Design Principles of the Encyclopedia of Arabic Rhetoric

Page information
🔗 DOI	10.64393/balagha.design.principles
Date	14 February 2026

Introduction

The Encyclopedia of Arabic Rhetoric is a curated compendium of Arabic rhetorical devices. It is designed primarily as a structured resource to assist in the identification of Arabic rhetorical devices in a way that is:

• Accurate - a rhetorical device should be correctly identified (a separate document specifies the Principles for Identifying Arabic Rhetorical Devices),
• Consistent - the same rhetorical device should be identified in the same way each time, and
• Reproducible - different analysts should agree on their identification of the same rhetorical device.

Because the Encyclopedia departs in certain ways from the organisation of classical pedagogical manuals, this page explains the eight overarching principles that govern its design. Each of these principles interacts to form a methodological pipeline that transforms classical Arabic rhetorical science into an operational analytical framework for Arabic rhetoric.

Principle 1: The Encyclopedia is Based on Classical Sources

The Encyclopedia does not invent new rhetorical devices; it organises those already established in the classical tradition. It is grounded in published works of classical Arabic rhetoric, and references are provided. The Encyclopedia preserves the classical understanding of each device's mechanism and scope; what changes is only the organizational framework for accessing them.

The Encyclopedia does not aim to innovate new rhetorical devices, but to organise, clarify, and structure the classical rhetorical tradition in a way that supports systematic analysis. This includes preserving the internal conceptual boundaries of Arabic rhetorical terminology without subordinating them to approximate categories inherited from Greek, Latin, or modern Western rhetorical traditions.

Principle 2: The Encyclopedia has Criteria for Inclusion

The purpose of the Encyclopedia is not merely to catalogue every named phenomenon in the rhetorical tradition, but to provide a stable, operational framework for the identification and analysis of Arabic rhetoric. Accordingly, the Encyclopedia's inclusion criteria ensure that each device contributes meaningfully to:

• Clear rhetorical identification,
• Structural analysis,
• Consistent annotation, and
• Computational or corpus-based operationalisation.

A device is included only if it satisfies methodological standards that ensure analytical clarity, structural distinctiveness, and reliable identification. Devices that lack definitional precision, overlap extensively with broader interpretive phenomena, or cannot be annotated with reasonable reliability are not included as independent categories. Exclusion does not deny a device's historical attestation or rhetorical value. Rather, it reflects a principled commitment to taxonomic coherence and operational clarity.

Inclusion Criteria

A rhetorical device must:

1. Be rooted in classical Arabic rhetorical tradition (see above, Principle 1: The Encyclopedia is Based on Classical Sources).

   • An extension of "The Encyclopedia does not invent new rhetorical devices" (stated in Principle 1), the Encyclopedia cannot rely on just one or two sources for rhetorical device information.

   • The classical Arabic rhetorical tradition must attest to a rhetorical device through multiple sources for it to be considered for inclusion.

2. Have a clearly defined mechanism (see below, Principle 4: Rhetorical Devices Are Atomic Units).

   • The rhetorical device must be based on a specific and identifiable rhetorical operation, not merely a general stylistic effect.

3. Be structurally distinguishable from other rhetorical devices.

   • It must have defining features that prevent misidentification with similar devices.

4. Be operationally identifiable.

   • Classical Arabic rhetorical tradition should attest to the existence of the device with sufficient definitional clarity and exemplification to permit stable operationalisation.

   • Independent analysts should be able to recognise the device using defined criteria.

5. Allow reliable annotation in corpus analysis.

   • The device must be capable of consistent identification in applied rhetorical or computational analysis.

The Encyclopedia maintains a list of devices which have been excluded, along with the reasoning for exclusion.

Principle 3: The Encyclopedia Recognises Four Consecutive Levels of Every Rhetorical Action

The Encyclopedia distinguishes between four analytically distinct, sequential levels of every rhetorical action:

1. The Rhetorical Intent Level - the pragmatic effect that the communicator wishes to invoke in the addressee.
2. The Rhetorical Mechanism Level – the rhetorical operation executed by the communicator through language, to achieve the Rhetorical Intent.
3. The Rhetorical Relationship Level – the shared cognitive relationship established between the communicator and the addressee through the communicator's execution of the rhetorical action at the Mechanism Level, and the addressee's reception of it.
4. The Rhetorical Effect Level – the rhetorical effect invoked in the addressee through the cognitive relationship that the communicator established through the execution of the rhetorical action at the Mechanism Level.

For example, when the communicator intends to to explain how brave Ahmed is. The communicator could say, "Ahmed is brave" (Aḥmad shujāʿun). This would be a truthful and correct statement. However, if the communicator had the intent of intensifying the addressee's perception of Ahmed's bravery, a Simile such as, "Ahmed is like a lion in bravery" (Aḥmad kal-asad fī al-shajāʿa) could be used. In this rhetorical action, Ahmed's bravery is stated to be similar to a lion's bravery.

This rhetorical mechanism of explicit comparison establishes a cognitive relationship between the communicatorand the addressee. The relationship depends on a crucial shared understanding: that lions are exceptionally brave. The relationship causes the addressee to equate Ahmed's bravery with a lion's bravery. If all the links in the chain have worked, the communicator will have evoked a heightened impression of Ahmed's bravery in the mind of the addressee.

The Encyclopedia's four-level model of rhetoric can be applied to any rhetorical action. As an example, in Metaphor, the rhetorical mechanism is semantic transfer from one word to another, the rhetorical relationship is projection of attributes from one semantic domain to another, and the rhetorical effect may be emphasis, compression, or emotional force - which may or may not have been the communicator's intent.

The model also illustrates the exceptional communicative power of rhetoric. From the communicator's intent, through the selection of a context-appropriate rhetorical mechanism and its execution, to the establishment of a many possible cognitive rhetorical relationships between communicator and addressee, to the multitude of rhetorical effects in the addressee - the efficient use of one rhetoric mechanism causes an exponential increase in communicative power as it progresses and amplifies through each level.

From the rhetorical critic's point of view, the model also exposes each discrete level for independent analysis. The distinction between levels is necessary because rhetorical analysis otherwise collapses these levels into one, causing confusion and disagreement amongst annotators and critics. Multiple rhetorical mechanisms and devices may establish a relationship between communicator and addressee. The resulting rhetorical effects may or may not correspond with the communicator's intent. A rhetorical device is not identical to the relationship it creates, nor to the effect it produces, which in turn may not be the effect that was intended.

The Encyclopedia primarily catalogues rhetorical devices at the rhetorical mechanism layer, as described below in Principle 4: The Encyclopedia Uses a Mechanism-First Typology. Rhetorical device descriptions include possible rhetorical effects where useful. A separate page describes common Rhetorical Relationships.

The Principles for Identifying Arabic Rhetorical Devices operationalise the four-layer model for practical rhetorical analysis. The relationship between the theoretical model and the identification framework is shown below:

Design Principle layer	Identification consequence
Rhetorical intent	Excluded from direct analysis (Principle 1)
Rhetorical mechanism	Evaluated in Step 1 ("Identification of a hallmark rhetorical mechanism") and Step 2 ("Substitution with non-rhetoric equivalent")
Rhetorical relationship	Evaluated in Step 3 ("Identification of positive contextual indicators of rhetorical usage")
Rhetorical effect	Evaluated in Step 4 ("Evidence of possible rhetorical effect")

Principle 4: The Encyclopedia Uses a Mechanism-First Typology

The Encyclopedia classifies rhetorical devices according to the mechanism by which they operate, rather than the pragmatic effect they produce. This is because a rhetorical effect – such as intensification, glorification, lamentation, amazement, rebuke, or emotional emphasis – may arise from multiple independent mechanisms. For this reason, identifying the effect alone is not sufficient for consistent and reproducible analysis.

Instead, the analyst identifies the linguistic transformation that produces the effect, such as:

• Deletion
• Substitution
• Repetition
• Contrast
• Projection

Once the mechanism is identified, the rhetorical device can be classified consistently and reproducibly.

For example, rather than listing "Glorification" (taʿẓīm) as a standalone device, the Encyclopedia identifies the specific syntactic, lexical, or semantic deviation that produces the effect of glorification. The emphasis is therefore placed on how the effect is created, not merely on the fact that it occurs.

Consider the sentence: "He is a lion in battle." (huwa asadun fī al-maʿraka). The effect may be glorification or intensification. However, the mechanism itself is identified: semantic transfer from one domain (lion) to another (human). The device is therefore classified as Simile, because the classification is based on the mechanism (semantic comparison), not on the emotional effect it produces.

The Encyclopedia distinguishes between mechanism-driven rhetorical devices and rhetorical effects. Only those rhetorical phenomena that introduce a distinctive and reproducible structural transformation of linguistic or semantic structure are classified as devices. Effects such as praise, blame, intensification, sarcasm, or irony, which depend upon mechanisms already accounted for elsewhere, are not treated as independent entries.

The Encyclopedia's mechanism-first approach enhances clarity, precision, and reproducibility in rhetorical analysis. The mechanism-first typology has direct consequences for rhetorical identification. The corresponding Principle 2: Mechanism-First Identification operationalises this design principle by requiring analysts to classify rhetorical devices according to their underlying rhetorical mechanism rather than their rhetorical effects.

Source Example

﴿يَٰحَسۡرَةً عَلَى ٱلۡعِبَادِ﴾

Ah! Alas for my servants.

Qur'an 36:30

Explanation

This Quranic verse illustrates why the Encyclopedia adopts a mechanism-first approach. Classical sources differ themselves regarding the rhetorical effect produced by the vocative particle yā, variously describing it as astonishment, lamentation, grief, or emotional intensification. An effect-first typology would result in this single rhetorical event being tagged by different annotators according to a multitude of pragmatic effects. This would reduce reproducibility of annotation.

The Encyclopedia addresses this by foregrounding objective linguistic mechanisms over subjective interpretation. It distinguishes (a) the observable structural transformation within the language performed by the communicator, from (b) the interpretive rhetorical effects that addressees may infer from it. The Encyclopedia describes the rhetorical effects as an aid to identifying the transformation, but the rhetorical effects themselves are not treated as independently annotatable rhetorical events.

In this verse, the underlying rhetorical mechanism is consistent across interpretations: the vocative particle yā is used non-literally and departs from its primary function of calling. Within the Encyclopedia's mechanism-first typology, the rhetorical event is consistently identified as VOCATIV: The Vocative, regardless of interpretive variation in rhetorical effects, which are not themselves treated as independently annotatable devices.

Principle 5: Rhetorical Devices Are Atomic Units

Each rhetorical device in the Encyclopedia corresponds to a single underlying mechanism such as deletion, substitution, repetition, contrast, or projection.

Some classical works describe compound devices that combine multiple mechanisms. For example, Parallel Antithesis (muqābalah) combines semantic contrast (Antithesis) and structural parallelism (Parallelism).

From an analytical perspective, this presents a difficulty. Both Antithesis and Parallelism must also exist independently as devices in cases where they occur separately. Treating a compound as a single device can obscure the fact that multiple distinct linguistic transformations are taking place.

The Encyclopedia therefore defines devices as atomic units: one mechanism per device. If multiple mechanisms can occur simultaneously in the same phrase or sentence, they are treated as separate devices. If an analyst encounters Parallel Antithesis (muqābalah), the Encyclopedia requires them to tag both the Antithesis and Parallelism separately, allowing for more precise analysis and clearer understanding of which mechanisms are actually present. This reflects the Encyclopedia's objective of presenting rhetorical devices as discrete atomic units.

Where subtypes share the same underlying mechanism and are mutually exclusive realisations of the same rhetorical act, they are grouped under a single device (for example, the various subtypes of Metaphor, which all involve semantic transfer). Where similar phenomena involve distinct mechanisms that can co-occur, they are treated as separate devices (for example, different structural forms of repetition). This approach allows layered rhetorical analysis without conceptual overlap.

The analytical consequences of this atomic approach are described in the Principles for Identifying Arabic Rhetorical Devices, Principle 4: Rhetorical Effects Can Be Multi-Layered, which explains how multiple atomic devices may be identified simultaneously when they contribute independent rhetorical mechanisms.

Principle 6: The Encyclopedia Operationalises Specifications of Arabic Rhetoric Devices

Classical Arabic rhetorical theory is rooted in a decentralised, genre-specific, historically and geographically fragmented scholarship which commenced in late antiquity and the early Islamic period, and continued until the post-Abbasid period. Unlike other Islamic and Arabic sciences that underwent later phases of formal standardisation, Arabic rhetoric remained largely descriptive and interpretive, allowing multiple overlapping frameworks to coexist without consolidation. This presents several problems when used for reproducible analytical and computational applications:

1. Rhetorical device names are not always standardised. Some rhetorical phenomena appear under multiple names, while others are described functionally without a stable technical label. This makes it difficult to compare definitions across sources and determine whether two scholars are describing the same device.

2. Rhetorical device definitions are inconsistent. Different scholars use different terminology to describe the same rhetorical mechanisms and effects. One scholar's taxonomy of rhetorical mechanisms and effects does not match another scholar's.

3. Rhetorical device definitions overlap. One scholar may state that a rhetorical device includes rhetorical mechanisms A, B, C, and D, while another scholar may state that C is an independent device and D is a subtype of A.

4. Rhetorical device definitions are incomplete. Every scholar's taxonomy of rhetorical devices is influenced by the breadth and depth of their exposure and analysis.

5. Rhetorical device definitions lack a consistent hierarchy or taxonomy. The boundaries between devices are not well-defined.

6. Rhetorical device definitions conflate rhetorical mechanisms and pragmatic effects. Classical definitions often do not clearly distinguish between what the rhetorical device does structurally and what effect it produces.

7. Rhetorical device definitions lack operational criteria. Classical sources rarely provide step-by-step recognition criteria such as decision rules, boundary tests, or reproducibility criteria.

8. Rhetorical device definitions use inconsistent levels of abstraction. Some devices such as Paronomasia may involve the change of a single letter or vowel. Other devices such as al-muqābala (Parallel Antithesis) combine multiple devices in one device.

9. Rhetorical device definitions sometimes use examples instead of rules.

10. Rhetorical device definitions are sometimes evaluative rather than analytical. Some scholars described rhetorical devices in terms of effects on style, eloquence, impact, beauty, or linguistic efficiency, without defining reproducible tests for these effects.

The Encyclopedia does not seek to modify or "fix" classical Arabic rhetorical theory. Rather, it seeks to render it operational for systematic analysis, through a process we describe as "controlled abstraction".

Controlled Abstraction

Controlled abstraction distils classical rhetorical concepts into clearly bounded, operationally definable units, while preserving their core semantic identity. Where classical descriptions are implicit, overlapping, or insufficiently defined for consistent identification, controlled abstraction seeks to introduce the best approximation of explicit and standardised formulations grounded in attested usage. These interventions are minimal, transparent, and methodologically constrained, enabling reproducible analysis while preserving the underlying structure of the tradition. The process results in consistent and standardised operational definitions of Arabic rhetorical devices. This provides the necessary contemporary foundation for reproducible device annotation, computational analysis, and eventual automation of device identification.

Controlled abstraction also avoids imposing external rhetorical taxonomies onto Arabic rhetorical devices through approximate synonymy or inherited equivalence. Where non-Arabic rhetorical traditions contain partially overlapping concepts, these may be noted for comparative purposes, but they do not govern the operational definition of the device. The Encyclopedia defines Arabic rhetorical devices on their own structural and semantic terms, grounded in the Arabic rhetorical tradition itself.

Further reading

See the following blog post for a methodological reflection on why the Encyclopedia removed Greco-Roman labels for rhetorical devices: “A Roman, a Hellene, and an Abbasid walk into a café...” (3 February 2026)

The same principle applies to illustrative material. Examples are selected for their explanatory value in demonstrating the operational mechanism of a device. The Encyclopedia does not seek to catalogue every attested example found in the rhetorical tradition.

Principle 7: Organisation Aids Identification of Rhetorical Devices

The primary objective of the Encyclopedia is to support rhetorical analysis of Arabic discourse, rather than offering training in the use of rhetorical devices when composing new texts. Classical books of Arabic rhetoric are structured for teaching; the Encyclopedia is structured for identification. This difference is reflected in the Encyclopedia’s organisation.

For example, in classical manuals, ‘ilm al-badī‘ (Linguistic Embellishments) is typically divided into:

• Lexical embellishments (al-muḥassināt al-lafẓiyya)
• Semantic embellishments (al-muḥassināt al-maʿnawiyya)

While this subdivision is suitable for teaching, it is not granular enough for analysis, considering that there are around 70 rhetorical devices in this field. The Encyclopedia therefore subdivides ‘ilm al-badī‘ into 11 thematic subcategories that group together devices which operate in similar ways, making them easier to locate and compare. If the analyst recognizes how a device operates (through sound, structure, or meaning transformation), a more granular classification helps to locate the device within the Encyclopedia faster.

In some cases, devices are arranged according to where an analyst might encounter them in discourse. For example, within the "Paragraph Structure & Flow" subcategory of ‘ilm al-badī‘, devices are ordered from those typically found at the beginning of a text, to those in the middle, and finally to those occurring at the end.

Similarly, classical manuals often treat certain devices in paired form, such as:

• Conjunction and Disjunction (al-waṣl wa al-faṣl)
• Definiteness and Indefiniteness (al-taʿrīf wa al-tankīr)
• Brevity, Verbosity, and Moderation (al-’ījāz wa al-’iṭnāb wa al-musāwāh)
• Quotation and Poetic Incorporation (al-iqtibās wa al-taḍmīn)

In pedagogical contexts, grouping these together is useful. For analytical purposes, however, the analyst must determine which specific mechanism is operating in a given sentence. Annotating "Conjunction and Disjunction" as a single unit is not precise enough. The Encyclopedia therefore treats these as distinct devices, because meaningful analysis requires that they be identifiable separately.

At the same time, the Encyclopedia avoids both extremes of taxonomic design: an excessively flat hierarchy, where every device and subtype is exposed at the top level, and an excessively deep hierarchy, where users must navigate through multiple layers of nested subcategories to reach individual devices. Instead, the Encyclopedia adopts a balanced organisational model: enough hierarchy to preserve conceptual relationships, but enough surface visibility to aid discovery and identification.

This means that some mechanisms treated in classical sources as subtypes are elevated to independent entries where their distinctiveness or operational recognisability warrants it. For example, Personification is given independent treatment despite its close relationship to Metaphor, and certain sound-based devices such as Alliteration and Palindrome are surfaced separately even though classical treatments often subsume them under broader categories such as Paronomasia. This organisational choice reflects the Encyclopedia's analytical aim: to make rhetorically salient and operationally identifiable mechanisms easier to find, compare, and annotate.

Principle 8: The Encyclopedia Is the Source of the Arabic Rhetorical Device Taxonomy

The Encyclopedia is a live and evolving compendium of Arabic rhetorical devices.

At defined intervals, snapshots of the Encyclopedia are published as the Arabic Rhetorical Device Taxonomy (ARDT) – a fixed, versioned, and citable taxonomy intended for use in Digital Humanities research and computational analysis.

To facilitate this, the Encyclopedia uses:

• Standardised formatting and layout
• Clear internal structure
• Version control
• Digital-first publication
• Persistent identifiers such as document object identifiers (DOIs)
• Predominantly English-language presentation.

The Encyclopedia is therefore both a scholarly reference resource and the foundational layer from which the ARDT is derived.

Summary

Together, these eight principles ensure that the Encyclopedia of Arabic Rhetoric serves as a bridge between classical Arabic rhetorical tradition and modern analytical needs, maintaining scholarly rigor while enabling systematic, reproducible analysis.

Page history

• 2026-04-30 - Updated Design Principles with addition of The Encyclopedia Recognises Four Consecutive Levels of Every Rhetorical Action
• 2026-03-27 - Updated Design Principles with re-organisation and addition of The Encyclopedia Operationalises Specifications of Arabic Rhetoric Devices
• 2026-02-14 - Page created: Design Principles of the Encyclopedia of Arabic Rhetoric.