Abstraction, though the two often occur together. Abstraction is the process of selectively filtering and retaining essential features while discarding inessential detail—a circuit diagram abstracts away the physical substrate of electronics and retains only functional relationships. Abstraction is about reduction and compression. Emergent Formalization is about explication and structure-building—taking implicit, intuitive understanding and making it explicit, formal, and rigorous. Abstraction often precedes formalization (you abstract to identify the essential features you want to formalize), but they are not the same operation. An artist might abstract visual patterns in nature into essential shapes; a mathematician might then formalize those abstractions into geometric structures. Abstraction prepares the conceptual ground; formalization builds explicit structure. You can have pure abstraction without formalization (a poet abstracting emotion into metaphor without mathematical structure), and you can have formalization without prior abstraction (writing down rules for a process you already understand implicitly, without abstracting away inessential details). The two concepts are most powerful in combination but remain distinct.

Emergent Formalization differs from Canonicalization, though the two can coincide. Canonicalization is the process of establishing a standard, reference, or official version of something—a canonical text, a canonical form of an equation. Canonicalization is about standardization and reference-setting. Emergent Formalization is about the process and structure of making tacit knowledge explicit. A formal mathematical framework might emerge from practice, and then be canonicalized (enshrined in textbooks, made official, adopted by a field). Canonicalization follows and institutionalizes formalization; it is not formalization itself. The two often co-occur—a field that formalizes its knowledge through emergent discovery often then canonicalizes that formal system—but they are distinct operations. Canonicalization is about authority and institutionalization; emergence is about discovery. The distinction matters because a formal system can emerge and be widely used without being canonical (people adopt it informally without institutional decree), and a canonical system can be imposed top-down without having emerged from practice (though such systems often fail to gain traction because they lack grounding in actual usage).

Finally, Emergent Formalization is distinct from Specification, though specifications can be formal and can emerge. Specification is the process of defining requirements, constraints, or detailed behavior of a system or artefact: a software specification details what the code should do, an engineering specification defines material properties and tolerances. Specification is about defining requirements and acceptance criteria. Emergent Formalization is about discovering and articulating the underlying logic and structure of a domain. They can coexist—a software specification often formalizes emergent patterns from prior code and usage—but they address different questions. Specification asks "what should this system do?" and is forward-looking (it guides creation). Emergent Formalization asks "what logic and structure does this activity or domain actually embody?" and is retrospective (it articulates what exists). A specification might be formal and precise; an emergent formalization might also be formal and precise; but one is prescriptive, the other is descriptive. The distinction matters because confusing them leads to over-specification (writing requirements that constrain beyond what is essential, locking in arbitrary choices) or under-specification (failing to capture essential patterns that users rely on). Emergent formalizations often inform specifications, translating tacit practice into explicit requirements.

Solution Archetypes¶

Solution archetypes in the catalog that build on this prime — directly (this prime is a source ingredient) or as a related prime.

Built directly on this prime (6)

Also a related prime in 20 archetypes

Backfire-Aware Suppression Design: Handle harmful or unwanted information without making the act of suppression more newsworthy than the information itself.
Cascaded Hierarchical Recognition: Recognize complex cases by moving attention through a hierarchy of coarse filters and fine discriminators instead of trying to inspect every possible feature at once.
Contingency-Visibility Across Scales: Compare micro-level detail with macro-level aggregation so local contingency is not erased and broad structure is not ignored.
Contrastive Differentiation: Clarify a concept, option, signal, or identity by making its differences from nearby alternatives explicit.
Distraction Minimization for Deep Engagement: Reduce avoidable interruptions and competing attentional demands so people can enter, maintain, and recover deep engagement with the target task.
Event–Narration Order Decoupling: Separate what happened when from the order in which it is shown, told, taught, or argued, then keep the two orders explicitly mapped so presentation can be optimized without corrupting chronology.
Figure-Ground Structuring: Structure a field so the intended figure stands forward against a supportive ground that supplies context without competing for the same attention.
Material Literalness Foregrounding: Foreground the literal material or medium of a thing by suppressing decorative, symbolic, and illusionistic cues that would make the viewer treat it as something else.
Narrative Construction Audit: Audit how a story selects events, assigns causality, centers actors, creates themes, and omits alternatives.
Negative Priming Avoidance: Do not let the warning, prohibition, or correction make the unwanted idea the easiest thing to think about.

▸ Show 10 more

Notes¶

Linguistic-pragmatics origin (Bühler's Sprachtheorie 1934 introduced the "origo" of speaker/here/now; Fillmore 1971/1997; Levinson 1983). Software-systems transfer is well-established in template engines, internationalization frameworks, and log-schema design. Companion to #322 contextual_mode_switching (which concerns switching whole communicative modes based on context, while deixis operates on individual referential expressions within any mode). Companion to #320 cooperative_principle_gricean_maxims (which governs how hearers infer speaker meaning, including deictic resolution). Companion to #315 speech_act_theory_illocution_perlocution (which describes the pragmatic force of utterances; deixis specifies how context is bound into those utterances' reference).

References¶

[1] Bolinger, Dwight L. (1972). "Accent Is Predictable (If You're a Mind-Reader)". Language, 48(3), 633–644. Foundational argument that sentence-accent placement is governed by semantic and emotional highlighting rather than syntax — supports the prosodic-stress/emphasis-vehicle claim and the 'John ATE the apple' worked example. registry ↩a ↩b

[2] Krifka, Manfred (2008). "Basic Notions of Information Structure". Acta Linguistica Hungarica, 55(3–4), 243–276. Defines focus (as indicating alternatives), givenness, and topic within a common-ground model — supports the focus/information-structure role in the structural signature. registry ↩

[3] Selkirk, Elisabeth O. (1995). "Sentence Prosody: Intonation, Stress, and Phrasing." In John A. Goldsmith (ed.), The Handbook of Phonological Theory, 550–569. Blackwell. Establishes that prominence is realized through multiple prosodic mechanisms (intonation, stress, phrasing) — supports the claim that stress is one channel of emphasis among many, not the whole concept. registry ↩

[4] Bringhurst, Robert (2004). The Elements of Typographic Style (Version 3.0). Hartley & Marks. Canonical typography reference codifying visual rhythm, hierarchy, and emphasis (bold, italic, scale, spacing) as a design discipline — supports the written-typography emphasis claim. registry ↩

[5] Pierrehumbert, Janet B. (1980). The Phonology and Phonetics of English Intonation. Ph.D. dissertation, MIT. Autosegmental-metrical account of English intonation — supports the prosodic-channel side of the prosodic-vs-typographic emphasis tension (T1). registry ↩

[6] Lambrecht, Knud (1994). Information Structure and Sentence Form: Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge University Press. Foundational treatment of how sentence form encodes focus/topic relative to hearer knowledge — supports the information-structure framing; only weakly bears on the specific cross-linguistic stress-system claim in T2 (see flag). registry ↩

[7] Zhai, Shuangfei; Likhomanenko, Tatiana; Littwin, Etai; et al. (2023). "Stabilizing Transformer Training by Preventing Attention Entropy Collapse". ICML 2023 (arXiv:2303.06296). Documents attention entropy collapse — attention distributions becoming degenerate/peaked — and shows regularization/normalization mitigations; supports the machine-learning 'attention collapse' half of T3 (emphasis inflation / collapse to no-emphasis). registry ↩

[8] Halliday, Michael A. K. (1967). "Notes on Transitivity and Theme in English, Part I". Journal of Linguistics, 3(1), 37–81. Foundational treatment of theme/information distribution in the English clause. Real and correctly cited, but it concerns spoken/written English clause structure, not sign-language visuo-spatial emphasis (the T4 claim it is attached to — see flag). registry ↩

[9] Mass, Yosi; Shechtman, Slava; Shmueli, Moran; et al. (2018). "Word Emphasis Prediction for Expressive Text to Speech". INTERSPEECH 2018, 2868–2872. Deep-neural model that learns which words to emphasize for expressive synthesis — supports the T5 claim that computational systems must learn to weight some tokens as emphasized and de-weight others. registry ↩

[10] Prince, Ellen F. (1981). "Toward a Taxonomy of Given/New Information." In Peter Cole (ed.), Radical Pragmatics, 223–255. Academic Press. A taxonomy of given/new (assumed-familiarity) information. Real and correctly cited, but it does not substantiate the T6 claim about cultural and historical variation in emphasis conventions (see flag). registry ↩

[11] Tufte, Edward R. (2001). The Visual Display of Quantitative Information (2^nd ed.). Graphics Press. Codifies data-graphic emphasis through contrast, weight, size, and position (data-ink discipline) — supports the Tufte data-visualization emphasis example. registry ↩

[12] Chafe, Wallace L. (1976). "Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View." In Charles N. Li (ed.), Subject and Topic, 25–55. Academic Press. Bibliography-only; foundational information-packaging source. registry

[13] Vallduví, Enric (1992). The Informational Component. Garland Publishing (Outstanding Dissertations in Linguistics; University of Pennsylvania). Bibliography-only; argues topic/focus are primitives of an informational component. registry

[14] Büring, Daniel (2007). "Semantics, Intonation, and Information Structure." In Gillian Ramchand & Charles Reiss (eds.), The Oxford Handbook of Linguistic Interfaces, 445–473. Oxford University Press. Bibliography-only; corrected from the wrong title/venue ('Linguistics and Philosophy 30(6), 699–728') given in the prime. registry

[15] Müller, Clemens (2018). Visual Hierarchy in Human-Computer Interaction: Principles and Patterns. International Journal of Human-Computer Studies, 115, 122–135. withdrawn registry

[16] Cresswell, M. J. (1973). Logics and Languages. Methuen. Bibliography-only; applies formal-logic semantics (including indexical/context-dependent expressions) to natural language. registry

[17] Harris, D. (2020). Speech Act Theory and Artificial Intelligence. In Handbook of Pragmatics and Language (pp. 812–838). Springer. Harris speech act theory and artificial intelligence LLMs dialogue agents intent classification. registry