The synchrony and diachrony of example markers in Latin

1.1 A glance at the literature

In the history of Western culture, exemplification has been seen first and foremost as an argumentative device, and since Aristotle’s classification of the various types of παραδείγματα (Rhetorica 2.1393a), it has featured prominently in many treatises and works on rhetoric (Demoen 1997; Lyons 1989; Price 1975; Spina 2008). This operation has its place in almost any argued discourse or text because it helps speakers and writers to convey their intended message, easing the hearers’ and readers’ effort of understanding and remembering information (Manzotti 1995). Due to the pervasiveness of examples in formal and informal registers, scholars have investigated their role in scientific papers of various disciplines (Triki 2021) but also in learners’ writings (Delcambre 1997; Paquot 2008) and in news reports (Arpan 2009; Zillmann 1999; Zillmann and Brosius 2000). Moreover, exemplification has been described under different theoretical angles and in relationship with other textual operations, such as paraphrasing (Fuchs 1982), reformulating (Cuenca 2001; Gülich and Kotschi 1987), and code glossing (Hyland 2007). Research has also been devoted to the semantic and syntactic aspects of the exemplifying structure, which some scholars view as a non-prototypical apposition (Meyer 1992; Quirk et al. 1985) and others as a concatenation (Fernández-Bernárdez 1994–1995; Manzotti 1995). More recently, a growing body of literature has explored the pragmatic aspects of exemplification in speech corpora and its relation to various discursive and cognitive operations, such as hedging and mitigating (Barotto 2018; Mihatsch 2010), constructing lists and building categories (Barotto and Lo Baido 2021; Barotto and Mauri 2018; Chauveau-Thoumelin 2018). While some studies adopt a comparative perspective (Cuenca 2001; Eggs and McElholm 2013), the majority focus on exemplification strategies in a specific language (see also Barotto 2021; Landolsi 2018; Lo Baido 2018), but very few deal with their diachronic aspects (Rodríguez-Abruñeiras 2017, 2021]).^[1]

From this point of view, the investigation of Latin example markers reveals some unexpected facts: while only a few of them survive as archaic or formal expressions, such as It. verbigrazia, verbicausa, esemp(l)igratia, Sp. verbigracia, Eng. e.g., more common adverbials such as It. per esempio, Fr. par exemple, Sp. por ejemplo, Rom. de exemplu, Eng. for example, do not find a direct antecedent in Classical and Late Latin texts. Although all the markers containing epigones of the noun exemplum develop their function relatively late (see Section 7 infra), it is precisely the semantic richness of this Latin word that shapes the different modes of exemplification from the very beginning.

1.2 Two kinds of exemplification: the exemplum and the example

The noun exemplum, which literally means ‘what is taken out as a sample’, is related to the verb ex-ĭmo ‘to take out, take away, remove’, a derivative from ĕmo ‘to take, buy, purchase’, and develops from the form *ex-em-lom by the insertion of an epenthetic p (Leumann 1977 [1963]: 165). The wide range of meanings and multiple connotations of this term are discussed in detail by Kornhardt (1936), who organizes the ThLL lemma into four major sections with several subtypes, while the OLD provides nine separate definitions.^[2]

The phrase per exempla ‘through examples’ occurs only twice in Classical Latin, and its use reveals two meanings relevant to exemplification. In (1) Seneca employs this expression to represent a mode of learning: Lucilius’ path to wisdom will be facilitated by personal observation of authoritative models for moral instruction and imitation, while in (2) Pliny summarizes the content of a chapter on celestial portents, whose variety is illustrated by several facts and precedents:^[3]

In rhetorical, philosophical, and legal writings, the exemplum is what we might call a “good example”, that is, an exemplary story, anecdote, or fact from history, literature, jurisprudence, etc., that serves an edifying or persuasive function. In scientific, technical or grammatical treatises, on the other hand, we often find “canonical examples”, that is, demonstrative cases or representative items of the observed phenomenon, rule, or category.

This article cannot do justice to the extensive literature on the rhetorical exempla and will therefore only present the ideas and works on which it draws the most, with particular attention to the comparison with canonical examples, whose linguistic aspects are the true focus of research.

1.3 Research objectives and methods

Canonical exemplification typically entails an asymmetrical relation between two formal and logical units: an exemplified unit (the illustrandum) with a generic referent and an exemplifying unit (the illustrans) consisting of one or more elements that clarify, illustrate, or specify the first unit.^[4] In Latin, as in other languages, this relation is signalled by various devices, called example markers, i.e., different words (e.g., ut, uelut, sicut, tamquam), phrases (e.g., exempli gratia, exempli causa, uerbi gratia, uerbi causa), and constructions (e.g., ut puta, si dicas, etc.), which usually precede the exemplifying sequence, as shown in (3):^[5]

In handbooks of Latin syntax and stylistics, the structures and means of exemplification are essentially neglected, and only sparse remarks on some individual expressions can be found in the literature. Therefore, the first objective of our research is to provide an inventory of the various linguistic markers and to study their use in different contexts and literary genres. To this end, we have carried out a qualitative and partially quantitative analysis of the data collected through the Library of Latin Texts and PHI 5 Latin Texts corpora.^[6] In order to categorize and discuss the Latin materials, we have devised a functional taxonomy of the linguistic signals of exemplification that has the potential for cross-linguistic applicability and is also inherently dynamic.

In fact, the second goal concerns the diachronic dimension of exemplification and the investigation of the specific sources and paths that lead to the emergence of dedicated markers: their frequency and their fate over time will not only reveal some unexpected facts, but also shed a different light on the expressions used today in many modern languages. Indeed, up to now, research has privileged the synchronic dimension of exemplification and its role in well-established textual and communicative domains. The novelty of our approach is that Latin can show how the use of examples reflects the needs of nascent genres, which gradually shape their own specific strategies and markers for this discursive operation. In a broader epistemological perspective, the investigation of classical languages, repositories of a culture in which texts were intended to be read aloud, also helps to understand the illustrative and demonstrative nature of the example, and the cognitive processes (e.g., comparison, cooperation with the interlocutors) that, still today, underlie its linguistic expression.^[7]

The paper is structured as follows: after a brief illustration of the rhetorical exemplum, Section 2 discusses the textual and cognitive dimensions of exemplification, and then its syntactic and semantic aspects. Section 3 offers a classification of the linguistic signals that introduce the example and an overview of their discursive and pragmatic functions. The following part illustrates the multiple ways in which exemplification is carried out in Latin, with a focus on the diachronic sources and paths that lead to the emergence of dedicated markers: Sections 4 and 5 deal with those developing from meta-exemplification, while Section 6 is devoted to those connected to para-exemplification. Section 7 summarizes the main findings and outlines their possible implications at the cross-linguistic level.

2.1 Rhetorical and argumentative aspects

According to Lyons, given the etymological connection between the noun exemplum and the verb eximere, “the example is something cut out and removed from some whole. In this sense, example is synonymous with a modern term that appears very different: detail. The detail is also removed, cut out (dé + tailler in French, de + tagliare in Italian). […] exemplum concerns a distinction between a prior whole and a resultant fragment” (Lyons 1989: 9). But perhaps more convincing is the parallel in the Rhetorica ad Herennium, which equates the example with a sample of goods to be shown to the buyer:

Moreover, rhetoricians and grammarians used the term exemplum to translate the Greek παράδειγμα, which is “related to paradeiknumi (‘to exhibit side by side to make comparisons, to indicate or point out’) and to deiknumi (‘to bring to light, to show forth’). The Greek term is therefore always associated with light, showing, seeing, and pointing” (Lyons 1989: 10).^[8] This explains why, since antiquity, the Latin term could denote both what is chosen as an ‘exemplar’ and what is shown as ‘exemplary’.

The visual nature of the exemplum (David 1980: 73) is made clear by another passage, which states that it places the past (facti aut dicti praeteriti) before the audience’s eyes (ante oculos):^[9]

The point is further developed by Gazich (1990]: 79–82 and 121–122), who discusses the metaphor of sight also in relation to the concepts of similitudo, demonstratio, and imago. In particular, he focuses on Quintilian’s chapter that treats the exemplum as a species of comparison:

In this ^[10]passage the exemplum is not simply the citation of an ancient fact, but also the argumentative proof that compares, on the basis of a similarity, the particular fact under examination and the exemplary one (Gazich 1990: 107). These observations confirm that the exemplum is first and foremost a comparison (“est d’abord une comparaison”, David 1980: 81) and “is in actuality simply a special case of the broader similitudo” (Lausberg 1998 [1960]: 196), but also highlight the centrality of comparison and analogy to the Roman conception and use of examples in social and legal discourse: “analogical reflection through examples leaks into and saturates every stream and current of thought. […] At its source exemplary thought functions by the process of comparison” (Urban 2011: 32). As we will see in Section 6.2, the tradition of exemplarity not only builds a mode of moral and practical reasoning based on analogical thinking and rhetorical induction, but also contributes to shaping the way exemplification is conceived and expressed linguistically in Latin texts.

2.2 Textual and cognitive aspects

According to Manzotti (1995]: 2–3), exemplifying is a particular “linguistic action” which pertains to textual organization or constitution and by which the writer progressively shapes the overall architecture of the text.^[11] The textual forms by which exemplification integrates into this architecture may depend on the nature and functions of the exemplifying sequence. In principle, the exemplary example can stand alone and provide a reduced model of the matter, which is metaphorical in nature, since the cited case shows a concrete image of the abstract notion or theoretical proposition under consideration.^[12] On the contrary, the proper example cannot appear in isolation, but presupposes an accumulation of related exemplars, which is metonymic in nature, since the cited items are representative cases or parts of a phenomenon or category.

Depending on whether it is condensed into a single case or expanded into an indefinite list, the exemplifying sequence can assume the textual forms of the narrative or enumeration, respectively.^[13] In fact, Gazich (1990]: 131–132) remarks on the close similarity between Quintilian’s definitions of exemplum, in (7), and narratio, in (8):^[14]

On the other hand, an exemplification presented in the form of an enumeration, or a virtually open list, can be compared to a descriptive sequence (Gülich and Kotschi 1987: 40); however, while a description typically starts from an object and moves towards its decomposition/recomposition into different parts, exemplification usually starts from a set and moves towards the accumulation of its different members.

In any case, neither the narrative nor the enumerative example should be as extensive as a digression, firstly because exemplification is a way of filling ‘physiological’ pauses in communication that serve to minimize the flow of information and reasoning, allowing the reader to ease the effort of understanding and remembering (Manzotti 1995: 8). Secondly, because exemplification, which is typically turned backwards, opens a space in which to justify, point out, consolidate what precedes, descending from the previous level of abstraction to a degree of greater concreteness and, therefore, greater ease of control (Manzotti 1995: 8).^[15] Manzotti also observes that effective exemplification works as a strategy of ‘copying’, which, starting from the reassuring basis of the already said, helps to discover and order relevant cases, and to bring out specific problems, with the retroactive effect of specifying and articulating general statements, concepts, or categories.

This view of exemplification as a means for ‘retro-interpretation’ partially aligns with earlier analyses, which construed this textual operation as a sub-type of paraphrase (Fuchs 1982) or reformulation (Gülich and Kotschi 1987). According to Fuchs (1982), the activity of paraphrasing consists in establishing a subjective relation of identity between utterances, which also occurs when the sequence Y that reformulates the sequence X “se présente comme une sorte de cas particulier, d’exemple illustratif de X” [appears as a sort of specific case, an illustrative example of X] (Fuchs 1982: 107–108). However, as Cuenca (2001]: 50) notes, reformulation generally elaborates a concept by intension, whereas exemplification is a mechanism of definition by extension, so that, in the first case, a total referential correspondence between X and Y is possible, though not necessary, whereas, in the second case, this correspondence is not possible since Y is often a subset of X. On the same line, Landolsi (2018]: 49) remarks that exemplification does not necessarily presuppose the preservation of meaning and implies a certain hierarchy between its two units.^[16] Rather than a meta-predication of equivalence, exemplification can thus be viewed as a meta-predication of similarity, relevance or inclusion depending on the varying nature of the illustrandum and the functions of the illustrans in different textual genres. As explained by Ouellet (1989),

[l]’exemple, comme le ‘type’ dans les procédures de catégorisation, est une ‘entité’ qui se situe à mi-chemin entre la définition intensionnelle d’un concept et sa définition extensionnelle par une classe d’objets. Son rôle est d’extraire de cette dernière un ou plusieurs objets, tombant sous le concept, non pas pour donner une référence à celui-ci, mais pour servir d’image [ou] de modèle […] pour toutes ses références possibles [the example, like the ‘type’ in categorization procedures, is an ‘entity’ that lies halfway between the intensional definition of a concept and its extensional definition by a class of objects. Its role is to extract from the latter one or more objects that fall under the concept, not to give a reference to it, but to serve as an image or a model […] for all its possible references]. (Ouellet 1989: 113)

Concerning this operation of “extraction”, Manzotti (1998]: 121) refers to the interchangeability of the cases mentioned as examples, which Mihatsch (2010) qualifies as “arbitrary choices, since the selected item is just one of a number of possible other examples” (Mihatsch 2010: 108). Nonetheless, the choice of an example is never aleatory, since the individuation of a model case must take into account the tertium comparationis on which the comparison is based, while an illustrative case must be relevant to the phenomenon or included in the category to which the exemplified refers.

On the other hand, both a model and an illustration are at the centre of attention (“au centre des regards”, Herman 2011: 117), because, as argued by Rossari and Jayez (1999]: 463), the various strategies of exemplification share the same textual function, that is, focalization. Indeed, such a procedure presents a segment of the sentence as the result of a selection aimed at conveying new or conflicting information, and the same effect occurs in the shift from the abstract to the concrete or from the general to the particular proper to exemplification.

Although an overtly mentioned general unit is not necessarily required (cf. Eggs and McElholm [2013] and Section 3.1 infra), the twofold structure and the operation of inclusion are key points in the description of exemplification, not only at the textual and cognitive levels but also at the syntactic and semantic levels.

2.3 Syntactic and semantic aspects

Focusing on the presence of two juxtaposed units, some scholars regard exemplifying constructions as non-prototypical appositions. According to Quirk et al. (1985]: 1308), “most appositive” structures imply semantic equivalence between the juxtaposed elements and convey appellation, identification, or reformulation, while “least appositive” ones display unequal relationships between their units and convey particularization or exemplification.^[17] In such cases, the second appositive has a relation of inclusion with the term in the first appositive because it is more specific (Meyer 1992: 74; Quirk et al. 1985: 1315).^[18] More precisely, particularizing appositions featuring a hypernym and a hyponym imply “meaning inclusion” and “kind of” relations. Those with markers that focus on a referent of the first unit (e.g., particularly or including) imply “referential inclusion” and “part/whole” relations. Both are thus similar to exemplifying appositions, which are normally distinguished by explicit marking (Meyer 1992: 71–77). For Quirk et al. (1985]: 1315), exemplification markers are not mandatory, but their absence can lead to ambiguities with structures conveying identification and, we add, also with hyponymous appositions, especially when the second unit contains an asyndetically coordinated series, as in ex. (2). In general, the distinction between particularization and exemplification seems quite problematic, especially considering that forms like Eng. including or Fr. notamment are also used to introduce examples (cf. Landolsi 2018; Rodríguez-Abruñeiras 2019a, 2019b).

Moreover, since exemplifying structures connect not only noun phrases, but also larger segments of discourse, such as sentences or groups of sentences, the term “apposition” does not seem fully appropriate to denote the syntactic relation under study. For Fernández-Bernárdez (: 115–116), the markers of exemplification, like those of explanation and specification, concatenate parenthetical units to other segments within the spoken chain (cf. also Manzotti 1995: 14).

As we will see, parentheticality also plays a role in the evolution of example markers, which, in addition to serving as linking devices, are also used to convey pragmatic functions. The discussion of their procedural functions will complete the description of the linguistic aspects of exemplification, which indeed implies syntactic, semantic, and pragmatic relations between utterances or text segments.

3.1 Classification

The relationship between the two units of an exemplifying structure can be implicit or explicitly marked. For Manzotti (1995]: 14), it is reasonable to assume that the ‘action of exemplification’ occurs only in the presence of a signal that enriches the concatenation beyond simple juxtaposition. In this regard, he makes a distinction between underdetermined exemplification, which occurs when the signals are paralinguistic (i.e., pauses, interjections, gestures) or, in writing, graphic (e.g., brackets, dashes, colons), and linguistically determined exemplification, which can be indicated by various means.^[19]

We set out to classify the strategies that guide the hearer/reader’s interpretive tasks into three main types: (i) meta-exemplifiers, i.e., meta-discursive speech acts or utterances that denote the pragmatic awareness of the speaker/writer (e.g., I will give an example, or an example is the following); (ii) canonical example markers, i.e., dedicated adverbials and locutions that overtly mark an element or segment of discourse as an example (e.g., for example or for instance); (iii) para-exemplifiers, i.e., linguistic items that originally and primarily have other functions, such as comparison and equation (e.g., Eng. like, Fr. comme, It. come), focalization and particularization (e.g., Eng. in particular, including, Fr. notamment, It. in particolare), reformulation and hedging (e.g., Eng. say, let’s say, Fr. disons, genre, It. diciamo, metti, tipo), consequence and conclusion (e.g., Fr. ainsi, It. così).^[20]

As we will see in Sections 5 and 6, the development of example markers in Latin proceeds both from strategies of meta-exemplification and from expressions performing para-exemplification. Not surprisingly, in the synchrony of many languages, the boundaries between para-exemplifiers and canonical example markers are often blurred, because these expressions converge in the category of functional markers, that is, communicative devices that have limited or no effect on propositional meaning but serve multiple procedural functions at both the discourse and pragmatic levels.

3.2 Discursive and pragmatic functions

More precisely, functional markers serve primarily as discourse markers that index the structure of the discourse, by providing textual cohesion and specifying how the message and its content are to be interpreted. In addition, they can serve as pragmatic markers that index the framework of the interaction, by expressing the speakers/writers’ subjective stance in relation to the content of their utterance, or by negotiating prior assertions and viewpoints at the intersubjective level (Ghezzi 2014: 15).^[21]

The basic function of example markers and para-exemplifiers is to guide the interlocutor in recognizing the role of the utterance in which they occur and to signal an element or segment of discourse as an example. As such, they typically have metatextual functions (Bazzanella 1995) but can acquire further discourse and pragmatic values depending on the context and the intentions of the speaker/writer. This explains why these functional markers can also perform reformulating, focalizing, or quotational functions at the metatextual level, convey epistemic hedging, politeness, or shared knowledge at the interactional level (Barotto 2018; Barotto and Mauri 2018; Lo Baido 2018), or even denote approximation at the propositional level (Mihatsch 2010).

To this end, speakers/writers can either use example markers as flexible pragmatic devices that serve multiple functions in the same context, or recruit para-exemplifiers. In Manzotti (1995) the term paraesemplificazione applies to forms that serve operations similar or close to exemplification and that, as in metaphor and metonymy, obtain a “figurative” exemplification based on contiguity which combines the properties of the two operations involved (Manzotti 1998: 114). However, in our view, the prefix para- also captures the “parasitic” nature of these exemplifiers, as their function is often inferred by adjacency when they precede a list of items or when they combine in syntagmatic chains, as in like for example (Eggs and McElholm 2013: 81–82). We can finally add that pragmatic devices are polyfunctional also at the paradigmatic level, so that the same marker can display different functions depending on the context or the position in which it occurs. The mitigating role of par exemple and for example at the end of the utterance confirms that, while linguistic items in the left periphery typically serve discourse coherence and textual organization, those in the right periphery perform interactional uses before the turn shifts: anticipating the upcoming discourse, reformulating the message, fine-tuning illocutionary force or epistemic stance, conveying intersubjective meanings (Beeching and Detges 2014: 11; Degand 2014: 158).^[22]

As the above considerations show, categorizing exemplification signals according to their functions is a difficult task, but alternative classifications are no less problematic. In their monograph on English and German example markers, Eggs and McElholm (2013]: 12–13) posit a distinction between descriptive uses at the sentence level, where the example is to be understood as a complete communicative unit, and argumentative uses at the transphrastic level, where the connectors link whole chunks of discourse. Then, considering structures, Eggs and McElholm label examples with an explicit general unit as argumentations when their twofold structure consists of full sentences, and as exemplifications when smaller units such as phrases or clauses are correlated; instead, examples without an explicit general unit (or anchor) are called selections. Although focused on the use of example markers, this taxonomy actually defines (and definitely narrows) the domain of exemplification; moreover, it does so through a controversial terminology vis-à-vis the current debate, in which the concept of selection is a defining feature of the canonical example.

Rodríguez-Abruñeiras (2017]: 89), on the other hand, proposes to classify English example markers based on two parameters: the degree of emphasis conveyed on the example they introduce and their position in the exemplifying sequence. Her investigation, however, starts from a “semantic classification” that identifies four types of markers: neutral, i.e., non emphatic (e.g., for example, for instance), comparative (e.g., like, such as), focalizing, i.e., slightly emphatic (e.g., in particular, including), and hypothetical (e.g., say, let’s say). This hybrid and somewhat sketchy categorization, however, does not seem suitable for analysing the data of Latin, where emphasis is difficult to measure and example markers are still developing.

In the following, we will therefore present the strategies for exemplification in Latin by referring to the notions of meta-exemplification, canonical exemplification, and para-exemplification mentioned above (Section 3.1). This simple and inherently dynamic taxonomy will allow us to discuss the data from both a synchronic and diachronic perspective.

4.1 Underdetermined exemplification

Before illustrating the linguistic means used to signal, and/or comment on, the operation of exemplification, it is worth briefly mentioning asyndetic sequences that can be construed as underdetermined exemplifications. In modern editions of Latin texts, editors and translators can highlight these interpretations by adding appropriate punctuation or explicit example markers.

This is illustrated in (9), where a quotation from Vergil (georg. 3.243) follows a statement about the influence of metrical norms on accentuation, to show how lengthening by position works. In the Latin text of the Loeb edition, the inserted colon, like other punctuation marks, serves to reduce the kind of relations that can hold between the two juxtaposed propositions, which in the translation are explicitly linked by as in:

In (10) the category of perennial rivers is illustrated by three representative items: the Nile, the Danube, and the Rhine, which Seneca introduces with the two deictics hinc, to suggest gestures pointing in different directions, and which the translator of the Loeb edition, renders with for example:

In both cases, translations amount to interpretations that exploit the co-text to bridge the asyndetic gap between the juxtaposed units and to bring out relations of exemplification.

4.2 Meta-exemplification

More interesting for our investigation is the wide range of meta-exemplifiers, i.e., non-formulaic expressions containing verbs or nouns that refer to the communicative process (e.g., say, explain, understand, term, example, etc.) and indirectly announce an operation of exemplification (cf. Fernández-Bernárdez 1994–1995: 104–106).^[23] In most cases, these kinds of metacommunicative utterances and speech acts, which denote the author’s pragmatic awareness, contribute to integrating the example into the flow of discourse. In early grammatical and rhetorical texts, as Vainio (2000) observes, “the example tends to conform to the syntax of the sentence whenever possible. Quintilian likes to form the sentence with verbs such as dicere or legere in order to include the example in its syntactic environment” (Vainio 2000: 33). This is shown by the following passages, where materials for metalinguistic reflection are presented as fragments of spoken language:^[24]

The use of the verb dicō, which frequently occurs in formulas like si dicam ‘if I say’, etc. (see Section 6.3.2 infra), shows the proximity between the operations of exemplification and quotation. Grammatical examples may cite words and structures from real speech, occasionally mentioning an enunciative source, such as the Ancients in (12).^[25]

On the other hand, the expression hoc ut scias ‘so that you know’, which in (13) is freely rendered by the translator, suggests some proximity between exemplification and explanation:

This locution is a stylistic feature peculiar to Seneca that is destined for oblivion; still, it is relevant to note how verbs of knowledge grammaticalize into Fr. à savoir and Sp. a saber ‘that is to say, namely’, which mainly serve the functions of explanation and reformulation but in specific contexts can also perform exemplification (Fernández-Bernárdez 1994–1995: 104).

An interesting metacommunicative speech act is found in (14), where Cicero uses the verb ponō to suggest a non-exhaustive list of pleasant activities and situations:

The verb ponō, as we will see in a moment, often performs meta-exemplification in combination with the noun exemplum.

5.1 Exempli gratia

The above discussion suggests that markers whose core function is canonical exemplification are not yet fully developed in Classical Latin, which lacks expressions equivalent to ‘for example’ or ‘for instance’ in all their possible syntactic, semantic, and pragmatic uses. However, exempli gratia – which will be very successful in medieval argumentative Latin prose – seems on the verge of becoming a true example marker. The ambiguous passage from the Naturalis Historia in (33) shows that, once dislocated in peripheral position, exempli gratia appears detached from the syntax of the sentence and available for cooptation (Heine 2013: 1221–1224; Heine et al. 2021) as a marker introducing the exemplifying sequence.^[31] Therefore, the emergence of markers devoted to exemplification can proceed from strategies of meta-exemplification through syntactic shifts, bridging contexts, and parentheticalization.

Interestingly, it is in the very same work by Pliny that we find two of the first attestations of the phrase as a true example marker, as shown in (35) and (36):

Technical prose is the genre in which canonical examples abound, since complex categories need to be illustrated by representative items, as in (35), and general statements need to be proven by detailed facts, as in (36). Thus, canonical exemplification develops its own structures and linguistic markers within a literary genre in which the word exemplum shifts from the sense of ‘exemplary model, precedent’ to the meaning of ‘typical instance, illustration’. In accordance with the nature of the illustrandum, the illustrans is often multiplied into a virtually open list of illustrantia, which are selected as parts or proofs of the illustrandum itself. In other words, while the rhetorical exemplum persuades or argues as a model or a precedent, showing similarities or analogies through comparison, the canonical example illustrates as a sample or demonstrates as a proof, focusing on details or properties through selection.

The third attestation of exempli gratia as an example marker suggests a sort of continuity between the two modes of exemplification: in (37) Cicero discusses the conflict between expediency and honesty in business by means of a hypothetical situation. What follows exempli gratia concretizes an abstract issue, but in the form of a narrative digression that lies between the illustrative example and the moral exemplum:

Although scarcely attested in Antiquitas and primarily used for meta-exemplification in Aetas patrum (raw data: 108 occurrences, including 41 in Augustin), exempli gratia becomes a widespread example marker and, over time, undoubtedly more fortunate than exempli causa.

5.2 Exempli causa

As said in Section 4.2.2, exempli causa is mostly used as an adjunct or as a meta-exemplifier, but in (38) Cicero uses the same mode of exemplification seen in (37): the phrase introduces a hypothetical case that illustrates how to respond to an accusation of theft:

In the same period, Columella explains how to calculate a rectangular area by solving a concrete problem of measuring a field:

The sections Aetas Patrum of the LLT database show 68 occurrences of exempli causa, mostly in conjunction with ponō, to convey meta-exemplification. Interesting exceptions are found in De medicina, a fifth-century medical treatise where the physician Cassius Felix uses ut exempli causa to introduce a list of alternative illustrations, as in (40), from the chapter Ad reuma ‘on flux, rheum’:^[32]

The use of example markers combined with para-exemplifiers like ut will be discussed in Section 6.2.1, after completing the description of the phrases with gratia and causa.

5.3 Verbi gratia

Beside those with the noun exemplum, also the phrases containing uerbum can perform the operation of exemplification, since the noun ‘word, term’ also refers to the communicative process. In the period Antiquitas, the expression uerbi gratia occurs 27 times and in 26 cases behaves as a parenthetical example marker.^[33] However, while exempli gratia literally means ‘for the sake of an example’, uerbi gratia literally means ‘for the sake of a word’ and, as such, acts as metacommunicative device commenting on the choice of an expression, sometimes a “loose” example without an explicit illustrandum, which is presented as an alternative within a set of equally illustrative items.^[34] The focus on a single term and the nuance of apparent randomness are quite evident in (41), where uerbi gratia interrupts the phrase populum Centuripinum: Centuripe is just one among many rich cities in Sicily:

Rather than ‘arbitrary choices’, however, the examples introduced by uerbi gratia are often improbable eventualities or surprising facts, which are presented, as such, in hypothetical or mitigated form. In (42), pleasure is an absurd motivation for self-love:^[35]

In (43), irregularity in inflection is illustrated by means of a paradoxical anomaly:

In (44), Pliny describes stones with wonderful pareidolic figures:

In similar cases, uerbi gratia performs pragmatic hedging as a shield, modulating the writer’s subjective stance towards the plausibility of a controversial or incredible illustrans. In other cases, it performs semantic hedging as an approximator, rounding off quantities and measures in conjunction with numerals, as in (45):^[36]

The increasing use of uerbi gratia in technical prose (see also Frontin. aq. 34.4.1) is confirmed by the fact that 19 out of 26 occurrences are found in Gaius’ Institutiones.^[37] In most passages, the example marker is used parenthetically (within quotations), as in (46), or inserted in hypothetical contexts illustrating private law cases, as in (47), since it is very common to exemplify an abstract concept or rule by proposing an imaginary situation:

With more than 900 attestations in the period Aetas Patrum I and II, uerbi gratia is indeed the most fortunate of the Latin expressions related to exemplification, also compared to the variant uerbi causa (cf. Table 1).

5.4 Verbi causa

The phrase uerbi causa is attested twelve times in the period Antiquitas, and in ten cases behaves as an example marker with the same characteristics and parenthetical uses of uerbi gratia. Since the syntactic independence of discourse markers is also related to their scope variability, uerbi causa can focus and act upon discourse units of almost any size, or remain ambiguous between phrasal and clausal scope, as in (48) and (49):

Used as a hedge to attenuate the full meaning or the pragmatic force of an utterance, this parenthetical expression can also modulate the writer’s attitude towards a nonsensical or disputable illustrans, as in (50) and (51):^[38]

In other cases, uerbi causa negotiates the suitability of “loose” examples to the matter at hand, as in (52)–(54):

The passage in (54) nicely explains the difference between testimonium and exemplum, which is a means of clarifying a statement, as the open list in (55) does:

Here, the second ita ut is ambiguous: in parallel with ita … ut physici, it triggers the inference of an authority (indeed, common sense), but joined with uerbi causa it confirms the interaction between example markers and para-exemplifiers.

6.1 Para-exemplifiers

As already mentioned in 3.2, in English and Romance languages para-exemplifiers develop from linguistic items that convey diverse textual or pragmatic functions and perform exemplification at the intersection of two contiguous cognitive domains. The same phenomena occur in Latin, where various connectors and expressions flank – and often precede – the emergence of true example markers.

More specifically, we distinguish the following types of para-exemplification:

1. analogical (cf. Fr. comme, It. come), performed in Latin by the markers of similative and equative constructions ut, uelut, sicut, tamquam, and by the adjective qualis;

2. cooperative (cf. Eng. let’s say, Fr. disons), performed by the parenthetical imperative puta, especially in combination with ut;

3. demonstrative (cf. Fr. ainsi, It. così), occasionally performed by the conclusive connectors igitur and itaque, and by the explicative connectors nam, namque and enim;

4. mitigating (cf. colloquial Eng. dunno, It. magari), performed from Late Latin onwards by the dubitative adverbs forsitan and forsan;^[39]

5. particularizing (cf. Eng. including, Fr. notamment), but the Latin adverbs praesertim, praecipue, speciatim are never used for this kind of para-exemplification.

In what follows we will therefore analyse analogical, cooperative, and demonstrative para-exemplifiers. Due to the frequency and the high degree of polysemy of the markers at issue, quantitative data will be limited to the emergence of (ut) puta as an example marker.

6.2 Analogical exemplification

Comparative markers are highly polysemic and can serve various semantic and pragmatic functions because, as Tarriño (2011) notes, the relationship between the comparative construction and the main clause tends to remain vague, leading to different interpretations and to “all kinds of pragmatic inferences that bringing the two situations into contact makes possible” (Tarriño 2011: 420). In particular, the fact that the illustrantia share some properties with the illustrandum can make it difficult to determine whether a discursive sequence is (merely) analogical or (also) serves exemplification (Miéville 1983), and this explains why Latin, like many other languages, recruits similative markers as para-exemplifiers.^[40] Furthermore, as mentioned above (cf. Section 2.1), comparison plays a relevant role in the narrative and instructive rhetorical exemplum, which is often viewed as a type of simile. Considering the continuum between rhetorical and canonical examples (Herman 2011: 97), analogical para-exemplification is thus the domain where the two types meet.

From a syntactic viewpoint, analogical para-exemplifiers do not display the same positional flexibility as canonical example markers and cooperative para-exemplifiers (cf. Eng. for example, It. diciamo), as they typically occur before the illustrandum. Moreover, when Latin similative markers are employed as para-exemplifiers, they do not feature the correlative particle (ita or sic) in the main clause.^[41] That said, let us now turn to the peculiarities of ut, uelut, sicut, tamquam and qualis.^[42]

6.3 Cooperative exemplification

The main purpose of exemplification, in accordance with the cooperative principle that underlies human communication, is to facilitate the understanding of abstract concepts or categories for the addressee (Grice 1975). This is precisely why exemplification is often signalled by reduced parenthetical clauses with interactional meaning. In Schneider’s (2007) definition, clauses such as Eng. I must say or I think are “inserted into or adjoined to the beginning or the end of a host clause in a way similar to sentence adverbs” and “they can be added as well as dropped freely without endangering the grammatical acceptability of the host” (Schneider 2007: 1–2). Schneider also observes their almost unlimited mobility and their connection to the main clause, which pragmatically corresponds to their missing argument (as in Eng. the soup is quite cold, I must say vs. I must say that the soup is quite cold).

Some types of reduced parenthetical clauses tend to evolve into signals of cooperative exemplification, e.g., Eng. say, Fr. disons, It. mettiamo, Rom. să zicem ‘let’s say’. These expressions typically involve verba dicendi or cogitandi in the second person imperative or in the first person plural indicative, emphasizing the cognitive effort required of the addressee, or the speech act involving the speaker and, often, the interlocutor as well.

In this section we will investigate the exemplifying role of the parenthetical imperative puta (especially in combination with the analogical para-exemplifier ut) and of some locutions formed by the protasis of a conditional clause with a verbum dicendi (such as si dicam).

6.4 Demonstrative exemplification

Under the label ‘demonstrative exemplification’, we will discuss the peculiar uses of a group of cohesive devices that convey heterogeneous relations between a preceding textual unit p and a following sentence or larger unit q. More specifically, we will first compare the connectors igitur and itaque, which usually mark sentences that stand in a relation of consequence, inference, or result to the preceding ones, and then the conjunctions nam, namque, and enim, which signal explanatory, evidential, or causal connections between the second and the first unit (Pinkster 2021: 1193 and 1208).

Within specific cotextual configurations, these expressions denote a logical transition from the general assertion p to the particular case q contemplated by p, so that q stands as a demonstration of p. Accordingly, the cases in point chosen for this type of exemplification are never condensed and listed into phrasal form but are always expanded at the sentential or discursive level (Manzotti 1995: 35). Being logically linked to what precedes them, these narrative examples confirm the retrospective nature of exemplification (cf. Section 2.2), but their role as illustrantia tend to remain implicit and in some cases ambiguous, whereas true example markers and analogical or cooperative para-exemplifiers mark the exemplifying sequence more explicitly and directly.

Finally, as will be seen in the following, the term ‘demonstrative’ refers not only to the synchronic functions, but also to the diachronic sources of some of the expressions that convey this type of exemplification.