Clitic Placement and the Grammaticalization of the Future and the Conditional in Old Catalan

Abstract

The romance future and conditional tenses are the result of the grammaticalization of Latin periphrasis, mainly cantāre habeō. In some medieval Romance languages, including Catalan, two types of forms existed: synthetic forms (faré ‘I will do’) and analytical forms (fer-lo he ‘I will do it’). Analytical forms do not present univerbation and are thus less grammaticalized than synthetic forms. The present work aims to study the distribution of synthetic and analytical forms diachronically. A diachronic corpus (11th c.–16th c.) was compiled to serve this purpose. According to the syntactic restrictions of clitic placement, analytical forms could appear in the same syntactic environments than synthetic forms with postverbal pronouns (faré-lo ‘I will do it’). Therefore, only those contexts are analysed to assess the degree of grammaticalization. Some recent works point out that the grammaticalization of future and conditional was more advanced in the eastern languages of the Iberian Peninsula, such as Catalan, than in the western ones. The results from our corpus confirm these differences. In addition, the data show another grammaticalization process: the evolution of clitic placement towards a fixed preverbal position.

1. Introduction1

The study of clitic placement in Old Romance languages is of great importance as regards the understanding of the formation of two verbal tenses: the future and the conditional (FC). These tenses were the result of the grammaticalization of a Vulgar Latin periphrasis, cantāre habeō (‘I have to sing’), in most Romance languages, including Catalan (cf. Coleman 1971; Fleishman 1982; Lehmann 1995; Pérez Saldanya 1998; Hopper and Traugott 2003; Company 2012). Nevertheless, this grammaticalization was not completed in some Romance varieties: the so-called synthetic future or conditional (SFC) (1a) coexisted beside the analytical future or conditional (AFC), which was possible only with mesoclisis, i.e., with the presence of a clitic pronoun between the infinitive and the auxiliary2 (1b) (Castillo Lluch 1996, 2002; Eberenz 1991; Moll [1952] 2006; Badia i Margarit 1984):

(1)	a.	si	Déus	li	ajudarà	(Costums: 490)
		if	God	him.DAT	help.fut.3sg
		‘if God will help him’
	b.	e	Déus	ajudar	-li	à	(Desclot, Crònica: III.85)
		and	God	help.inf	him.dat	have.aux.3sg
		‘and God will help him’

This formal variation in the FC shows two different stages from the grammaticalization of the original Latin periphrasis. The SFC construction attained univerbation, and the infinitive and auxiliary, therefore, lost their boundaries and coalescence took place (cf. Lehmann 1985, 2020).3 The AFC construction is, on the contrary, a less grammaticalized form, since the infinitive and the auxiliary are separated by at least a clitic pronoun. In the Middle Ages, this analytic construction could be found in western Romance varieties, such as Galician-Portuguese (2a), Astur-Leonese (2b), Spanish (2c), Aragonese (2d), Catalan (1b), Occitan (2e), and some Italian dialects (cf. Pérez Saldanya 1998; Paden 1998, pp. 183–86; de Andrés Díaz 2013, pp. 496–97, 499–500; Company 2006; Buridant 2019, pp. 371–72; Primerano and Bouzouita 2023, among others). Nowadays, it can only be found in Portuguese, in which it is preserved in formal registers (Cunha and Cintra 1992, p. 281).

(2)	a.	Et	eu	dar	vos		ey		este caualo
		and	I	give.inf	you.pl.dat		have.aux.1sg		this horse
		‘And I will give you this horse’
		(apud Martins 1994, p. 159)
	b.	A	ti		fazer	te		lo	é
		to	you.sg.dat		make.inf	you.sg.dat		it.acc	have.aux.1sg
		uerdat
		true
		‘To you I will make it come true’
		(apud Primerano et al. 2022)
	c.	E	tornar		los	é		a sus tierras	e
		and	bring back.inf		they.acc	have.aux.1sg		to their lands	and
		a sus heredades […]
		to their landed properties
		‘And I will bring them back to their lands and their landed properties’
		(apud Bouzouita 2016: 270)
	d.	Dizir	uos		hemos		quoales	horas
		tell.inf	you.pl.dat		have.aux.1pl		which	hours
		deue	auer	el infançon
		must.3sg	have.inf	the nobleman (minor nobility)
		‘I will tell you which hours the nobleman must have’
		(apud Primerano 2019: 117)
	e.	mas	servir	l’	ai		dos ans o tres
		but	serve.inf	him.dat	have.aux.1sg		two years or three
		‘but I will serve him two years or three’
		(apud Jensen 1994, p. 243)

The variation between the SFC and the AFC is not accidental. However, there is disagreement between the various proposals, which mainly focus on Old Castilian and whose hypotheses are different. On the one hand, for some scholars, the AFC displays a pragmatic function in which the clitic seems to reinforce the presence of an important entity in the discourse, since it appears in a central position in the predicate (Company 2006, p. 383). Therefore, the AFC is seen as a topicalization strategy (Company 1985, 2006; Company and Medina 1999). This pragmatic approach explains the loss of the AFC by means of the pragmatic restrictions of the form, which is considered marked, in contrast with the SFC, a less restricted construction that progressively starts to cover the space of AFC (Company 2006, p. 410). According to this view, the syntactic restriction of clitic placement in Old Romance languages do not play a role in the variation between SFC and AFC.

On the other hand, other scholars have precisely highlighted the connection between clitic placement and the distribution of the different forms of FC. In the Middle Ages, clitic placement with the FC is governed by the same syntactic–pragmatic restrictions than those with other tenses (Eberenz 1991; Castillo Lluch 1996, 2002; Bouzouita 2011, 2012, 2016; Sentí and Bouzouita 2022). The three following points support this perspective against the pragmatic approach and the perception of AFC as a marked configuration: (i) the AFC and SFC with postverbal clitics (SFC-p, also known as postposition or enclisis) can exclusively appear in the syntactic environments that require a postverbal clitic; (ii) in some linguistic varieties, such as Old Castilian, the AFC is (almost) the only available construction for the environments which demand postposition; (iii) the SFC with preverbal clitics (SFC-a, also known as anteposition or proclisis) are permitted in those syntactic contexts in which preverbal clitics are expected with other verbal tenses. The syntactic–pragmatic restrictions of clitic placement will be discussed in more detail in Section 1.1.

Additionally, some authors have studied the AFC as a modal periphrasis with a marked configuration, a special case of fronting (cf. Fischer 2005; Mathieu 2006; Elvira 2015). Octavio de Toledo (2015), for instance, relates the AFC with fronting configurations of modal periphrasis in Spanish (i.e., hacer lo debo ‘I must do it’), whereby the infinitive and the clitics of the AFC display an informative function, which results in a weak focus (cf. Batllori and Hernanz 2015). Thus, although AFC and SFC-p appear in the same syntactic environments, both constructions would not share the same informative function. In her formal analysis, Batllori (2015) argues that the AFC expresses different modal, evidential and speech act meanings due to the movement from S_v to the left periphery, into Mood or Modality nodes. According to these authors, the loss of AFC (and SFC-p) is not because of the increase in SFC-a, but because of a more general linguistic change in Old Spanish: the various information structure options that were coded using word order fade away and they are replaced by syntactic marking (Octavio de Toledo 2015, p. 210). Although these proposals offer an explanation of the AFC and SFC-p variation, we do not consider the AFC as a marked configuration. As has been underlined before, in Old Castilian the AFC is (almost) the only possible construction in some syntactic environments. Modal or informative differences would thus not be expected in this form as no other form of FC takes up its place. Furthermore, in this paper, the AFC is not considered a periphrasis because, among other reasons, modal meanings have not been attested in our data.4 In addition, important differences have been attested in the evolution of these forms between some Romance languages (i.e., the Catalan data of this paper, see Section 3, in contrast with Old Castilian). In fact, we aim to provide empirical data in favour of a language contact hypothesis (Bouzouita 2016) to explain the diatopic propagation of the SFC (Primerano et al. 2023; see Section 1.2).

1.1. Clitic Placement in Old Catalan

The position of pronominal clitics with respect to the verb in Old Catalan (and Old Romance languages) is motivated by syntactic–pragmatic restrictions (Fischer 2002; Batllori et al. 2005; Sentí and Bouzouita 2022)—unlike in the present-day language, in which morphosyntactic restrictions related to the type of verb form govern clitic positioning (cf. GIEC 2016, Section 8.3.3; Bonet 2002).5

The syntactic–pragmatic environments in which the finite verb appears determine the position of clitics. Clitics generally occur in proclisis in subordinate clauses, while there are three possibilities in main clauses: (i) clitics can be proclitic in certain syntactic contexts, such as negative sentences, (ii) they appear only enclitically in others, and (iii) they admit variation in cases such as when the verb follows the subject. According to the literature, and particularly studies on Old Castilian, the following three groups of syntactic environments that determine the position of pronominal clitics have been established (cf. Eberenz 1991; Castillo Lluch 1996, 2002; Bouzouita 2008a, 2008b, 2011, 2012, 2013, 2016). The three groups are detailed in

Table 1. Syntactic–pragmatic environments governing clitic placement.

Syntactic–Pragmatic Environment		Preverbal Clitics	Postverbal Clitics
Main clauses: Group 1	Wh-pronoun	*
	Negation	*
	Non-coreferential NP	*
	Prepositional complement	*
	Predicative complement	*
	Contrastive ans ‘but’	*
	Infinitive	*
	Participle	*
Main clauses: Group 2	P1 (verb in first position)		*
	Paratactic clause		*
	Vocative		*
	Contrastive mas, però ‘but’		*
Main clauses: Group 3	Subject	*	*
	Coreferential NP	*	*
	Adverbial	*	*
	Coordinative conj. e, y, i ‘and’	*	*
	Disjunctive conj. o ‘or’	*	*
	Non-root and absolute clause	*	*

and will be explained below.

Group 1 always leads to proclisis. Certain preverbal constituents always trigger proclisis, such as a wh-word, a negation marker, a preverbal NP non-coreferential with the clitic, a preverbal predicative complement, the contrastive conjunction ans (‘but’) or an infinitive or a participle dislocated before the verb (fronting).

Group 2 has only postverbal clitics. The syntactic–pragmatic environments in which a non-preverbal clitic is expected in all verb tenses are those that appear with the verb in the first position, paratactic clauses, after a vocative, and in coordinated adversative clauses after the conjunction mas or però (‘but’).

Finally, Group 3 is made up of those environments in which variation in clitic placement occurs: following a preverbal subject, after a dislocated preverbal NP coreferential with the clitic pronoun, after a preverbal adverbial phrase, following a coordinative or disjunctive conjunction and following a non-root or absolute clause. The default position in these syntactic contexts is enclisis. However, some (pragmatic) marked configurations can trigger proclisis. Preverbal clitics appear when the preverbal constituent proves to be emphatic or prominent in the information structure. Following Granberg (1999), the emphatic constituent is considered focus if it receives informative prominence in the discourse.6 Even though analysing information structure in written medieval texts is not always reliable, some discourse phenomena can indicate the presence of emphasis (for example, a contrastive subject regarding a different previous entity in discourse). Regarding clitic placement, some studies on present-day Galician (Álvarez Blanco et al. 1986) and Old Castilian (Granberg 1988, 1999; Castillo Lluch 1996; Bouzouita 2008a, 2008b, 2011, 2012, 2016) have observed that this pragmatic function (emphasis) accounts for the presence of preverbal clitics in the aforementioned syntactic environments. Similar behaviour has also been recently observed in Old Catalan (Bouzouita and Sentí 2022; Sentí and Bouzouita 2022; Torres-Latorre 2023). Even so, a detailed exploration of the consequences of information structure in clitic placement in Old Romance languages is still pending.

With regard to the grammaticalization of the FC, and according to the above-mentioned syntactic restrictions of clitic placement, the AFC could appear in the same syntactic contexts as the SFC with postverbal pronouns (Group 2 and 3). These contexts will, therefore, be examined in this work.

1.2. Objectives

According to the summarised state of the art, two different processes of linguistic change related to clitic placement are dealt with in this paper. First, the variation permitted between a more grammaticalized construction, i.e., SFC-p, and a less grammaticalized version of the FC, i.e., the AFC, since these forms appear in the same syntactic contexts. As previous studies have shown, both constructions are frequent in Old Catalan (Bouzouita and Sentí 2022; Sentí and Bouzouita 2022; Torres-Latorre 2023). However, this was not the case in Old Castilian. As already urged by Bouzouita (2016), SFC-p should be scrutinised in more detail in Ibero-Romance, because, whereas this form was rarely found in Old Castilian texts, it was quite a lot more frequent in some eastern texts written in this language, as already noted by Saralegui (1985). Some recent works (Bouzouita 2016; Bouzouita and Sentí 2022; Primerano et al. 2022) point out that the synthetic forms could have been spread from Catalan to other languages and varieties in the Iberian Peninsula during the 11th–13th centuries (from Catalan to Navarro-Aragonese, from Navarro-Aragonese to Castilian, etc.), signifying that language contact may have played an important role. The empirical findings of the grammaticalization of the FC in Old Catalan, therefore, prove crucial to understand the Old Ibero-Romance.

The objective of this paper is to study the distribution of the SFC and the AFC diachronically and quantitatively and to analyse the grammaticalization process from the 11th to the 16th century. In this respect, an increase in the frequency of SFC-p, with univerbation and a more grammaticalized form, and a consequent decrease in the AFC would be expected, bearing in mind that the AFC is no longer used in contemporary Catalan. In fact, this development has been noted in Old Castilian and in Old Navarro-Aragonese in the 13th and 14th centuries. In Old Castilian, SFC-p was almost non-existent in the 13th century, and it remained uncommon in the 14th century, but the increase in its presence appears to be statistically significant (Primerano et al. 2023, Section 4.1). Although SFC-p was not as rare in Old Navarro-Aragonese as it was in Old Spanish, an increase in its use is also attested in these two centuries (Primerano and Bouzouita 2023).

Furthermore, another grammaticalization process participated in the evolution of Old Catalan: the evolution of clitic placement from the above-mentioned syntactic distribution towards a fixed preverbal position (as in present-day Catalan). In spite of the previous knowledge on pronominal clitics in Old Catalan (Fischer 2002; Batllori et al. 2005; Bouzouita and Sentí 2022; Sentí and Bouzouita 2022), this topic has yet to be studied quantitatively and qualitatively. There is lack of knowledge on which variation contexts favour preverbal placement and on the extent to which proclisis is linked to emphasis (or whether it can also appear in a neutral, i.e., non-emphatic environment, and therefore be used in innovative contexts). The same applies to its evolution throughout history and the fixing of word order with the clitics in proclisis to the finite verb, or its dialectal and discursive variations. This is the second purpose of this paper: we wish to contribute to explaining this process with data regarding the FC, again from the 11th to the 16th century.

In the following section (Section 2) we explain our selected corpus and some methodological issues. The empirical results and a discussion of the corpus-based study are then addressed: Section 3 shows the distribution of the AFC and SFC-p in Old Catalan, while Section 4 summarises the evolution of clitic placement towards anteposition. Finally, we present our conclusions in Section 5.

2. Methodology and Corpus

Old Catalan usually covers the period between the first attested Catalan texts (11th century) and the beginning of the modern age (16th c.) (Martines and Pérez Saldanya 2009).7 In order to carry out the study of the grammaticalization of the FC and clitic placement in this period, we have used the Corpus Informatitzat del Català Antic (CICA 2006) (Claveria and Torruela 2012), which is the only available corpus that is devoted to the study of this period. Although the CICA (2006) corpus is not a morpho-syntactically annotated corpus, the extraction of the FC with clitics has been carried out semi-automatically by seeking the forms through a search for verb endings. Moreover, the analytic forms were found thanks to the diacritic accentuation available in the electronic edition of the texts.

The high token frequency of the FC made it necessary to restrict ourselves to select some texts from the CICA (2006) for this study.8 We have selected only those texts that were written in the same period as the preserved manuscript. We have selected texts from the second half of each century in order to analyse the linguistic change from one period to another. Diatopic variation has also been taken into account: we have established a balanced subcorpus for each group of Catalan dialects with a similar number of tokens (eastern and western Catalan). Despite having paid attention to the extraction of a similar number of tokens for each dialect, it was not easy to obtain these results for all the periods considered (see

Table 2. FC tokens retrieved for each period in each dialectal group.

Dialect	11th–12th	13th	14th	15th	16th
Eastern Catalan	-	732	925	213	360
Western Catalan	-	991	333	953	357
No dialectal adscription	218	1186	28	-	-
Total	218	2909	1286	1166	717

).9 Finally, a reasonable range of textual typologies has also been considered, such as religious, legal, narrative, historiographic texts and letters, although attempting to preserve the representativeness of each discursive tradition in each period was quite challenging (Kabatek 2005, 2013).

Those examples of the SFC and AFC with adverbial clitic pronouns (e.g., en < ĭnde and hi < ĭbi) have been analysed separately. This decision made it easier to compare Catalan and other Ibero-Romance languages that barely conserve these pronouns (e.g., Primerano 2021 for Navarro-Aragonese). Moreover, it also allowed a comparison between the behaviour of personal clitic pronouns and that of adverbial clitic pronouns, which has proved relevant (Section 3.1).

3. The Distribution of AFC and SFC-p in Old Catalan

The analytical and synthetic forms in Old Romance do not always coexist for pragmatic–syntactic reasons (Section 1). In fact, the AFC and SFC with postverbal pronouns appear in the same pragmatic–syntactic environments, whereas the SFC with preverbal pronouns are used in other contexts. Thus, the degree of grammaticalization of the FC can be observed only by means of the variation between the AFC and SFC-p. The use of SFC-p rather than the AFC shows a more grammaticalized construction of these two verbal tenses, since synthetic forms display univerbation (Lehmann 1985, 2020). Previous studies have already attested SFC-p from the 13th c. in Old Catalan (Bouzouita and Sentí 2022; Sentí and Bouzouita 2022; Torres-Latorre 2023). Although the AFC is more frequent, the presence of SFC-p in early periods of the language shows a high degree of grammaticalization. In fact, the aforementioned studies have found SFC-p to be more frequent in Old Catalan than in Old Navarro-Aragonese and in Old Spanish.

The empirical results regarding the distribution of the AFC and SFC-p (with personal pronouns) in Old Catalan are summarised in

Table 3. The distribution of the AFC and SFC-p with personal pronouns (11th–16th c.).

Chronology	AFC	SFC-p
Preliterary Catalan (11th–13th c.)	50% (1/2)	50% (1/2)
13th century, 2nd half	61.2% (180/294)	38.8% (114/294)
14th century, 2nd half	60.2% (50/83)	39.8% (33/83)
15th century, 2nd half	58.1% (25/43)	41.9% (18/43)
16th century, 2nd half	80% (12/15)	20% (3/15)
Total	61.3% (268/437)	38.7% (169/437)

. This table considers only two of the three possible constructions of the FC: the AFC and SFC-p. The data contained in it, therefore, originate from the syntactic environments of Groups 2 and 3 explained previously (Section 1.1). The results with adverbial pronouns (en, hi) will be presented hereafter (Section 3.1).

At first sight, both forms, the AFC and SFC-p, are present throughout the period. The distribution of the two forms does not have the exact expected evolution, signifying that SFC-p does not appear to gain ground, while the AFC disappears over time. The distribution of the AFC and SFC-p does not change in the following periods. Leaving aside preliterary Catalan, which is deemed unusable since only two tokens have been attested,10 the differences between the remaining centuries shown in Table 3 are not statistically significant (χ² = 2.4328, p = 0.487565; significant at p < 0.05). The AFC always predominates, with 61.3% (268/437) of the total of instances, over SFC-p, with 38.7% (169/437). There appears to be a preference for the AFC in the 16th century (80%, 12/15), but the limited number of both constructions in this century makes the statistical differences between periods non-significant. It would, therefore, appear that the more grammaticalized construction (SFC-p) does not increase in frequency over time, as would be expected in this kind of linguistic change process (Bybee 2003).11

However, some empirical findings may be relevant in order to understand the grammaticalization of the FC. First, it is important to point out that an example of FC with postverbal clitics (SFC-p) is found in the 11th c. (3). The presence of a clitic cluster could have triggered this appearance of the univerbated form, which shows the grammaticalized FC construction at an early date, quite a long time before its appearance in other Ibero-Romance languages. In fact, SFC-p is already a productive construction in the 13th century, considering the 38.8% (114/294) of instances, whereas it is a rare form in Old Castilian, as will be seen later.

(3)	et	ajudaré	·ls	te		a	tener
	and	help.fut.1sg	them.acc	you.dat		to	have.inf
	contra	tots	homines	et	contra	totas	fèminas	(Documents feudals 3: 77)
	against	all	men	and	against	all	women
	‘and I will help you have them against all men and against all women’

Leaving aside preliterary Catalan (see Note 8), the data show a progressive decrease in both the AFC and SFC-p from the 13th c. onwards.

Table 4. The distribution of the AFC and SFC-p with personal pronouns (11th–16th c.) as regards relative frequency (%) per million words.

Chronology	AFC	SFC-p
Preliterary Catalan (11th–13th c.)	39.72 (1/25176)	39.72 (1/25176)
13th century, 2nd half	212.55 (180/846841)	134.62 (114/846841)
14th century, 2nd half	146.84 (50/340517)	96.91 (33/340517)
15th century, 2nd half	58.29 (25/428921)	41.97 (18/428921)
16th century, 2nd half	36.69 (12/327032)	9.17 (3/327032)

(below) shows the relative token frequency of the AFC and SFC-p per million words (in relation to the number of words in each period). The AFC represents 215.55 tokens per million words in the 13th century, whereas it represents 36.69 in the 16th century, that is, a seventh of what it was in the 13th century. The decrease in both the AFC and SFC-p, which can already be observed in the 14th, 15th and 16th centuries, is owing to an increase in SFC-a and will be studied in the following section (Section 4). The limited number of tokens in the 16th century is not owing to differences in the size of the corpus. The reason why only 15 tokens are found in our corpus can be explained by the relative frequency, which indicates the general tendency to decrease of both the AFC and SFC-p constructions. There appears to be a preference for the AFC in the century in question (80%, 12/15), but the limited quantity of both constructions in this century make the statistical differences between periods non-significant.

The comparison of the empirical data from Old Catalan with Old Castilian and Old Navarro-Aragonese also provides some important findings to understand the grammaticalization of the FC in the Iberian Peninsula. Old Catalan differs from the behaviour of Old Castilian and Old Navarro-Aragonese: despite the fact that the AFC is the main form throughout the period, it is not as frequent as the two other varieties, as summarised in

Table 5. AFC and SFC-p in Old Catalan, Old Navarro-Aragonese and Old Castilian (13th–14th c.).

Language	Chronology	AFC	SFC-p
Catalan	13th century	61.2% (180/294)	38.8% (114/294)
	14th century	60.2% (50/83)	39.8% (33/83)
	Total	61% (230/377)	39% (147/377)
Navarro-Aragonese12	13th century	89.1% (49/55)	10,9% (6/55)
	14th century	75% (51/68)	25% (17/68)
	Total	81.3% (100/123)	18.7% (23/123)
Castilian13	13th century	99.8% (1827/1830)	0.2% (3/1830)
	14th century	97.3% (359/369)	2.7% (10/369)
	Total	99.4% (2186/2199)	0.6% (13/2199)

.

In the case of Old Castilian, Primerano et al. (2023, Section 4.1) have discovered that AFC are by far the most frequent in the 13th and the 14th centuries (99.4%) when compared to SFC-p (0.6%); in that of Old Navarro-Aragonese, Primerano and Bouzouita (2023) have found 81.3% of AFC, which is considerably more than the 18.7% obtained for SFC-p in the same two centuries. With regard to Old Catalan, SFC-p represents 38.8% (114/294) and 39.8% (33/83) in the 13th and the 14th centuries, respectively, and it is, therefore, much more frequent. As expected, Old Catalan has the highest degree of grammaticalization of the FC. Furthermore, an increase in SFC-p is perceived in the 14th century in both Old Castilian and Old Navarro-Aragonese, but this is not the case of Old Catalan, in which the variation between the two forms is stable. The 15th century has yet to be studied in these two other languages.

The differences between Catalan on the one hand and Castilian and Navarro-Aragonese on the other can be partially explained by the intervention of certain morphological factors that contribute to the choice of AFC or SFC-p. It has been suggested that certain morphological and even syntactic factors could induce the choice of SFC-p rather than AFC. These factors have been studied principally for Old Castilian (Bouzouita 2016), but also for Old Navarro-Aragonese (Primerano and Bouzouita 2023). Since SFC-p is an uncommon construction in both languages, these works sought the causes of the presence of SFC-p and not of the AFC, which is the construction by default. The inductors of SFC-p would also be triggers of the grammaticalization process of the FC. The factors analysed in these works are the verbal tense (if it is a future or conditional), the conjugation class, the presence of a syncope in the verb and the introduction of a non-finite verbal form after the FC.

A more detailed study of these factors is still necessary for Old Catalan (Torres-Latorre n.d.a). However, some differences between Old Catalan and Old Castilian and Old Navarro-Aragonese can be noted only by observing our preliminary empirical data. First, the presence of a syncope in the FC is much more common in Catalan than in the other two languages. This syncope is the omission of the thematic vowel of the verb and can be found in some verbs of the second conjugation (volerà > volrà ‘he will want’) and a few verbs of the third conjugation too (partirà > partrà ‘he will leave’).14 A verb with a syncope always prefers SFC-p over the AFC (Bouzouita 2016, p. 284; Primerano and Bouzouita 2023), as (4a) shows. That is for phonetic reasons: mesoclisis cannot be conducted if there is a syncope. A form such as *volr-la à (‘he will want her’) is not possible when following the phonetic rules of these languages, but the same form without a syncope (voler-la à ‘he will want her’) is indeed possible. In our data, every instance of non-preverbal clitics with the FC has the SFC-p form, as expected.

During the Middle Ages, the presence of a syncope was optional in the three Romance languages mentioned. Nevertheless, it was much more frequent in Old Catalan. In fact, in contemporary Catalan the presence of a syncope is mandatory in the verbs of the second conjugation which formerly varied (voler ‘to want’ > voldrà ‘he will want’, with an epenthetic d; saber ‘to know’ > sabrà ‘he will know’, etc.). Contemporary Spanish has, on the contrary, preferred to maintain the omitted vowel in some of the verbs which had an optional syncope (saber ‘to know’ > sabrá ‘he will know’, with a syncope; but perder ‘to lose’ > perderá ‘he will lose’, without a syncope). The differences in frequency in the presence of a syncope in this group of verbs could explain a slight tendency in favour of SFC-p in Old Catalan with respect to the rarity of this construction in Old Castilian and Old Navarro-Aragonese.

Second, there is one particularity of Catalan’s verbal conjugation that distinguishes Old Catalan from Old Castilian and Old Navarro-Aragonese. In Catalan, in the second conjugation, most verbs take a rhizotonic form (Pérez Saldanya 1998, pp. 40–48), that is, they are verbs whose stressed syllable is in the stem, such as conèixer (‘to know’, ‘to meet’) or perdre (‘to lose’). Alsina (2022, p. 192) finds that rhizotonic verbs also prefer SFC-p over the AFC, but, although a preference for SFC-p is confirmed in our corpus (4b), some tokens of the AFC are also attested (4c). The inclination towards SFC-p is for phonetic reasons: a rhizotonic verb with mesoclisis is not unpronounceable in Old Catalan, but in a form such as defendre-nos hien (4c) the main stress remains far from the word ending, in the syllable -fen-, and in Catalan, oxytones and words stressed on the penultimate syllable are far more common (Oliva and Serra 2002). This factor could once more entail a slight preference in Catalan for SFC-p when compared with Old Castilian and Old Navarro-Aragonese. However, it is still necessary to quantitatively explore both factors (Torres-Latorre n.d.a).

Moreover, instances of SFC-p in which none of the factors mentioned above (verbal tense, conjugation, syncope, rhizotonic verbs or non-finite verbal forms after the FC) are attested. The first conjugation (verbs with infinitives finished with -ar, i.e., cantar ‘to sing’) is characterised by a total preference for the AFC in Old Castilian (Bouzouita 2016) and in Old Navarro-Aragonese (Primerano and Bouzouita 2023), but in Old Catalan there are examples such as (4d), in which a verb of the first conjugation appears with the SFC-p construction. The variation between AFC and SFC-p, therefore, requires an explanation. Nevertheless, with regard to the evolution of the variation in the two constructions, these factors appear constant throughout the period studied. For instance, the case of (4d) is from the 13th century, and it does not contain any of the factors that favoured SFC-p. It is possible that there were no cases of SFC-p without the presence of any of these factors in a previous stage of the language, and SFC-p later spread to other verbs, such as alegrar from example (4d). Unfortunately, this can only be hypothesised, since previous stages of the Catalan language have not been preserved.

(4)

a.

Nostre

Senyor

no

vol

la

mort

del

Our

Lord

neg

want.pres.3sg

the

death

of-the

peccador,

e

volrà-

la

lo

rey?

(Tirant: 268)

sinner

and

want.fut.3sg

it.acc

the

king

‘Our Lord does not want the death of the sinner, and the King will want it?’

b.

e

conexerets

-ho

(Lletres reials: 564)

and

know.fut.2pl

it.acc

‘And you will know it’

c.

si

s

crexien

de

gent,

defender

-nos

if

refl

increase.imp.3pl

of

people

deprive.inf

us.acc

hien

la

Ciutat

(Fets: 48r)

have.aux.3pl

the

City

‘If they increased in number, they would deprive us of the City’

d.

Alegrarà

·s

lo

just

(Vides: 104)

Be happy.fut.3sg

refl

the

just

‘The just will be happy’

To sum up, the distribution of the AFC and SFC-p with personal pronouns is stable from the 11th c. to the 16th c., that is, the presence of the AFC does not decrease in favour of SFC-p. Nevertheless, another change occurs in the same period: the grammaticalization process of pronominal clitics and, in particular, their fixation in the preverbal position (Section 4). Both the AFC and SFC-p eventually cease to exist when anteposition becomes the only possible position of clitics. Although signs of evolution towards SFC-p have not been observed in the period studied, the data from Old Catalan show a more advanced stage of the degree of grammaticalization of the FC in the 13th and the 14th c. than occurs in other peninsular languages. Old Catalan frequently exhibits SFC-p and the AFC is used to a lesser extent when compared with Old Navarro-Aragonese and Old Castilian, as has been shown quantitatively. The morphological differences between the languages (syncope, rhizotonic verbs) could partially justify this contrast, but a more detailed study of these factors is still required (Torres-Latorre n.d.a).

3.1. AFC and SFC-p with Adverbial Pronouns

Thus far, the distribution of the FC with non-adverbial pronouns has been studied. Adverbial pronouns in Old Catalan are mainly en (<ĭnde) and hi (<ĭbi), but also hic (<hīc) (Ribera 2020). The decision to analyse them separately was made for two reasons: (i) en, hi and hic may have different degrees of pronominalization in Old Catalan and could, therefore, behave differently to personal clitic pronouns; (ii) the grammar of Old Romance languages varies in the case of these pronouns, signifying that it is not possible to compare the languages in this respect with the same reliability as in the case with the remaining pronominal system.

The pronouns en and hi are preserved in contemporary Catalan, unlike in Spanish, from which they disappeared in the 15th c. (Badia i Margarit 1947, p. 128). In the Middle Ages, their behaviour was more clitic-like in Old Catalan than in the other peninsular Romance varieties. In Old Castilian, they are often considered to be pronominal adverbs rather than adverbial pronouns (Badia i Margarit 1947; Meilán García 1994; Polo Cano 2006; Matute 2016). In Old Navarro-Aragonese, they appear to be more similar to clitics than to adverbs (Matute 2016, p. 221). Hic appears to be the least pronominalized form of the three and is not conserved in contemporary Catalan. Unfortunately, no instance of the FC with a non-preverbal pronoun hic has been attested.

Table 6. The distribution of the AFC and SFC-p with adverbial pronouns (11th–16th c.).

Chronology	AFC	SFC-p
Preliterary Catalan (11th–13th c.)	0% (0/2)	100% (2/2)
13th century, 2nd half	45.8% (38/83)	54.2% (45/83)
14th century, 2nd half	25% (5/20)	75% (15/20)
15th century, 2nd half	28.6% (2/7)	71.4% (5/7)
16th century, 2nd half	50% (2/4)	50% (2/4)
Total	40.5% (47/116)	59.5% (69/116)

provides a summary of the data obtained for the AFC and SFC-p with regard to the pronouns en and hi.

The lower number of tokens contained in Table 6 in contrast with the total number of occurrences with personal pronouns was to be expected, as only two pronominal clitics were studied. The total number of 437 instances of clitics in the AFC or SFC-p with non-adverbial pronouns (see Table 3) now contrasts with the total of 116 that have been attested. The general distribution of the AFC and SFC-p with adverbial pronouns is almost the opposite of the distribution of the two constructions with non-adverbial pronouns. As a reminder, the AFC represents 61.5% (268/436) of tokens, while SFC-p has 38.5% (168/436) as regards personal clitics. With adverbial pronouns, SFC-p is the main form, with 59.5% (69/116) of tokens, and the AFC follows with 40.5% (47/117). The differences between the two groups of pronouns are statistically significant (χ² = 16.4142, p = 0.000051; significant at p < 0.05).

With regard to the evolution of the distribution of the AFC and SFC-p with adverbial pronouns, there appears to be some variation, although it is not quantitatively important. In preliterary Catalan, the only two instances attested are SFC-p. However, a greater presence of the AFC has been found in the 16th c., with 50% (2/4), although the total number of instances is highly reduced and it is not, therefore, possible to draw strongly founded conclusions.15 The differences between the other periods studied are not statistically significant (χ² = 3.4742, p = 0.324119; significant at p < 0.05), and the AFC does not, therefore, disappear with adverbial pronouns either. Although there is a preference for SFC-p from the very beginning, its presence does not increase in the period studied. A decrease in the frequency of both constructions is also found with adverbial pronouns, mainly in the 15th and the 16th centuries.

The different status of the pronouns en and hi may possibly provide an explanation of the difference between the distribution of the AFC and SFC-p. In Old Catalan, adverbial pronouns are less grammaticalized than personal pronouns such as el (‘him’), la (‘her’) or li (‘to him, to her’), because the transition from adverbs to pronouns is not yet complete. In fact, the difference between the distribution of the AFC and SFC-p and that of adverbial pronouns has also been observed in Old Navarro-Aragonese (Primerano 2021, p. 418), a language in which adverbial pronouns behave as normal clitic pronouns to an even lesser extent. However, this explanation cannot be completely confirmed without considering the aforementioned morphological and syntactic factors which favour SFC-p rather than AFC (verbal tense, conjugation class, syncope, rhizotonic verbs, non-finite form after the FC). For example, the adverbial pronoun hi frequently collocates with the verb haver as in hi haurà (‘there will be’) and haver has a rhizotonic form with FC (haure ‘to have’). A review of the factors in relation to this hypothesis is, therefore, also required (Torres-Latorre n.d.a).

In conclusion, the AFC and SFC-p have a stable distribution from the 11th c. to the 16th c. as regards both non-adverbial and adverbial pronouns. There are two differences between Old Catalan, on one hand, and Old Castilian and Old Navarro-Aragonese on the other: (i) the grammaticalization of the FC is more advanced in Old Catalan since SFC-p is found in the above-mentioned syntactic environments in main sentences (Group 2 and Group 3) from the 11th c. onwards, and with an important frequency during the 13th and the 14th c.; (ii) the diverging evolutions of the variation in the distribution of the AFC and SFC-p, which is stable in Catalan rather than in favour of SFC-p. The progressive symmetrical disappearance of AFC and SFC-p is owing to another grammaticalization process, that of pronominal clitics, which is discussed in the following section (Section 4).

4. The Evolution of Clitic Placement towards Anteposition

In the Middle Ages, the position of pronominal clitics with finite verbs was owing to syntactic–pragmatic restrictions (for Old Catalan, see Fischer 2002; Batllori et al. 2005; Sentí and Bouzouita 2022). It followed the distribution into three groups of syntactic–pragmatic environments summarised in Table 1 (Section 1.1). However, in present-day Catalan, clitic placement follows only morphological restrictions: (i) finite verbs (apart from the imperative) can be used only with preverbal clitics, (ii) the imperative, the infinitive and the gerund can be used only with postverbal clitics, and (iii) verbal periphrases can be used with either preverbal or postverbal clitics (Bonet 2002, p. 937; GIEC 2016, Section 8, pp. 200–1).

In the case of finite verbs (with the exception of the imperative), clitic placement has, therefore, gone through a fixation process to the preverbal position. This fixation process has not yet been studied sufficiently. In this section, the results obtained for the FC will be analysed by considering the three possible constructions of these two verbal tenses: SFC-a, the AFC and SFC-p. An evolution towards anteposition is expected from the 11th c. to the 16th c. Because of space limitations, only the results obtained for personal pronouns will now be considered and also because adverbial pronouns do not behave differently in the frequencies of SFC-a (Torres-Latorre 2022). The results will be examined first quantitatively and then qualitatively.

Table 7. The position of personal pronouns with the FC (11th–16th c.).

Chronology	SFC-a	AFC or SFC-p
Preliterary Catalan (11th–13th c.)	98.7% (151/153)	1.3% (2/153)
13th century, 2nd half	86.3% (1854/2148)	13.7% (294/2148)
14th century, 2nd half	91.6% (904/987)	8.4% (83/987)
15th century, 2nd half	95.4% (889/932)	4.6% (43/932)
16th century, 2nd half	97.6% (616/631)	2.4% (15/631)
Total	91% (4414/4851)	9% (437/4851)

shows the total number of frequencies of the three constructions, without making any distinction according to the groups of syntactic–pragmatic environments. All the attested instances of the FC with personal pronouns are included. Since the variation between the AFC and SFC-p has already been reviewed, both options are considered together from this point on.

At first glance, it will be noted that SFC-a is by far the most frequent throughout the periods studied. Nevertheless, a more detailed analysis shows differences between centuries (statistically significant: χ² = 126.8946, p < 0.00001; significant at p < 0.05). Surprisingly, the period in which the frequency of SFC-a is greatest is that of preliterary Catalan, with 98.7% (151/153) of the instances in our corpus, but methodological issues oblige us not to consider these results (see Note 10). The remaining centuries do show the expected development towards anteposition: the smallest percentage of SFC-a is in the 13th c. (86.3%, 1854/2148), followed by the 14th c. (91.6%, 904/987), the 15th c. (95.4%, 889/932) and finally, the 16th c., in which this percentage reaches 97.6% (616/631) of tokens. The differences among the four centuries are still statistically significant (χ² = 110.4827, p < 0.00001; significant at p < 0.05).

The data from all syntactic-pragmatic environments could be conditioned by a larger or smaller presence of a particular environment in each century, i.e., subordinate clauses, in which the required pronouns are preverbal. The results in Table 7 must, therefore, be viewed with caution. From this point on, attention will consequently be focused on: (i) the main sentence environments in Group 2, i.e., the group in which no token should take the SFC-a form, according to the grammar of the 13th–14th centuries (§Section 4.1), and (ii) the environments in Group 3, which are characterised by the possibility of variation between anteposition and postposition governed by pragmatics (information structure) (Section 4.2).

4.1. The Evolution of Clitic Placement in Group 2

Group 2 is distinguished by the fact that it permits only postverbal clitics (Section 1.1). The expected position of clitics in this environment with the FC would be either the AFC or SFC-p. However, the results from our corpus indicate a different distribution, since SFC-a is possible in some environments from the 13th c.

Table 8. The position of personal pronouns with the FC in Group 2 (13th–16th c.).

Chronology	Environment	SFC-a	AFC or SFC-p
13th century, 2nd half	P1	0% (0/16)	100% (16/16)
	Paratactic	0% (0/0)	0% (0/0)
	Vocative	69.2% (9/13)	30.8% (4/13)
	Contrastive mas, però	0% (0/7)	100% (7/7)
14th century, 2nd half	P1	0% (0/5)	100% (5/5)
	Paratactic	0% (0/2)	100% (2/2)
	Vocative	50% (2/4)	50% (2/4)
	Contrastive mas, però	0% (0/3)	100% (3/3)
15th century, 2nd half	P1	33.3% (2/6)	66.7% (4/6)
	Paratactic	0% (0/0)	0% (0/0)
	Vocative	85.7% (6/7)	14.3% (1/7)
	Contrastive mas, però	0% (0/0)	0% (0/0)
16th century, 2nd half	P1	85.7% (6/7)	14.3% (1/7)
	Paratactic	100% (1/1)	0% (0/1)
	Vocative	100% (1/1)	0% (0/1)
	Contrastive mas, però	0% (0/3)	100% (3/3)
Total		36% (27/75)	64% (48/75)

provides a quantitative summary of the evolution of the environments in Group 2. The table does not contain preliterary Catalan because no token from this group has been found for this period.

Table 8 proves that anteposition is gaining ground both quantitatively and qualitatively from the 13th century to the 16th century. In Group 2, 0% of this construction was to be expected according to the bibliography for Old Castilian and Old Romance in general (Eberenz 1991; Castillo Lluch 1996, 2002; Bouzouita 2008a, 2008b, 2011, 2012, 2013, 2016). In spite of this, examples of preverbal clitics in this group have been attested even in the 13th century. SFC-a is found only after a vocative in the 13th and the 14th centuries (5a); in the 15th century, it appears for the first time with the verb in the first position of the phrase (5b); in the 16th century, SFC-a is also attested in a paratactic clause (5c), with the presence of a contrastive conjunction mas or però (‘but’) being the only environment in this group in which SFC-a is never found. Only the context after a vocative, therefore, appears to be sufficiently innovative for the presence of preverbal clitics in the first two centuries, but a clear change can be noted in the 15th, and especially in the 16th, centuries, since other environments in Group 2 have the SFC-a construction.

(5)	a.	E	tu,	diable,		l’	esperaràs		en	aquest
		and	you.sg	devil		him.acc	wait.fut.2sg		in	this
		loc	entrò que	El	vinga		(Vides: 119)
		place	until	he	come.subj.3sg
		‘And you, devil, will wait for him in this place until he comes’
	b.	M’	esforçaré			al	meu	poder	per	serveys
		refl	make an effort.fut.1sg			in.the	my	power	by	services
		(Tirant: 1371)
		‘I will make the biggest possible effort by services’
	c.	Lo	acòlith	encendrà […],			servirà […],		portarà […],
		the	acolyte	light.fut.3sg			serve.fut.3sg		carry.fut.3sg
		ministrarà […],		se		pararà		ab	camís,
		manage.fut.3sg		refl		prepare.fut.3sg		with	shirt
		amit	y	cordó		(Baccallar: 267)
		amice	and	lace
		‘The acolyte will light […], will serve […], will carry […], will manage […], will prepare himself with a shirt, an amice and lace’

Of these three instances provided as examples, the first (5a) can be explained for pragmatic reasons: the vocative, diable (‘devil’), is preceded by an emphatic subject, tu (‘you’), signifying that the anteposition of the clitic could be triggered by emphasis (Torres-Latorre 2023, Section 3.1). This pragmatic cause explains the high frequency of SFC-a after a vocative, since all the tokens in this environment have another element before the vocative which could be subject to emphasis. Furthermore, (5a) originates from the Vides de sants rosselloneses, a text which has been observed to be highly innovative as regards clitic placement (Torres-Latorre 2023). The reason for this innovation appears to be its diatopic adscription: Vides de sants rosselloneses belongs to septentrional Catalan, the Catalan dialect closer to Occitan, and Old Occitan may be more advanced than Catalan as regards both the grammaticalization process of the FC and the fixation of clitic placement in the preverbal position. This hypothesis should be verified in future research. Whatever the case may be, the instances of SFC-a in the first position (5b) and in a paratactic clause (5c) can be understood only as innovative examples of anteposition.

4.2. The Evolution of Clitic Placement in Group 3

The results regarding all environments presented in Table 7 include subordinate clauses and Group 1, that is, the syntactic contexts in which only SFC-a is possible. The difference between preverbal clitics and others, therefore, can be better observed if only the data from Group 3 are taken into account, since this group is made up of those environments in which variation in position is possible (Section 1.1). The quantitative review of Group 3 in

Table 9. The position of personal pronouns with the FC in Group 3 (11th–16th c.).

Chronology	SFC-a	AFC or SFC-p
Preliterary Catalan (11th–13th c.)	95.8% (23/24)	4.2% (1/24)
13th century, 2nd half	41.8% (166/397)	58.2% (231/397)
14th century, 2nd half	74.6% (191/256)	25.4% (65/256)
15th century, 2nd half	84.8% (179/211)	15.2% (32/211)
16th century, 2nd half	96.3% (129/134)	3.7% (5/134)
Total	67.3% (688/1022)	32.7% (334/1022)

is more reliable as regards the expansion of anteposition than the general results seen in Table 7.

However, this reasoning is not yet conclusive, since the variation within Group 3 can sometimes be explained by a more accurate classification in each of the syntactic–pragmatic environments. For instance, emphasis can cause the appearance of a preverbal clitic, as mentioned briefly in Section 1.1. If this is the case, it is necessary to examine the instances of SFC-a one by one in order to determine whether or not they are innovative examples of anteposition. A review of each of the tokens belonging to Group 3 is, therefore, necessary. Only some examples of this group will be commented on in this paper owing to space constraints (see Torres-Latorre n.d.b; cf. Bouzouita and Sentí 2022 and Sentí and Bouzouita 2022 for a tentative study on the effect of pragmatics in these contexts in the 13th and 14th c.), and the focus will be solely on the quantitative view of this phenomenon. Table 9, therefore, provides a summary of the evolution of clitic placement in this group of variation.

The results obtained for the group of variation obey the same development as the total of tokens included in the other groups (Table 7). Preliterary Catalan again differs from the other periods, with SFC-a being attested with by far the most frequency (95.8%, 23/24). The evolution from the 13th c. to the 16th c. shows a noticeable increase in SFC-a: it starts with 41.8% (166/397) in the 13th c. and concludes with 96.3% (129/134) in the 16th c. Indeed, the differences between the four centuries are statistically significant (χ² = 201.6922, p < 0.00001; significant at p < 0.05). Group 3 clearly shows the evolution towards anteposition and the fixation of clitics in this position, regardless of the syntactic context or pragmatic motivations.

It is first necessary to explore the case of preliterary Catalan. This period has the highest level of SFC-a in both the group of variation (see Table 9) and all the possible environments (see Table 7). However, all the preverbal clitics are to be expected, since emphasis can explain anteposition in all the instances, as can be seen in the following example (6), in which the adverb of manner sí (‘yes’, ‘this way’) triggers anteposition, as usually occurs with adverbs of manner, not only in Old Spanish (Granberg 1988, pp. 155–94), but also in Old Catalan (Bouzouita and Sentí 2022).

(6)	Sicut	superius	és escrit	sí
	as	above	write.pass.3sg	this way
	t’	ó	tenrei	(Jurament feudal 3: 71)
	you.dat	it.acc	have.fut.1sg
	‘As above it is mentioned, I will have it this way to you’

The gradual increase in SFC-a in the remaining centuries is linked to an increase in the possible environments in which SFC-a can be found. This was the case of Group 2 (Section 4.1) and also of Group 3. Merely as an illustration of this increase, four examples of innovative cases of anteposition are discussed, three after an adverbial phrase (7a–c) and the last after a subordinate clause or an absolute clause (7d). According to the literature on Old Castilian (Granberg 1988, pp. 155–94), adverbs which express a cause and temporal adverbs of chronological progression are followed by postverbal clitics. Nevertheless, from the 14th c. onwards, cases of anteposition are attested in Old Catalan after a causal adverbial phrase (7a) and anteposition is already found after a temporal adverb of chronological progression in the 13th c. (7b), although it appears in only one text, which is highly innovative for dialectal reasons, as mentioned previously (Torres-Latorre 2023). Examples of preverbal clitics after these temporal adverbs are once again found in the 15th c. (7c). With regard to innovation after subordinate and absolute clauses, the first examples of anteposition are found after an absolute clause in Old Catalan in the 15th c. (7d).

(7)

a.

et

Déus,

per

la sua

bonea,

vos

ó

and

God

because-of

his

goodness

you.dat

it.acc

aministrarà

(Muntaner, Crònica: 14v)

provide.fut.3sg

‘and God, because of his goodness, will provide it to you’

b.

Eu

derocaré

aquest

temple,

e

après

lo

I

demolish.fut.1sg

this

temple

and

then

it.acc

redificaré

(Vides: 373)

rebuild.fut.1sg

‘I will demolish this temple, and then I will rebuild it’

c.

Aprés

vos

farà

tres

figues

e

les

Then

you.dat

do.fut.3sg

three

figs

and

them.acc

vos

posarà

en

la

barba.

Aprés

vos

you.dat

put.fut.3sg

in

the

beard.

Then

you.dat

dirà

(Tirant: 848)

say.fut.3sg

‘Then he will flip you off in your face. Then he will say’

d.

E

fet

lo

dit

anniversari,

ne

And

hold.part

the

mentioned

anniversary

part

romandrà

a

la

Seu

lo

X_e.

remain.fut.3sg

in

the

Cathedral

the

10th

(Solemnitats: 357)

‘And held the mentioned anniversary, it will remain in the Cathedral the 10th’

These four examples illustrate the expansion of SFC-a to environments in which the AFC or SFC-p were previously the usual constructions. A thorough analysis of all of the occurrences of the environments in Group 3 is still necessary in order to review the role played by emphasis in the selection of a preverbal or postverbal clitic position (Torres-Latorre n.d.b). Nevertheless, the data shown in this section confirm the expected evolution towards anteposition in both quantitative and qualitative terms.

5. Conclusions

The study of the grammaticalization of the FC in Old Catalan contributes to understanding the formation of the FC in Ibero-Romance languages. Unlike Old Spanish, in which the SFC with postverbal clitics are rare, the empirical findings in our Catalan corpus attest SFC-p from an early time in the 11th c. and an important token frequency in the 13th and 14th c. (38–39% of the tokens in those syntactic environments that prevent preverbal clitics). The corpus-based data presented in this paper show a more advanced stage of the degree of grammaticalization of the FC in the 13th and the 14th c. in Old Catalan than in other peninsular languages. SFC-p appears frequently in Old Catalan and less use is made of the AFC when compared with Old Navarro-Aragonese and Old Castilian, as has been shown quantitatively. Although some morphological differences among the languages (syncope, rhizotonic verbs) have been suggested in order to partially justify this contrast, the high frequency of the univerbated FC in Old Catalan, even in examples without those morphological features, should be highlighted, since it indicates a more grammaticalized FC. As suggested by Bouzouita (2016), diatopic variation in Old Ibero-Romance could explain the differences in the distribution of the AFC and SFC-p in these languages. The results obtained for Old Catalan appear to confirm a language contact phenomenon in which Catalan (and probably Occitan) spread the more grammaticalized FC (i.e., SFC-p) to Navarro-Aragonese and Castilian (see Primerano et al. 2023).

However, our corpus is unable to confirm the expected increase in frequency of the more grammaticalized construction, i.e., SFC-p in Old Catalan. In fact, the distribution of the AFC and SFC-p with personal pronouns is stable from the 13th c. to the 16th c. This could be accounted for by another ongoing linguistic change process in diachrony that has been addressed in this paper: clitic positioning. In the 13th–14th centuries, syntactic–pragmatic restrictions governed clitic positioning, and clitics were able to occupy various positions in Old Catalan, unlike in modern and contemporary Catalan (Sentí and Bouzouita 2022). However, this syntagmatic variability disappears in the 16th c., and clitics occupy a fixed slot, mainly a preverbal position, according to the morphological features. This process, which Lehmann (1985) terms as fixation, is part of the grammaticalization of clitic pronouns. This paper outlines the general evolution, although this process should be addressed in more detail in order to, for example, understand the loss of pragmatic conditioning (Torres-Latorre n.d.b). Our empirical corpus data make it possible to draw several conclusions. The first innovative examples with preverbal clitics are attested in the 13th and 14th centuries (preverbal clitics following a vocative), which show the beginning of a change; more examples of preverbal clitics in main sentences (Group 2) are found in the 15th c. (preverbal clitics in the first position of a sentence). It appears that the fixation is almost accomplished in the 16th c., despite some examples of the AFC or SFC-p. This process explains why the SFC-p does not increase in frequency in the Middle Ages. In fact, both the AFC and SFC-p decrease in frequency from the 13th c. onwards in favour of preverbal clitics (SFC-a). The grammaticalization and coalescence of the FC become mixed with another simultaneous grammaticalization process in Old Catalan: that of clitics and the fixation of clitic placement.