1 Introduction

The proposal of a Romance grammaticalization cline (RGC, also referred to as the pace, rhythm or gradualness of grammaticalization in the Romance languages) has given rise to a vast body of research since the publication of pioneering work by Lamiroy (1999). The findings of these studies have recently been summarized in Lamiroy and De Mulder (2011), Carlier et al. (2012a), and De Mulder and Lamiroy (2012).

The RGC hypothesis suggests that languages belonging to the same linguistic family, such as the Romance languages, grammaticalize at different speeds. For the majority of grammaticalization phenomena, such as word order, the encoding of motion events, adversative connectives, the use of the subjunctive, the evolution of the prepositional system, etc. (see Carlier et al. 2012b; Fagard and Mardale 2012; Giacalone Ramat and Mauri 2012; Iacobini 2012; Lahousse and Lamiroy 2012), Romance languages can be situated along a scale, where French is ahead of Italian and Italian is ahead of Spanish. These findings are visualized in (1) (adapted from Lamiroy and De Mulder 2011; Carlier et al. 2012a: passim; De Mulder and Lamiroy 2012: 200):

(1)

more grammaticalized		less grammaticalized
French	Italian	Spanish	(Latin)

To date, however, less attention has been paid to non-national Romance languages, although they are numerous and, in some cases, extremely vital across the Romance territory. In the continental European tradition of sociolinguistics, independent languages that are “spoken in a given place or region in concomitance with a more prestigious superimposed […] standard language” (Berruto 2010: 230), such as Piedmontese, Sicilian or Venetan (vis-à-vis Italian), are usually referred to as dialects (dialectes, Dialekte, dialetti). For English-speaking and non-European linguists, on the other hand, the term dialects does not refer to separate languages, but rather to (sub)varieties of a single language (see, e. g., Trudgill 2002: 165). For the sake of clarity, I will use the label (regional or minority) languages throughout this paper to refer to those linguistic systems that are different from the official language(s) of a given nation state, and that are traditionally used by a group of nationals smaller than the rest of that nation state’s population. ^[1] Varieties of the same language will be referred to as dialects. Dialects may be geographical or social (see Trudgill 1994: 2) or situational (also called styles, see Trudgill 1994: 10). By way of example, American English is a geographical dialect of English, Black English is a social dialect of English, and colloquial French is a situational dialect of French. Varieties of dialects will be indicated with the term sub-dialects.

This article will focus on the pace of grammaticalization in (a number of dialects of) the Piedmontese language. Piedmontese (or Piemontese, or, as its speakers call it, Piemontèis) is a regional language spoken in the Piedmont region of northwestern Italy (for an overview of Piedmontese, see Telmon 1988; Parry 1997; Clivio 2002). ^[2] The region’s chief town is Turin, which was also the capital of the entire Kingdom of Italy from the Unification of Italy in 1861 until 1865. Until the second half of the 20th century, Piedmontese was the mother tongue of nearly all Piedmontese people (except for those living in the Provençal, Walser and Franco-Provençal valleys and in the environs of Novara), who often also had some command of French, Italian, and other local non-Piedmontese varieties (see, e. g., Telmon 2001: 37–38, Marazzini 2012). Nowadays, Piedmontese is probably spoken by 700,000 people in Piedmont and along the borders of that region (Regis 2012a: 93). However, in everyday life, virtually all Piedmontese speakers also, or predominantly, use Italian now, which was previously, at least until the 1950s, limited to cultural, literary, and scientific domains. Despite the fact that Piedmontese has a standard script, used, for example, in the regional edition of Wikipedia, a rich literature, and a grammatical tradition that dates back to the mid/late-18th century, it has never been an official language of any nation state, nor has it ever been taught in schools as part of compulsory education. Recent efforts to revitalize the language have had little success. It therefore still lacks a proper standard variety. Nonetheless, during the 18th century, Piedmontese developed a koiné, mainly corresponding to the Turinese variety (Clivio 2002: 151–152; Regis 2012b: 10–15). Moreover, its geographical dialects, such as High and Low Piedmontese, and their sub-dialects, such as Asti Piedmontese, Biella Piedmontese, Monferrato Piedmontese, etc. (see Telmon 1988, http://glottolog.org/resource/languoid/id/piem1238, http://www.ethnologue.com/language/pms), are also identifiable and of interest: some of them will be taken into account in the course of this paper. The map in Figure 1 (courtesy of Francesco Rubat Borel) shows the geographical extension of Piedmont, the regional and/or minority languages spoken there and in the neighboring areas, along with the Piedmontese names of towns and some sub-dialect areas. The latter are in small caps. The names of the regions bordering on Piedmont are in capital letters. ^[3]

Figure 1:

Map of the regional or minority languages spoken in Piedmont and neighboring areas.

In the following sections, I compare Piedmontese to the geographically closest languages conventionally discussed in the literature on the RGC: French and Italian. Specifically, I examine the speed of grammaticalization in Piedmontese in four morphosyntactic domains which have already been investigated for national languages in the literature on the RGC, namely demonstrative adjectives (see Section 2), negation (Section 3), perfective auxiliaries (Section 4), and plural indefinite articles (Section 5). The analysis relies on data from my own fieldwork and from previous studies on Piedmontese, as well as from a selected corpus of Piedmontese texts ranging from 1321 to the present day.

The data shows that the Piedmontese koiné or other Piedmontese varieties are frequently to be situated at the same stage as, or are even slightly ahead of, French on the RGC. These findings complement De Mulder and Lamiroy’s (2012: 219–221) view about the role of language-external factors in grammaticalization processes: the view proposed here (Section 6) is that contact with other languages and social ties among speakers are more important in favoring grammaticalization than early urbanization and the size of the speaker community. Finally, I also discuss the relationship between literacy in (official) languages and grammaticalization.

2 Demonstrative adjectives

Demonstratives are deictic or anaphoric elements used “to orient the hearer outside of discourse in the surrounding situation [and] to organize the information flow in the ongoing discourse” (Diessel 1999: 2). Classical Latin had a four-term system for demonstrative pronouns and adjectives, made up of proximal hic, medial iste and distal ille, as well as anaphoric is. In this section, I focus on the evolution of the Piedmontese adjectival system and then examine how it compares to developments in the French and Italian systems. My own fieldwork, which partially confirms Lombardi Vallauri’s work (1986), suggests that present-day Piedmontese features a system with three demonstrative adjectives, which emerged via a number of steps that I now summarize. The first attestations of demonstrative adjectives in full-fledged Piedmontese are found in the 14th-century Statuti della Compagnia di San Giorgio a Chieri (Salvioni 1886: 347–350). In the Statuti, the Latin system had already been reduced to two terms: proximal cost and distal col. These derive from the Latin iste and ille, respectively, prefixed by the exclamative eccum < ecce: (ec)cuist(e); (ec)cuill(e). ^[4]

A third form appeared in 17th-century Canzoni (Clivio 1974; see (3a)), namely st, a phonetic reduction of cost and/or of its allophonic variant cas(t). St carried the same meaning as cost, being a proximal demonstrative, while col, with its variant cal, remained the distal adjective: ^[6]

Throughout the 18th century, st became more widespread in the Piedmontese koiné, at the expense of cost. As an adjective, cost was rare in 18th- and 19th-century texts, mainly featuring in plays that mimicked the sermo popularis (4a–b). During the 20th century, another demonstrative form entered Piedmontese grammar: (ë)s (feminine sa, 4c). (Ë)s, derived from the Latin ipse, is quite clearly a first-person demonstrative, although it is used inclusively to indicate all the possible referents in a given context (Lombardi Vallauri 1986: 57). The co-existence of three forms with almost the same proximal meaning probably explains why cost specialized, reinforced by the adverb sì ‘here’, as a demonstrative pronoun. Today, Piedmontese speakers rarely use cost as an adjective (Bonato 2003–2004: 190–217). Col is still the only distal adjective.

Table 1 sums up the system of demonstrative adjectives in use in contemporary Piedmontese. It is a three-term and two-distance system:

According to Lombardi Vallauri (1986), (ë)s may have entered the koiné grammar from peripheral, dialectal varieties. Indeed, (ë)s is attested, in the form is, in non-koiné (i. e., non-Turinese/non-peri-Turinese) texts from the 16th century (e. g., in Alione 1521). In peripheral varieties, the demonstrative adjectival system has since been heavily reduced: several dialects of Piedmontese today feature only one demonstrative adjective, e. g., (ë)s/s(ë) in Cairo (Parry 1991; 2005: 150–156), Biella (Vv.Aa 2000), and Asti (Musso 2003: 68–69), ca in Pinerolo and Avigliana (Calosso 1973: 142–149), with the postposition of deictic adverbs to represent different degrees of (psychological or spatial) distance from the speaker.

Furthermore, from a diachronic point of view, the grammatical changes in this domain seem to have first taken place in the peripheral varieties, and only subsequently in the koiné. While 17th-century Turinese songs contain very few instances of st, the Historia della guerra del Monferrato (1613), written in a Monferrato dialect, already featured the two competing proximal forms cost and st/s, as in (5). In addition, while De’ Conti’s (1798)Gerusalemme Liberata in Monferrino featured a strictly two-distance, two-term system (st vs col), similar to that of the contemporary Turinese variety, the third term cost is still attested in the 19th-century koiné.

Table 2 summarizes the evolution of the demonstrative adjective system in the Piedmontese koiné with respect to the peripheral varieties: ^[7]

My data are mainly drawn from written texts, which are less accurate in the representation of phonetics and more conservative by nature. It is thus possible that (ë)s became undistinguishable from st via phonetic reduction, and that these two forms have ultimately merged into one demonstrative, in the koiné as well as in the peripheral varieties. If this is indeed what happened, then the contemporary koiné system would also be a two-term system, with one proximal adjective st (and its phonotactic variants) and one distal adjective, col. French has developed a demonstrative adjective system that is very similar to those of the peripheral Piedmontese varieties: it “has reduced the paradigm of the demonstrative determiners maximally, so that it no longer marks differences related to distance” (De Mulder and Lamiroy 2012: 217; Guillot and Carlier 2015). Contemporary French has distinct adjectives and pronouns (ce vs celui/ceux/celle(s)). Note that the only adjectival term in the current French system is ce [sə], now a deictically neutral demonstrative, phonetically identical to the forms attested in Cairo, Biella, and Asti. Just as in Cairo Piedmontese, Biella Piedmontese, and Asti Piedmontese, French ce is reinforced by adverbial suffixes (-ci and -là placed after the noun following the demonstrative), but these “are no longer used systematically” (De Mulder and Lamiroy 2012: 217), so that, for example, -là does not imply that the referent is far away (6a).

Contemporary standard Italian, on the other hand, has a two-term demonstrative system (with both terms used as adjectives and as pronouns), expressing person-related features in the following way: questo, occasionally shortened to (’)sto in informal speech and in allegro forms, i. e., phonetically reduced forms used in rapid speech (see Matthews 2007: s.v.), means ‘close to the speaker’; quello means ‘away from the speaker’ (6b–c). Codesto, a second-person oriented demonstrative that means ‘away from the speaker but close to the hearer’ may still be alive in Tuscan Regional Italian, but has disappeared from everyday communication in standard Italian (Sabatini 1985: 158; Berruto 1987: 78).

In sum, with regard to demonstrative adjectives: Italian has a two-term person-related system (questo/quello). The Piedmontese koiné has a two/three-term person-related system with two distances: (ë)s/st vs. col. As suggested earlier, the first two of these forms might possibly be regarded as allomorphs of a single term. Piedmontese dialects have one neutral demonstrative adjective ((ë)s (+N-sì/lì/là)). ^[8] French has one neutral adjective (ce (+N-ci/là)). French and Piedmontese dialects thus appear to be more grammaticalized than the Piedmontese koiné and Italian, because they have reduced the paradigm to the maximum extent, which is typical of grammaticalization processes (Lehmann 2002: 118). The Piedmontese koiné appears to be on a par with Italian, although it should be borne in mind that the former is a grammaticalization cycle ahead of the latter. This assessment does not take into account the Italian (’)sto (reduced form of questo, corresponding to the Piedmontese st, reduction of cost), although it is frequently used in everyday communication. In practice, the development of a single dedicated proximal adjective form, i. e., (’)sto, which is different from the pronoun form, i. e., questo, is still in progress in Italian, while it is virtually complete for the Piedmontese koiné (s(t) vs. cost(-sì)). Thus, Italian seems closer to Latin than the Piedmontese koiné, because although both daughter languages have reduced the system to two terms, Italian still does not distinguish between adjectives and pronouns. The RGC with respect to the morphological microsystem of demonstrative adjectives is represented in (7):

(7)

more grammaticalized	less grammaticalized
French Piedmontese dialects	Piedmontese koiné	Italian	(Latin)

3 Negation

In this section, I examine how negative forms have evolved in Piedmontese, and their pace of development in relation to negation in French and Italian (Mosegaard Hansen and Visconti 2012). My analysis is confined to standard negation. ^[9] In a number of languages and especially in the Romance area, standard negation follows a cyclical pattern known as Jespersen’s cycle:

The original negative adverb is first weakened, then found insufficient and therefore strengthened, generally through some additional word, and this in its turn may be felt as the negative proper and may then in course of time be subject to the same development as the original word. (Jespersen 1917: 4)

The development of the cycle has been studied in detail by Van Der Auwera (2010), and, with a special focus on Piedmontese, by Parry (1996, 2013), who proposed a five-stage model (Parry 2013) that was complemented by a sixth stage argued for in Mosegaard Hansen and Visconti (2012: 455) (see Table 3):

In early Piedmontese texts, the descendant of the Latin preverbal non (=[neg1 V]) is the most frequent negator, exhibiting conditioned allophony (non vs no). Nonetheless, some instances of discontinuous negation (=[neg1 V neg2]) already appeared in the Sermons Subalpins, while, alongside [neg1 V] and [neg1 V neg 2] sentences, a single occurrence of post-verbal negation (=[V neg3]) is attested in a song from Carmagnola as early as the 14th century (8b).

Stages 2 and 3 in the development of negation (Table 3) were reached in the 16th century, as attested in the work of Alione (9a), while 17th-century Turinese songs and plays already exhibited Stage 4, with n … nent/pa (9b) as the most usual form of negation and the postverbal nen (9c) rising to more than 25 % of all negations in Tana’s ‘L Cont Piolet (a play written during the 17th century but not published until 1784). ^[10] Analysis of Bersezio’s (1863)Le miserie ’d Monssù Travet (see 9d) shows that Stage 5, [V neg3], had been reached in the Piedmontese koiné before the 19th century.

Bordering varieties, such as Cairese (i. e., Cairo Piedmontese, 10a–b) and Roccavignale Piedmontese (10c), still display some degree of variation today in that they can exhibit, along with [V neg3], preverbal n’ and discontinuous n’ … nen (see Parry 2014: 221–222; Duberti and Regis 2014: 104–105). Other conservative varieties spoken along the political regional borders, however, consistently use the [V neg3] form (see 10d for an example of so-called Kje spoken in Roccaforte Mondovì). The data suggest that the dialects linguistically oriented towards Ligurian have tended to maintain a [neg1 V] or [neg 1 V neg2] strategy, because of the contact with Ligurian, a [neg1 V] language, much more so than the dialects linguistically oriented towards the koiné.

The diachronic development of French negation has been investigated, among others, by Ashby (1981), Mosegaard Hansen (2009), and Mosegaard Hansen and Visconti (2012): “Colloquial spoken French is currently very firmly Stage 4” (Mosegaard Hansen and Visconti 2012: 455). The [neg1 V neg2] construction ne … pas had grammaticalized as the standard negation by the end of the 17th century (Mosegaard Hansen 2009: 229), but French is now moving towards Stage 5, given that pas is in the course of grammaticalizing as the only negator. ^[11]

Standard Italian, on the other hand, is still at Stage 1 in Table 3, as the Latin preverbal non is still the standard negation in non-emphatic sentences (12a). Northwestern Italian dialects (i. e., socio-geographical and situational varieties of Italian) are developing [neg1 V neg2] and [V neg3] strategies (12b–c), possibly under the influence of regional languages spoken in that part of Italy, such as Lombard or Emilian:

In light of the data discussed in this section, the position of the three languages with respect to standard negation may be summarized as follows ^[12]:

(13)

more grammaticalized	less grammaticalized
Piedmontese koiné	(colloquial) French (Northern-)Italian
Piedmontese dialects	(Latin)

Piedmontese is ahead of all other languages in that it has been consistently at Stage 5 since the 19th century, while French is currently still at Stage 4. Standard French, in turn, is ahead of Italian, whose standard variety has not left the Latin stage yet. Nevertheless, northwestern spoken Italian appears to be moving towards Stage 2 or 3. ^[13]

While it is impossible to predict whether Piedmontese will display Stage 6 constructions with finite verbs in the future, contemporary varieties of Piedmontese already negate infinitives by means of a preverbal nen (14a). The same holds for French infinitives (15a). In relation to gerundives, the Piedmontese edition of Wikipedia shows that both pre-verbal and post-verbal negation may be used, although the latter strategy is strongly preferred (14b–c). French is still a [neg1 V neg2] language as far as the negation of the gerundives is concerned (15b).

These data show that Piedmontese is also ahead of French as far as the negation of non-finite verb forms is concerned.

4 Perfective auxiliaries

Perfective auxiliaries, i. e., the light verbs with which present perfect and other compound tenses are formed, go back in Romance to the Latin verbs esse ‘to be’ and habere ‘to have’, which gradually lost their referential meaning. As is well known, Romance languages usually display split intransitivity: while active-voice transitive verbs consistently use have as a perfective auxiliary, intransitives are divided into two subclasses: unergatives, which preferably use have, and unaccusatives, which preferably use be. ^[14] Nonetheless, recent developments in the study of this topic suggest that the selection of have or be is in fact a gradual phenomenon, with core unaccusatives and core unergatives displaying less variation than other verbs of the same class (Sorace 2000). Furthermore, the choice of have or be for an individual verb may change over time. “This is particularly evident in the Romance languages, which have been undergoing a diachronic change leading to the progressive replacement of be by have” (Sorace 2000: 860). Sorace (2000), Cennamo (2001), Cennamo and Sorace (2007), and Cennamo (2008) showed that cross-linguistically the spread of one perfective auxiliary at the expense of the other follows a lexical-semantic path. Accordingly, verbs can be divided into subclasses (e. g., verbs indicating a change of state/location, verbs indicating a state, verbs indicating motional/non-motional processes, etc.), and have has been shown to spread over time from the “core-unergative” verb class (i. e., controlled non-motional processes) to “less core-unergative”, or “peripheral”, verb classes (i. e., motional processes, change of state/location, etc.). “Change proceeds from the center towards the periphery in the case of changes involving the constitution of a category with a radial structure [like split intransitivity, EM], but follows a reverse path (from the periphery towards the center) in the case of the (partial) cancellation of a category with a radial structure” (Cennamo 2008: 137; see also; Lazzeroni 2005: 18).

The contemporary Piedmontese koiné is a split-intransitivity language, its perfective auxiliaries being avej ‘to have’ and esse ‘to be’ (see Burzio 1986; Villata 1997: 168–170):

Nonetheless, in varieties such as Biella Piedmontese (the dialect spoken in the city of Biella, in north-eastern Piedmont), as well as in Southern Piedmontese dialects, the aforementioned replacement of esse by avej is ongoing, and indeed, in the areas around Biella and Cuneo, change has already proceeded to the extent that split intransitivity has disappeared. In Vallanzenghese and Cigliese, for instance, avej is the only remaining auxiliary for all verbs (Cerrone and Miola 2011: 196–198).

In contemporary French, only about a dozen verbs use être ‘to be’ as a perfective auxiliary (Rowlett 2007: 40). These verbs belong to two verbal subclasses indicating a definite/indefinite change of location (e. g., arriver ‘to arrive’) or a definite/indefinite change of state (devenir ‘to become’; Sorace 2000: 863–867). All other French verbs use avoir ‘to have’.

Standard Italian, on the other hand, uses avere ‘to have’ with transitive and intransitive verbs indicating uncontrolled (squillare ‘to ring’) and controlled processes (lavorare ‘to work’), although in the case of uncontrolled processes essere ‘to be’ may be used instead of avere without any variation in meaning (17a–b).

Interestingly enough, in the small Piedmontese towns of Coggiola (see 18) and Niella Tanaro (Nicola Duberti, p.c.), only be is used to form the perfective compound tenses (even with transitive verbs, see 18):

The choice of perfective auxiliaries in the languages and varieties discussed so far is schematized in Table 4. ^[15]

The development of auxiliary selection is in fact a counterexample to the RGC hypothesis (Lindschouw 2010: 186), because “Spanish reduced its auxiliary alternation, thus showing more paradigmaticity (Lehmann 2002), and hence a higher degree of grammaticalization […] than French and Italian” (De Mulder and Lamiroy 2012: 201), whereas in other domains, French and Italian are consistently more grammaticalized than Spanish. It is still not clear why Spanish and other Ibero-Romance languages have grammaticalized faster in this domain (but see Rosemeyer 2014), but they are beyond the scope of the current paper. However, the same degree of grammaticalization as in Spanish is displayed by the peripheral Piedmontese dialects Vallenzenghese and Cigliese, which are thus ahead of French, the Piedmontese koiné, and Italian. With regards to the Coggiola and Niella Piedmontese dialects, the pace of grammaticalization of the perfective auxiliaries may even have been faster than that of the other Piedmontese dialects. Following Cennamo (2008), we might speculate that all verb classes first adopted have (as in the Biella Piedmontese dialects), with be subsequently once more superseding have. In other words, the Coggiola and Niella varieties may have completed a double auxiliary cycle in the same timespan that it has taken the other varieties to complete only one.

The diagram in (19) represents the degree of grammaticalization of perfective auxiliaries in French, Italian, and Piedmontese:

(19)

more grammaticalized		less grammaticalized
Piedmontese dialects	French	Italian

Piedmontese koiné

5 Plural indefinite articles

Indefinite and partitive articles are a key topic that has been widely discussed in recent linguistic research on the Romance languages. My discussion is informed by previous work by Herslund (2012), Carlier and Lamiroy (2013), Luraghi (forthcoming), and, with a special focus on Piedmontese, Bonato (2004). The indefinite “article”, which in Latin did not exist as such, developed from the Latin preposition de ‘of, from’ plus the definite “article” ille, -a, -um. In all three languages at issue here, this development began during the 12th/13th centuries. Originally, de(+ille) had a strictly partitive meaning, as in (20).

Whereas there is no evidence for the use of de as an indefinite article in the Sermons Subalpins (12th/13th century), it appears in the poems of Alione (1521) with both partitive and indeterminate meanings, although the latter function could also be fulfilled by Ø. At that time, however, articulated forms were more frequent than unarticulated ones, that is, forms derived from the Latin de+ille were used more than the bare de.

In the course of the following centuries, the use of indefinite articles and partitives became more frequent, although they continued to alternate with Ø. With regard to article function, both articulated and unarticulated forms were used. The latter, however, came to be increasingly preferred, so that by the mid-18th/early-19th century the unarticulated form had been almost fully grammaticalized and had become obligatory for both the indefinite article and the partitive functions (see Bonato 2004, whose examples are quoted in 21).

Whereas in spoken Turinese and peri-Turinese varieties (i. e., the varieties upon which the koiné is modeled) the unarticulated form dë/ëd is nowadays obligatory (see 22a, and all the spontaneous interactions registered by Bonato 2004: 185), in contemporary (peripheral) Piedmontese dialects, both articulated and unarticulated forms are still in use (and also alternated with Ø), as shown in examples (22b–c):

This shows that some Piedmontese syntactic structures “contemporaneamente riflettono stadi più arcaici del dialetto stesso, all’italiano più affini di quelle odierne. È quindi possibile cogliere non solo l’impulso all’italianizzazione, ma anche la tensione tra koiné e varietà periferiche” [simultaneously reflect more archaic stages of the regional language that were more similar to Italian than the contemporary stage. It is thus possible to capture not only the drift towards Italianization, but also the variation between the koiné and peripheral varieties] (Bonato 2004: 190; my translation).

French has followed a pattern very similar to Piedmontese’s. After 13th-century “exploratory expressions” (Harris and Campbell 1995), de+ille gradually developed into a full-fledged article, mainly with direct objects at first (15th and 16th century), but becoming increasingly obligatory from 1700 onward in all syntactic positions, including subjects (Carlier and Lamiroy 2013; see 23a).

As for Italian, the de+ille+NP construction (initially used with mass nouns and direct objects) has always appeared in alternation with Ø (see 23b). Today, its use is still not obligatory, but is sensitive to diatopy, with Northern Italian varieties displaying it more frequently than Central and Southern ones (Tekavčić 1980: 118; Luraghi forthcoming). In other words, the regional languages of Italy and (Regional) Italians (in this case Northern Italians) appear to mutually influence one another with respect to the use of partitive/indefinite articles. The fact that Northern speakers of Italian seem more prone to use the partitive than Southern speakers (Carlier and Lamiroy 2013) might result from their different regional linguistic adstrata (i. e., regional and minority languages spoken throughout Italy, as well as French, German, and/or other neighboring languages).

With respect to the development of the plural indefinite article, the different degrees of grammaticalization displayed by the languages under investigation are schematically represented in (24):

(24)

more grammaticalized	less grammaticalied
Piedmontese koiné	French	(North-)Italian
		Piedmontese dialects

A remark is in order here to explain why I have positioned the Piedmontese koiné ahead of French. While French and Italian developed their indefinite article from the articulated Latin preposition de, the competition between the articulated and unarticulated forms in Piedmontese came to an end in favor of the latter during the 19th century. Given that phonetic reduction is a key feature of grammaticalization (Lehmann 2002), Piedmontese is arguably more grammaticalized than the other languages examined here.

6 Conclusion: How do language changes spread throughout a linguistic community?

For the majority of the morphosyntactic domains discussed above, Piedmontese, in its dialect varieties or in its koiné or in both, is on a par with, or even slightly ahead of, French on the RGC, while French is consistently ahead of Italian. It is beyond the scope of this paper to address the question as to why some grammatical domains seem to constitute counterexamples to the usual grammaticalization cline for Romance languages. Take, for example, perfective auxiliation, where Spanish and Portuguese are ahead of French on the RGC, but are constantly behind it in other domains. Similarly, the Piedmontese koiné is at the same stage as Italian with respect to perfective auxiliaries, but is consistently ahead in all other domains. A possible explanation might be that perfective auxiliation is not regarded by (naïve) speakers as a morpho-syntactic feature, but rather as a lexical feature. For bilingual speakers, lexical features are more salient, which is why the difference between the two languages is easily identified, and leveled out or completely eliminated, possibly on account of the prestige of the more standardized language, or of the so-called “roof language”, in the terms of Kloss (1967). This may have happened with Piedmontese, given that it was, and is, spoken alongside Italian, which is generally perceived as more prestigious by Piedmontese speakers. ^[16] In any case, the higher degree of grammaticalization of Piedmontese in the Romance area seems to represent, at least with respect to French and Italian, a strong pattern. ^[17] These findings may help explain why some languages appear to be more grammaticalized than others within the same genealogical family. First, they complement De Mulder and Lamiroy’s (2012: 219–221) view about the role of language-external factors in the grammaticalization processes. As argued by De Mulder and Lamiroy, early urbanization and the size of the speaker community may play a role in favoring grammaticalization. However, neither of these factors can account for the fact that regional languages, such as Piedmontese (and even its local dialects), which by definition are spoken by a smaller population than national languages, are grammaticalizing at least as fast as the most grammaticalized language investigated by De Mulder and Lamiroy, namely French, and consistently faster than Italian. What exactly is it, then, that speeds up grammaticalization in Romance languages?

I suggest a tentative answer to this question on the basis of three parameters. Further studies, taking into account other regional languages of Europe, are needed to confirm, refine or change my speculative conclusions. Given the sometimes extremely different sociolinguistic and socio-historical status of each minor or regional language present in Italy, it is beyond the scope of this paper to investigate the position of other regional languages on the RGC. However, at least as far as the regional languages of Northern Italy are concerned, the data available appears to display a very similar pattern to the Piedmontese data presented here, for example, in relation to demonstrative adjectives, negation (Vai 1996), and other grammar domains (Cerruti 2008; Fedriani and Miola 2014; Loporcaro et al. 2014).

The three factors that I propose may be responsible for speeding up grammaticalization are (i) the strength of social ties among speakers, (ii) the amount of contact with other languages, and (iii) the absence of a fixed linguistic norm (enforced by schools).

6.3 The role of literacy and standardization

One might wonder, then, why Italian did not evolve as fast as the regional languages spoken on Italian territory, and why French has been faster to grammaticalize, even though it was spoken diglossically alongside regional languages until the last century (Lodge 1993: 193–194), a situation somewhat similar to Italian.

A key point here is that Italian was spoken as a native language by almost no one until the mid-20th century (De Mauro 1991: 43). It had been the written language of the entire Italian peninsula since 1500 (Testa 2014), but, although used by intellectuals and writers, Italian was never consistently taught to (and above all, never learnt by) large portions of the population until the end of the 19th century. In some Italian regions, the rate of illiteracy was still over 25 % in the mid-20th century (De Mauro 1991: 95). Until the last century, Italian was only used in “high” domains by cultured people, while the majority of Italians learned it as an L2 and scarcely, if ever, spoke it. Thus, Italian has remained extremely conservative with respect to Latin because of this peculiar sociolinguistic situation: “standard” norms were enforced and were rarely transgressed, simply because nobody actually spoke Italian, apart from highly cultured adult individuals, who mainly used it in written/formal contexts, and whose language is – as is well known – highly conservative in any case. Only when other strata of the population, such as youngsters, started to learn Italian and it was, so to speak, vernacularized, could the pace of grammaticalization begin to speed up. Furthermore, the standardization and scripturalization of a language usually has the effect of slowing down its rhythm of grammaticalization (“language reformation often aim[s] at halting, regulating or redirecting […] grammaticalization”, Laitinen 2004: 247). Italian was codified in the mid-16th century by Pietro Bembo, based on the prose of three 14th-century writers (Dante Alighieri, Francesco Petrarca, and Giovanni Boccaccio). For this reason “in Italia […] il processo di formazione della lingua nazionale coincise con la formazione di una lingua letteraria, basata sul prestigio del modello fiorentino, per secoli utilizzata prevalentemente come lingua scritta e come strumento di comunicazione tendenzialmente alta” [in Italy the process of forming a national language coincided with the formation of a literary language, based on the prestige of the Florentine model, which, for centuries, had mainly been used as a written language and as a means of cultured communication] (Banfi 2014: 26; my translation). Indeed, few changes were made in prescriptive grammars until the mid-20th century.

France, on the other hand, evolved towards a new sociolinguistic situation during the 16th century, when, “for a sizeable proportion of the population, Parisian French functioned as a colloquial L[ow] language in addition to its H[igh] role” (Lodge 1993: 150). When it comes to standardization, then, French is positioned at least some two hundred years after Italian: while the first Italian codification may be said to be based on written texts from 1300–1375 (when Boccaccio, the last of the Tre Corone, died), for French, the elaboration of function began during the 16th century and was not fully codified until the 18th century (Lodge 1993: 149). Moreover, French was standardized based on a spoken dialect, that of Paris, rather than on the language used in literary works, as Italian was. For French, these conditions resulted in less restrained grammaticalization processes.

As regards Piedmontese, it remained a spoken-only Romance variety until recent years: the absence of conservative factors, such as a codified and enforced set of norms (see Parry 1996: 233) may have played a major role in increasing its grammaticalization pace, yielding the following summarizing outline of the RGC for the three languages studied here.

(25)

more grammaticalized	less	grammaticalized
Piedmontese	French	Italian	(Latin)