Communicative methods and target vocabulary selection

In recent decades, the teaching of Latin has experienced a notable shift towards new communicative pedagogical approaches. As Rambsy (Reference Ramsby2020) notes, this change in focus is associated with various terms, such as Active Latin or Comprehensible Input, but they all emphasise the importance of using and speaking the target language as part of the learning process. In their introduction to the recent volume Communicative Approaches to Ancient Languages, Lloyd and Hunt (Reference Lloyd and Hunt2021) identify the goal guiding this shift as ‘escaping the image of ancient languages as mere codes to be deciphered and moving towards students experiencing them as normal means of communication, both spoken and written’ (p.1).

This trend is not the first attempt to depart from the traditional grammar-translation method, with which it remains associated in many instructional contexts. The natural (or direct) method is now more than a century old, and its origins can be traced back to medieval times, when Latin served as a medium of both oral and written communication in academic settings (Wingate, Reference Wingate2013). Acknowledging this, Manning (Reference Manning, Lloyd and Hunt2021) characterises the current shift as a ‘return to communicative emphasis’ (p. 9).

Compared with previous efforts, however, new communicative approaches benefit from extensive research conducted in the field of second language acquisition (SLA), which has primarily focussed on the teaching of modern languages (Carlon, Reference Carlon2013). This represents a profound paradigm shift, inviting a review not only of teaching practices and strategies, but also of the appropriateness of instructional materials and course structures.

Some educators argue that communicative activities should not be seen as an end in themselves, but rather as a means to help students develop greater competence in reading written texts (Lloyd, Reference Lloyd2017; Sinclair, Reference Sinclair2018). Nevertheless, a communicative approach introduces goals that may diverge from those associated with enhanced reading comprehension. Illustrating this tension, Venditti (Reference Venditti2021) recalls that, having been trained through the grammar–translation method, she lacked familiarity with basic structures related to everyday life – structures that would typically be introduced early in a communicative course. The dilemma is clear: the linguistic resources required to read classical texts are not the same as those needed for basic communication, and neither fully supports the acquisition of the other.

One aspect of teaching that clearly exemplifies this problem is the selection of target vocabulary. In the field of second language acquisition (SLA), it is widely acknowledged that, given the vast number of lexical units and the significant time and effort required for their acquisition, it is essential to determine which words should be learned first (Nation, Reference Nation2013). However, no unified criteria exist for making this selection, an issue that already presents challenges in modern foreign language (MFL) instruction. Contemporary researchers emphasise the need to apply multiple criteria, both objective and subjective, including lexical frequency, lexical availability, and the perceived usefulness or difficulty of specific items (He and Godfroid, Reference He and Godfroid2019; Hidalgo Gallardo & Rufat Sánchez, Reference Hidalgo Gallardo and Rufat Sánchez2022; Laufer and Nation, Reference Laufer, Nation, Gass and Mackey2013).

Lexical frequency is often considered the most suitable metric for vocabulary selection in the teaching of classical languages. Since frequency lists are derived from the corpus of available texts, they align naturally with the needs of students whose long-term goal is to read that corpus. As Gruber-Miller and Mulligan (Reference Gruber-Miller and Mulligan2022) note, high-frequency words significantly enhance students’ ability to recognise and comprehend the language. Ash (Reference Ash2019) further underscores the value of restricting vocabulary to high-frequency terms within a comprehensible input framework, enabling learners to better acquire and retain the language through repeated exposure to relevant and meaningful items.

However, the applicability of frequency lists becomes more problematic when selecting vocabulary for courses with a communicative focus. Since these lists are derived from ancient texts, many high-frequency words have limited relevance for everyday interaction. For example, the Latin Core Vocabulary list compiled by Dickinson College (Francese, Reference Francese2021) includes terms such as deus (ranked 42), rex (60), iubeō (84), bellum (86), hostis (93), and arma (98), which are unlikely to be useful in basic conversational contexts. Conversely, highly practical words – particularly those referring to contemporary objects or concepts – are notably absent.

Another limitation of frequency lists is that they do not, in themselves, offer a criterion for sequencing vocabulary. For example, the Dickinson College list includes 1,000 terms – roughly the number of words a learner is expected to acquire for basic communicative competence – yet it does not indicate the order in which these terms should be introduced. This is a significant issue, as not all words in the list are equally useful: a term like vestīgium (ranked 716) is arguably less essential than rēs (ranked 38). One possible solution is to establish cut-off levels that create hierarchical groupings – for instance, assuming that words ranked 1–99 represent a more basic vocabulary than those ranked 100–199. However, as Hashimoto (Reference Hashimoto2021) observes, this method is arbitrary and lacks empirical justification. There may be little difference in usefulness between words ranked 99 and 100, whilst terms within the same range (e.g. ranks 1 and 99) can vary significantly in frequency.

When evaluating the usefulness of a word for learners, subjective criteria often play a significant role. Dang and Webb’s (Reference Dang, Webb, Griffiths and Tajeddin2020) study of 17 English as a foreign language (EFL) teachers in Vietnam found that most based their vocabulary selection primarily on the textbooks they used, followed by their own experience and intuition. However, such criteria are not always reliable. McCrostie (Reference McCrostie2007) observed that teachers were accurate in estimating word frequency for very common or very rare terms, but less so for mid-frequency vocabulary. In light of this, Sánchez Gutiérrez et al. (Reference Sánchez-Gutiérrez, Robles-García and Serrano2025) conclude that neither corpus-based lists nor teacher intuition alone provide sufficient guidance for selecting relevant vocabulary. In the case of Latin, as noted earlier, the challenge stems less from the subjective side of the selection process than from the objective one: resources are far more limited than in modern language teaching, and frequency lists prove inadequate for courses with a communicative focus.

The purpose of this article is to offer an objective alternative for Latin vocabulary selection on the basis of the natural semantic metalanguage (NSM). As we will argue, this approach not only provides us with a concise core vocabulary list, but also offers a viable model for lexical sequencing. Another advantage of this approach is that, by being based on a finite inventory of universal concepts, it provides a model applicable to all languages, making it easier to transpose developments in MFL to the teaching of Latin and ensuring maximum translatability and comprehensibility of the input.

In the following sections, we will present the theoretical principles underlying our proposal, with a summary of the NSM theory. Next, we will introduce Minimal Latin (ML), including its vocabulary and syntax. Finally, we will demonstrate several ways in which the use of ML can be applied in Latin teaching under a communicative method, both as instrumental vocabulary and in explaining complex concepts.

The natural semantic metalanguage

The natural semantic metalanguage (NSM) approach originated in Wierzbicka’s seminal work Semantic Primes (Reference Wierzbicka1972). Its central hypothesis is the existence of a finite set of universal, indefinable concepts, known as semantic primes, which can be combined to express all complex meanings in natural languages (Wierzbicka, Reference Wierzbicka2021). Wierzbicka initially identified 14 such primes; subsequent research by her and other scholars has expanded the inventory to a current total of 65, as detailed in Table 1.

Table 1. Semantic Primes

Semantic primes, typically written in small caps, are abstract entities. However, the theory predicts that these primes, being universal, are linguistically represented in all natural languages. Such exponents may take the form of words, morphemes, or phrases (Fernández, Reference Fernández2020). This universality ensures that both the list of exponents and the definitions constructed with them are cross-translatable between any two languages, with minimal loss or addition of meaning (Gladkova and Larina, Reference Gladkova and Larina2018).

The metalanguage possesses not only a lexicon but also a syntax, which is not arbitrary but determined by the meaning of each semantic prime (Goddard, Reference Goddard2018b, p. 60). Each prime has a specific valency – that is, a set number and type of arguments it requires to form a well-structured clause. For instance, a predicate such as DO needs, in its basic form, two arguments, as illustrated in (a).

1. a. someone DOES something

This clause can be further extended to reveal additional aspects of the situation denoted by the predicate:

1. b. someone DOES something to someone

2. c. someone DOES something with something

3. d. someone DOES something with someone

These additional arguments can be categorised using traditional semantic terminology: in sentence (b), the underlined argument is a patient; in (c), an instrumental; and in (c), a comitative. Additionally, the predicate may include evaluators or adjuncts of time and place, as exemplified in the following sentences:

1. e. someone DOES something (good/bad)

2. f. someone DOES something (there)

With its vocabulary and syntax, NSM purportedly allows for the semantic explanation of any complex concept expressed in natural languages. For example, Goddard and Wierzbicka (Reference Goddard and Wierzbicka2014b) define the English term happiness as follows:

From a theoretical perspective, NSM addresses the classical semantic problem of circular definitions (Fernández, Reference Fernández2020). Contrary to the assumption that all concepts can be defined, Wierzbicka (Reference Wierzbicka1996) contends that semantic primitives are inherently indefinable. This view not only avoids the risk of infinite regress, but also sidesteps a subtler and more frequent form of circularity, in which simple concepts are defined using more complex ones.

Furthermore, NSM has been proposed as a remedy for ethnocentrism – particularly Anglocentrism – which, according to Goddard and Wierzbicka (Reference Goddard and Wierzbicka2014a), has historically permeated linguistic research. The dominance of English as an academic lingua franca has fostered the mistaken belief that it can function as a culturally neutral tool of analysis. This misconception becomes especially evident when terms such as politeness or kinship are employed to describe categories in other languages. The problem is twofold: first, it neglects the fact that such English terms are themselves socially and culturally conditioned, and therefore not universal; second, it introduces concepts that may be foreign or irrelevant to the languages being studied.

Given its cross-translatable nature, NSM helps mitigate ethnocentric bias, making it a valuable tool for intercultural communication. In recent decades, the growing focus on pragmatics within NSM research has given rise to ethnopragmatics, a closely related framework (Goddard, Reference Goddard2018b, p. 286). This approach emphasises the use of semantic primes to develop cultural scripts – semantic explanations of norms, values, and culturally specific practices.

This field has expanded with the development of minimal languages, first proposed in a volume edited by Goddard (Reference Goddard and Goddard2018a). The authors argue that, despite its role as a global lingua franca, English is saturated with concepts deeply rooted in Anglo cultural frameworks, which can hinder effective cross-cultural communication. Minimal English, a radically simplified version of English, aims to overcome these limitations by enhancing translatability into other languages.

Whilst grounded in NSM, Minimal English is not identical to it. Its lexicon comprises a core set of approximately 200 terms, including the 65 semantic primes, along with a selection of semantic molecules – definable, low-complexity words that are generally universal or near-universal (e.g. mother, we, eye, animal, day, wood). Once a semantic molecule has been defined using NSM, it can be employed to construct more complex meanings. This makes Minimal English a flexible tool for bridging linguistic and cultural divides. For instance, the molecule we is defined by Goddard (Reference Goddard1995) as follows:

Subsequently, we can be used in more complex definitions in combination with semantic primes. For instance, Goddard (Reference Goddard2020) defines country as:

The inventory of Minimal English includes not only universal semantic molecules, but also certain more complex terms that are nearly universal or unavoidable in contemporary global discourse – such as money, government, mosquitoes, plastic, or computer. This lexicon is flexible and expandable, allowing for the incorporation of new terms as they gain relevance across cultures. Moreover, Minimal English admits local adaptations that include context-specific vocabulary. For example, for someone living in the Arctic, it might be important to include terms such as seal or whale (Goddard, Reference Goddard2018b, p. 277).

It is important to emphasise that the focus on English in this theoretical framework does not reflect any intrinsic superiority of the language, nor does it suggest any advantage beyond its current global usage. As Goddard (Reference Goddard2018b, p. 269) notes, Minimal English aspires to be “minimally English.” (Wierzbicka, Reference Wierzbicka2013, p. 195) prefers the term Basic Human to highlight that this is not simply a reduced version of English, but rather the English instantiation of a universal minimal language. The emphasis on English and other global languages in this strand of NSM research is pragmatic: these languages already function as vehicles of intercultural communication, and Minimal English is intended as a tool for enhancing clarity and mutual understanding within those same contexts (Goddard, Reference Goddard2021, p. 2).

Latin NSM and Minimal Latin

The first step in constructing the Latin version of NSM is to define the inventory of exponents of semantic primes in Latin. The principal method for identifying these exponents is the use of canonical sentences – sentences or fragments composed predominantly, or entirely, of semantic primes. These are formulated for each prime according to its expected syntactic pattern and can be tested cross-linguistically (Goddard and Peeters, Reference Goddard, Peeters and Peeters2006). A representative example is the list compiled by Goddard and Wierzbicka (Reference Goddard and Wierzbicka2017), which features 150 simple sentences in English. Translating these into the target language allows researchers to determine the language’s semantic prime exponents and to map its core lexicogrammatical features.

In the case of Latin, a key limitation of this approach is that, unlike with living languages, we cannot rely on native speaker intuition to assess the validity of canonical sentences or to confirm the proposed exponents. As a result, our primary source has been original Latin texts, complemented by the pedagogical intuition of experienced Latin instructors. Additionally, prior NSM research conducted in other languages – especially Romance languages (e.g. Peeters, Reference Peeters2006) – has offered valuable insights that informed the development of our list.

Table 2 presents the resulting inventory of semantic prime exponents for Latin, which serves as the foundation for the approach proposed in this article.

Table 2. List of Latin Exponents of Semantic Primes

For our list, we have adhered to conventions commonly used in Latin to facilitate its integration with other teaching methods. Verbal forms are presented in the present tense, first-person singular, whilst nouns are listed in the masculine singular, except in cases where the primitives carry an intrinsically plural meaning, as is the case with homines, multī, and paucī.

As can be observed in the list, there is a notable case of polysemy with the exponent sum, a Latin verb that lexicalises three different semantic primes. This case is analogous to to be in English. To resolve the ambiguity, we list the three exponents accompanied by synonyms or by the arguments they can take.

The list also accounts for cases in which a single semantic prime is represented by multiple exponents, which Wierzbicka called allolexy (Goddard and Wierzbicka, Reference Goddard and Wierzbicka2021). In our list, following standard NSM practice, we include allolexes only when they correspond to two or more distinct lemmas. Variants of a single lemma, such as me and mihi, which are allolexes of ego, are not listed individually, but are considered part of the usable metalanguage and used according to its grammatical rules.

Notably, amongst the 65 exponents we identified, 6 are not included in the Dickinson College Latin Core Vocabulary list: vēritās, paulīsper, aliquamdiū, mōmentum, īnfrā, and intus. As anticipated, although there is a correlation between an item’s universality and its frequency, the overlap is only partial. This serves as further evidence that the two criteria cannot be used interchangeably in the selection of pedagogical vocabulary.

In some cases, universal items appear to be replaced in the frequency list by semantically related but more complex terms. For instance, the Dickinson College list includes vērus and vērē, whose semantic relationship to the absent vēritās is evident. Whilst more frequent terms may prove more useful in certain contexts, when constructing our NSM we must prioritise the more universal exponent, as it ensures the highest degree of cross-linguistic translatability.

We can test the effectiveness of our Latin version of NSM by generating semantic explications or translating existing ones, such as the definition proposed by Goddard in [3], which can be rendered as follows:

Constructing a minimal language requires expanding its lexicon beyond semantic primes by incorporating semantic molecules. To ensure consistency, it is useful to rely on pre-established inventories in which the molecular status of these items has already been demonstrated. Following this approach, we propose, as an initial stage in the development of Minimal Latin, the incorporation of 270 semantic molecules on the basis of the list provided by Bullock (Reference Bullock2014) in his LTWF project. This yields a Minimal Latin with a total of 335 vocabulary items (see Appendix: Semantic Molecules in Minimal Latin). The total number of lexical items in the minimal language should be considered provisional and flexible, subject to modification depending on specific contexts of use.

Regarding the grammar of our metalanguage, the guiding principle is to preserve, as far as possible, the rules of Classical Latin, provided that doing so does not compromise the cross-linguistic translatability of the resulting utterances. Following the criteria established by Travis (Reference Travis, Goddard and Wierzbicka2002, Reference Travis2003) in her adaptation of NSM to Spanish, the sentences generated by the metalanguage must be both grammatically correct and pragmatically neutral. In other words, they should avoid morphosyntactic constructions which, though acceptable in the language, introduce implicit meanings, modal nuances, or culturally specific presuppositions that cannot be reliably rendered in other languages.

Thus, for instance, in Latin resultative clauses introduced by ut, the use of the subjunctive, is grammatically mandatory and poses no problem for translatability; it should therefore be incorporated into the metalanguage. By contrast, we avoid the subjunctive in contexts where it alternates with the indicative – such as in causal clauses introduced by quod, quia, or quoniam – since this alternation often introduces a subjective or inferential nuance that is not universally interpretable.

As can be seen, simplification in many cases consists in reducing the range of grammatical options available to a native speaker. We therefore propose using only the SOV word order, as it is the unmarked order in Latin. Similarly, and following the same principles, our metalanguage preserves the case system, gender and number agreement between nouns and modifiers, the use of high-frequency prepositions, and the verbal system, with particular emphasis on the contrast between the perfect and imperfect tenses.

This proposal should be understood as a work in progress, open to adjustment, expansion, or revision as practical experience with its implementation accumulates. Rather than offering a prescriptive or final set of rules, our aim is to establish a flexible and functional framework upon which local variants, contextual adaptations, or broader developments may be constructed in accordance with specific pedagogical goals.

Pedagogical applications

To conclude, we will summarise the potential applications of ML in the teaching of Latin. We begin with the practical problem that originally inspired this article – the selection of target vocabulary – and then move on to more exploratory uses of ML.

Explaining cultural keywords

In NSM theory, cultural keywords are words or expressions in a language that are deeply embedded in the culture of its speakers and reflect culturally significant ideas, values, or practices (Wierzbicka, Reference Wierzbicka1997). They are ‘culturally-specific words around which whole discourses are organised’ (Leivsen and Waters, Reference Levisen and Waters2017, p. 7). These terms often lack direct equivalents in other languages, as they are closely tied to the particular worldview and social norms of the culture in which they emerge. The identification and explication of cultural key words using the metalanguage has long been a central concern of NSM research. As Goddard notes, this emphasis represents a return to words, rather than syntactic structures, as the primary focus of linguistic inquiry.

This theoretical framework can be applied to several culturally salient Latin terms, such as virtūs, pietas, or dignitas, whose meanings are rooted in Roman social life and cannot be fully captured by a single English equivalent. As Schrader (Reference Schrader2016) argues in the case of virtus, and as is normally the case with cultural keywords, translating such terms with a single word – such as virtue or excellence – fails to reflect their full semantic and cultural scope.

What follows is a semantic explication of virtūs in Minimal Latin, intended to illustrate how the NSM framework can be used to clarify such culturally complex concepts using a small set of universal elements:

This explication of virtus in ML seeks to capture the core features discussed by Schrader (Reference Schrader2016). The expression X habet virtūtem is unpacked through a series of minimal propositions that reflect not only what X does (X facit rēs), but also why X does them: namely, because X is a certain kind of person or thing (X est Y). This captures the normative dimension of virtus in Roman thought, where excellence was linked to fulfilling one’s expected role. The following embedded thought – ‘Est bonum si Y facit has rēs’ – introduces the social evaluation of this behaviour: the idea that it is desirable for a person or thing of type Y to do such things – and that doing them well is what constitutes virtus.

For pedagogical purposes, a more concise version of the explication can be formulated as follows, using semantic molecules such as debeo and valdē:

This tool lends itself particularly well to classroom exploration, as it allows students to engage with the cultural logic of a term without relying on the inevitably reductive process of translation.