Semantic Discourse Analysis

Compiled and edited by Ross Woods, 2026

In Discourse Analysis (DA) and text linguistics, semantic categories bridge the gap between individual sentence meanings and the overarching global meaning of a text (macrostructures). Instead of looking at dictionary definitions, discourse semantics analyzes how concepts relate to one another to build coherence across a whole text.

Discourse Analysis (DA) based on semantic categories examines how meaning is organized across stretches of language, not just individual words or sentences.

Key semantic categories often used in DA include:

Category	Focus	Example
Participants	Who is involved	teacher, student, government, citizens
Actions / Processes	What is happening	argue, explain, blame, request
Objects / Topics	What is being discussed	education, policy, identity
Time	When events happen	yesterday, now, in the future
Place	Where discourse is located	school, workplace, nation
Evaluation	Attitudes or judgments	good, dangerous, unfair
Modality	Certainty, obligation, possibility	must, might, should
Cause–Effect	Reasons and consequences	because, therefore, leads to
Comparison / Contrast	Similarities or differences	however, unlike, more than
Identity / Roles	How people are represented	expert, victim, outsider, leader

A DA based on semantic categories typically asks:

What meanings are emphasized?
Who or what is represented positively or negatively?
What actions are linked to which participants?
What values, assumptions, or ideologies appear in the discourse?
How do semantic choices shape the reader’s interpretation?

Example sentence: “The government must protect vulnerable citizens from rising prices.”

Semantic analysis:

Participant: the government, vulnerable citizens
Action: protect
Modality: must
Evaluation: vulnerable
Cause/context: rising prices
Ideological meaning: government is framed as responsible for social protection.

Three Distinct Levels

Depending on whether you are analyzing text microstructures (clause-to-clause relations) or ideological structures (Critical Discourse Analysis), these semantic categories generally fall into three distinct levels:

1. Local Semantic Relations (Microstructures)

These categories describe how consecutive propositions or sentences connect logically. Teun van Dijk and other text linguists categorize these regional links into several primary semantic operations:

Cause / Consequence: One proposition states the cause, the next states the effect. Example: "The river swelled. The nearby village flooded."
Specification / Clarification: A general statement is followed by more detailed information or an explicit example.
Contrast / Opposition: Two ideas are juxtaposed to highlight differences or unexpected outcomes. Example: "The policy looked perfect on paper, but it failed completely in practice."
Functional Addition: Adding info that expands the mental model of the situation without necessarily being a direct cause or consequence.

2. Global Semantic Structures (Macrostructures)

Discourse analysis looks at how individual semantic categories build up into overall themes, topics, and mental models.

As shown above, discourse relies on the interaction between formal grammar, semantics (meaning), and pragmatics (context). To manage this interaction at a global level, we use:

Macropropositions: The core thematic categories that summarize blocks of text (what a paragraph or chapter is "about").
Implication and Entailment: Implicit semantic information. Often, what is left unsaid or assumed to be common knowledge forms the strongest semantic category in an analysis.

3. Ideological & Critical Semantic Categories

In Critical Discourse Analysis (CDA), semantic categories are used to uncover power dynamics, bias, and underlying ideologies within a text.

Semantic Category	Definition	Discourse Function
Actor/Agent Roles	Who is performing the action vs. who is passive.	Can downplay responsibility (e.g., using passive voice: "Mistakes were made").
Lexical Overlexicalization	Using an abundance of different words for a single concept.	Indicates an area of high preoccupation, anxiety, or ideological focus.
Euphemism & Dysphemism	Replacing harsh words with neutral ones, or vice versa.	Softens negative realities or weaponizes plain facts to bias the reader.
Us vs. Them Polarisation	Grouping descriptions into positive self-presentation and negative other-presentation.	Solidifies in-group solidarity and marginalizes out-groups.

Multimodal Semantic Categories

Modern discourse analysis frequently deals with texts that combine both words and visuals. When analyzing multimodal documents, semantic categories expand across different mediums to maintain structural meaning:

As categorized in the table above, semantic categories like synonymy (same meaning), antonymy (opposing meanings), and homonymy (same form, different meanings) do not just apply to spoken or written words–they operate across visual elements and physical interactions as well, allowing researchers to track cohesive ideas through imagery and layout layout.

To demonstrate how this works in practice, let’s analyze a short, hypothetical excerpt from an academic policy proposal regarding university curriculum restructuring. First, we will look at the raw text, and then we will map out its local semantic relations (microstructures) and its global semantic theme (macrostructure).

The Sample Text

The current undergraduate curriculum has remained largely unchanged for over fifteen years. Consequently, our graduates face increasing difficulties when entering a highly digitized workforce. We must immediately integrate advanced data literacy and practical coding elements into every major field of study. For example, history students could use automated document analysis tools to track language shifts in historical archives, while biology majors could process environmental datasets using basic Python scripts. This systemic updates will bridge the gap between traditional theory and modern professional realities

1. Local Semantic Relations (Microstructures)

At the micro-level, we look at how each proposition (clause or sentence) logically connects to the next. The text builds a clear argumentative chain using specific semantic operations:

Sentence 1 → Sentence 2: Cause / Consequence
- Sentence 1: The curriculum hasn't changed.
- Sentence 2: Graduates struggle in a digital job market.
- Discourse Function: Establishes the problem. The linguistic marker "Consequently" explicitly flags this local semantic dependency.
Sentence 2 → Sentence 3: Problem / Solution (Specification)
- Sentence 3: We need to integrate data literacy and coding.
- Discourse Function: Directly addresses the consequence identified in the previous sentence by proposing a mandatory action.
Sentence 3 → Sentence 4: General / Particular (Exemplification)
- Sentence 4: Gives specific examples for history (document analysis) and biology (Python scripts).
- Discourse Function: Grounding abstract policy ("data literacy") into concrete, recognizable academic scenarios. The marker "For example" signals this shift from general to particular.
Sentence 4 → Sentence 5: Purpose / Evaluation

Sentence 5: This update will bridge the gap between theory and practice.
Discourse Function: Synthesizes the examples back into a final justification, assigning a positive value ("bridge the gap") to the proposed changes.

2. Global Semantic Structures (Macrostructures)

The global semantic structure reduces the text down to its core semantic themes, stripping away the specific examples to reveal the underlying "mental model" the text wants the reader to adopt.

We can compress the text into two primary macropropositions:

Macro-Problem: Outdated university curricula are harming student employability in a digital economy.
Macro-Solution: Integrating practical data skills across all disciplines will restore institutional relevance.

When we combine these, the overarching global semantic category (the Core Macrostructure) of the entire passage is: Curriculum Modernization as an Employability Imperative.

3. Ideological & Critical Semantic Categories

Even in a brief policy text, a critical discourse analysis reveals subtle semantic positioning through word choice and grammar:

Us vs. Them / Agent Roles: Notice the shift in responsibility. The curriculum is described passively as having remained largely unchanged (no one is explicitly blamed for the stagnation). However, when it comes to the solution, an active, collective agent is introduced: "We must immediately integrate... This positions the current leadership as proactive problem-solvers rather than creators of the original issue.
Lexical Polarization: The text sets up a binary contrast between two semantic fields:

Negative/Stagnant: "Current," "unchanged," "fifteen years," "difficulties," "traditional theory."
Positive/Dynamic: "Advanced," "practical," "immediately," "modern professional realities."

Implication / Presupposition: The text presupposes that "traditional theory" and "modern professional realities" are currently disconnected. By using the phrase "bridge the gap," it implies a deficit exists without needing to explicitly prove it.

While both Teun van Dijk an Norman Fairclough are founding figures of Critical Discourse Analysis (CDA) who share the goal of exposing power abuse, dominance, and inequality, they approach text analysis through radically different theoretical lenses.

The core difference lies in how language connects to society . Fairclough maps language directly to social structures and institutions, whereas van Dijk argues that society and language cannot interact without a middleman: human cognition.

The Core Theoretical Difference

Teun van Dijk’s Socio-Cognitive Approach

Van Dijk argues that there is a cognitive gap between social structures and discourse. A social system (like institutional racism or corporate power) cannot directly write a text or influence a speaker. Instead, it must first influence people's minds.

The Interface: Cognition (mental models, social representations, attitudes, and ideologies).
How it works: People understand, produce, and analyze discourse based on their personal Mental Models of a situation. These personal models are shaped by broader, socially shared ideologies (e.g., feminist, neo-liberal, or racist ideologies).
Key Focus: Analyzing how elite groups control the mind of the public by controlling the access to discourse (e.g., who gets to speak in the media, what topics are highlighted in headlines).

Norman Fairclough’s Socio-Cultural Approach

Fairclough views discourse as a form of social practice. He is heavily influenced by Marxist theory (particularly Gramsci's concept of hegemony and Michel Foucault. For Fairclough, language usage is inherently tied to power struggles and social change, acting simultaneously as a product of social structures and a tool to reshape them.

The Interface: Discourse Practices (production, distribution, and consumption of texts).
How it works: Texts are produced within specific institutional contexts that dictate what can be said and how. Over time, these practices become naturalized as "common sense," maintaining unequal power dynamics.
Key Focus: Intertextuality (how texts draw upon other texts or existing genres) and how changes in language reflect shifts in broader socio-cultural power relations (e.g., the "marketization" of higher education discourse).

Analytical Framework Comparison

When it comes to analyzing a concrete text, their methodological toolkits differ significantly.

Dimension	Teun van Dijk (Socio-Cognitive)	Norman Fairclough (Socio-Cultural)
Analytical Dimension	The Triangulation: Discourse, Cognition, and Society.	The Three-Dimensional Model: Text, Discourse Practice, and Social Practice.
Micro-level focus	Discourse Semantics: Macrostructures (global themes), microstructures (local meanings), semantic propositions, and mental models.	Linguistic Description: Grammar, vocabulary, syntax, cohesion, text structure, and metaphor.
Meso-level focus	Cognitive Operations: Schemas, social stereotypes, in-group/out-group categorization ("Us vs. Them").	Intertextual Analysis: Interdiscursivity (how genres mix), how texts are distributed, transformed, and consumed.
Macro-level focus	Social Power & Access: Who controls the means of communication and how elites manufacture public consent.	Social Explanation: Ideological effects, institutional power dynamics, hegemony, and cultural reproduction.

A Practical Example: Analyzing a News Headline

To see how their frameworks diverge in practice, imagine analyzing this headline about university budget cuts: University Announces Measures to Streamline Faculty Allocations

Using Van Dijk's framework, you will look for the underlying mental model and ideological structures.

You would note the use of the euphemism "Streamline Faculty Allocations" to mask the reality of firing professors. You would argue that this is an elite strategy to shape the reader’s mental model of the event—transforming a harmful social action into an objective, administrative necessity. You would analyze the semantic strategy of positive self-presentation (the university acts responsibly to "streamline") while marginalizing the negative consequences for teachers and students.

Using Fairclough's framework, you will use his Three-Dimensional Model to trace social power.

Text (Description): Analyze the grammar. It uses nominalization ("allocations") and passive concepts to obscure who is actually getting hurt.
Discourse Practice (Interpretation): Analyze the interdiscursivity. The headline imports corporate, managerial language "streamline," "allocations") into an educational institution. It shows how business discourse has invaded academia.
Social Practice (Explanation): Explain how this text serves a broader hegemony. By framing education in corporate terms, the text reinforces the neo-liberal social shift where universities function as market-driven businesses rather than public goods.

Summary Rule of Thumb

If you want to analyze how texts manipulate beliefs, biases, and mental representations, lean toward van Dijk.
If you want to analyze how texts reflect institutional shifts, ideological hegemony, and cultural power struggles, lean toward Fairclough.

CC BY-NC-ND
This work is released under a CC BY-NC-ND license, which means that you are free to do with it as you please as long as you (1) properly attribute it, (2) do not use it for commercial gain, and (3) do not create derivative works.
Disclaimer: Artificial Intelligence (AI) was used in some parts of this book. If AI plagiarized your content, contact the author with evidence to initiate changes.