Arabic Lexical Atlas
Rob Rob
Can large-scale lexical data change the way we understand Arabic morphology?
After several years building the Arabic Lexical Atlas (currently over 11,000 quadriliteral lexemes), I wanted to share one of the first corpus-based observations emerging from the project.
Hello! Here's the latest article from my Arabic Lexical Atlas (ALA) newsletter. I would be delighted to hear comments from colleagues working in Arabic linguistics, historical linguistics, lexicography, or NLP: https://www.linkedin.com/pulse/arabic-lexical-atlas-ala-rethinking-quadriliteral-through-rougeaux-ottse
#Arabic #Linguistics #NLP #Morphology #HistoricalLinguistics
Arabic Lexical Atlas (ALA)
Rethinking Arabic Quadriliteral Morphology through Large-Scale Lexical Analysis
This article presents one of the first findings emerging from the ongoing Arabic Lexical Atlas (ALA) project, a large-scale investigation of Arabic lexical evolution based on more than 11,000 quadriliteral lexemes, together with related rare triliteral forms, diachronic lexicographic evidence, and dialectal data.
Introduction
Arabic lexical structure has traditionally been studied through the interaction of roots, morphological patterns, and semantic fields. The Arabic Lexical Atlas (ALA) is an ongoing research initiative that seeks to document and analyze more than 11,000 Arabic quadriliteral lexemes, together with related rare triliteral forms, diachronic evidence extracted from the major Arabic dictionaries, morphological classification, lexical origin, productivity, geographical distribution, and quadriliteral formations attested across the principal Arabic dialect groups (Levantine, Egyptian, Maghrebi, and Peninsular Arabic).
Rather than simply cataloguing lexical items, ALA aims to uncover the mechanisms that drive Arabic lexical evolution: reduplication, consonantal substitution, borrowing, semantic change, lexical innovation, and morphological productivity.
The present study initially focused on a relatively simple question: does the consonantal sequence KR, occupying the first two positions of a quadriliteral root (C₁C₂), exhibit a recurrent semantic orientation?
The initial question itself originated from a discussion with my friend and fellow researcher Jean-Claude Rolland, who encouraged me to explore the KR sequence. The investigation subsequently evolved in directions that neither of us had anticipated.
As the analysis progressed, however, the question itself gradually evolved. The investigation ultimately became less about the KR sequence than about the mechanisms through which quadriliteral lexemes emerge, develop, and interact throughout the history of Arabic.
The observations presented here do not claim to provide definitive answers. Rather, they suggest that several assumptions commonly taken for granted in the study of Arabic morphology deserve to be reconsidered in light of large-scale lexical evidence.
1. Formal similarity does not necessarily reflect historical relationship
The first observation is that Arabic quadriliteral lexemes do not constitute a historically homogeneous category. Behind their apparent formal unity lies a remarkable diversity of formation processes, including complete reduplication, partial reduplication, consonantal substitution, metathesis, borrowing, compounding, back-formation, and formations whose origin remains uncertain.
In other words, the mere presence of a quadriliteral structure tells us very little about the historical process that produced it.
A consonantal configuration such as C₁C₂C₁C₃, for example, may indeed result from genuine reduplication, but it may equally arise through consonantal substitution or metathesis. From a synchronic perspective, these forms appear structurally similar; historically, however, they may represent entirely different evolutionary pathways.
This distinction between observable morphological structure and historical mechanism of formation is, in my view, fundamental. Morphological form alone cannot reliably reconstruct lexical history. Any attempt to infer historical development directly from surface structure therefore requires considerable caution.
2. Consonantal variation may be morphologically significant but semantically neutral
A second observation emerged from the corpus. Several lexical families exhibiting only minor consonantal differences were found to share an almost identical semantic core. Pairs such as karkara / karkasa, karbaḥa / karmaḥa, and karbaša / karmaša provide particularly revealing examples. These correspondences suggest that certain consonantal substitutions may be morphologically significant while remaining semantically neutral.
This observation deserves particular attention because it challenges an assumption that is often taken for granted in the study of Semitic roots: namely, that each consonant carries part of the semantic identity of the root. Under this view, changing a single consonant, or merely altering its position within a quadriliteral configuration, would normally be expected to produce a semantic change. The corpus suggests that this expectation is generally justified, but not universal. This does not imply that consonantal structure is devoid of semantic relevance. Rather, it indicates that semantic interpretation cannot be inferred directly from formal similarity alone.
The evidence therefore calls for a clear distinction between three complementary levels of analysis:
1. the observable morphological structure;
2. the historical mechanism of formation;
3. semantic evolution.
These dimensions sometimes converge, but they should never be conflated.
3. Lexical substitution as a mechanism of Arabic lexical evolution
The analysis also highlights the importance of lexical substitution as a genuine mechanism of Arabic lexical development.
Some cases are explicitly documented by the medieval lexicographers. Karmaḥ, for example, is presented as a variant of karbaḥ, while karmaṣ is associated with karṣam. Other relationships emerge only through the systematic comparison of numerous lexical families across the corpus.
Taken individually, such correspondences might appear anecdotal. Considered collectively, however, they suggest that consonantal substitution represents a recurrent mechanism in the early history of the Arabic lexicon rather than a collection of isolated irregularities.
These alternations are overwhelmingly found among ancient inherited lexemes. One possible explanation is that they preserve traces of linguistic varieties that coexisted before the stabilization of Classical Arabic. Under this interpretation, lexical substitution would not necessarily reflect later innovation, but rather the survival of competing forms inherited from different varieties of early Arabic.
If this hypothesis is correct, the subsequent standardization and geographical expansion of Arabic may have progressively reduced much of this lexical variation. As Arabic became the language of increasingly large non-Arabic-speaking populations, lexical competition may have given way to greater standardization, leading to the disappearance of many formerly coexisting variants.
The present corpus cannot yet demonstrate this scenario conclusively. Nevertheless, it offers a plausible historical framework that deserves further investigation and points toward a more dynamic view of the formation of the Classical Arabic lexicon.
4. Borrowings become fully productive Arabic lexemes
The corpus also revealed an unexpected phenomenon. Borrowed words might initially appear peripheral to the study of quadriliteral morphology, since they do not directly illuminate the internal history of Arabic lexical formation. Yet they ultimately proved to be among the most revealing observations.
Words such as kārāmīl ("caramel") demonstrate that Arabic does not simply borrow foreign lexical items. Instead, it incorporates them into its own derivational system, allowing them to function as fully productive morphological bases.
The case of kārāmīl is particularly illustrative. From a single borrowed noun, Modern Arabic has generated several new derivatives, including:
· karmala – to caramelize;
· takarmala – to become caramelized;
· takarmul – caramelization.
The foreign origin of a lexical item therefore does not prevent its integration into Arabic morphology. On the contrary, borrowed forms may become fully productive components of the Arabic derivational system.
From both a linguistic and a computational perspective, these examples are especially significant. They demonstrate that lexical origin and morphological behaviour are largely independent dimensions. A borrowed lexeme may participate as naturally in Arabic word formation as an inherited one.
The importance of such forms therefore lies less in their etymological origin than in their capacity to become active participants in Arabic lexical creativity.
5. The semantic profile of the KR pattern
The original objective of this study was to determine whether the consonantal sequence KR exhibits a preferential semantic orientation.
The corpus certainly does not support the existence of a single, fixed meaning associated with this sequence. It does, however, reveal several remarkably coherent semantic domains whose recurrence deserves attention.
The first concerns motion. Many of the verbs examined express notions such as running, advancing, pursuing, rolling, turning, returning, descending while rolling, or progressing in an unstable or hesitant manner. Although these meanings differ in detail, they repeatedly converge on the broader semantic domain of movement.
A second domain concerns the restriction or constraint of movement. Several lexemes denote binding, restraining, compressing, contracting, or otherwise limiting motion.
Particularly noteworthy is the fact that certain verbs meaning to restrain or to hinder give rise to nouns describing the gait of a person whose movement is impaired or constrained. The transition from free movement to constrained movement therefore appears as a natural semantic extension rather than an isolated development.
A third domain brings together notions of accumulation, density, and thickness. Here the corpus includes references to compact masses, dense clouds, thick vegetation, bulky structures, and various heavy or compact objects.
Considered individually, none of these semantic domains is unexpected. Considered together, however, their recurrence suggests that the distribution of meanings within the KR corpus is unlikely to be entirely accidental.
6. Toward a dynamic model of Arabic lexical organization
Taken separately, each of these observations is informative.
Taken together, they suggest the existence of genuine semantic attractors: not a single meaning encoded by the consonantal sequence KR, but recurrent semantic tendencies whose frequency and internal coherence call for explanation.
This conclusion must nevertheless be interpreted with caution.
The corpus also contains numerous modern borrowings, which clearly cannot be used to reconstruct any hypothetical ancient semantic value associated with KR. Their importance lies elsewhere. They demonstrate the remarkable capacity of Arabic to absorb foreign lexical material and integrate it into its own derivational system. Any investigation of semantic attractors must therefore focus primarily on inherited lexemes or on forms whose historical development can be reconstructed with reasonable confidence.
A working hypothesis: could the true invariant be dynamics?
Looking at the corpus as a whole, one broader idea gradually emerged.
The three principal semantic domains identified—motion, constraint, and accumulation—may not, in fact, be independent. Rather, they may represent different manifestations of a more general underlying principle.
Across the inherited lexemes, a recurring notion of dynamic transformation repeatedly appears.
· A body moves, rolls, advances, or pursues.
· A body is compressed, restrained, or contracted.
· A substance accumulates and becomes dense.
· A body part rises, swells, or bristles.
· Even certain tools are primarily associated with mechanical action.
From this perspective, KR may not encode movement as such. Instead, it may be associated with the activation of a dynamic process: the initiation of movement, the application of force, or, more generally, a transformation affecting a body, a substance, or an object.
I do not present this as a conclusion. At this stage, it remains a working hypothesis suggested by the corpus rather than demonstrated by it.
Nevertheless, it offers a coherent framework capable of bringing together the principal semantic attractors identified during the analysis while opening new directions for future research.
If comparable semantic tendencies were found to characterize other consonantal sequences, it might become possible to describe the organization of the Arabic lexicon from a different perspective—not by seeking a fixed meaning attached to each consonant or root, but by investigating the recurrent statistical and morphological dynamics associated with consonantal configurations.
Such an approach could provide a productive bridge between traditional Arabic lexicology and contemporary computational models of morphology, lexical organization, and semantic representation.
For Arabic Natural Language Processing (NLP), this perspective raises a broader question. Rather than treating roots exclusively as discrete symbolic units, could computational models also benefit from incorporating recurrent semantic tendencies associated with consonantal configurations, while fully respecting historical and morphological constraints?
In my view, this is the most promising direction opened by the present investigation.
About the Arabic Lexical Atlas (ALA)
The Arabic Lexical Atlas (ALA) is an evolving lexical knowledge base designed for research in historical linguistics, Arabic morphology, lexicography, and Arabic Natural Language Processing (NLP).
It integrates data from Classical Arabic, Modern Standard Arabic, and the major Arabic dialects while documenting relationships between rare triliteral and quadriliteral formations. Beyond lexical description, ALA seeks to provide a corpus-based framework for investigating lexical innovation, morphological productivity, semantic evolution, and the long-term dynamics of the Arabic lexicon.
Robert R.
Founder, Arabic Lexical Atlas (ALA)
Independent Researcher in Arabic Morphology and Lexical Evolution