2.1. Negation: Terminology and the Case of Standard French

Traditionally, a distinction is made between standard and non-standard negation. Standard negation is “non-emphatic negation of a lexical main verb in a declarative main clause” (van der Auwera & Krasnoukhova, Reference van der Auwera, Krasnoukhova, Déprez and Espinal2020: 91), such as:

Standard French exemplifies the Jespersen’s cycle for standard negation. The study period (Medieval and early Pre-Classical French) covers stages 1–3 (see Table 1), where ne originates as an unmarked preverbal clausal negator (Hansen, Reference Hansen, Allan and Robinson2012, Reference Hansen, Carlier and Guillot-Barbance2018; GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020 vol.2: 1681) that can function independently before being doubled by an emphasizer. These emphasizers later evolve into mandatory second negators in bipartite negation.

Table 1. Jespersen’s Cycle in French (based on Hansen in GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020 , vol. 2: 1680)

All other instances are considered non-standard negation, encompassing numerous types. Among these, the negation of indefinites is most crucial for our study, such as:

Regarding non-standard negation, negative polarity (henceforth NP), describing the relationship between a clause and semantic negation, is a key concept. Since Klima (Reference Klima, Fodor and Katz1964), strong NP, or proper negative clausesFootnote ¹ (examples 1 and 2), is distinguished from weak NP—interrogative, comparative, and hypothetical clauses (examples 3, 4, and 5); affirmative clauses possess positive polarity.

Some authors, notably de Swart (Reference de Swart2010), extend NP to technically non-negative clauses, such as privative negation through prepositional phrases headed by without (henceforth without-type):

Items appearing regularly in weak and strong NP contexts without being inherently negative are Negative Polarity Items (henceforth, NPIs), the textbook Standard English examples including anybody, anything, any as in 3-6.Footnote ² Conversely, items and constructions appearing freely in both negative and positive polarity contexts are polarity neutral.

According to Hansen (Reference Hansen, Carlier and Guillot-Barbance2018: §4.2, in GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020 vol. 2: 1682), Old French saw ne as the sole clausal negator in approximately 50% of cases. It was accompanied by a post-verbal emphasizer (pas ‘step’, mie ‘crumb’, goutte ‘drop’) in 20% of cases, and an indefinite in 30%. From MF, often passing through an NPI stage, these indefinites formed the French NI set.

In Modern Standard French, NIs serve two functions: negative quantifiers that express the negative meaning independently,Footnote ³ as in

and negative concord (henceforth NC), as in

NC is defined as two negative expressions conveying a single semantic negation (Auwera & Van Alsenoy, Reference van der Auwera, Van Alsenoy, Turner and Horn2018: 108);Footnote ⁴ typologically, French is an NC language (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020 vol. 2: 1688). Although, since Laka Mugarza (Reference Laka Mugarza1990) negative word/N-word (e.g. Déprez & Martineau, Reference Déprez, Martineau, Auger, Clements and Vance2004; Hansen, Reference Hansen, Mosegaard Hansen and Visconti2014, Reference Hansen, Carlier and Guillot-Barbance2018; Larrivée & Kallel, Reference Larrivée and Kallel2020) or else Negative Concord Item (Breitbarth et al., Reference Breitbarth2020) are common terms, we adopt van des Auwera & Van Alsenoy’s (Reference van der Auwera, Van Alsenoy, Turner and Horn2018) term negative indefinite (NI) encompassing both negative quantification and concord, fitting our article’s investigation into personne’s not only negative, but also indefinite value.Footnote ⁵

The French NI’s set originated in the Medieval period (Martineau & Déprez, Reference Martineau and Déprez2004; Déprez, Reference Déprez, Sleeman and Perridon2011; Larrivée & Kallel, Reference Larrivée and Kallel2020; GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 717-720). Gradually eliminating alternative expressions (e.g., nient ‘nothingness’, chose ‘thing’ as competitors of rien, see GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2: 1692), it reached its modern form by the 17^th century. While initial assumptions posited “French N-words started out as indefinite positive expressions that gradually acquired a negative value” (Déprez & Martineau, Reference Déprez, Martineau, Auger, Clements and Vance2004: 2), which is true for aucun (Prévost & Schnedecker, Reference Prévost and Schnedecker2004; Déprez & Martineau, Reference Déprez, Martineau, Auger, Clements and Vance2004), this claim was subsequently refined. The development is now described as progressing from polarity-neutral, but not necessarily indefinite (e.g., rien, from lat. noun rem ‘thing’, used in Early Old French and Old French as a noun with etymological meaning) through weak polarity to strong polarity (Martineau & Déprez, Reference Martineau and Déprez2004; Déprez, Reference Déprez, Sleeman and Perridon2011). Personne is traditionally assumed to parallel this evolution, often compared with rien (Déprez, Reference Déprez, Sleeman and Perridon2011: 277-279; Hansen, Reference Hansen, Carlier and Guillot-Barbance2018; GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2: 1692-1693). This view echoes the model proposed by Haspelmath (Reference Haspelmath1997: 230ff) for indefinites, illustrating what has been called the Jespersen’s Argument cycle (Ladusaw, Reference Ladusaw, Guenter, Kaiser and Zoll1993)Footnote ⁶ or the Quantifier cycle (e.g., Willis et al., Reference Willis, Lucas, Breitbarth, Willis, Lucas and Breitbarth2013). Nevertheless, while the Quantifier cycle is based on examples such as French, where some indefinites chronologically follow Jespersen’s cycle, it appears language specific rather than universal (van der Auwera et al., Reference van der Auwera, Krasnoukhova, Vossen, Veselinova and Hamari2022: 615). Moreover, it is not a cycle stricto sensu, as the final stage does not return to the initial one. Instead, van der Auwera et al. suggest a broader understanding of the Quantifier cycle illustrating it by Latin nemo and French personne:

Thus, it represents one of the Indefiniteness cycles (van der Auwera, Reference van der Auwera2024), being independent from the negative system evolution itself.

Finally, Hansen demonstrated that the temporal/aspectual adverbs (ja) + mais ‘more/no more/never’ and plus ‘more/no more’ directly grammaticalized as NIs in strong polarity contexts, developing weak polarity uses only subsequently (Hansen, Reference Hansen, Mosegaard Hansen and Visconti2014). Comparing their evolution to that of aucun, rien and personne, she conceptualizes the set of French negative words as a Gestalt figure “where either the paradigm or any individual member can become foregrounded at any given time” (Hansen, Reference Hansen, Mosegaard Hansen and Visconti2014: 208) and argues that French negation is marked by a “heterogeneity that cannot easily be reduced to the operation of a few simple principles” (ibid: 209). Furthermore, Hansen considers the set of French NIs as a family of negative constructions (Fr. famille de constructions GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2 2024). This term is based on Gisborne and Patten’s (Reference Gisborne, Patten, Narrog and Heine2011: 101) concept of macro-construction whose elements did not evolve at the same pace nor in the same direction, but resulted in a functional system with diverse subgroups. In our case study, we adopt Hansen’s perspective and will provide further evidence of the varied evolutionary paths within the macro-construction of French NIs.

2.2. Innovation in Speech and Where to Find It

The main issue in historical linguistics is, of course, the inaccessibility of native speaker competence and natural speech. This is particularly problematic when tracing innovations that typically originate in speech before passing in written production. One way to mitigate this is to integrate the diamesic perspective into the analysis, positioning texts along the immediacy-distance continuum.

Proposed by Koch and Oesterreicher (Reference Koch and Öesterreicher1985), the communicative immediacy-distance theory positions a continuum along which different forms of communication can be placed, based on the degree of shared context and planning involved. The communicative immediacy pole, representing conceptual orality, features high shared context and less planning; the communicative distance pole, representing conceptual scripturality, exhibits the opposite characteristics. Thus, natural, everyday communication leans toward immediacy, whereas a formal prepared public speech, though oral, is closer to distance. Similarly, written communication can be closer to the distance pole (e.g., cover letter) or to the immediacy pole (e.g., text message in a family chat). While the immediacy-distance theory does not solve the core problem of the speech inaccessibility, it allows for a more accurate analysis of ancient texts’ discourse than the simple speech/writing dichotomy.

A complementary approach is Marchello-Nizia’s (Reference Marchello-Nizia, Guillot, Combettes, Lavrentiev, Opperman-Marsaux and Prévost2012) represented speech perspective (fr. oral représenté). While historical pragmatics has long studied dialogue in distance-proximal texts such as romance, Marchello-Nizia suggested contrasting direct/indirect speech segments with narrative segments, based on the assumption that representing speech involves different linguistic strategies compared to narration. This holds true morpho-syntactically, revealing distinct patterns for represented speech and narrative segments (e.g., Guillot et al., Reference Guillot, Heiden, Lavrentiev, Pincemin, Jeppesen Kragh and Lindschouw2015; see also Lefeuvre & Parussa, Reference Lefeuvre and Parussa2020; GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 44-45). In the field of negation, Donaldson (Reference Donaldson2018) demonstrated that in Old French fiction, more innovative variants containing negative emphasizers pas, mie, and goute are significantly specific to direct discourse. Therefore, our analyses specifically considered both the texts’ position on the immediacy-distance continuum and the represented speech vs narration parameter.

4.1. Nominal Function

Studies of French negation that consider personne as an NI rarely focus on its preceding nominal function, limiting themselves to citing its Classical Latin origin (persona ‘mask’) as evidence of its original neutral polarity and stating its Old French meaning ‘people’/‘person’ (e.g. Déprez & Martineau, Reference Déprez, Martineau, Auger, Clements and Vance2004). It appears important, however, to provide a detailed overview of personne‘s nominal function in MF, as this has implications both for classifying it as a (negative) indefinite and for the dating of this change.

Originating from the Latin persona (in Classical Latin, ‘actor’s mask’ Reference WartburgFEW, VIII: 271a), personne is first attested in the 12^th century (Reference WartburgFEW,Footnote ⁷ VIII: 268b). An overview of major MF dictionaries (Reference GodefroyGdf, vol. 6: 115; Reference GodefroyGdfC, vol. 10: 323, Reference Tobler and LommatzschTLel entry persone, DMF entry personne) suggest the following meanings of personne in nominal use:

• “human being, person, as individual entity or as a physical body (by metonymy, face)”;

• “cleric, priest”;

• “one of the hypostases of the Holy Trinity”;

• grammatical category: the one who speaks, is spoken to, is spoken about.

As Geylikman (Reference Geylikman2024) demonstrated using the BFM19 Footnote ⁸ data, throughout the Medieval period, as a noun, personne is primarily attested in 12^th century text types that originally appeared in Medieval Latin—religious, legal, and didactic texts; conversely, the occurrences are initially very rare in vernacular literature—epic texts, romance. Table 3 shows personne’s distribution per text type in the BFM19 (Geylikman Reference Geylikman2024: 18, Table 3):

Table 3. Distribution of personne in BFM19 per text-type ^{Footnote 9}

According to Geylikman (Reference Geylikman2024: 18), personne emerges in religious texts as a semantic loan from Latin. Initially, it carries two main ‘religious’ meanings: ‘one of the hypostases of the Trinity’ as in 9:

and ‘cleric of a high rank’ as in 10:

From the 13^th century onward it appears with increasing frequency in its most common modern meaning—‘human being’—with extensive use in legal texts, still under Latin influence:

Tables 4-5 show the three meanings’ distribution per period according to Geylikman (Reference Geylikman2024: 19):

Table 4. Personne-noun meanings’ distribution per period in BFM19 (12^th–13^th centuries) Footnote ¹⁰

Table 5. Personne-noun meanings’ distribution per period in BFM19 (14 ^th –15 ^th centuries)

4.2. Towards Pronominalization: Personne as a General Noun

The meaning ‘human being’, places personne within the general human nouns lexicon—shell nouns (Schmid, Reference Schmid2000) referring to human beings in general, as a species (Schnedecker, Reference Schnedecker2018: §1). This general semantic nature fosters the comparison between rien and personne, both having evolved from general nouns toward indefinite pronouns in French diachrony (Classical Latin res ‘thing’ to Old French rien ‘thing’ to rien ‘nothing’).

According to Haspelmath (Reference Haspelmath1997: 182), general nouns tend to grammaticalize universally: this phenomenon occurs in 42 languages on a 100-languages sample encompassing three stages:

1. 1. The general noun is used without modifiers with a ‘general’ meaning, e.g. ‘someone’;

2. 2. The new item diverges from the initial form on phonological, morphological and syntactical levels;

3. 3. The general noun ceases to be employed in the original meaning (Haspelmath, Reference Haspelmath1997: 182).

While an anonymous reviewer noted the third stage is not technically a stage of grammaticalization, naming all the stages highlights personne‘s exceptional nature with respect to a universal tendency. Within French diachrony, its development differs from two other salient examples of grammaticalization from general noun to indefinite pronoun. Indeed, ContF on (indefinite pronoun ‘one, they’/personal pronoun ‘we’, already grammaticalized to indefinite pronoun in Early Old French, Marchello-Nizia, Reference Marchello-Nizia2009, ch. 3: § 25) and homme (ContF ‘man’) both derive from Latin general noun homo,-inis. Formally, the former comes from Nominative homo, while the latter descends from Accusative hominem. Although Early Old and Old French forms did not perfectly distinguish pronoun and noun uses, later graphic stabilization solidified this bimorphism, definitively separating the two uses.

Conversely, personne underwent no phonological change distinguishing the noun from the pronoun form. Unlike rien, the initial noun use did not disappear with the indefinite use emergence; instead, personne as a noun continues to function successfully in ContF across the immediacy–distance continuum (see Cappeau, Reference Cappeau2018: §15-§16).

The grammaticalization of general human nouns into indefinite pronouns is sometimes considered an area-specific feature of the Standard Average European Area (Giacalone Ramat & Sansò, Reference Giacalone Ramat, Sanso, Ramat and Roma2007). Mihatsch (Reference Mihatsch2017, in particular: 92), on the example of French and German, argues these nouns are inherently “on the edge of grammaticalization”, fulfilling traditionally grammatical roles: templates for modification/nominalization, counting units, and impersonal/indefinite pronoun equivalents in bare form.

Analyzing personne’s noun-to-pronoun shift, its general semantic nature makes defining the threshold between general noun and indefinite pronoun challenging. Moreover, in ContF, although nominal personne tends to be extensively modified (Cappeau & Schnedecker, Reference Cappeau and Schnedecker2014, in particular: 3036), the use of unmodified personne with an indefinite article is also possible, including weak polarity contexts. Thus, examples 12 and 13 are (almost)Footnote ¹¹ semantically equivalent:

whereas 15 is the negative equivalent of 14:

While the indefinite value is conveyed by the indefinite article, these examples’ parallelism is striking. Moreover, they demonstrate that ContF provides contexts where the noun/pronoun interpretation depends entirely on the presence/absence of the article. Therefore, as in other general noun-to-pronoun studies, we will rely primarily on morphosyntactic parameters to distinguish nominal use from potential NI use.

4.3. Grammaticalization toward an NI: state-of-the-art

Existing research papers, manuals, and dictionaries offer rather a heterogeneous picture of personne‘s indefinite use in MF.Footnote ¹²

The Reference WartburgFEW dates the first attestation of personne as an indefinite pronoun ‘somebody’, ‘anybody’ to 1226, without citing the source or providing evidence on the context polarity. From Middle French it supposedly occurs with the same value in various NP contexts (FEW, VIII: 270b), remaining rare before the 15^th century (FEW, VIII: 271b). The Trésor de la Langue française informatisé (TLFi) places personne‘s first NI attestation in a late 13^th century text (TLFi, entry personne, s.f.). The Reference Tobler and LommatzschTLel (entry persone) and the Reference GodefroyGdfC (vol. 10: 323) suggest the use of personne as a pronoun both in negative and positive polarity contexts with three isolated examples, although the GdfC’s example allows for a more plausible noun reading in the meaning of ‘cleric of a high rank’:

Indeed, interpreting it as ‘someone’ appears questionable; instead, ‘cleric of a high rank’ renders this sentence a metalinguistic reflection, consistent with Medieval scholars’ conception that form conveys meaning. Thus, the abbot’s “name” would be personne (i.e., ‘cleric’), given that phonetically the form ‘contains’ ‘pere’ (i.e., ‘father’). This aligns with the classic “father – children” metaphor for clergy and their flock. Finally, personne‘s predominant use in religious meaning in the 12^th century (see 4.1.), the example’s dating strongly supports a noun reading.

The Dictionnaire du Moyen français (DMF, entry personne) states an indefinite use of personne in Middle French, distinguishing two cases:

• in sentences with “negative aspect” – indefinite “anyone” (that is, NPI in weak NP);

• in association with ne – negative indefinite – “no one, nobody” (see DMF, entry personne) (that is, NI in strong NP).

Marchello-Nizia notes that, although personne appears in negative clauses from the early 14^th century and, in this use, replaces certain concurrent NIs by the 16^th century, its grammaticalization is not completed before the late 17^th century: until then bare personne in indefinite contexts can still show feminine agreement with a past participle (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 720). Déprez (Reference Déprez, Sleeman and Perridon2011), on whose paper the assumption of a similar evolution for rien, aucun and personne as NIs is largely based, cites two examples for the Medieval period. Regarding the grammaticalization dating—specifically, the point when personne could no longer have feminine agreement—she states it persists in Early Old French, illustrating this with a 11^th century example from the Vie de Saint Alexis (ibid: 273). The issue, however, is that Déprez’s example, borrowed from Nguyssaly (Reference Nguyssaly2003), neither comes from the original 11^th century Vie de Saint Alexis nor is in Early Old French:

The example’s language clearly reflects a later stage of French (e.g. the form nous), whereas the verse is an alexandrine, a metrical structure of 12 syllables that only emerges in the second half of the 12^th century. Nguyssaly’s paper, based on Paris & Pannier’s edition of the Vie de Saint Alexis (Reference Paris and Pannier1872), draws this example from a 14^th century version of the hagiography (ibid: 340, example on page 351). Moreover, as for the original 11^th century version of the Vie de Saint Alexis, neither Paris & Pannier’s edition, nor the BFM22 contain any instance of the item personne or any segment resembling (17).

Déprez further states that, while feminine agreement can occur until the late 17^th century, masculine agreement is attested as early as the 14^th century illustrating it by the following example:

However, searching 14^th century BFM22 texts for other instances of vivant reveals occurrences where it modifies feminine nouns that never exhibited signs of grammaticalization:

Although GGHF (Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 654; 854-855) notes analogical feminine agreement with -e for present participles and gerundives between the 14^th and 17^th centuries, it remains mostly invariable. Therefore, we argue that example 18 provides no evidence regarding the possibility of masculine (or feminine) agreement with personne.

Only TLel (entry persone) provides a valid–although single–14^th century example suggesting bare personne could show masculine agreement in NP context:

Finally, quantitative corpus analyses of personne are rare in existing research; whenever undertaken, the scarcity of data obtained is said to “disallow firm conclusions” regarding personne‘s evolution toward an NI (Larrivée & Kallel, Reference Larrivée and Kallel2020: 446).

5.1. Filtering nominal function

The automatic search by lemma personne produced the following quantitative results (Table 6):

Following the considerations in 4.1. and 4.2., we first automatically filtered the occurrences to retain only those that were formally bare (i.e., undetermined) and not immediately modified by an adjective. Occurrences modified by a prepositional phrase headed by de or a relative clause were included. Indeed, other NIs partly synonymous with personne, such as nul, appear in such constructions during this period, without their indefinite status being questioned.

Furthermore, Martineau & Déprez (Reference Martineau and Déprez2004: 41) note that in Pre-Classical and Classical French, when rien already was an established NPI, in the object function, it was more often modified by a prepositional phrase headed by de or by a relative clause than left bare. Finally, for personne it is still possible in ContF:

For the three corpora this stage has yielded the following results (Table 7):

Our next step involved separating the examples of clear nominal use from those potentially allowing an indefinite reading. For the historical period studied, zero article for a singular form does not automatically signify pronominal use. Indeed, common nouns could use zero article in several cases such as negative, hypothetical, or comparative sentences in Old French (Moignet, Reference Moignet2002[1976] : 105–109) – i.e., NP contexts – or for general reference that does not require specification (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2: 1548-1549) until the 15^th, or even 17^th century. In order to ascertain whether an unmodified and undetermined occurrence of personne qualified for an indefinite pronoun reading, we examined not only the morphosyntactic characteristics, but also other personne occurrences in comparable contexts.

As Table 7 shows, four occurrences of personne from BFM22’s 12^th-century texts meet the criteria of being undetermined and unmodified by an adjective. Two are easily filtered out: they appear in religious texts, specifically discussing the Trinity’s unity as part of the noun phrase unité de personne (‘unity of person’; see BFM22, Psautier de Cambridge, 1155-1160, p. 291 and BFM22, Li sermon saint Bernart, end of the 12^th century, p. 38). These occurrences clearly display personne’s nominal use, meaning ‘one of the hypostases of the Holy Trinity’.

Table 6. personne’s frequency in BFM22, MICLE1316 and ConDÉ1316

Table 7. Bare personne after filtering clear nominal examples

The remaining occurrences are from Vie de Saint Thomas Becket, a 12 ^th -century hagiography, and also exhibit clear nominal function, aligning with the second religious meaning: ‘cleric of high rank’. One example is particularly challenging for a ContF speaker:

While ContF linguistic competence suggests a negative indefinite reading of personne, such a reading would be anachronistic. As stated in 4.1., the general meaning ‘human being’ only becomes widespread in the 13^th century; in contrast, the 12^th century shows a considerable proportion of the meaning ‘cleric of high rank’, including lists of clerical titles, as in 10. Therefore, personne is employed here with this meaning and is paired with another clerical title, prelat (‘prelate’). This type of common noun pairings, either synonymous or complementary (as in 24 and 25 below), is a well-known stylistic figure in early French literature (see Buridant, Reference Buridant1980; Geylikman, Reference Geylikman2020); complementary pairings appear regularly in negative clauses, with both nouns undetermined and at least the second preceded by ne, as in 25:

The two 13^th-14^th century occurrences from the ConDÉ1316 were also excluded, being modified by the prepositional phrase “de sainte église” ‘of Holy Church’ and, therefore, meaning “cleric of high rank”.

Several examples have clearly a noun reading, appearing in a list of common nouns, such as lieu ‘place’, temps ‘time’, opinion ‘opinion’, (e.g. BFM22, Jean D’Antioche, De l’invention, p. INVI082-1), domicile ‘residence’ (e.g. ConDÉ1316, terrien, 1578).

Law texts of the corpora contain several bare, unmodified personne occurrences, notably three in Coutumes de Beauvaisis. Due to its discursive specificity, the legal genre can be considered a weak NP macro-context. Indeed, defining different legal procedures for various social interactions (e.g., the rules of inheritance), it extensively uses hypothetical clauses (e.g. if X happens/Y does Z, then A should do B). Therefore, a potential indefinite reading may be supposed for occurrences of the following type:

Were the indefinite reading accurate, personne would appear as a positive indefinite, meaning ‘someone’. However, a closer look at personne in the same text in comparable contexts reveals singular, undetermined personne can be modified by feminine-agreeing adjectives:

Moreover, some examples show personne in similar conditions, but in the plural (e.g., p. 501).

Furthermore, the Coutumes de Beauvaisis contains a total 133 personne occurrences. Except the three cited above, all are determined (by articles/indefinite determiners such as nulle or aucune), or modified by adjectives with feminine agreement, or plural. This suggests the three exceptions are nominal personne uses, meaning ‘human being’, typical of legal texts.

A 15^th-century Coutumier des Forêts contains six personne occurrences, either in contexts similar to 26, where personne is the antecedent of a relative clause, or in the noun phrase personne pour elle/lui:

Like the Coutumes de Beauvaisis, this text contains personne examples in similar contexts, but modified by adjectives with feminine agreement or in the plural. The MICLE1316 and the ConDÉ1316 corpora also feature occurrences of personne pour elle/lui in comparable contexts, that can appear with modification or in plural elsewhere; they are also to be interpreted as nominal uses characteristic of legal coutumes discourse.

While our previous analysis focused on a potential positive indefinite reading, the Registre Criminel du Châtelet, a large collection of court proceedings, contains three occurrences a ContF speaker might interpret as NI uses, as they appear in strong polarity contexts, as 29:

Nevertheless, a closer look at the text segment shows these examples are directly preceded by a similar occurrence with personne modified by the indefinite determiner quelconque:

These occurrences describe two cases of theft by the same person and come from the same testimony; quelconque prevents us from concluding firmly to a pronoun reading of personne in the bare example. Still, the optional nature of the determiner in (29) suggests that it was no longer perceived as mandatory for the expression of indefinite value in strong polarity contexts by the late 14^th century.

After filtering out nominal uses of personne, we have obtained a reduced number of potentially indefinite occurrences (Table 8):

Those examples were further annotated for following parameters: context nature with respect to NP, preverbal/postverbal position, modifier type (if any), text segment’s place on the immediacy-distance continuum, and the “represented speech” factor.

5.2. Negative Polarity (NP)

The distribution of the potential indefinite examples with respect to NP is summarized in Table 9.

Table 8. Potentially indefinite use of personne

Table 9. Potentially indefinite occurrences of personne with respect to NP

These results require several comments. First, aligning with Marchello-Nizia’s dating (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 720), no potentially indefinite occurrences are found before the 14^th century. Second, positive and neutral polarity examples are absent. Third—contrary to suggestions of a similar change with aucun and rien (Martineau & Déprez, Reference Martineau and Déprez2004; Déprez, Reference Déprez, Sleeman and Perridon2011; Hansen, Reference Hansen, Mosegaard Hansen and Visconti2014), strong NP contexts are largely predominant from the beginning. This is even more pronounced if without-type is considered strong polarity (de Swart, Reference de Swart2010: 213ff). Incidentally, the without-type attestations only occur in the 15^th century, thus being chronologically secondary in comparison to proper strong polarity.

Examples 31-35 illustrate these NP types. While we cannot yet clearly state the most plausible reading—nominal ‘person’, NPI or NI—we deliberately use the earliest possibility while glossing (noun person), incorporating broad translation options.

1. 1. Weak NP

2. • interrogative:

• conditional:

• comparative:

2. Privative negation (without-type)

3. Strong NP:

5.3. Morpho-syntactic patterns

As stated in 5.1., we allowed several types of modifiers while separating nominal from potentially indefinite uses. In most of the the cases (18 of 20 modified examplesFootnote ¹⁶ ), personne is thus modified by a hypothetical relative clause, as in (31), or by a prepositional phrase headed by de:

Tables 10-11 summarize quantitative data with respect to preverbal/postverbal position and the presence/absence of modifiers, uniting the data from the three corpora:

Table 10. Potentially indefinite occurrences of personne: position regarding the finite verb

Table 11. Potentially indefinite occurrences of personne: modification

Without yielding comprehensive patterns combining all variables, these results warranted an observation: generally speaking, potentially indefinite occurrences of personne tend to appear postverbally and without modification. Overall, strong NP contexts are predominant throughout the corpus.

Therefore, we focused on occurrences exhibiting a strong NP context, no modification, and postverbal position. As subjects could still appear postverbally during this period, we annotated the dataset for syntactic function, yielding 47 object examples out of 81 total postverbal occurrences. This final step enabled a more precise identification of the personne’s predominant pattern for the potentially indefinite use. Illustrated by (35) and representing almost half of all occurrences (45 out of 96), this predominant pattern combines

• strong polarity

• object function in postverbal position

• no modification.

5.4. Immediacy-distance continuum and represented speech

Table 12 summarizes the distribution of the potentially indefinite examples per text-type in BFM22.

Table 12. Distribution of the potentially indefinite occurrences of personne in BFM22.

The primary distinction from nominal use is the preponderance of fiction in the 14^th-century examples. In the 15^th century, the historical genre’s significance stems entirely from Philippe de Commynes’ Mémoires, written in the late 15^th-early 16^th centuries (8 on a total of 9 occurrences in historical genre). Commynes had not received a literary education that would have imposed a Latin influence on his writing, and his texts are known for their “orality” and the “modern face” of their sentence structure (see Blanchard, Reference Blanchard, Hahn, Melville and Röcke2006).

While text-type distribution alone yields no firm conclusions, a closer look reveals the high frequency of the studied use in fiction is primarily due to the represented speech segments—both direct and indirect speech: 44 occurrences (70%) on a total 63 (100%). Moreover, comparing numerical data for clearly nominal vs. potentially pronominal functions regarding the direct speech vs. narration dichotomyFootnote ¹⁷ in BFM22 for the 14^th and 15^th centuries reveals a statistically significant difference in distribution:

14^th century, χ²(1, N=639)= 38.5618, p-value is < 0.00001 (see Table 13).

Table 13. Distribution of the two uses of personne regarding the direct discourse vs narration parameter in the 14 ^th century subcorpus of BFM22

15^th century, χ²(1, N=729)= 27.4664, is < 0.00001 (see Table 14).

Table 14. Distribution of the two uses of personne regarding the direct discourse vs narration parameter in the 15 ^th century subcorpus of BFM22

Table 15. Distribution of the potentially indefinite occurrences of personne per text in MICLE1316

Table 16. Distribution of the potentially indefinite occurrences of personne regarding the immediacy-distance continuum

Those results demonstrate that, whereas the nominal function occurs massively in narration, the potentially pronominal use is more likely to appear in represented speech.

Similarly, the MICLE1316 corpus allows for an interesting observation. It contains two types of legal production: court proceedings that record the words of the accusation and the accused (henceforth, trials) and sets of laws containing procedural instructions (henceforth, legal). With respect to the text type, we obtain the following distribution of potentially indefinite occurrences (see Table 15):

Combining these results with the distance-proximal ConDÉ1316 corpus, with only five potentially indefinite personne instances in a 1578 text, clearly demonstrates that from the 14^th century to late 16^th century, personne as a potential NI occurs only in texts representing speech—namely, trials.

Broadly considering all speech representation (direct/indirect speech, trials) as immediacy-proximal, and adding the 8 occurrences from the Commynes’ Mémoires marked by “orality” and using innovative constructions, a clear pattern emerges for personne: potential NI use is largely associated with the continuum’s immediacy pole (Table 16):

The results can be summarized as follows:

• personne’s potentially NI occurrences emerge in the 14^th century, remaining rare but increasing in frequency throughout the period studied;

• No attestations of the potentially indefinite use exist in positive contexts. In NP, strong contexts are preponderant from the beginning;

• For strong NP contexts, the predominant pattern is a postverbal position without modification;

• Finally, the potentially indefinite use appears as immediacy-proximal.

6.1. Indefinite Indeed?

Summarizing personne’s indefinite use dating, it seems clear that this development could not precede the 13^th century. At the same time, the increased use of personne in the general meaning of ‘human being’ throughout the 13^th century undoubtedly paved the way for its later NI evolution. One might wonder whether personne’s extensive use as a general human noun in weak polarity contexts within coutumes initiated subsequent grammaticalization. However, this appears unlikely, given the genre’s distance-proximal discourse under considerable Latin influence. Conversely, no current evidence suggests widespread personne use as a general human noun in spoken language that would trigger grammaticalization.

In any case, we observe the following chronology:

• 12^th century: personne-noun begins appearing in distance-proximal religious texts;

• 13^th century: personne-noun is used extensively in distance-proximal law texts and spreads throughout vernacular literature appearing at various points of the immediacy-distance continuum;

• 14^th century: personne exhibits signs of an indefinite use.

Moreover, the distinction between nominal and potentially indefinite use correlates with the distribution of occurrences along immediacy-distance continuum. Nominal use, remaining under strong Latin influence, tends to appear in distance-proximal texts, whereas the potentially NI use—or the innovative variant—is clearly associated with communicative immediacy.

But is personne’s indefinite value in Medieval and early Pre-Classical French real, or are we biased by our ContF competence? When translating the examples above, we chose to make no claims about whether personne should be considered a noun, an NPI or an NI. Given the difficulty of distinguishing the semantic threshold after which a bare, unmodified noun acquires a pronominal interpretation, one way is to investigate other elements in comparable contexts during the period studied.

Searching for such elements, we focused on items synonymous with a potential indefinite reading of personne—that is, referring to human beings in weak/strong NP contexts—and whose indefinite interpretation is undisputed: nul and aucun. Footnote ¹⁸ This yielded comparable occurrences in both NP contexts, suggesting a parallelism between personne and clearly indefinite forms.

We also investigated whether other general human nouns or their contextual synonyms were used in similar conditions. The two most salient examples of general human nouns in MF are homme and gent (gens). Having a collective meaning, gent was directly excluded. As stated earlier, the grammaticalization of on, coming from the same etymon as homme, had begun long before the 14^th century (Marchello-Nizia, Reference Marchello-Nizia2009, ch. 3, §25). By the Middle French stage, is it not easy to find bare homme or on in uses similar to personne. Instead, comparable strong NP reveal the noun phrase nul homme ‘no man’:

While Old French exhibited grammaticalization potential for nul homme, in Middle French, its frequency declined, and it failed to become an NI in French language history (Geylikman, Reference Geylikman2025). The 14^th century thus saw nul homme’s decline, alongside personne’s rise.

According to Hansen (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2: 1692), âme ‘soul’ was another personne’s concurrent that didn’t prevail. In BFM22, only a few occurrences of âme as an NPI appear during the period studied, with just one in the 14^th century:

Finally, nulle personne, the last competitor of personne’s potential indefinite use, warrants a comment. Given its occurrence only in strong NP contexts, we compare its use with that of personne within this type of context only:

As Table 17 shows, while nulle personne is attested earlier, bare personne have prevailed by the 15^th century. The 14^th century is most interesting, as the frequency is comparable.

Table 17. nulle personne and personne in strong NP contexts in BFM22

A fascinating example of the nulle personne vs. personne competition appears in two 14^th-century prose romances: Bérinus (1370, ms. early 15^th, 104497 tokens) and Mélusine (1392, ms. early 15^th, 124938 tokens). Despite their similarities in date of composition, date of manuscript and dimensions, Bérinus exclusively uses nulle personne (7 occurrences), while Mélusine opts solely for bare personne in comparable contexts (6 occurrences).

It appears the author/scribe of Mélusine considers the indefinite use already acceptable, while Bérinus’s author/scribe leans toward its noun interpretation, or prefers securing the expression with a clearly indefinite item, nul.

According to Marchello-Nizia, in the 14^th century bare personne in strong polarity contexts still agrees in feminine with past participles, and one have to wait until the 17^th century for Vaugelas to recommend a masculine agreement (GGHF, Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 1: 720). However, Marchello-Nizia’s 14^th century example comes from a text by Jean Corbechon, absent from the main corpus on which GGHF is based, suggesting such examples are rare. Déprez agrees with Marchello-Nizia’s view, acknowledging variation since Middle French, although, as shown in 4.3., the illustrative example appears problematic.

Our corpora provide no context for an anaphorical reprise or a past participle agreement, which aligns with the preponderant pattern: object function in postverbal position. Therefore, without invaliding Déprez’s dating (2011: 273), we must also consider that, at least in the 14^th–16^th centuries, the examples for bare personne’s clear masculine/feminine agreement in NP contexts are quantitatively scarce and appear qualitatively marginal.

Overall, acknowledging that we are dealing with an ongoing grammaticalization, and that variation is possible throughout the studied period, the assumption that personne can develop indefinite value since Middle French appears legitimate.

6.2. Just NPI or NI Already?

Our corpora provide no occurrences where personne carries the negative meaning alone. Therefore, since ne can still function as the sole negator, deciding whether a given occurrence classifies as NI or merely an NPI is particularly challenging given the inaccessibility of native linguistic competence.

Indeed, while Hansen (Reference Hansen, Mosegaard Hansen and Visconti2014, chapter in GGHF Reference Marchello-Nizia, Combettes, Prévost and Scheer2020, vol. 2: 1688-1691) considers N-word use for the entire indefinite macro-construction from the beginning, Martineau & Déprez (Reference Martineau and Déprez2004) argue that, in Pre-Classical and Classical French rien and aucun are inherently NPIs, not yet NIs.

Comparing personne with Martineau & Déprez (Reference Martineau and Déprez2004) findings on rien and aucun yields interesting results. They argue that in Pre-Classical and Classical French, in strong polarity contexts, postverbal aucun and rien in object function are typically modified by a prepositional phrase or a relative clause (Martineau & Déprez, Reference Martineau and Déprez2004: 41), which they analyse as nominal traits. From this perspective, personne appears closer to a quantifier than rien and aucun from the beginning, as the postverbal bare unmodified use is preponderant for strong negative contexts. Moreover, unlike rien and aucun, for which this use appears regular, our corpora contain only one example of a three-element negation ne…point + indefinite in late 16^th century text (MICLE1316, Guernesey, 1563). For Martineau & Déprez, this is a sign of an inherent NPI value of aucun and rien; conversely, the rareness of such co-occurrence suggests an NI value for personne from MF.

As Section 5 states, indefinite personne exhibits two major patterns: a preponderance of strong negative contexts from the beginning, and, within strong polarity, a majority of unmodified postverbal examples. The first pattern invalidates assimilating personne’s evolution to that of rien and aucun, which typically showed a majority of weak polarity occurrences as an intermediate stage between positive indefinite and NI (Martineau & Déprez, Reference Martineau and Déprez2004; Prévost & Schnedecker, Reference Prévost and Schnedecker2004). Hansen (Reference Hansen, Mosegaard Hansen and Visconti2014) demonstrated aspectual/temporal adverbs grammaticalized directly as NIs in MF and only developed weak polarity uses afterward. Conversely, our data suggests personne had its own path toward NI, as weak and strong polarity contexts appear simultaneously.

Lacking native competence for ancient stages of languages, we can still search for metalinguistic evidence for a given period. Thus, Robert Estienne’s Dictionnaire François Latin Footnote ¹⁹ a French-Latin dictionary dated to 1549 offers various examples of personne in weak and strong NP, occupying half of the entry (see Table 18).

Table 18

The rather idiosyncratic translations by Estienne represented in Table 18, while not glosses in the framework of modern linguistics, are valuable insofar as they provide metalinguistic insight based on author’s linguistic intuition.

Thus, all examples are clearly understood as containing at least an indefinite and a negative element; strong polarity contexts predominating over weak ones. They can be grouped into several types:

• (b): nominal function, the NI value being encompassed by nulle;

• (c): a hesitation between nominal (Nemo homo) and indefinite use (Nemo) of personne;

• (d) to (e): indefinite value encompassed by personne, allowing an NPI or an NI reading;

• (f): either an NPI reading, jamais having an already established negative value by the mid-16^th century (Hansen, Reference Hansen, Mosegaard Hansen and Visconti2014), or an NI endorsing NC reading;

Example (a) remains the most interesting. Containing neither ne nor any NI, it is still translated by a negative sentence. Had Estienne perceived it as NPI ‘anyone’ or a positive indefinite ‘someone’, the translation would rather be Quis hic est? (gloss: anyone_here_be. prs.3sg ) using the NPI quis. Although we may allow for some translational liberty, we argue it reflects the strong association of personne with negativity, possibly even an inherently negative meaning, according to the author’s linguistic feeling. If this interpretation holds, the other occurrences appear as NIs rather than NPIs.

Overall, Estienne’s dictionary suggests the significance of the association of bare personne with strong NP contexts by the mid-16^th century. It also allows the hypothesis that personne’s indefinite—and possibly negative—value was already widespread enough to be thus extensively included in a dictionary.

Incidentally, personne in these occurrences—whether we consider it NPI or NI—is in most cases translated by the Latin NI nemo. This choice reminds us of the indefinite quantifier cycle suggested by van der Auwera et al. (Reference van der Auwera, Krasnoukhova, Vossen, Veselinova and Hamari2022) and van der Auwera (Reference van der Auwera2024) and illustrated precisely by Latin nemo and French personne (van der Auwera et al., Reference van der Auwera, Krasnoukhova, Vossen, Veselinova and Hamari2022: 615).