Building a Setswana Thesaurus: Challenges and Solutions

Setswana, one of the eleven official languages of South Africa, is spoken by roughly five million first-language users and several million second-language speakers across the country and Botswana. Comprehensive reference works for Setswana remain sparse when measured against the resources available for English or major European languages. A thesaurus, the kind of controlled vocabulary that lists synonyms, antonyms, and related terms in semantic clusters, has never been produced at scale for Setswana. The absence of such a tool affects translators, educators, lexicographers, and software developers.

The gap is felt acutely in language-technology contexts. Machine translation systems, search engines, and educational platforms all require structured lexical data, and where that data is missing, languages tend to fall behind. Researchers working on less-resourced languages elsewhere face the same pressures. In Australia, for instance, the maintenance of Aboriginal languages has prompted comparable efforts to build dictionaries, thesauri, and corpora, drawing on expertise at institutions like the Australian National University and the Australian Institute of Aboriginal and Torres Strait Islander Studies.

A project to build a Setswana thesaurus must combine methods drawn from general lexicography with sensitivity to the specific structure of Bantu languages. Setswana's noun-class system, agglutinative morphology, and rich system of ideophones create challenges with no direct counterpart in English thesaurus design. Overcoming these requires a mix of corpus analysis, expert consultation, and iterative refinement.

This article surveys the principal obstacles encountered in the early stages of compiling a Setswana thesaurus, and the practical strategies that have emerged. The work draws on the experience recorded on T.J. Otlogetswe's professional site, where ongoing research into Setswana etymology and language contact provides much of the methodological foundation.

Mapping the semantic terrain of Setswana

The first step in any thesaurus project is gathering a corpus of running text large enough to reflect actual usage. For Setswana, this means drawing on newspapers, literary works, government publications, educational materials, and transcribed oral sources. The resulting corpus can be queried for word frequencies, collocations, and patterns of co-occurrence, allowing lexicographers to identify natural synonym clusters rather than imposing them from outside.

Corpus design in a Bantu language is more complicated than it might appear. Setswana words carry prefixes that signal noun class, and these prefixes affect how words are indexed and retrieved. A lemma list for a thesaurus must handle the morphological variability that any search engine would also need to handle. The corpus built for the project includes both written and spoken registers, with attention to regional variation between South Africa and Botswana.

Comparable approaches have been used for Australian Aboriginal languages, though on a smaller scale. Researchers at universities in Melbourne and Brisbane have shown that even modest corpora can support meaningful lexical analysis when paired with careful elicitation sessions with speakers. The lesson is that volume matters less than representativeness.

Defining semantic fields in a Bantu language

A thesaurus organises words by meaning, and meaning in Setswana is shaped by grammatical categories that do not exist in English. Noun classes group together nouns that trigger the same agreement patterns on verbs, adjectives, and pronouns. A thesaurus entry cannot ignore these patterns, because two nouns that appear synonymous in translation may belong to entirely different noun classes and therefore govern different verb forms.

Consider the relationship between ntlo (house) and ntlwana (small house or hut). Speakers sometimes treat these as interchangeable, but the morphological structure reveals a clear distinction: ntlwana is a diminutive derived from ntlo. The distinction matters for a thesaurus, because users searching for words related to domestic dwelling need to know whether they are looking for the general concept or a specific size or function. The finer points of such distinctions are explored further in why ntlo and ntlwana are not the same, a discussion that highlights how a single English gloss can mask several Setswana realities.

Semantic fields in Setswana also reflect cultural priorities. Domains such as cattle husbandry, traditional governance, and kinship carry dense terminology that may have no direct English equivalent. A thesaurus must therefore include these culturally specific clusters, even when they appear peripheral from an English-language perspective.

The problem of polysemy and semantic networks

Polysemy, the phenomenon by which a single word carries several related meanings, presents a perennial headache for lexicographers. In Setswana, words frequently combine concrete and abstract senses in ways that English separates into distinct lexemes. The verb go thusa, for instance, covers assistance, usefulness, and the state of being helpful, and a thesaurus entry must reflect these layered meanings without collapsing them into a single gloss.

Abstract concepts are particularly difficult. Setswana draws on metaphor and bodily imagery to express ideas that English handles with abstract nouns, and these metaphorical extensions do not always map neatly onto the target language. The challenges of capturing such nuance are addressed at length in defining abstract concepts in African languages, a discussion that resonates with similar debates in the documentation of Aboriginal Australian languages.

Resolving polysemy in Setswana relies on corpus evidence more than on intuition. Concordance lines showing how a word behaves in different contexts allow lexicographers to separate the senses and assign each to a distinct semantic node in the thesaurus. This evidence-based approach reduces the risk of imposing an English template onto Setswana semantics.

Cultural concepts and untranslatable terms

Every language encodes cultural knowledge in its vocabulary, and a thesaurus must accommodate terms that resist straightforward translation. Setswana has words for specific cattle colours, traditional brewing processes, and kinship roles that carry connotations no English phrase can reproduce. Listing such terms alongside English approximations would distort their meaning, yet excluding them would impoverish the resource.

The question of how to render culturally specific vocabulary is one shared by translators worldwide. Discussions of translating Japanese proper names, for example, highlight the choices facing anyone who works between languages with different writing systems and naming conventions. The same trade-offs appear when Setswana names and titles are carried into English texts, where choices about transliteration, explanation, and cultural fidelity affect how the source culture is represented.

A Setswana thesaurus can take several approaches to this problem. It can include a cultural note with each entry that carries specific cultural weight, flag terms for which no clean English equivalent exists, and cross-reference related terms that together approximate the missing concept. These strategies preserve the integrity of the source vocabulary while still serving users who need quick access to a workable English rendering.

Digital infrastructure and open data

The infrastructure for publishing a thesaurus has never been more accessible, but Setswana still faces gaps in the foundational data needed to support digital tools. Spell-checkers, predictive text models, and morphological analysers all require validated word lists, and many of these lists do not yet exist for Setswana at the scale required.

Building this infrastructure means investing in open, machine-readable datasets that researchers and developers can build upon. Standards such as the Lexical Markup Framework and TEI provide templates that any new project can adopt. The thesaurus itself will be released in formats that allow both human consultation and programmatic querying, contributing to a broader ecosystem of language technology.

The Australian experience offers a useful parallel. State governments and language centres across New South Wales and Western Australia have invested in digital archives and open-access platforms for Indigenous languages, recognising that language data is a form of cultural infrastructure. Setswana would benefit from a similar coordinated approach.

Collaboration with speakers and institutions

No thesaurus can succeed without the active involvement of language speakers. Expert lexicographers provide the methodology, but only speakers can confirm which synonyms feel natural, which terms are falling out of use, and which meanings carry emotional or cultural charge. The project includes structured consultation with teachers, writers, broadcasters, and community elders across the Setswana-speaking region.

Institutional partnerships are equally important. The Pan South African Language Board, the Botswana Centre for African Studies, and various university departments have all expressed interest in contributing to the resource. International collaboration with groups working on related languages, including projects on Zulu and Sesotho, helps ensure methodological consistency across the Bantu languages of southern Africa.

The process of documenting word histories and tracing etymologies often requires looking beyond the language itself. The Latin phrase curriculum vitae, for instance, has an etymology explored here that illustrates how borrowed terms enter and reshape target languages. Similar borrowing processes have shaped Setswana, especially through contact with Afrikaans and English, and a thesaurus must acknowledge these layers.

The shape of a Setswana thesaurus will ultimately depend on the communities it serves. What users should take away is that building a thesaurus is not simply a matter of listing synonyms in alphabetical order. It requires careful attention to grammar, cultural nuance, digital infrastructure, and the voices of speakers themselves. The result, when achieved, is a resource that supports translators, educators, and technologists while preserving the richness of Setswana for future generations.