Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Linguistics. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 18 February 2025

Non-Finite Verb-Forms in the Romance Languageslocked

Non-Finite Verb-Forms in the Romance Languageslocked

  • Nigel VincentNigel VincentThe University of Manchester

Summary

The distinction between finite and non-finite forms, although a standard part of traditional grammatical descriptions, is not easy to establish and is not necessarily the same from language to language, not least because of the different ways the forms in question have developed over time and the different constructions that they are part of. Nor is terminology in this domain used consistently in the literature. The present article therefore begins with a brief overview of these issues before moving on the consider the Latin patterns which lie behind modern Romance developments. These in turn will be examined form by form and construction by construction with exemplification drawn, as appropriate, from different Romance varieties. The approach blends synchrony with diachrony. Space does not permit exhaustive coverage of all forms in all languages, but nonetheless examples are drawn from across the family with a view to demonstrating significant regional and historical differences. Where relevant, cross-reference will also be made to other articles in the present collection which address in greater detail some of structures considered here.

Subjects

  • Language Families/Areas/Contact

Editor’s Note

This article was commissioned as part of the Oxford Encyclopedia of Romance Linguistics, edited by Michele Loporcaro. The Oxford Encyclopedia of Romance Linguistics is published as a collection within the Oxford Research Encyclopedia of Linguistics, edited by Mark Aronoff.

1. Introduction

Grammars of Latin and individual Romance languages, both traditional and modern, commonly take for granted a set of verbal forms called ‘non-finite’. However, this usage is not without its problems. In particular, it presupposes answers to two questions on which there is by no means a theoretical consensus, namely: How is finiteness to be defined? And how do the individual forms fit within a theory of grammatical categories?

Traditionally, finiteness has been defined in terms of either form or function, or of a mixture of both types of criterion (Nikolaeva, 2007a). On the form side, a classic definition going back to the times of the ancient grammarians is that the finite items are so-called because they are ‘finished’ by the person/number inflections which instead are absent from participles, gerunds, and infinitives. However, while this is true of Latin and many of the standard Romance languages, it does not hold for European Portuguese, old Neapolitan, and some Sardinian varieties (see Sections 3.5 and 5.5). Another criterion, which blends both form and content, is that finite items express absolute tense linked to the context of speech or writing in contrast to the relative tense expressed by the so-called present, past, and future participles (Vincent, 2011). Once again, this holds true for Latin but the system of tensed participles does not survive into Romance. It is a short step from the tense-based account to the purely functional definition that finite forms are those which have the potential to occur in independent utterances, and this is the approach adopted here albeit with a special proviso for so-called root or historic infinitives (Section 3.10). It should be noted, however, that it follows from this approach that imperatives also need to be considered as finite (pace Heine, 2016), even though in certain circumstances they are suppletively complemented by non-finite forms, as with the Italian negative singular imperative which consists of the negative marker plus the infinitive: non piangere ‘don’t cry’—lit. ‘not cry.inf’. By contrast, subjunctives, even though characterized by person and number and in some contexts tense, would then fall into the non-finite category, a conclusion which has sometimes been argued for, not least in contexts where infinitives are displaced by subjunctives as in Romanian and some southern Italian dialects, an issue to be discussed in Section 3.6. A challenge to this approach is also posed by the ‘historic’ or ‘root’ infinitive, a topic addressed in Section 3.10. The categorial question arises because of the fact that non-finite forms often share properties typically associated with different syntactic classes. Not without reason did the Latin grammarian Varro (116–27 bce) explain, quod simul habent casus et tempora quo vocantur participia “because they have both cases and tenses for that reason they are called participles” (De lingua latina 8.58). That said, the extent to which this mixing of features holds true varies not only from language to language and construction to construction but also depends on the theoretical accounts that underpin them (Lowe, 2020).

A further difficulty is posed by the way traditional labels like ‘supine’, ‘participle’, and ‘gerund’ are often used to refer to phenomena with different properties in different languages at different stages in their historical development. We will address this issue at the relevant points in our discussion in both diachronic and synchronic terms, questions to which we now turn our attention.

2. The Latin Background

As background, both empirical and terminological, the relevant Latin forms are reviewed here and the labels standardly attached to them, as displayed in Table 1 for a representative set of verbs:

Table 1 Latin Non-Finite Verb Forms

‘praise’

‘do’

‘come’

‘have’

‘be’

infinitive

laudare

facere

venire

habere

esse

present participle

laudan-s/tis

facien-s/tis

venien-s/tis

haben-s/tis

past participle

laudat-us/a/um

fact-us/a/um

vent-us/a/um

habit-us/a/um

gerund

laudand-um/i

faciend-um/i

veniend-um/i

habend-um/i

gerundive

laudand-us/a/um

faciend-us/a/um

veniend-us/a/um

habend-us/a/um

supine

laudat-um/ū

fact-um/ū

vent-um/ū

habit-um/ū

Not included in Table 1 are forms which have no continuation in Romance. These include the perfect and passive infinitives, as for example laudavisse ‘to have praised’, laudari ‘to be praised’. This is not of course to say that the relevant feature combinations cannot be expressed in the Romance languages but the realization is periphrastic rather than inflectional as for example French avoir loué and être loué. Latin also had a future participle in -uruslaudaturus, facturus, venturus, habiturus, futurus—but this only survives in occasional adjective formations such as the word for future itself (Fr. futur, Pt./Sp./It. futuro, etc.) or the Italian near synonym venturo ‘coming’.

Worthy of particular note here is the verb ‘to be’, which lacks all the non-finite forms apart from the infinitive esse (where -se is the infinitival suffix in its original non-rhotacized form), the perfect infinitive fuisse and the already mentioned future participle futurus. A present participle -sens is found only with compound verbs: praesum ‘I am here’ hence praesens ‘being here, present’ and absum ‘I am away’ hence absens ‘being away, absent’. However, with the development of the various periphrastic constructions mentioned in what follows, the ‘be’ verb needs a full paradigm of both finite and non-finite forms. This is achieved either by invoking suppletive forms such as the past participles French été and Italian stato < Lat. statum ‘stand.pst.prt’ and Spanish and Portuguese sido < late Lat. *seditum ‘sit.pst.prt’ or by new analogical formations as in Romanian fost built on the Latin f- stem. A similar strategy is to be seen in gerunds like Italian essendo and Catalan essent on the model of the Latin V+nd- pattern and parallel with forms like Ital. lodando and Cat. elogiant ‘praising’.

Evidence of mixed category status is to be seen in the participles and gerundives, which inflect for case, number, and gender and agree with their head nouns just as adjectives do. By contrast, the supine and the gerund, being nominalized verbs, have case marking appropriate to their syntactic context, although the details need not concern us here. The precise relation between the gerund and gerundive is a complex one (see Jasanoff, 2010, for discussion and references) that also need not delay us. It should be noted, however, that it is the source of conflicting Romance terminology in that French forms in -ant like étant ‘being’ and mangeant ‘eating’ are referred to in the French grammatical tradition as gérondif (< Lat. gerundivum) while elsewhere corresponding Italian, Spanish, Portuguese, Catalan and Romanian forms are labelled gerundio or gerunziu (< Lat. gerundium).

3. Infinitives

Etymologically the Latin infinitival inflection derives from the locative case of a deverbal noun in *-si, traces of which can be seen in posse ‘be able’ and esse ‘be’ (Weiss, 2020, pp. 473–475). Elsewhere, when intervocalic, -s- has rhotacized to yield the suffix -re which survives in different realizations across the majority of Romance varieties: Fr. manger, It. mangiare, Sard mandigare, Cat menjar, Sp./Port/Gal comer. Analogical changes have served to replace those forms of the Latin infinitive which did not involve a theme vowel, thus French pouvoir, vouloir, and Italian potere, volere, and cognate forms instead of Latin posse, velle. With the ‘be’ verb an alternative was to add the -re ending to the existing infinitive, hence essere, still the form in Italian, and the source of French être and Catalan (es)ser.

The situation of the infinitive in Romanian is exceptional in two respects. In the first place there are two forms, the so-called ‘long’ form with the inherited -re as in mâncarea ‘eat, eating, food’, exclusively nominal in the present-day language and feminine in gender unlike nominalized infinitives elsewhere (Section 3.8), and the short form in which -re has been elided, and which has a verbal function. Second, this short form is generally found with the prefixed particle a from Lat ad ‘to’: a mânca ‘eat’, a fi ‘be’. The question of short form infinitives is addressed in Section 3.9.

While the etymology reflects an original mixed categorial status for the infinitive, synchronically in Latin and subsequently in Romance it is arguably the prototypical non-finite form since it retains all the properties of the verb in terms of the assignment of argument structure, thus heading the clause and indeed in the case of the historic infinitive constituting the head of the main clause (Section 3.10). Even so, it is reasonable to call it non-finite since in the vast majority of contexts it cannot occur by itself but is of necessity linked to a governing item which may itself be verbal, nominal, adjectival, or prepositional. The combinations with a finite verb demonstrate various degrees of structural closeness, as explored in the following sections.

3.1 The Prolative Infinitive

In Latin when a governing verb, in both control and raising constructions, shared its subject or object argument with its complement the latter was typically a plain infinitive, sometimes called the prolative as with conari ‘try’, incipere ‘begin’, iubere ‘order’, persuadere ‘persuade’, cogere ‘force’, videri ‘seem’, and many others (Pinkster, 2015, chapter 9). By contrast, it is generally the case across the Romance languages that such infinitives in control constructions are introduced by reflexes of the Latin prepositions ad ‘to’ and de ‘(down) from’, commonly known in the recent literature as prepositional complementizers (Schulte, 2007a, for an overview) as in (1):

(1)

In origin it is likely that the contrast between a/à and de/di plus infinitive is semantically motivated, with constructions with a/à expressing purpose or intent and those with de/di related to source or reason (Schulte, 2007a, chapter 7). Traces of this source that are still evident in contrasts such as French inviter à beside décider de, where invitations are open and future oriented and decisions are completive or past oriented. Similarly, there are cross-linguistic consistencies in the semantics of verbs falling into these two classes even when the etyma differ: verbs of finishing or completing typically take de (e.g., Spanish acabar, cesar, dejar) while verbs of persuading typically take a (e.g., Spanish incitar, impulsar). That said, there can also be contrasts and overlaps: for example, Italian provare a and cercare di both mean ‘try’; French décider de sits beside the reflexive se décider à ‘make up one’s mind’ while commencer ‘begin’ allows both de and à with a difference of register rather than meaning. In short, languages vary in the detail; for instance Jones (1993, p. 260) says of Sardinian, “in many cases the choice between a and de is largely arbitrary and subject to dialectal or idiolectal variation.”

A different way in which the opposition has been grammaticalized is to be seen in Romanian where, as noted above, a has become a marker of all infinitives and is thus akin to English infinitival to, while de serves to introduce the supine (see Section 6). That said, it is still possible for a Romanian infinitive with a to be the complement of a preposition as in înainte de a veni ‘before de + a come.inf, before coming’.

3.2 Restructuring

Generally speaking infinitival complementizers like a and de serve to mark a boundary between the main and the embedded clauses and the arguments of one do not migrate to the other. Thus, in Italian (2a), where the clitic ci expressing the destination is attached to the movement verb, is grammatical but (2b), where it raises to attach to the governing verb, is not:

(2)

In the examples in (3) on the other hand this clitic climbing is indeed possible:

(3)

These examples exhibit the effect of what, following Rizzi (1978), is standardly known as restructuring and indicate a closer bond between the governing finite verb and its infinitival complement; compare Wurmbrand (2003) who argues that such infinitives must be tenseless even though she postulates tensed infinitives in other contexts. Italian is the language with which the effect is most frequently associated, hence the examples here, but there are related if not precisely parallel constructions in European Portuguese, Spanish and Catalan (Cinque, 2006, Chapter 2). The determining factor here seems to be not the presence or absence of the complementizer but the semantic class of verbs that permit it: modals (dovere ‘must’, potere ‘be able’, volere ‘want’), aspectuals (finire di ‘finish’, cominciare a ‘begin’), and deictics (andare ‘go’, venire ‘come’). Here too there are regional differences so that for example provare a ‘try’ is less readily accepted as a restructuring verb by speakers in northern Italy than it is in the south.

3.3 Complex Predicates

An even closer bond is to be found in complex predicate constructions with verbs meaning ‘make, let’ as in (4):

(4)

Here too the ‘make’ verb faire takes an embedded infinitival clause but, unlike with the restructuring pattern, any clitics associated with the embedded verb must attach to faire, hence se raser ‘shave oneself’ and se faire raser ‘have oneself shaved’. Note further that this rule holds despite the fact that modern French does not display the restructuring pattern discussed above. If the embedded verb is transitive, as in (4a), then the postverbal position is occupied by that verb’s object with its subject marked as indirect object with the à ‘to’ preposition, a distributional pattern which argues in favor of this structure being analyzed as monoclausal, unlike for example the English equivalent make read. This conclusion is reinforced by the fact that the structure cannot be reiterated *elle fait faire lire le livre aux étudiants au professeur beside English she made her professor make the students read the book.

This type of complex predicate is attested across the whole family (Sheehan, 2016), though ever since the earliest texts it has not survived in modern Romanian. The compilation of early examples in (5) is taken from Vincent (2016):

(5)

Vincent (2016) also argues that the original Latin construction with facere ‘make’ involved a full accusative and infinitive (AcI) complement and hence that there has been a diachronic shift from bi- to monoclausality. However, the modern Spanish data adduced by (Davies, 1995) attest to the emergence of a new biclausal variant as the contrasts in (6) evidence:1

(6)Modern Spanish

In the first member of the pair in (6a) the clitic attached to hicieron ‘make.pst.3pl’ is the dative le, analogous to the pattern seen in the French (4a), while the second member has the accusative la, which would be consistent with both the ‘make’ verb and the embedded verb having their own objects, as would be expected in a biclausal structure. A similar conclusion derives from the differential placing of the object clitic lo in (6b), the presence versus the absence of the reflexive se in (6c) and the different linear position of the full NP object a Pedro in (6d). Davies shows how these properties vary over time with the forms in the lefthand column being older and those on the right more recent developments; in other words a new diachronic shift from monoclausal to biclausal. This fits with the conclusion in Sheehan (2016, 2020) that there is a scale within Romance, with Italian and European Portuguese as the most conservative, while Spanish and Catalan have moved toward biclausality and with French at an intermediate stage.

A construction with verbs of seeing and hearing that bears some parallels to the complex predicates discussed here will be the topic of Section 5.3.

3.4 Periphrases

A property that the restructuring construction and complex predicates share is that the governing verb may contribute some or all of its arguments to the interpretation of the sentence as a whole. With periphrases instead the non-finite form is the source of the whole argument structure of the clause with the associated auxiliaries expressing tense, aspect, mood and voice (for a somewhat wider definition of what constitutes a periphrasis see the article by Laca, “Non-Passive Verbal Periphrases in the Romance Languages”). Typical in this respect are the various Romance periphrases to express the future. In combination with the postposed have auxiliary it is the source of the now synthetic futures such as French je manger-ai ‘I will eat’, Portuguese ver-á ‘(s)he will see’, and the like. It is also to be found as part of an analytic future in early Romanian texts as in (7) (= Nedelcu, 2016, example (18a)):

(7)

By contrast, the prevalent future construction in modern higher register Romanian follows a cross-linguistically well-attested path and consists of forms derived from the Latin want verb combined with the infinitive but without the usual infinitival particle a to be seen in (7): voi merge ‘I will go’, va merge ‘(s)he will go’. Colloquial Romanian instead has a construction with the particle o (itself a grammaticalized reflex of third singular va) plus the subjunctive: o să mergemo sbjv go.prs.1pl, we’ll go’ (Zafiu, 2013, pp. 38–41).

Also well known from the grammaticalization literature are futures built with go verbs as in French je vais manger ‘I am going to eat’. And in this respect there is a striking contrast between the development of the Romance go + infinitive, as the minimal pair in (8) demonstrates:

(8)

While the French pattern expresses future, and in particular near future, meaning, the Catalan pattern, which is also shared with older Occitan varieties and Aragonese, expresses definite past. The historical routes by which these different values have emerged are usually considered to be distinct: the French construction is a classic example of grammaticalization and semantic bleaching while the Catalan pattern has a narrative origin similar to English he went and did it (Detges, 2004). However, Paoli and Wolfe (2022) suggested a partially shared diachronic trajectory. The different regional developments of the two constructions also deserve mention. Canadian French has developed a restriction to near future meaning and a distinction between the negative and positive usage (Mooney, 2020, pp. 266–271 and references therein). Meanwhile Occitan has lost the past value and replaced it with the future meaning due to contact with French (Mooney, 2020). However, the original Occitan value is preserved in the modern variety Guardiolo still spoken in the southern Italian village of Guardia Piemontese and brought there by a community of religious exiles in the 14th century (Micali et al., 2023).

A third variant is the Spanish ir a + infinitive: va a llover ‘go.prs.3sg to rain.inf, it’s going to rain’ (Bravo, 2008). This has the same future value as the French aller construction but differs both in requiring the linking particle a and in allowing optional clitic climbing as in the synonymous pair in (9):

(9)

3.5 Inflected and Personal Infinitives

The constructions considered in the previous sections share the property that there is no overt embedded subject, the subject of the infinitive being instead co-referential with one of the arguments of the governing verb. Some Romance languages, however, also exhibit constructions in which the subject of the infinitive is overt, these in turn falling into two types according to whether the infinitive does or does not also bear person/number markers (see Mensching, 2000, Chapter 1, for a compilation of examples from across the family; Ledgeway, 2000, Chapter 4; Sitaridou, 2002). Thus, in the Logudorese variety of Sardinian there are examples such as those in (10) (Jones, 1993, pp. 278–282), with the person-marked infinitives venneres and esseret:

(10)

It is instructive to compare (10a) with the old Sicilian (11) (Bentley, 2014):

(11)

They are similar in that both contain a verb of volition controlling an infinitival clause with a non-coreferential nominative subject, but they differ in that a non-phrasal subject may precede the infinitive in Sicilian but is required to follow it in Logudorese, and the Sicilian infinitive does not bear a person/number inflection. This so-called personal infinitive is still attested in modern Sicilian, as in the adverbial clause in (12a) and the purpose clause in (12b) (both cited after Bentley, 2014):

(12)

The inflected infinitive is also characteristic of European Portuguese (Madeira & Fiéis, 2020; Sheehan, 2018) and Galician (Sheehan et al., 2020). In both the impersonal construction in (13a) and the control context in (13b) the infinitives aprovarem and terem have the same person/number suffix as the finite forms parecem and adiaram in (14a) and (14b):

(13)

14

The examples in (14) also demonstrate that the inflected infinitive does not occur in raising constructions, hence its absence in (14b) and the ungrammaticality of (14c).

With control verbs there is too a difference between exhaustive (15a) and partial (15b) control (examples from Madeira & Fiéis, 2020, p. 429), where the notation % indicates judgment differences between native speakers.

(15)

The same construction is also attested in Old Neapolitan, although it has not survived into the modern language (Loporcaro, 1986; Vincent, 1996, 1998). In (16a) the inflection on the infinitive is determined by the subject of the governing verb and in (16b) by its object:

(16)

In general, in all these languages the inflected variant is excluded from raising constructions and complex predicates and is only rarely attested with modals (Vincent, 1996, pp. 397–402). The cross-linguistic distribution of inflected infinitives thus respects the already noted scale of tightness of bonding between main verb and infinitive, however that is to be represented in theoretical terms.

3.6 Loss of Infinitives

Whereas inflected and personal infinitives are found in contexts where other languages would require a finite form, the reverse situation is attested in Romanian and some southern Italian dialects. Thus, compare the Romanian examples (17a) and (17b) with their French equivalents in (18):

(17)

18

When the subjects of the main and embedded clauses are not co-referential, as in (17b) and (18b) there is a close structural parallelism with the embedded clause being introduced by the complementizer ca/que and the verb in the subjunctive. The small difference here is that the subjunctive in Romanian is introduced by a special particle (< Lat. si ‘if’ ). There is, however, a notable contrast between (17a) and (18a) in that Romanian has a finite form also when the two subjects are coreferential. The same effect is to be seen in Salentino, spoken in the south-east of Italy (Calabrese, 1993):

(19)

Here (19a) is parallel to Romanian (17a), but with the additional property that the clitic object lu can be raised (19b) as in the restructuring constructions discussed in Section 3.2. Salentino also allows the full embedded in clause as in (19c) with the overt complementizer ku, but in that case clitic climbing is blocked, hence the ungrammaticality of (19d). Here too the distribution of the finite alternative to the infinitive obeys the same scale of structural adjacency identified above. It is noteworthy as well that the one such structure in which an embedded infinitive has survived is with the modal potere/putea ‘be able’ in both Salentino and Romanian (Dragomirescu, 2013a, pp. 194–196). It is no coincidence that this is the same context in which there are rare examples of a modal combining with an inflected infinitive in some southern Italian dialects (Loporcaro, 1986; for a fuller discussion of these regional patterns see Ledgeway, 2009).

Taken in a broader linguistic context the patterns reviewed here are typical of the Balkan Sprachbund and analogous structures are to be found in Albanian and Greek (see the article “Balkan Romance” by Dragomirescu, 2020; Friedman and Joseph, in press; Ledgeway et al., in press).

3.7 The Accusative and Infinitive

The accusative and infinitive (AcI) construction is of special interest for the way it provides evidence of a different kind of relation between Latin and Romance, not historical continuity but rather a construction which was lost from the informal spoken registers which are the usual source of linguistic change only to re-emerge as a learned borrowing in the written language. The shift is thus not just diachronic but also diamesic. In Latin the AcI played a central role in the system of verbal complementation particularly with verbs of saying and thinking as in (20) and (21) but also with verbs of volition (22) and cause (23):

(20)

(21)

(22)

(23)

There is a considerable literature on how in Romance this construction, which is built around a non-finite nucleus, has been replaced by a system of finite clauses with overt complementizers but less attention has been paid to the way parallel structures came to be used in Romance. An exception here is the study of the AcI in Castilian by Pountain (1998), from which the examples in (24) are taken:

(24)

Parallel examples are to be found elsewhere as he notes in his wider pan-Romance study (Pountain, 2011):

(25)

(26)

This pattern had its heyday in the learned prose of the 15th and 16th centuries.

Apparently similar examples are found in higher registers of modern usage, most frequently with the embedded be and other stative verbs and a postposed subject, as in the following from the 20th-century Italian novelist Elsa Morante:

(27)

In such structures, however, there is good reason to believe that the subject of the infinitive is nominative since that is the form that any replacement pronoun would take (see discussion and references in Ledgeway, 2000, pp. 295–297). An exception here, as Mensching (2017, p. 384) notes, is Occitan. He cites the following examples after Sauzet (1989) and argues that here the subject is indeed accusative:

(28)

3.8 The Nominalized Infinitive

So far this section has focused on the various uses of the infinitive in which it retains its purely verbal status. However, nominalized uses with an accompanying definite article are also attested as in the Italian saying in (29a) and in Sardinian (29b) (= Jones, 1993, p. 282, example (115)):

(29)

These structures typically show the mixed category effects discussed by Lowe (2020), as considered in detail for Italian by Vanvolsem (1983), who also explores the Latin background and the broader Romance context. Thus, (29b) is introduced by an article but the nominalized infinitive is then accompanied by the adverb semper ‘always’. Similarly, in the Spanish example (30a) the verb hacer, despite occupying the subject slot and being accompanied (optionally in the spoken language) by the definite article el, has a direct object and not the dependent genitive that one would expect of a full noun. By contrast, the Romanian (30b) exemplifies the dedicated nominal infinitive plecarea with its definite affix -a and an argument, in this instance an unaccusative subject, which is obligatorily genitive:

(30)

3.9 Second-/Short-Form Infinitives

In both Romanian and in the dialects of the region of Campania in Italy we find two forms of the infinitive: a long-form ending in the inherited suffix -re and a short- or second-form (abbreviated here as SFI) from which the -re suffix has been lost. However, both their histories and their morphosyntactic distribution are distinct. The Romanian version of the short form is already attested in the earliest texts (Nedelcu, 2016), where it has an exclusively verbal function as complement of control and raising verbs and as part of the periphrases discussed in Section 3.4, and in that sense, as already noted, has come to be the functional equivalent of the -r(e) forms elsewhere in Romance. The Campanian SFI, by contrast, emerges in the 19th century as the complement to the imperative of deictic verbs and in due course comes to be reanalyzed as an imperative in Neapolitan (Ledgeway, 1997). An intermediate stage in this development is attested in the Valle Caudina (Groothuis & De Sisto, 2024), where the shift to imperative status has still not happened and where the long and short forms of the infinitive are in alternation as they were in earlier stages of Neapolitan and unlike the relation between the two forms in Romanian.

3.10 Root or Historic Infinitives

This section concludes with a phenomenon which challenges the criterion that independent use is the defining property of non-finite forms (Nikolaeva, 2007b). In (31) the infinitive is the only verb in a simple main clause and the subject argument is in the nominative:

(31)

This usage is traditionally labeled the historic infinitive (Pinkster, 2015, pp. 527–531; Rosén, 1995) and more recently the root or narrative infinitive (Mensching, 2017, p. 374). It is found in Latin prose from the earliest writers onward in both declaratives (31a) and interrogatives (31b). It is also attested in narrative work in Romance up to the present day, but usually with an overt subject and introduced by one of the infinitival particles as in (32) (cited after Eloundou Eloundou, 2022):

(32)

A borderline case is the Italian (33), where Maiden and Robustelli (2007, p. 397) suggest the infinitive is embedded as a complement of the exclamative ecco rather than being the head of the main clause:

(33)

4. Past Participles

In Latin the forms called past or perfect participle (terminology varies) are inherently passive as is evident from their use in the ablative absolute construction:

(34)

It is a clear instance of a mixed category expressing the typically verbal features of tense and voice but at the same time agreeing with the noun to which it is attached in gender and number. Formally it comes in two variants traditionally labeled regular and irregular or, more recently, long and short.2 The regular/long forms have the template root + theme vowel + t + case/number/gender as in (34a) depugn-a-t-o, while the irregular/short forms retain the c/n/g suffix but have undergone stem contraction as in (34b) accep-t-is (from accipere). In the development into Romance the structure of the regular forms is still evident in pairs such as Spanish pensado ‘thought’ vs. sentido ‘felt’ or Italian cantato ‘sung’ vs. uscito ‘gone out’ while the irregular forms have undergone numerous analogical and other changes—for a thorough and detailed overview see Laurent (2000). In some languages, however, the regular/irregular split has developed into a more systematic structural opposition, as will be explored in Section 4.3 after the major morphosyntactic contexts in which the past participle can appear have been reviewed.

4.1 Absolute Constructions

The label ‘absolute’ has been retained in some grammars to refer to structures of the type exemplified in (35):

(35)

The the syntax of these constructions will not be considered in great detail (see Gunnarson, 1994; Loporcaro, 2003, for some discussion), but it should be noted that the adjectival agreement properties are retained as is the relative tense value and that for the most part they have the patient-subject feature shared by passives (35b,c) and unaccusatives (35a); at the same time instances like (35d) are also attested (Maiden & Robustelli, 2007, pp. 311–312). While the use as complement of a preposition as in (35c) is frequent and unmarked, the adverbial type in (35a,b) has clear parallels with Latin examples like (34), hence their being labeled ‘absolute’ even though traditionally that term is restricted to structures that are not co-referential with the subject of the main clause unlike in (35a,d). The pattern tends to be restricted to higher and written registers and has been considered a learned borrowing rather than a direct inheritance, in that sense akin to the AcI (Pountain, 2011, pp. 650–651). See also the forthcoming article in this encyclopedia “Participial Clauses in the Romance Languages” by Soare.

4.2 Periphrases

Undoubtedly the most frequent role played by the past participle is as part of the periphrases with the ‘have’ and ‘be’ auxiliaries which are characteristic of all Romance languages (see the chapters by Ledgeway “Passive Periphrases in the Romance Languages” and Schaden “Perfects in the Romance Languages”). This in turn has necessitated the creation of past participles not only, as already noted, for the ‘be’ verb but also for others which did not have one in Latin such as posse ‘be able’ and velle ‘want’. In all these constructions too the agreement properties are preserved, though with differences in the contexts in which they are triggered. Thus, while French and Italian share the property of agreement between participle and clitic objects which is not found in Spanish and Romanian, they differ when it comes to agreement with the head of a relative clause: Ital. la casa che ho visto/*a vs. Fr. la maison que j’ai vue/*vu. Beside these pan-Romance patterns there are also individual local developments such as the Italian passive constructions with ‘go’ plus past participle exemplified in (36) (Mocciaro, 2014):

(36)

A further difference concerns the availability of compound periphrases. Thus, (36a) is acceptable but this construction is only found with verbs with a negative sense such as ‘destroy’ or ‘lose’ but not with ‘build’ or ‘find’; (36b) by contrast is not found with a compound auxiliary. These differences also vary from language to language. For instance, Italian, beside the be passive, also has a come passive, but only in the non-compound forms, whereas come is the only available passive auxiliary in some Raeto-Romance varieties and then the compound form is admitted (Vincent, 2014, pp. 18–19). Finally, given our discussion above of the possibility of a diachronic trajectory biclausal > monoclausal > biclausal in the case of complex predicates, mention should be made of the Lombard dialect Borgomanerese, described in Tortora (2014), in which clitics attach not only to causatives as in (37a) but also to past participles in periphrases as in (37b,c):

(37)

In the light of examples such as these Tortora (2014, p. 106) states, “I will argue that the facts exhibited strongly suggest a bi-clausal analysis of the compound tenses.” This pattern is also attested in many adjacent Piedmontese varieties.

4.3 Past Participle as Independent Adjective

Both present and past participles can develop into freestanding adjectives: compare French absent vs. présent and Italian aperto ‘open’ vs. chiuso ‘shut’ as applied to shops or doors. The difference here is that the Italian forms are also the past participles of the relevant verbs aprire ‘to open’ vs. chiudere ‘to close’. This raises the question of how to tell whether the adjective forms are independent lexical items or still in some way linked to the verb (Schwarze, 2017). In this connection, Bentley (2018, p. 286) gives the following example, where there is a contrast between the resulting state—asciutto ‘dry’—and the activity involving the verb asciugare ‘to dry’:

(38)

In fact, she glosses asciutto as ‘dry.ptcp.msg’ since on her analysis derived adjectives of this kind retain their morphologically derived status even in the modern language.

In Italian there are sporadic pairs of this kind but in some languages there is a systematic pattern. Thus, in Portuguese there are the contrasts exemplified in (39), where the regular (thematic, arhizotonic) forms are traditionally labeled ‘weak’ and the irregular ones ‘strong’ (Loporcaro et al., 2004):

(39)

There are similar pairs in Sicilian (Bentley, 2018; Bentley & Ledgeway, 2014, 2015):

(40)

There is not space here to go into the theoretical accounts that have been advanced to deal with these patterns but in essence they are built on a distinction between verbal and predicative bases, and contrast the resultant state with the activity expressed by the verb. In Portuguese a key test is the distribution with two copulas ser and estar, where the former is the passive auxiliary and the latter co-occurs with both participial formations and adjectives to express the state of the argument in question, typically the subject. Since the contrast in (39) is lexically determined, some verbs have only one participle, either weak or strong, to do both jobs, but crucially even when a verb like those in (39) does have two participles, no verb uses the strong form with both ser and estar and the weak form in active periphrastic perfects. There is thus a clear distinction between verbal and predicative uses.

That change can continue is to be seen in the discussion of the subsequent development of the pattern in Brazilian Portuguese by Schwenter et al. (2019). They note the parallelism between the strong participles and the first-person singular: pago can either mean ‘I pay’ or be the strong participle of pagar ‘to pay’ for which the weak participle is pagado as in tenho pagado ‘I have been paying’. Their statistical study goes on to show the effect of analogical developments based on this parallelism which overrides contrasts based on syntactic context and leads to forms such as chego and trago instead of chegado ‘arrived’ and trazido ‘brought’ regardless of whether the first-person singular form is itself regular, as with chego beside the infinitive chegar, or irregular as with trago and infinitive trazer.

In addition, some southern Italian varieties allow participles, called ‘resultative-stative’ by Bentley and Ledgeway (2014), as in (41) (= their 1):

(41)

Once again, albeit in a different structural context, there is an overlap between a form which has the semantic properties of the verb to which it is related and the distribution appropriate to an adjective.

5. Present Participle/Gerund

As seen in Section 2, the Latin present participle in -ns/-ntis inflects as an adjective and in general in Romance only survives in lexicalized adjectives or nouns such as French plaisant ‘pleasant’, charmant ‘charming’, déterminant ‘determiner’, or Italian convincente ‘convincing’, amante ‘lover’, dirigente ‘manager’ (though (59a) is an instance of the occasional verbal use in early texts). Its use to mark concurrent activity has instead been taken over by the gerund. This is not formally evident in French where the two endings have converged for verbs of all classes in -ant but elsewhere there are at least two distinct suffixes according to the conjugation of the verb as for example Spanish pensando ‘thinking’, comiendo ‘eating’, Occitan anant ‘going’, disent ‘saying’, Romanian lucrȃnd ‘working’, fugind ‘running’. In origin the ablative in -ndo of the Latin gerund expressed cause or means as in (42):

(42)

However, over time this meaning was bleached out and the value came to be simply one of an accompanying activity. The example in (43) is instructive in this respect:

(43)

This is the Latin text which introduces a witness’s statement; there are two present participles tenens and tangens followed by the gerund testificando, but it is hard to identify the latter as having a different function from the former. That the statement is itself in vernacular and constitutes the earliest attestation of an Italian variety is indicative of the stage and direction of development.

5.1 Gerund as Adverbial

The principal function of the gerund in modern Romance is adverbial; it serves to express an action or a state of affairs that relates to the action or state expressed in the main clause. As can be seen in the examples in (44), there is considerable variety in the type of adverbial meaning (see Ramat & Da Milano, 2011, from where the following examples are taken, for a comprehensive overview of both the cross-linguistic patterns and the relevant literature):

(44)

In these examples there is a range of different adverbial functions: means in (44a), concessive in (44b), cause in (44c), manner or accompanying activity in (44d). Such semantic diversity is to be found right across the family since the core meaning of the gerund is an activity which accompanies or precedes another with the precise connection between the two being determined by context. There is too a parallelism with the adverbial use of infinitives in combination with prepositions (Schulte, 2007b).

Beside the patterns in (44) there are also cases with a compound gerund as in (45), the Italian example in (45a) showing co-reference with the subject of the main clause while the French (45b) has an independent subject within the gerundival clause. In general, structures of both kinds are grammatical in all the Romance languages, except Romanian.

(45)

At first glance examples such as these might seem a natural non-finite concomitant of the Romance periphrastic finite perfect, but as Menoni (1982) shows, their historical profile is different and they are first attested in medieval translations of Latin texts where they seem to have been coined as a way of rendering the classical ablative absolute, as in her example (46), in which (46a) is the original text from Valerius Maximus (4.6.1) and (46b) is Boccaccio’s 14th-century rendering of it:

(46)

Perhaps not surprisingly, to this day this construction tends to be restricted to the higher registers.

5.2 Co-occurrence of Gerunds and Prepositions

The co-occurrence of gerunds and prepositions is of interest for both descriptive and theoretical reasons. At one extreme lies modern French where the preposition en seen in (44d) necessarily accompanies the gerund (gérondif) in all its uses (Halmøy, 1982). Although this is not so systematic in the other languages, parallel examples are attested such as Italian (47a). Old Italian also has a ‘to’ and di ‘of’ with a gerund (47b,c)—examples from De Roberto (2013):

(47)

Other instances of the combination preposition + gerund are:

(48)

(49)

Examples such as these tend to undermine the proposal by Gallego (2010) that gerunds contain an inherent but unexpressed prepositional head, a conclusion that he argues for precisely on the grounds that they do not occur with overt prepositions.

A more general theoretical question is whether the fact that a gerund can be the complement of a preposition means that it has nominal as well as verbal properties, as is standardly argued for the English gerund in contexts such as after eating breakfast or without doing any work. Since in other respects the gerunds in Romance do not display nominal features, unlike their Latin ancestors, it seems preferable to say simply that they are non-finite verbal complements akin to infinitival uses such as French avant de partir or Italian prima di partire ‘lit. before of leave.inf = before leaving’.

5.3 Gerunds Versus Infinitives

One context in which gerunds and infinitives can both be found, although with significant differences from language to language, is as the complement of perception verbs as in the Spanish pair in (50a,b), with a difference in meaning very close to that derivable from the English translations and where a third alternative is the type in (50c) with a complementizer and a finite verb commonly referred to as a pseudo-relative:

(50)

Casalicchio (2013, 2016) describes and contrasts a similar pattern in the Ladin variety Gardenese. In particular, Gardenese has the gerund and the pseudo-relative as in (51a,b) and also allows the construction with impersonal verbs, something which is ungrammatical in Spanish, hence the contrast between (52a) and (52b):

(51)

(52)

And one can compare here the Italian sento piovere ‘I hear it rain.inf’ vs. *sento piovendo ‘I hear rain.ger’. The pattern is also found in Romanian:

(53)

As Nicula (2013) notes, an example like (53) contrasts with the Italian (54) since in Romanian the gerund links to the object while in Italian, a language which does not have the perception verb + gerund construction, it necessarily refers back to the subject (Maiden & Robustelli, 2007, p. 310):

(54)

Once again space forbids discussion of the details of the theoretical proposals designed to account for these patterns, as for example by Casalicchio and Herbeck (2024), but the similarities and differences across the whole family are instructive when it comes to understanding how related but not identical patterns can develop.

5.4 Periphrases

Given that the gerund in its adverbial and absolute uses frequently expresses ongoing and accompanying activity it is perhaps not surprising that it also forms part of a range of periphrastic formations in combination with verbs of location and movement as well as with the simple copula be verb which express progressive or continuing events (Bertinetto, 2000). To take the last first, compare the examples in (55):

(55)

Sardinian has retained this pattern to the present day, French has lost it entirely and modern Romanian only has the presumptive pattern seen in (56):

(56)

More common among the modern languages is a similar construction built with the etymological stand verb stare/istare/estar, as in Italian stavo lavorando, Sardinian istaio travallande, regional and Brazilian Portuguese estava trabalhando ‘I was working’. A notable difference between Sardinian and the other languages which have this construction is that the compound form so istatu travallande ‘I have been working’ is also available, a possibility that does not exist elsewhere. However, the fact that, in contrast for example to (55a), here the clitic must attach to the gerund and cannot precede the finite form, and hence the difference in grammaticality between (57a) and (57b), suggests that there is a different morphosyntactic structure at work here (Jones, 1993, p. 141; Remberger, 2006, pp. 272–281):

(57)

The third type of periphrasis involves deictic support verbs as in the Italian examples in (58), where the parentheses around the clitic reflexive si indicate that it may occur either before the support verb or attached to the gerund, and where (58c) shows that with both andare ‘go’ and venire ‘come’ the compound forms are acceptable.

(58)

It is difficult to render in the English translations the way the difference in the choice of ‘come’ versus ‘go’ in these examples reflects the different perspectives of speakers or participants, as is to be expected given that these are items whose core lexical content is deictic. Similar patterns are also found in Catalan, Spanish, and Portuguese although there are subtle cross-linguistic differences in the use of these constructions which space does not permit us to consider here (see Bertinetto, 2000; Squartini, 1998, for thorough discussion and exemplification and further references). It should be emphasized, however, that with these constructions both compound and non-compound forms of the support verb are available across the board. And here too is a construction that is well attested from the earliest stages of all the languages in question (see, e.g., for Italian Giuliani, 2012). It was also available in old and middle French but in that instance has not survived into the modern language (Gougenheim, 1929, ch. 1).

5.5 Inflected Gerunds and Participles

As already discussed, some Romance varieties have extended the use of person/number markers to infinitives but Old Neapolitan is distinctive for the way these inflections also appear optionally with gerunds and participles as the following examples demonstrate:

(59)

Note in particular the fact that in (59c) the gerund portandomo bears the first-person plural inflection -mo, as does the finite verb affaticamo, but the other gerund caminando does not. Inflected gerunds akin to (59c) have also been reported for some dialects of European Portuguese and Galician. The generalization that emerges here is that there is an implicational hierarchy: person/number markers are only found on non-finite forms if they are also on finite forms, and they are only to be found on participles or gerunds if they are also found on infinitives.

6. Supine

There is in Latin a verbal noun traditionally called ‘supine’. The only domain in Romance where this term continues to be used is in Daco- and Istro-Romanian, although it is a matter of debate whether the form in question is a continuation of Latin or an independent innovation to which the traditional label has been reapplied (see Pană Dindelegan, 2013, sec. 4.4, for references to the relevant literature and a more detailed overview of the constructions involved than space allows here and Dragomirescu, 2013b, for an exploration of the wider Romance context).3 From the perspective of form the Romanian supine is identical to the masculine singular of the participle, while in function it has two distinct uses—the so-called nominal supine, which takes the typically Romanian suffixed definite article, can be governed by a preposition and assigns genitive case to its nominal arguments, and the verbal supine whose arguments are marked in the same way as they would be if associated with a finite verb. Both uses of the supine are introduced by their own dedicated complementizer de. Thus compare (= examples (101a,b) from Pană Dindelegan, 2013, p. 235):

(60)

While the meaning is identical, the syntactic structures are distinct with the nominal use adopting nominal syntax and the verbal use verbal syntax. Neither use would therefore count as a mixed category in the sense of Lowe (2020).

From a historical perspective (Dragomirescu, 2016), the nominal supine is older, having been attested since the earliest 16th-century texts, and has been argued to be one of the Romance reflexes of the Latin ‘third stem’ (Maiden, 2013). The verbal pattern, on the other hand, does not have parallels in Latin usage and is a Romanian innovation which emerges at the end of the 17th century and is the more frequent in the modern language. There are close parallels with a similar formation in Albanian suggesting that here too there is evidence of the effects of the Balkan Sprachbund.

7. Conclusion

In conclusion, then, it seems reasonable to accept and adopt the view expressed by Lowe (2019, p. 324) that finite verb forms constitute a prototype with the non-prototypical features of non-finite forms varying from language to language both across different language families and, as seen here, within one family. That said, the focus in the present article has been on the empirical side of the question and on the wide range of morphosyntactic variation to be found within Romance. Notable too is the fact that such variation is not simply diatopic and diachronic but also diamesic and diastratic.

Acknowledgments

Many thanks for their comments and suggestions to Adam Ledgeway, Martin Maiden, and the anonymous reviewers (one of whom subsequently revealed himself to be Giampaolo Salvi). Special thanks to Adina Dragomirescu both for her specific help with the Romanian material and for her general advice on the organization of the article.

Further Reading

A natural place to start is with the other ORE articles that are linked in the present article, which develop in more detail some of the specific themes addressed here, as do several of the chapters in Dufter and Stark (2017). For the broader Romance perspective, there are two valuable general handbooks: Ledgeway and Maiden (2016, 2022). The first provides a descriptive overview of the languages and constructions while the second aims to embed Romance data in a wider theoretical and typological context.

References

  • Bentley, D. (2014). The personal infinitive in Sicilian. In P. Benincà, A. Ledgeway, & N. Vincent (Eds.), Diachrony and dialects (pp. 25–40). Oxford, UK: Oxford University Press.
  • Bentley, D. (2018). Monotonicity in word formation: The case of Italo-Romance result state adjectives. Transactions of the Philological Society, 116, 285–319.
  • Bentley, D., & Ledgeway, A. (2014). Manciati siti? Les constructions moyennes avec les participes résultatifs-statifs dans l’italien et les variétés italo-romanes méridionales. Langages, 194, 63–80.
  • Bentley, D., & Ledgeway, A. (2015). Autour de la question des participes résultatifs-statifs dans les variétés romanes. In I. Mirto (Ed.), Le relazioni irresistibili (pp. 61–91). Pisa, Italy: Edizioni ETS.
  • Bertinetto, P. M. (2000). The progressive in Romance, as compared with English. In Ö. Dahl (Ed.), Tense and aspect in the languages of Europe (pp. 559–604). Berlin, Germany: Mouton de Gruyter.
  • Bravo, A. M. (2008). <Ir a + infinitivo> y los tiempos compuestos: Semejanzas y diferencias; La prospectividad y el paradigma temporal y aspectual del español. In A. Carrasco Gutiérrez (Ed.), Tempos compuestos y formas verbales complejas (pp. 403–442). Madrid, Spain: Vervuert.
  • Buridant, C. (2000). Grammaire nouvelle de l'ancien français. Paris, France: SEDES.
  • Calabrese, A. (1993). The sentential complementation of Salentino: A study of a language without infinitival clauses. In A. Belletti (Ed.), Syntactic theory and the dialects of Italy (pp. 28–98). Turin, Italy: Rosenberg & Sellier.
  • Casalicchio, J. (2013). Pseudorelative, infiniti e gerundi nelle varietà romanze: Affinità (solo) superficiali e corrispondenze strutturali. Munich, Germany: Lincom.
  • Casalicchio, J. (2016). The use of gerunds and infinitives in perceptive constructions: The effects of threefold parametric variation in some Romance varieties. In E. Bidese, F. Cognola, & M. C. Moroni (Eds.), Theoretical approaches to linguistic variation (pp. 53–87). Amsterdam, The Netherlands: John Benjamins.
  • Casalicchio, J., & Herbeck, P. (2024). Pseudorelative clauses, infinitives, and gerunds with Spanish perception verbs: A comparative view. Syntax. doi:10.1111/synt.12271
  • Cinque, G. (2006). Restructuring and functional heads. Oxford, UK: Oxford University Press.
  • Davies, M. E. (1995). The evolution of the Spanish causative construction. Hispanic Review, 63, 57–77.
  • De Roberto, E. (2013). Usi concorrenziali di infinito e gerundio in italiano antico. In E. Casanova Herrero & C. Calvo Rigual (Eds.), Actas del XXVI Congreso Internacional de Lingüística y de Filología Romániques (pp. 878–888). Berlin, Germany: De Gruyter.
  • Detges, U. (2004). How cognitive is grammaticalization? The history of the Catalan perfet perifràstic. In O. Fischer, M. Norde, & H. Peridon (Eds.), Up and down the cline: The nature of grammaticalization (pp. 211–227). Amsterdam, The Netherlands: John Benjamins.
  • Dragomirescu, A. (2013a). Complex predicates. In G. Pană Dindelegan (Ed.), The grammar of Romanian (pp. 191–201). Oxford, UK: Oxford University Press.
  • Dragomirescu, A. (2013b). Particularități sintactice ale limbii române în context romanic: Supinul. Bucharest, Romania: Editura Muzeului Național al Literaturii Române.
  • Dragomirescu, A. (2016). The supine and the supine construction. In G. Pană Dindelegan (Ed.), The syntax of old Romanian (pp. 249–259). Oxford, UK: Oxford University Press.
  • Dragomirescu, A. (2020). Balkan-Romance. Oxford, UK: Oxford University Press.
  • Dufter, A., & Stark, E. (Eds.). (2017). Manual of Romance morphosyntax and syntax. Berlin, Germany: Mouton De Gruyter.
  • Eloundou Eloundou, G. (2022). Sintaxis y traducción del infinitivo de narración en francés y en español. Estudios de Lingüística Aplicada, 74, 89–111.
  • Friedman, V., & Joseph, B. (in press). The Balkan languages. Cambridge, UK: Cambridge University Press.
  • Gallego, Á. J. (2010). On the prepositional nature of non-finite verbs. Catalan Journal of Linguistics, 9, 79–102.
  • Giuliani, M. (2012). L’immagine linguistica del movimento nello spazio e nel tempo: In margine alla voce andare del TLIO. In F. Benozzo et al. (Eds.), Culture, livelli di cultura e ambienti nel Medioevo occidentale (pp. 533–551). Rome, Italy: ARACNE editrice.
  • Gougenheim, G. (1929). Étude sur les périphrases verbales de la langue française. Paris, France: Librairie A.-G. Nizet.
  • Groothuis, K., & De Sisto, M. (2024). Microvariation in the second form of the infinitive in Campania: The case of the Valle Caudina. Isogloss: Open Journal of Romance Linguistics, 10(3/6), 1–22.
  • Gunnarson, K.-Å. (1994). Small clauses and absolute constructions in Spanish. Probus, 7, 125–171.
  • Halmøy, J.-O. (1982). Le gérondif: Eléments pour une description syntaxique et sémantique. Trondheim, Norway: University of Trondheim Press.
  • Heine, B. (2016). On non-finiteness and canonical imperatives. In C. Chamoreau & Z. Estrada-Fernández (Eds.), Finiteness and nominalization (pp. 243–268). Amsterdam, The Netherlands: John Benjamins.
  • Jasanoff, J. H. (2010). The origin of the Latin gerund and gerundive: A new proposal. In H. Goldblatt & N. Shields Kollman (Eds.), Rus´ writ large: Languages, histories, cultures; Essays presented in honor of Michael S. Flier on his sixty-fifth birthday (pp. 195–208). Cambridge, MA: Harvard Ukrainian Research Institute.
  • Jones, M. A. (1993). Sardinian syntax. London, UK: Routledge.
  • Laurent, R. (2000). Past participles from Latin to Romance. Berkeley, CA: University of California Press.
  • Ledgeway, A. (1997). Asyndetic complementation in Neapolitan dialect. The Italianist, 17, 231–273.
  • Ledgeway, A. (2000). A comparative syntax of the dialects of Southern Italy: A Minimalist approach (Publications of the Philological Society 33). Oxford, UK: Blackwells.
  • Ledgeway, A. (2009). Diachrony and finiteness: Subordination in the dialects of southern Italy. In Nikolaeva (Ed.), Finiteness: Theoretical and empirical foundations (pp. 335–365). Oxford, UK: Oxford University Press.
  • Ledgeway, A., & Maiden, M. (Eds.). (2016). The Oxford guide to the Romance languages. Oxford, UK: Oxford University Press.
  • Ledgeway, A., & Maiden M. (Eds.). (2022). The Cambridge handbook of Romance linguistics. Cambridge, UK: Cambridge University Press.
  • Ledgeway, A., Schifano, N., & Silvestri, G. (in press). Greek-Romance contact: The fading voices of southern Italy. Oxford, UK: Oxford University Press.
  • Loporcaro, M. (1986). L’infinito coniugato nell’Italia centro-meridionale: Ipotesi genetica e ricostruzione storica. L’Italia dialettale, XLIX, 173–240.
  • Loporcaro, M. (2003). The unaccusative hypothesis and participial absolutes in Italian: Perlmutter’s generalization revised. Rivista di Linguistica [Italian Journal of Linguistics], 15(2), 199–263.
  • Loporcaro, M., Pescia, L., & Ramos, A. M. (2004). Costrutti dipendenti participiali e participi doppi in portoghese. Revue de linguistique romane, 68, 15–46.
  • Lowe, J. (2019). The syntax and semantics of nonfinite forms. Annual Review of Linguistics, 5, 309–328.
  • Lowe, J. (2020). Mixed projections and syntactic categories. Journal of Linguistics, 56, 315–357.
  • Madeira, A., & Fiéis, A. (2020). Inflected infinitives in Portuguese. In L. Franco & P. Lorusso (Eds.), Linguistic variation: Structure and interpretation (pp. 423–437). Berlin, Germany: De Gruyter Mouton.
  • Maiden, M. (2013). The Latin “third stem” and its Romance descendants. Diachronica, 30, 492–530.
  • Maiden, M., & Robustelli, C. (2007). A reference grammar of modern Italian (2nd ed.). London, UK: Routledge.
  • Menoni, V. (1982). Formazione e storia del gerundio composto nell’italiano antico. Studi di Grammatica Italiana, 11, 5–88.
  • Mensching, G. (2000). Infinitive constructions with specified subjects: A syntactic analysis of the Romance languages. Oxford, UK: Oxford University Press.
  • Mensching, G. (2017). Infinitival clauses. In A. Dufter & E. Stark (Eds.), Manual of Romance morphosyntax and syntax (pp. 370–396). Berlin, Germany: Walter De Gruyter.
  • Micali, I., Vincent, N., & Wheeler, M. W. (2023). Contatto interlinguistico: Il perfetto perifrastico anar + infinito. Bolletino dell’Atlante Linguistico Italiano, 47, 79–108.
  • Mocciaro, E. (2014). Passive in motion: The Early Italian and are (‘to go’). In M. Devos & J. van der Wal (Eds.), COME and GO off the beaten grammaticalization path (pp. 45–68). Berlin, Germany: De Gruyter.
  • Mooney, D. (2020). Future temporal reference in French and Gascon: Aller/anar + infinitive periphrasis and structural transfer in the bilingual grammar. In S. Wolfe & M. Maiden (Eds.), Variation and change in Gallo-Romance grammar (pp. 258–278). Oxford, UK: Oxford University Press.
  • Nedelcu, I. (2016). The infinitive and the infinitival construction. In G. Panǎ Dindelegan (Ed.), The syntax of old Romanian (pp. 232–248). Oxford, UK: Oxford University Press.
  • Nicula, I. (2013). The gerund. In G. Panǎ Dindelegan (Ed.), The grammar of Romanian (pp. 245–253). Oxford, UK: Oxford University Press.
  • Niculescu, D. (2016). The gerund and the gerundial construction. In G. Panǎ Dindelegan (Ed.), The syntax of old Romanian (pp. 271–287). Oxford, UK: Oxford University Presss.
  • Nielsen, P. J. (2016). Functional structure in morphology and the case of nonfinite verbs. Leiden, The Netherlands: Brill.
  • Nikolaeva, I. (2007a). Introduction. In I. Nikolaeva (Ed.), Finiteness: Theoretical and empirical foundations (pp. 1–19). Oxford, UK: Oxford University Press.
  • Nikolaeva, I. (2007b). Constructional economy and nonfinite independent clauses. In I. Nikolaeva (Ed.), Finiteness: Theoretical and empirical foundations (pp. 138–180). Oxford, UK: Oxford University Press.
  • Pană Dindelegan, G. (Ed.). (2013). The grammar of Romanian. Oxford, UK: Oxford University Press.
  • Paoli, S., & Wolfe, S. (2022). The go-future and go-past periphrases in Gallo-Romance: A comparative investigation. In A. Ledgeway, J. C. Smith, & N. Vincent (Eds.), Periphrasis and inflexion in diachrony: A view from Romance (pp. 123–144). Oxford, UK: Oxford University Press.
  • Pinkster, H. (2015). The Oxford Latin syntax: Vol. 1. The simple clause. Oxford, UK: Oxford University Press.
  • Pountain, C. (1998). Learnèd syntax and the Romance languages: The ‘accusative and infinitive’ construction with declarative verbs in Castilian. Transactions of the Philological Society, 96, 159–201.
  • Pountain, C. (2011). Latin and the structure of written Romance. In M. Maiden, J. C. Smith, & A. Ledgeway (Eds.), The Cambridge history of the Romance languages: Vol. I. Structures (pp. 609–659). Cambridge, UK: Cambridge University Press.
  • Ramat, P., & Da Milano, F. (2011). Differenti usi di gerundi e forme affini nelle lingue romanze. Vox romanica, 70, 1–46.
  • Remberger, E.-M. (2006). Hilfsverben: Eine minimalistiche Analyse am Beispiel des Italienischen und Sardischen. Tübingen, Germany: Niemeyer.
  • Rizzi, L. (1978). A restructuring rule in Italian syntax. In S. J. Keyser (Ed.), Recent transformational studies in European languages (pp. 113–158). Cambridge, MA: MIT Press.
  • Rosén, H. (1995). The Latin infinitivus historicus revisited. Mnemosyne, 48, 536–564.
  • Sauzet, P. (1989). Topicalisation et prolepse en occitan. Revue des langues romanes, 93, 235–273.
  • Schulte, K. (2007a). Prepositional infinitives in Romance: A usage-based approach to syntactic change. Munich, Germany: Peter Lang.
  • Schulte, K. (2007b). What causes adverbial infinitives to spread? Evidence from Romance. Language Sciences, 29, 512–537.
  • Schwarze, C. (2017). A proposito delle restrizioni sulla conversione di participi in aggettivi. In R. D’Alessando, G. Iannàccaro, D. Passino, & A. Thornton (Eds.), Di tutti i colori: Studi per Maria Grossmann (pp. 301–317). HAL Open Science Archive.
  • Schwenter, S., Hoff, M., Christodulelis, E., Pflum, C., & Dauphinais, A. (2019). Variable past participles in Portuguese perfect constructions. Language Variation and Change, 31, 69–89.
  • Sheehan, M. (2016). Complex predicates. In A. Ledgeway & M. Maiden (Eds.), The Oxford guide to the Romance languages (pp. 981–993). Oxford, UK: Oxford University Press.
  • Sheehan, M. (2018). Control of inflected infinitives in European Portuguese. In A.-L. Santons & A. Gonçalves (Eds.), Complement clauses in Portuguese: Syntax and acquisition (pp. 29–58). Amsterdam, The Netherlands: John Benjamins.
  • Sheehan, M. (2020). The development of exceptional case marking in Romance with a particular focus on French. Probus, 32, 367–400.
  • Sheehan, M., Blokzijl, J., & Parafita Couto, M. C. (2020). Inflected infinitives in Galician. In I. Vogel (Ed.), Romance languages and linguistic theory 16: Selected papers from the 47th Linguistic Symposium on Romance Languages (LSRL), Newark, Delaware (pp. 260–273). Amsterdam, The Netherlands: John Benjamins.
  • Sitaridou, I. (2002). The synchrony and diachrony of Romance infinitives with nominative subjects [PhD dissertation, University of Manchester].
  • Squartini, M. (1998). Verbal periphrases in Romance: Aspect, actionality and grammaticalization. Berlin, Germany: Mouton de Gruyter.
  • Tortora, C. (2014). A comparative grammar of Borgomanerese. Oxford, UK: Oxford University Press.
  • Vanvolsem, S. (1983). L’infinito sostantivato in italiano. Florence, Italy: Accademia della Crusca.
  • Vincent, N. (1996). Appunti sulla sintassi dell’infinito coniugato in un testo napoletano del ’300. In P. Benincà, G. Cinque, T. De Mauro, & N. Vincent (Eds.), Italiano e dialetti nel tempo (pp. 387–406). Rome, Italy: Bulzoni.
  • Vincent, N. (1998). On the grammar of inflected non-finite forms (with special reference to Old Neapolitan). Copenhagen Studies in Language, 22, 135–158.
  • Vincent, N. (2011). Non-finite forms, periphrases and autonomous morphology in Latin and Romance. In M. Maiden, J. C. Smith, M. Goldbach, & M.-O. Hinzelin (Eds.), Morphological autonomy: Perspectives from Romance inflectional morphology (pp. 417–435). Oxford, UK: Oxford University Press.
  • Vincent, N. (2014). Similarity and diversity in the evolution of Italo-Romance morphosyntax. In P. Benincà, A. Ledgeway, & N. Vincent (Eds.), Diachrony and dialects: Grammatical change in the dialects of Italy (pp. 1–21). Oxford, UK: Oxford University Press.
  • Vincent, N. (2016). Causatives in Latin and Romance. In J. N. Adams & N. Vincent (Eds.), Early and late Latin: Continuity or change? (pp. 294–312). Cambridge, UK: Cambridge University Press.
  • Weiss, M. (2020). Outline of the historical and comparative grammar of Latin (2nd ed.). Ann Arbor, MI: Beech Stave Press.
  • Wurmbrand, S. (2003). Infinitives: Restructuring and clause structure. Berlin, Germany: Mouton De Gruyter.
  • Zafiu, R. (2013). Mood, tense, and aspect. In G. Panǎ Dindelegan (Ed.), The grammar of Romanian (pp. 24–65). Oxford, UK: Oxford University Press.

Notes

  • 1. In his study Davies considers a wider range of verbs than causative hacer, including in particular dejar ‘let’ and ver ‘see’. For simplicity, and because it does not affect the overall argument, I have restricted attention to hacer and have accordingly modified some of his modern Spanish examples.

  • 2. Other terms that are found in the literature for these pairs are ‘rhizotonic vs. arhizotonic’, referring to whether the stress is on the root or the ending, and ‘thematic vs. athematic’, depending on whether the form does or does not contain a theme vowel.

  • 3. The term ‘supine’ has come to be used in an entirely different sense in grammars of North Germanic, where it refers to the non-agreeing past participle as in Danish jeg har skrevet ‘I have.prs write.pstprt’ (Nielsen, 2016, p. 417). How that came to be need not concern us here.