Skip to content
BY-NC-ND 4.0 license Open Access Published by De Gruyter Mouton May 25, 2019

Gerunds become prepositional infinitives in Romance Small Clauses: The effects of later Merge to the syntactic spine

  • Jan Casalicchio ORCID logo EMAIL logo
From the journal Probus

Abstract

This article offers a comparative analysis of “predicative” gerunds (“PGs”) and prepositional infinitives (“PIs”), focussing on perception constructions in Spanish (PGs) and European Portuguese (PIs). I demonstrate that these two constructions are diachronically related and that they still have a similar syntax. The evidence discussed suggests that both constructions are Small Clauses headed by a preposition of central coincidence. In PGs, this preposition is merged in a low aspectual projection and incorporates into the verb. In PIs, an evolution of PGs, it is merged later in the structure. This leads to a layering process, i.e. the splitting of a single head into a series of heads and transferring the semantics of one functional head to more heads. Thus, PIs have a more expanded structure than gerunds. Finally, further empirical evidence for this analysis is discussed, such as the insertion of adverbs and the use of sin (“without”) to negate the perceived event.

1 Introduction

This article presents a comparative analysis of two Romance constructions formed with an uninflected verb and used as secondary predicates: so-called predicative gerunds (henceforth, “PGs”) and prepositional infinitives (“PIs”). [1] The two constructions are generally in complementary distribution within the Romance domain: either a variety uses PGs, like Spanish (1), or it uses PIs, like European Portuguese (“EP”), see (2).

(1)
ViaJorgellorando.(Spanish: PG)
I.sawdomJorgecrying
(2)
VioJorgeachorar.(EP: PI)
I.sawtheJorgeacry

‘I saw Jorge crying.’

I propose that the two constructions in (1) and (2) have a similar structure: they are both Small Clauses (“SCs”) with a truncated TP as complement, and both contain a P(reposition). This preposition incorporates into the Verb in gerunds, while in PIs (2) it does not, because it is merged later: Thus, it is higher in the structure, [2] and therefore it is kept separate from the verb, which appears in the infinitive. Thus, the structures for (1–2) are: [3]

(3)
(4)

The contribution of the present article is twofold: on the one hand, it broadens our knowledge of gerundial constructions, since the comparison it draws between PGs and PIs is new in the literature. [4] On the other hand, it extends the analysis of adverbial and periphrastic gerunds as PPs headed by a preposition of central coincidence to PGs (Hale 1986). [5] I support my analysis by adopting a comparative perspective which builds on a discussion of novel – and previously unnoticed – synchronic and diachronic evidence. This broader perspective leads me to reduce the Romance-internal variation between PGs and PIs to a single difference: the position of the preposition. As mentioned above, in PGs the preposition is in a low AspP and is incorporated into the verb that moves to the same projection; in PIs, it is found in a higher position of the TP space. This is possible due to the division of the phrase into several layers. [6]

The layering process, i.e. the splitting of a head into a number of heads, thereby rearranging the featural content of each functional head, is the result of a diachronic process that has led the preposition to be merged later (and thus to appear higher up in the structure) in PIs than in PGs: PGs are the older form – they were present all over Romance in the Middle Ages – and PIs evolved from them. Crucially, the opposite evolution, i.e. the passage from PIs to PGs, is not attested in any variety.

PGs and PIs thus differ not only in whether or not they incorporate the P, but also in their size: both PGs and PIs are truncated TPs, as the ungrammaticality of high adverbs or modal verbs in these constructions shows. However, merging the P later in PIs than in PGs results in the expansion of their structure. I show that PGs are merged in a low aspectual projection; PIs, on the other hand, have a richer structure, since they also contain (at least) NegP and some adverbial projections. To prove this difference in size, I discuss some new observations, such as the hitherto unnoticed fact that in some varieties of Spanish the event expressed by the PG can only be negated by the preposition sin (“without”) followed by an infinitive. So, even varieties with PGs – like Spanish – use a PI when an event is negated. This is an additional piece of evidence for the underlying form of gerunds as “P+V complexes”.

The article is structured as follows. Section 2 describes the distribution and syntactic properties of PGs and PIs, and demonstrates that these constructions must be distinguished from both bare infinitives (Vi a Jorge llorar – “I saw Jorge cry”) and English gerunds. Section 3 sums up the main analyses of PIs or PGs in the literature to date. Section 4 is the paper’s core: it first discusses different types of evidence for a unified analysis of PGs and PIs and then presents the analysis and explains each part of it; finally, it examines two predictions of the proposed analysis and briefly discusses adverbial gerunds. Section 5 contains the conclusions.

2 Properties of PGs and PIs

2.1 Description and cross-linguistic distribution

In Romance, either PGs or PIs are used as secondary predicates. As shown in Map 1, PGs are especially used in Spanish, Catalan, Portuguese varieties spoken in Brazil, in Southern Portugal and in the Azores and Madeira, French, [7] Sardinian and Rumanian, [8] while PIs are mainly used in two geographically distinct areas: Central and Northern Portugal (Pereira 2015), [9] and an area that includes Central and Northern Italy (including Standard Italian) and Southern France (with inter-speaker variation in Occitan). Only in the Swiss Romansh varieties, and (marginally) Catalan, are both PGs and PIs used: in Catalan, PIs are limited to some very restricted contexts selected by a perception verb; in Romansh, however, they seem to alternate quite freely with PGs. Finally, in the areas of Portugal where PGs are used, they are competing with PIs (Carrilho and Pereira 2011; Pereira 2015), probably because of the influence of the standard variety, which uses PIs. [10]

Lastly, Southern Italy is unique: in most of its varieties both PGs and PIs are ruled out, and speakers generally use another construction, namely “Pseudo-relative clauses” (“PRs”). PRs are used all over Romance, except for Rumanian (Radford 1975; Cinque 1992; Brito 1995; Casalicchio 2013, a.o.):

(5)
a.
HovistoGiannichemangiavalamela.(Italian)
I.haveseenGiannithatatetheapple
b.
ViaJuanquesecomíalamanzana.(Spanish)
I.sawdomJuanthatseatetheapple

‘I saw John eating the apple.’

In Italian, the distribution of PRs perfectly overlaps with PGs in Spanish or PIs in European Portuguese (“EP”), and they share the same aspectual properties (for a correspondence between PIs and PRs, see Raposo 1989; Guasti 1992; Brito 1995, for PGs and PRs, see Fernández Lagunilla 2011b, and for a correspondence between all three constructions, see Casalicchio 2013). In this article I do not discuss the structure of PRs in depth (for which, see Casalicchio 2013, Casalicchio 2016a), but I refer to them when any correlation is particularly important for the analysis of PGs and PIs.

In this article, I focus mainly on two varieties, which I take as exemplary: Spanish for PGs and EP for PIs. When other varieties offer data of particular interest for my analysis, however, I also refer to them.

Map 1: The distribution of PGs and PIs in Romance (Europe).
Map 1:

The distribution of PGs and PIs in Romance (Europe).

2.1.1 PIs and PGs are Small Clauses

PGs and PIs share common structural properties. First of all, both can be considered Small Clauses (“SCs”), as demonstrated by the fact that they have the same distribution as adjectival and prepositional SCs (exemplified in (6a) (6b) respectively): see Di Tullio (1998: 198) for Spanish PGs, Raposo (1989: 284) for EP (from which these examples were adapted).

(6)

The coordination test also confirms this analysis: both PIs and PGs can be coordinated with other types of SC (either with the same (7), or with a different (8), subject):

(7)
a.
ViaJuan[ muytranquilo ]y[ sonriendo ].(Spanish)
I.sawdomJuanveryquietandsmiling
b.
VioJoão[ muitosossegado ]e[ asorrir ].(EP)
I.sawtheJoãoveryquietandasmile

‘I saw John very quiet and smiling.’

(8)
a.
Vi[ aMaríaenlacama ]y[ a
I.sawdomMariainthebedanddom
Antonioviendolatelevisión ].(Spanish)
Antoniowatchingthetelevision
b.
Vi[ aMarianacama ]e[ oAntónioaver
I.sawtheMariain-the bedandtheAntonioAlook
televisão ].(EP)
television

‘I saw Maria in bed and Antonio watching TV.’

(Barbosa and Cochofel 2004: 395)

PGs and PIs generally express imperfective aspect; they are stage-level predicates in which a single moment of an ongoing event is singled out. Di Tullio (1998: 200) points out that a sentence like:

(9)
Vielbarcoatracandoenelpuerto.
I.sawtheshipdockingintheport

‘I saw the ship docking at the port.’

can be followed by the clause “but it went adrift because a hurricane came up” (for a similar aspectual reading of PIs, see Barbosa and Cochofel 2004). This is due to the fact that in (9) the speaker perceived the ongoing event of docking, but did not necessarily see its completion. PGs and PIs are therefore often considered to be ruled out when an SC with an individual-level predicate is required (e.g. when the matrix verb is consider):11[11]

(10)

The ungrammaticality of PGs and PIs is parallel to that of stage-level adjectives like “naked”, while individual-level adjectives like “intelligent” are ruled in. This shows that both PGs and PIs are inherently stage-level SCs.

As far as the overall structure of these constructions is concerned, both PGs and PIs are compatible with two different constituent types (Raposo 1989; Di Tullio 1998; Rafel 2000; Casalicchio 2013, Casalicchio 2014): they may form a single constituent together with their antecedent [12] (11), or may not (12): [13]

(11)

Vi [(a) Jorge llorando / a chorar]

(12)

Vi [(a) Jorgei] [PROi llorando/a chorar]

As shown by Casalicchio (2013), the choice between a mono-constituent and a bi-constituent PG/PI is not free, but depends on the syntactic context in which it is embedded: in most cases only one or the other construction is possible, while in some cases (e.g. with perception verbs) both constructions can occur.

2.1.2 Differences with bare infinitives and English gerunds

Before discussing the data of PGs and PIs, I want to exclude from my analysis two constructions that superficially look like PGs and PIs, but actually have a different structure: (i) bare infinitive constructions in Romance; (ii) English gerunds.

Bare infinitives are also used in Romance perceptive constructions, like PGs and PIs (cf. (13) with (1)–(2)):

(13)
a.
ViaJorgellorar.(Spanish)
I.sawdomJorgecry
b.
VioJorgechorar.(EP)
I.sawtheJorgecry

‘I saw Jorge cry.’

Despite their similarity, bare infinitives have a different semantics, which is related to a different syntactic structure (paceDi Tullio 1998; Duarte 1992; Duarte and Gonçalves 2002): [14] when a bare infinitive is used, the perception concerns an event perceived as a whole, and the logical subject of the infinitive is just part of it. With PGs/PIs, on the other hand, an individual (i.e. the antecedent) is perceived, at a time when he is involved in an event. As noted in Casalicchio (2016b: 61), this difference involves “the distinction between categorical and thetic predication”: only in PGs/PIs is the subject singled out from the event (as we expect in SCs). This different semantics is reflected syntactically because the logical subject of the infinitive, but not of the PG/PI, can be phonologically null (14). Furthermore, bare infinitives and PGs/PIs differ in their distribution, because the former can only be used with perception verbs, while PGs/PIs are used in all instances of stage-level SCs (15). Finally, recall that PGs/PIs can be coordinated with other types of SC, such as APs and PPs; bare infinitives cannot (cf. (16) with (7)–(8)): [15]

(14)
Oigollover/*lloviendo.
I.hearrain.infraining

‘I hear it rain(ing).’

(15)
EncontréaAna*llorar/llorando.
I.founddomAnacry.infcrying

‘I found Ana crying.’

(16)
*ViaJuan[ muytranquilo ]y[ sonreír ].
I.sawdomJuanverycalmandsmile.inf

(Di Tullio 1998: 202 fn. 8)

Finally, there is a language-internal difference between PIs and bare infinitives in EP: the antecedents of PIs can never appear in the nominative (17a), whereas in inflected bare infinitives they can (17b) (Raposo (1989):

(17)
a.
*Euvielesatrabalhar(em)(EP)
Isawthey.nomawork(3PL.)

‘I saw them working.’

b.
Euvielestrabalharem.
Isawthey.nomwork.infl

‘I saw them work.’

(examples from Raposo 1989: 288)

The other construction that I do not consider in this analysis is the English gerund. As argued in Felser (1999), in English perceptive constructions infinitives and gerunds share the same structure and only differ in their aspectual reading. If Felser is on the right track, English gerunds thus structurally correspond to Romance bare infinitives, but not to PGs and PIs (paceRafel 2000). [16]

2.2 Main contexts of use for PGs and PIs

The correspondence between the language-internal distribution of Spanish PGs and that of European Portuguese PIs is striking: these two verb forms are perfectly overlapping, because they can both be used as secondary, stage-level predicates. Descriptively, we can enucleate four groups of contexts in which they are used, according to the syntactic role of the antecedent (or of the whole PG): it may be the subject (18) or the object of the matrix verb (19)–(20), [17] the complement of a preposition (21), or used in “root contexts”, like captions or titles (22): [18]

(18)
a.
Mariovolvióacasaoliendoavino.(Spanish)
Mariocame.backtohomesmellingtowine
b.
OMariochegouacasaacheiraravinho.(EP)
theMariocame.backtohomeasmelltowine

‘Mario came back home smelling of wine.’

(19)
a.
aFelipecantando.(Spanish)
I.hearddomFelipesinging
b.
OuvioFelipeacantar.(EP)
I.heardtheFelipeasing

‘I heard Felipe singing.’

(20)
a.
Lapolicíalosorprendióentrandoporeltecho.(Spanish)
thepolicehimsurprisedenteringbytheroof

‘The police caught him while he was entering by way of the roof.’

b.
Eleapanhouosrapazesaroubar(em)galinhas.(EP)
hecaughttheboysasteal(3.pl)chickens

‘He caught the boys stealing chickens.’

(Raposo 1989: 292)

(21)
a.
LafotodeMaríahaciendolasgalletas
thepictureofMariamakingthebiscuits
nuncalahabíavistoantes.(Spanish)
neveritI.hadseenbefore
b.
AfotodaMariaacozinharosbiscoitos,
thepictureof-theMariaacookthebiscuits
nuncatinhavistoantes.(EP)
neverI.hadseenbefore

‘I had never seen the picture of Maria making biscuits before.’

(22)
a.
E. Delacroix,LaLibertadguiandoalpueblo(Spanish)
theLibertyleadingdom-thepeople
ALiberdadeaguiaropovo(EP)
theLibertyaleadthepeople
E. Delacroix, Liberty leading the people (title of painting)

3 Previous analyses

3.1 The analyses of PIs

The structure of Portuguese PIs is still debated in the literature, which divides along two lines in this regard. The first originates with Raposo (1989), who proposed that PIs are PP-Small Clauses headed by the preposition a. He interprets them as a particular type of control structure, where the antecedent (in my terms) is the controller and gets case from an element of the matrix sentence via Exceptional Case Marking (“ECM”). The proposed structures are shown in (23)–(24) for non-inflected and inflected PIs, respectively:

(23)
non-inflected PIs:[PP NPi a [CP [IP PROi I VP]]]
(24)
inflected PIs:[PP NPi a [IP proi I/Agr VP]]

Note that when the infinitive is not inflected, the preposition a, which is the head of the SC, selects a CP, because in GB-terms the preposition must be unable to govern inside IP – otherwise it would trigger infinitive inflection. As a consequence, the subject of the infinitive is a PRO. In (24), however, I with Agr is assigned case by the preposition a, and this in turn enables it to case mark the null pro within the PP. The analysis of PIs as PPs was taken up later by Rafel (2000). However, this analysis raises two problems: i) it assumes that the antecedent and the preposition are merged in the Specifier and in the Head of the same projection; ii) the subject is generated in Spec,PP, from where it controls the null subject. As we will see infra (§ 4.2), however, both these assumptions are contradicted by the data.

The second line of analysis for PIs in EP was suggested by Duarte (1992) and Duarte and Gonçalves (2002). They propose that PIs are neither PPs nor control structures, but reduced clauses, i.e. AspPs:

(25)

Verbperc. [AspP NP [Asp’ [Asp° a -r][AgrP [Agr’ [Agr] VP]]]]

(Duarte 1992: 152)

In this analysis, the preposition a and the infinitival ending -r are analysed as a discontinuous morpheme. However, this analysis, which is reminiscent of Felser’s (1999) analysis of English perceptive constructions, treats PIs and bare infinitives as the same structures (both are AspPs), without considering the important semantic and syntactic differences shown above (Section 2.1.2). Moreover, it makes the same wrong predictions as Raposo (1989) with regard to intervening material and the first-merge position of the antecedent.

Barbosa and Cochofel (2004) propose an account that aims to unify the main aspects of the two previous analyses: they share Raposo’s (1989) analysis of PIs as control clauses, and discuss supporting evidence. At the same time, they argue that the preposition a expresses a progressive meaning, and is therefore merged in an aspectual projection, like in Duarte’s (1992) and Duarte and Gonçalves’ analysis, forming a Small Clause with the antecedent of the PI: [19]

(26)

[VP Perception verb [SC NPi [AspP a [TP proi/PROi… V…]]]]

(Barbosa and Cochofel 2004: 398)

The authors further suggest that the a in PIs is a preposition of central coincidence, following Hale’s (1986) distinction between prepositions of central and non-central coincidence. This explains the progressive value of the construction, since central coincidence prepositions are generally associated with non-terminative (i.e. durative/progressive) meanings, because they focalise on a specific moment of the event. [20] I will come back to the role and position of the preposition and to the evidence against analysing PIs as control structures in Section 4, after discussing the main analyses of PGs.

3.2 The analyses of PGs and Rafel’s (2000) comparative analysis of PGs and PIs

Borgonovo (1996) and Di Tullio (1998) offer two analyses of Spanish PGs. While Borgonovo compares Spanish and English gerunds in perceptive and nominal constructions (see above), Di Tullio mainly discusses the overall structure of PGs, considering the various contexts in which they can occur and proposing a threefold analysis, which is reminiscent of Cinque’s (1992) threefold analysis of PRs. With regard to the internal structure of PGs, Di Tullio proposes that gerunds are AspPs, and that they are only distinguished from bare infinitives by aspectual properties: [21]

(27)
VP [NP [AspP –r / -ndo] VP ](Di Tullio 1998: 201)

This claim resembles Duarte (1992) and Duarte and Gonçalves’ (2002) claim for PIs, although it was proposed independently – Di Tullio does not consider PIs. Nevertheless, as in Duarte’s studies, the syntactic differences between PGs and bare infinitives are not taken into account, although these cannot be explained in purely aspectual terms, since they involve structural differences (see Section 2.1.2).

Rafel (2000) takes a different approach. As far as I know, he is the first scholar to have proposed a comparative analysis of PRs, PGs and PIs in perceptive constructions. He calls all these constructions “Complex Small Clauses”, i.e. SCs with a more complex internal structure than ordinary SCs: they are a CP for PRs, a PP for PIs and an FP for (English) PGs:

(28)
a.
HevistoaJuanquecorría.(Spanish; PR)
I.haveseendomJuanthathe.run

‘I saw Juan running.’

b.
Viosadvogadosatrabalhar(em).(EP; PI)
I.sawthelawyersawork(.infl)

‘I saw the lawyers working.’

c.
I saw John running.(English; gerund)
(29)
a.[CP Juani que[IP proexpl.[VP proi corría]]]]]
b.[PP Juani [P’ a[CP ø[IP proexpl./PROexpl.[VP proi/PROi correr]]]]
c.[FP Johni [F’-ing[CP ø[IP PROexpl.[VP PROi run(n)-]]]]

(simplified representation of Rafel 2000: 79, 115, 210 [22])

The common denominator of all these structures is that they are headed by an element that Rafel holds to be an aspectual marker: the complementiser in (29a), the preposition a in (29b) and the -ing ending in (29c). The subject of the Complex Small Clause [23] is base generated in the Spec of the Complex SC, and not moved from within the predicate; thus, PGs and PIs are control structures (as in Raposo’s 1989; Barbosa and Cochofel’s 2004 analyses).

Three critical points should be noted in Rafel’s analysis: first, it differs from Di Tullio’s (1998) and Barbosa and Cochofel’s (2004) in not being based on reduced structures (AspP), because even PGs and PIs contain a full CP: the aspectual value of these constructions is not syntactically encoded in a dedicated projection, but is directly expressed by the aspectual marker. Moreover, PGs and PIs are more expanded than PRs, because the PP has a full CP as complement. This is clearly contrary to facts, because PGs and PIs only have a reduced TP (e.g. modals and auxiliaries are excluded), as I show infra.

Second, Rafel’s analysis of PGs and PIs predicts that the antecedent and the infinitive or gerund must be adjacent. However, in both PIs (as noted above), and in PGs, some types of adverb can intervene between these two elements. Finally, Rafel’s comparison is skewed by the fact that he considers mainly English – and not Romance – gerunds (see Section 2.1.2).

4 A unified analysis of PGs and PIs

Having discussed the main aspects of the analyses proposed in the GB and Minimalist frameworks, I now present a new proposal, based on a unified analysis of both PIs and PGs. To justify it, I first discuss synchronic and diachronic evidence for a structural correspondence between gerunds and PIs. Then, I show that the difference between these two constructions just depends on the position of the preposition a, which is merged earlier in PGs (and thus lower, in a position which is reached by the non-finite verb, giving rise to incorporation), and later in PIs. The preposition is present in both constructions because it is responsible for the progressive reading of the PG/PI. The antecedent, on the other hand, is merged in Spec,vP as usual, and is then moved to the Specifier of a FP:

(30)
a.
ViaPablocomiendo(Spanish PG)
I.sawdomPabloeating
b.
VioPabloacomer(EP PI)
I.sawthePabloaeat

‘I saw Pablo eating’

(31)

In the last part, I discuss two predictions made by my analysis that are borne out by the data.

4.1 Arguments for a structural correspondence of gerunds and prepositional infinitives

In this section, I discuss empirical and theoretical arguments that all point to a structural correspondence between PGs and PIs.

4.1.1 Empirical arguments

In the previous sections, I discussed some of the arguments for a correspondence between PGs and PIs, and their nature as SCs, recapitulated below:

  1. In Romance, each variety uses either PGs or PIs. They only overlap in a few areas, where one of the two is usually the “traditional” form, while the other is “imported” (PIs in Southern Portugal, Azores and Madeira due to standard EP, some uses of PGs in Catalonia due to contact with Spanish), see § 2.1;

  2. PGs and PIs share the progressive aspect: both refer to an ongoing event, without implying its completion (9);

  3. finally, both PGs and PIs are SCs, as the following syntactic tests show:

    1. both PGs and PIs can be coordinated with other types of SC (APs and PPs, (7)–(8));

    2. PGs in languages like Spanish have exactly the same distribution as PIs in EP (§ 2.2);

    3. being SCs, both PGs and PIs require an overt antecedent (14).

Two further types of evidence support a unified analysis. The first is diachronic: in Romance, we find various cases of synthetic forms replaced by a PP headed by a functional preposition. This affected the case system in the passage from Late Latin to Early Romance, for example; not only in the case of nouns (32), but also of the Latin gerunds (gerundia), flected nominal forms of a verb that were replaced by infinitives headed by the preposition a or de (33), see Tekavčić (1972):

(32)
a.campi>de campo(genitive: ‘of the field’)
b.campo>a(d) campo(dative: ‘to the field’)
(33)
a.amandi>de amare(genitive: ‘of loving’)
b.amando>ad amare(dative: ‘to loving’)

As (32)–(33) show, in Late Latin the synthetic forms campi and campo evolved to de campo and a(d) campo. Similarly, the gerundial forms amandi and amando became de amare and ad amare.

A similar change affected PGs in Northern Italy between the Medieval and the Modern period: Casalicchio (2016c) shows that most Northern Italian varieties used PGs up to the XIV century. Later on, PGs were completely replaced by PIs. This process took place sooner in some areas and later in others, but the change appears to have been sudden: until a certain period we find only PGs, but then, at a certain point, only PIs are used. Consider, for example, Milanese and Friulian: in the first centuries of their attestation, we find only PGs ((34a) and (35a)). From as early as the XVII century onwards, PIs appear to have entirely replaced PGs ((34b) and (35b)):

(34)
a.
QuandhavintesPillato|lopopulzodigando || …
WhenhasheardPilatusthepeoplethissaying

‘When Pilate heard the people saying this, …’

(Bonv. S II, 41)

b.
Quandseramaræv’hanvistapiansc
WhenIwasillyou.clthey.haveseenacry

‘When I was ill you were seen crying.’

(Maggi, Rime XI 43)

(35)
a.
TuestchevedèRibaltvignintpevie…
as.soonasshe.sawRibaltcomingonway

‘As soon as she saw Ribaldo coming along that way, …’

(Trav. Orl. Fur. I, 79)

b.
Chochu‘lsoccorssintirinavignì
whenthatthehelpthey.heardacome

‘When they heard the rescuers coming, …’

(Trad. Orl. Fur. IV 71)

The examples in (34) come from Milanese: they are taken from a religious poem by Bonvesin de la Riva (XIII century) (34a), and from a poem by Carlo Maria Maggi (XVII century) (34b). In both cases we have a perceptive construction, but in the first the gerund digando (“saying”) is used, while in the second we find the infinitive piansc (“cry”) preceded by the preposition a. In (35), I quote two Friulian examples taken from different vernacular versions of Ludovico Ariosto’s famous Orlando Furioso, dating back to the XVI and XVII centuries, respectively. In both cases the predicate of the perceptive construction is the verb vignì (“come”): it occurs as a gerund in (35a), [24] as a PI in (35b). [25]

The same switch, from PGs to PIs, is also attested in EP (Cunha 1986; Barbosa 1999; Pereira 2015), a fact of considerable significance for my theory. However, even more remarkably, to the best of my knowledge, the opposite situation, namely the switch from PIs to PGs, is never attested.

The second type of evidence concerns gerunds and PIs in contexts other than secondary predication, i.e. in adverbial and periphrastic constructions. Adopting a pan-Romance perspective, we see the same complementary distribution of gerunds and PIs. In periphrastic constructions, the choice between the two forms depends both on the variety and the functional verb used: e.g. Spanish uses gerunds with continue, Italian and EP PIs (36). [26] In the progressive construction with stare/estar, however, PIs are limited to EP and some Italo-romance varieties (Casalicchio and Migliori 2018), witness (37):

(36)
a.Maríasiguetrabajando/*atrabajar.(Spanish)
b.Mariacontinua*lavorando/alavorare.(Italian)
c.AMariacontinua*trabalhando/atrabalhar.(EP)
theMariakeepsworkingawork

‘Maria keeps working.’

(37)
a.
Estoycomiendo.(Spanish)
I.stayeating
b.
Stomangiando.(Italian)
I.stayeating
c.
Estouacomer.(EP)
I.stayaeat
d.
Stenghǝammagna’.(Ariellese, Abruzzo)
I.stayaeat

‘I’m eating.’

A similar pattern holds for adverbial gerunds: we see the same complementary distribution, with some varieties resorting to gerunds, others to PIs. Although gerunds are the most widespread option (including EP, Lobo 2002, Lobo 2006; Brito 2003), in some cases EP and Italian admit the use of a PI as an alternative to a gerund (38–39), [27] or as the only option (40):

(38)
AAnarecitouopoemaquase{cantando/acantar }(EP/BP)
theAnarecitedthepoetryalmostsingingasing

‘Ana – almost singing – recited the poetry.’

(Móia and Viotti 2004: 136)

(39)
{Apensarci/pensandoci }bene,nonavrestidovuto
athink=of.it.clthinking=of.it.clwellnotyou.hadhad.to
farlo(It.)
do=it.cl

‘Considering the facts well, you shouldn’t have done it.’

(40)
a.
*Avendomagnàmàssatorta,elGiorgiol’èstamal.
havingeatentoo.muchcaketheGiorgiohe.clisbeenill
b.
Aavérmagnàmàssatorta,elGiorgiol’èstamal.
ahaveeatentoo.muchcaketheGiorgiohe.clisbeenill

(Trentino, ASIt 1.7, cited in Casalicchio and Cordin forthcoming)

‘Having eaten too much cake, Giorgio felt ill.’

So, the alternation between gerunds and PIs in Romance is not limited to their use as secondary predicates. Rather, it is found in all contexts, not only in diachrony, but also in synchrony. It is crucial to note that in all the Romance languages I have examined the only two forms that can appear are gerunds or infinitives selected by the preposition a, and never other forms, like past participles, or infinitives selected by other prepositions or bare infinitives, which are frequently used in other contexts.

4.1.2 Theoretical arguments

The idea that gerunds are structurally PPs has already been proposed in some formal studies, such as Mateu (2002); Panagiotidis (2010); Fábregas (2008); Gallego (2010); Gallego and Hernanz (2012), a.o. Unlike the present article, however, they only consider periphrastic or adverbial gerunds, and not PGs. Moreover, no study on Romance has ever compared PGs with PIs as secondary predicates.

A first proposal to consider gerunds as (locative) PPs comes from Mateu (2002), who suggests that “[…] the progressive construction must be regarded as implying a locative unaccusative structure over the corresponding argument structure lexically assigned to the verb.” (Mateu 2002: 134). This means that progressive constructions are locative adverbials (“a locative unaccusative structure”). To support this claim, he cites cross-linguistic evidence of progressive forms, realised with a preposition followed by the nominal form of the verb in various languages. Furthermore, he adds evidence from Bolinger (1971), [28] who observed that the English progressive derives from a prepositional phrase (working < aworking < on+working[29]). Finally, he notes that the progressive form can often be replaced by a prepositional expression, as these examples show:

(41)
a.He is at workHe is working
b.She is at prayerShe is praying

(Mateu 2002: 137)

Note that this property is not restricted to English, since it is also present in Spanish, as shown by Masullo (2008):

(42)
a.de fiesta (lit. of party)festejando (celebrating)
b.de caza (lit. of hunt)cazando (hunting)

(Masullo 2008, cited in Gallego 2010: 88)

However, in all their examples Bolinger and Masullo compare gerundial forms with PPs that take a nominal complement; neither considers the correspondence of gerunds with other verb forms (such as PIs).

Mateu further argues that the preposition involved in progressives is a preposition of central coincidence (Hale 1986), as independently proposed, two years later, by Barbosa and Cochofel (2004) for Portuguese PIs (see above). Thus, a sentence like “John was breaking the window” should receive the following transparent syntactic meaning:

(43)

  1. be [John centrally located in [event [cause [ the window TCR break]]]]

  2. =John was centrally located in the event of causing the window to become broken

(Mateu 2002: 142)

Panagiotidis (2010) follows a similar line of investigation in his analysis of Greek and Hebrew adverbial gerunds, although he does not propose that the preposition is incorporated, but that it is phonologically null and selects the gerundial clause:

(44)
irthe[PPtraghudh-ondas](Modern Greek)
he.camesinging

‘He came singing.’

(Panagiotidis 2010: 178)

Panagiotidis proposes that the null P expresses temporal value, independently of the value of the gerund, which may be modal, causal, or other. He explains this by means of an inclusion relation (⊆) of the type in (45): the event of the matrix clause τ(ME) is contained in the event expressed with the gerund T(GE):

(45)

τ(ME) ⊆ T(GE)

Gallego (2010) and Gallego and Hernanz (2012) also take up Mateu’s analysis (2002), focussing on Spanish gerunds. They, too, propose that gerunds incorporate a preposition of central coincidence. Gallego (2010), in fact, suggests that the preposition has a hybrid nature, because it also participates in some of the properties of the complementiser. Following a proposal by Pesetsky and Torrego (2004), Gallego posits that P and T are contextual versions (i.e. allomorphs) of the same abstract category, since they are both birelational predicates. He suggests the following analysis (46):

(46)

In (46), the P is first merged in TP, to which position the verb also moves. The P incorporates into V, and the whole P-V complex then moves to the PP. The nature of adverbial gerunds as “infinitives + P” is demonstrated by (examples adapted from Gallego 2010: 86 f.):

(i) gerunds cannot have a subordination marker (first noted in Hernanz 1994):

(47)
(*Por)habiendodemostradoeldomadorsuvalentía …
(*for)havingshownthetamerhisbravery…

‘The tamer having shown his bravery, …’

(ii) gerunds cannot be preceded by an overt preposition in verbal periphrases (in other contexts, however, they can be, see Section 4.2.2):

(48)
empezar(*a)gritando
beginashouting

‘to start shouting’

(iii) gerunds cannot be used in complement positions, unlike in English:

(49)
Quieroganar/*ganando.
I.wantwinwinning

‘I want to win.’

For these reasons, Spanish gerunds cannot be considered as correspondent to infinitives; the hypothesis that gerunds are a “P + infinitive” complex clearly explains the differences between these two verb forms. Gallego’s proposal can also be adapted to PGs, although there are some differences concerning especially the size of PGs vs. adverbial gerunds (see below).

Having discussed these different arguments in favour of a unified analysis, I now present my proposal that PGs and PIs are only distinguished by a parametric variation regarding the merging site of the preposition.

4.2 The structure of Prepositional Infinitives and Predicative Gerunds

As we saw in the previous section, the distribution of gerunds and PIs overlaps in Romance, sometimes even within the same variety. I suggest that this happens because there is a structural correspondence between the two constructions. The basic structure for PGs (51) and PIs (52) is:

(50)
a.
ViaJorgellorando.(Spanish PG)
I.sawdomJorgecrying
b.
VioJorgeachorar.(EP PI)
I.sawtheJorgeacry

‘I saw Jorge crying.’

(51)

(52)

The structures of PGs (51) and PIs (52) differ minimally: [30] what changes is the position of the preposition. In PGs, it is merged in a lower AspP (“LAspP”), [31] in the projection to which the verb also has to move. As a consequence of this movement, the verb incorporates into the preposition, resulting in the formation of a gerund. In (52), however, the preposition is merged later and is therefore in a higher AspP (HAspP), which is not reached by the infinitive (unlike the proposals of Duarte 1992; Duarte and Gonçalves 2002). The verb moves to a low AspP like in Spanish, but since it does not reach the projection in which P is hosted incorporation cannot take place. The P is thus spelled out as a. This analysis makes a clear prediction: in PIs, material can intervene both between the antecedent and the P and between the P and the infinitive. [32] Finally, the antecedent moves to the Specifier of the SC, whose Head is not realized (or possibly filled by a Relator, as proposed in den Dikken 2006).

In the next section I discuss the evidence for the different parts of the structures in (51) and (52), and then I come back to the predictions made by my analysis.

4.2.1 Evidence for the structure proposed for PGs and PIs

The PG and PI structures proposed in (51) and (52) contain three claims that I discuss in this section: i) the preposition in a PI is merged later, and thus in a higher AspP; ii) as a consequence, the complement of the SC is more reduced in PGs than in PIs: it is a LAspP in PGs, a HAspP in PIs; iii) PGs and PIs are not control structures.

First, in PIs the preposition is merged in HAspP, thus neither in the projection to which the verb moves (as claimed by Duarte 1992; Duarte and Gonçalves 2002; Barbosa and Cochofel 2004), nor in the head of the SC (52), as in Raposo’s (1989) and Rafel’s (2000) analysis. My account is based on the fact that adverbs can occur both on the right and on the left of the P, showing that it must be in a separate projection. If we now consider the derivation of the PI, the first element to be merged (i.e. moved) outside the vP/VP is the infinitive itself, as its position with respect to low TP-adverbs (Cinque 1999) shows: [33]

(53)
OuvioJorgea(*bem)cantar(bem).(EP)
I.heardtheJorgea(well)sing(well)

‘I heard Jorge singing well.’

In (53), the infinitive cantar must occur to the left of the low adverb well, which sits in Spec,VoiceP, at the boundary between TP and vP (Cinque 1999, 2006). Since Cinque shows that the lowest projections in the TP are dedicated to Aspect, the non-finite verb must reach (at least) a low AspP. Thus, the preposition must be higher than this projection, otherwise it would sit in the same position as the verb, and this would lead either to the unattested order *cantar a, or to the incorporation of the preposition into the verb (and thus to a PG, see infra). We thus find the following preliminary order of elements:

(54)

a> [LAspP Verb [VoicePwell]

Moreover, the preposition cannot be in LAspP because it occurs higher than the negation, which precedes the low aspectual projections (Zanuttini 1997; Cinque 1999, Cinque 2006):

(55)
TodosnósvimosDecoanãometerumgolo(EP)
allwesawDecoanotshootagoal

‘We all saw that Deco did not shot a goal.’

(real example, cited in Barbosa and Cochofel 2004: 388)

In addition, the sequence “a + inf.” can be interrupted by some adverbs (which are themselves higher than negation), as shown by Barbosa and Cochofel (2004):

(56)
EunãoestouaveroprimeiroMinistroa,derepente,
InotstaytoseethefirstMinisteraofsudden
cederpoderesaoPresidentedaRepública
cedepowersto-thePresidentof-theRepublic

‘I’m not seeing the First Minister suddenly ceding powers to the President of the Republic.’

(real example, cited in Barbosa and Cochofel 2004: 388)

Thus, the order of the elements in TP is:

(57)

a>[HAspPde repente [NegPnão [LAspP Verb [VoicePwell]

On the other hand, some adverbs can occur above the P, i.e. between the antecedent and a: [34]

(58)
a.
ViaAnadescaradamenteatiraraúltimafatiadobolo
I.sawtheAnabrazenlyatakethelastpieceof-thecake

‘I saw Ana brazenly taking the last piece of cake.’

b.
Euvioavô{já}a{já}nãorespirar
Isawthegrandpaalreadyaalreadynotbreathe
epercebique eleestavamorto.
andI.perceivedthat hestayeddead

‘I saw grandpa not breathing anymore and I realised that he was dead.’

Note that the adverb in (58b) can be placed both before and after the preposition (at least for some speakers I consulted), which shows that there is some variation. This shows that the P is still merged in AspP, although in a higher projection than the P of Spanish PGs: [35]

(59)

[HAspP [HAsp°a [HAspP [NegPnão [LAspP Verb [VoicePwell]

The second claim concerns the size of the SC hosting the infinitive and the gerund. In the structures in (51) and (52), they are both analysed as AspPs, unlike Raposo’s (1989) and Rafel’s (2000) proposals (cf. the structures in (23)–(24) and (29), respectively). I argue for this solution again on the basis of empirical evidence: not only do some adverbs occur higher than the preposition or the PG, but modal verbs, which occur in the higher portion of the TP (Cinque 1999, Cinque 2006), are ruled out in PGs/PIs. In (60), I give the relevant examples for PGs, but the same observations hold for PIs:

(60)
*ViaJuandebiendo/pudiendocantar.
I.sawdomJuanhaving.tobeing.ablesing

Furthermore, PGs and PIs have no temporal specification on their own (as we would expect if they had a full TP), and their temporal interpretation depends on the tense of the matrix verb. Again, the examples are given with PGs, but also hold for PIs: [36]

(61)
*HoyveoaPablo{trabajando/habiendotrabajado}ayer.
todayI.seedomPabloworkinghavingworkedyesterday

Finally, recall that according to Raposo (1989), PIs cannot check nominative case, as we would expect if they were full TPs (see, again, ex. (17a)). The same holds for PGs:

(17) a.
*Euvielesatrabalhar(em)(EP)
Isawthey.nomawork(infl.)

The properties noted here are valid for both PGs and PIs: both are AspPs. However, there are also differences between these two constructions: as we saw above, negation is possible in EP through the use of the standard negation não, but PGs cannot be negated with no (cf. (62) with (55)), Fernández Lagunilla (1999, 2011a):

(62)
*TodosnosvimosaDeconometiendoungol(Spanish)
allwesawdomDeconotmakingagoal

This difference shows that the structure of the complement of the SC is more extended in PIs than in PGs, because in the former it also includes NegP. Thus, I propose that the procrastination of the merge of a in PIs results in the expansion of the structure of the whole SC.

My last claim regarding the structures in (51) and (52) concerns the coindexation mechanism of the antecedent with the syntactic subject of the embedded clause. We have seen that Raposo (1989), Rafel (2000), and Barbosa and Cochofel (2004) propose that there is a control relation. For PRs, this claim has been put forward by Grillo and Moulton (2015) and Casalicchio (2016a), on the basis of the reconstruction test: reconstruction effects have previously been used by Belletti and Rizzi (1988) to distinguish between raising and control structures. In the case of Italian psych-verbs like preoccupare (“worry”), reconstruction effects are grammatical with raising (63), but not with control, verbs ((64), examples taken from Casalicchio 2016a): [37]

(63)
IpropriifiglisembranopreoccupareGiannii(Italian – raising)
theownchildrenseemworryGianni

‘His own children seem to worry Gianni.’

(64)
*IpropriifiglicredonodiPROpreoccupareGiannii(control)
theownchildrenbelieveofworryGianni

In PRs, reconstruction yields ungrammatical effects, thus Grillo and Moulton (2015) and Casalicchio (2016a) analyse them as control structures. [38] I have used the same test with PGs and PIs, and in this case the majority of consulted speakers found reconstruction acceptable (if sometimes a bit marginal):

(65)
EsunapenaverasuspropiosiniñospreocupandoaJuani
it.isapityseedomhisownchildrenworryingdomJuan

‘It is a pity to see that his own children worry Juan’(Spanish)

(66)
ImaginaosprópriosimeninosapreocuparoJoãoi (EP)
ImaginetheownchildrenaworrytheJoão

‘Imagine that his own children worry João.’

This indicates that – unlike in PRs – the subject does not control a PRO-subject from outside, but is first merged as the External Argument of the embedded verb in Spec,vP, and then moves to the Specifier position of the SC. [39]

4.2.2 Predictions of the proposed analysis

The contrastive analysis proposed in (51) and (52) makes two important predictions, which are verified in this section: first, the analysis predicts that even in languages with PGs, like Spanish, a PI shows up when incorporation is not possible (i.e. when a lexical preposition is used instead of the functional preposition a). Second, PGs may be preceded by another preposition (e.g. en durmiendo, lit. “in sleeping”). All two predictions are borne out.

The first prediction concerns the replacement of PGs by PIs even in languages like Spanish, when PGs are ruled out by some independent factor. I have already mentioned that in Spanish, a PG cannot be negated with the usual negative adverb no. However, this restriction does not hold for all gerunds. In fact, adverbial gerunds can be negated in two ways: either by using the negative element no before the gerund, or by using the preposition sin (“without”). In the latter case, however, the verb does not show up as a gerund, but as an infinitive:

(67)

Negation of gerunds in Spanish:

  1. no + gerund (no comiendo)

  2. sin + infinitive (sin comer)

  3. *sin + gerund (*sin comiendo)

Fernández Lagunilla (2011a) shows that the choice between (67a) and (67b) depends on complex semantic and syntactic constraints; in some cases both options are possible (usually with semantic/structural differences):

(68)
a.
Ganaronelpartidonocometiendofaltaspersonales.
they.wonthematchnotmakingfoulsindividual

‘They won the match because they didn’t make individual fouls.’

b.
Ganaronelpartidosincometerfaltaspersonales.
they.wonthematchwithoutmake.inffoulsindividuals

‘They won the match and didn’t make individual fouls.’

(Fernández Lagunilla 2011a: 247)

In (68a), the negated gerund explains the cause of the team’s victory, while the PP headed by sin in (68b) merely expresses a concomitant action.

In other cases, the construction “sin + infinitive” is preferred, or even the only available option. If we want to negate the event expressed by the gerund in (69), we may use either (70a) or (70b):

(69)
Maríasiguesoportándote(Spanish)
Mariakeepsbearing=you.cl

‘Maria still bears you.’

(70)
a.
?Maríasiguenosoportándote
Mariakeepsnotbearing-you
b.
Maríasiguesinsoportarte
Mariakeepswithoutbear=you

‘Maria still doesn’t bear you.’

In example (70) the gerundial construction is selected by a functional verb and forms a periphrasis; both negative forms are possible with the same meaning, although the “no + gerund” form is judged as marginal.

In other cases, however, an event expressed by the gerund can never be negated, whether with no or with sin. This is usually the case in PGs, which are often described as incompatible with negation. However, this assertion is often based on perceptive constructions, which do, indeed, usually reject negated PGs (but for semantic reasons, see infra). If, on the other hand, we look at PGs in other contexts, we see that the secondary predicate can be negated with sin when the following conditions apply: (i) the matrix verb is of movement, speaking or achievement (cf. Fernández Lagunilla 2011a: 250 f.) and (ii) the antecedent is the subject of the matrix clause. In these cases, only the negation with sin is possible:

(71)
a.
??/*Juanvolvióacasanooliendoavino
Juancame.backtohomenotsmellingtowine
b.
Juanvolvióacasasinoleravino
Juancame.backtohomewithoutsmelltowine

‘Juan came back home without smelling of wine.’

However, even with perception verbs things are less straightforward than they appear at first sight. In fact, with verbs like see or catch, events negated with sin are possible in some varieties of Spanish, in favourable contexts. [40] In fact, a sample Google search (12/9/2018) produced various results for “sin + infinitive” with these two verbs. Some examples are:

(72)
XDDDDDsiempreteveoconellospuestosdehechonunca
XDDDDDalwaysyouI.seewiththemput.oninfactnever
tehevistosinescucharmúsica
youI.have seenwithoutlisten.tomusic

‘[laughter] I always see you with them (sc. your headphones) on, in fact I’ve never seen you not listening to music (=Every time I see you you’re listening to music)’

(https://www.clubensayos.com/Historia-Americana/Ygtxgcvjb/2307407.html)

(73)
Silasnuevecampanadas,medije,mesorprendensintener
iftheninechimesmeI.toldmecatchwithouthave
lamanosobrelaaldabadelapuerta,
thehandontheknockerofthedoor
algofunestoacontecerá.
somethingill-fatedwill.happen

‘If the nine chimes catch me when I don’t have my hand on the doorknocker, I told myself, something ill-fated will happen.’

(A. Reyes, La cena)

(72) and (73) are formed with a perception verb, respectively, ver (“see”) and sorprender (“catch”), which take as their complement a PP headed by the preposition sin. In both examples it is clear that the subject of the PP is coindexed with the object of the perception verb, and not with the matrix subject, as the context shows: in (72) the addressee is listening to music whenever he is seen by the speaker. The example (73) is formed with the verb sorprender (“catch”): the antecedent of the PP headed by sin is caught not doing something he should be doing (keeping his hand on the knocker).

Complements with sin are SCs, like PGs and PIs; more precisely, syntactically they are PIs, and therefore they behave like Portuguese PIs, rather than like Spanish PGs. This is shown by the next examples, again taken from Google: [41]

(74)
Mientras,observoamitíoconlamiradaperdida,
in.the.meantimeI.observedommyunclewiththelooklost,
sinyacomer,negandoconlacabeza
withoutany.moreeatnegatingwiththehead
ysujetándoseelmentónconlasmanosentrelazadas.
andseizingthechinwiththehandsjoint

‘In the meantime, I observe my uncle and see that he has a lost look, that he has stopped eating, is shaking his head and grasping his chin with hands interlinked.’

(http://ita.calameo.com/books/004354687c0c555487c36)

(75)
Zeusrescato (sic!)asuhijotodavíasinnacer
Zeusrescueddomhissonyetwithoutbe.born

The example in (74) is particularly revealing, because the PI with sin is coordinated with three other SCs, two of them formed with a PG and one with a PP. This confirms the analysis of this type of sin-clause as a SC, just like PGs and PIs. Comparing (74) and (75) we find a second important clue, namely the position of the adverbs, which can follow (as is the case of ya in (74)) or precede (todavía in (75)) the preposition sin. This double pattern exactly mirrors what we found with PIs in EP ((55)–(56), and (58)) and confirms my analysis: the preposition is neither the head of the SC, nor is it merged in the projection to which the infinitive moves. Examples (74)–(75), which are not constructed but real, hence confirm the analysis proposed in (52). [42]

All the examples thus clearly show that secondary predicates can be negated with the preposition sin – in at least some Spanish varieties –, and that this eventuality is not ruled out syntactically, although it is usually considered odd on semantic grounds. In appropriate contexts, however, they are perfectly acceptable. Note that these examples correspond to secondary predicate PPs taking a nominal complement:

(76)
a.
Lovieronsinpantalones
himthey.sawwithouttrousers

‘They saw him without [his] trousers.’

b.
Losurprendieronsinpapeles
himthey.caughtwithoutdocuments

‘They caught him without documents.’

As demonstrated in the above discussion, the use of “sin + infinitive” as a negative counterpart to PGs is important evidence for my overall hypothesis. But why does Spanish resort to a PI in negative contexts? I propose that this is due to the fact that the negation with no is excluded from PGs for structural reasons (as we have seen, they have a reduced TP, where NegP is not available). Sin, however, is available, but since it is a lexical rather than a functional preposition, unlike a, it cannot incorporate into the verb. [43]

Finally, the proposed analysis also gives a straightforward explanation of why most Romance varieties have gerunds preceded by a preposition in restricted contexts, at both past and current language stages. The most well-known example is of course the French gérondif, which is used with adverbial value and is formed by the preposition en + the form of the participe présent (see fn. 7):

(77)
Illedisaitenriant.(French)
heit.clsaidinlaughing

‘He said it laughing.’

EP adverbial gerunds can also be preceded by the preposition em: [44]

(78)
EmelevindoaoPorto,falamosnisso (EP)
inhecomingto-thePortowe.talkon.that

‘When he’ll come to Porto we’ll talk about that.’

(Brito 1984: 431)

There are also examples of PGs preceded by prepositions, however. Here are some examples from different languages:

(79)
a.
AudistideSansone,|cum’elfoençegnao:
you.heardofSamsonhowhewasdeceived
lamoierendormando|lecrenelitaiao.(Old Venetian)
thewifeinsleepingthehairhimcut

‘Have you heard how Samson was deceived? His wife cut his hair off while he was sleeping.’

(Proverbia, 93-94)

b.
El…vesevagiaamadirondsiastschereschaspiglonn
Hesawyettoripeninghischerriesfor-theyear
vegnent.
coming

‘He … saw his cherries already ripening the next year.’

(Sursilvan Romansh: Liver 1991: 72)

c.
Lpartìakantando.(Istriot)
heisleavedtosinging

‘He left singing.’

(Cernecca 1976: 238)

(79a) is an example of Old Venetian (XIII century), where the PG en dormando (“sleeping”) refers to the indirect object li (“him”, i.e. Samson). The examples (79b-c), on the other hand, come from currently spoken varieties: (79b) is Sursilvan, a Romansh dialect spoken in Switzerland. In this case, the prepositional gerund is used in a perceptive construction. Finally, (79c) is a PG on the subject from Istriot, spoken in south-western Istria. The link between a preposition and the gerund may even lead to a full grammaticalization of the “P + gerund” complex, like in the French example (77) and in some dialects spoken in Trentino (north-eastern Italy), where the preposition in and the gerund have become one morphological unit: in Val Cismon, there are some fixed forms which derive from a prepositional gerund: incantèn (> Latin in cantando, “singing”), incorèn (< Latin in currendo, “running”), Tissot (1976).

In my analysis of PGs (51), the presence of a preposition is accounted for by the fact that it is possible for an additional preposition to be merged, with the function of an aspectual marker, in the same projection in which a is merged in EP. [45]

4.3 A final note on adverbial gerunds

The analysis presented in this paper is based on PGs and PIs used as secondary predicates. It is legitimate to wonder, however, whether and to what extent this analysis can be extended to other contexts – especially adverbial gerunds – which differ syntactically from PGs in various ways. [46]

First of all, consider that there are two types of adverbial gerunds: internal modifiers, which are similar to PGs in many respects and also have a reduced TP structure, and external (or peripheral) modifiers, which are full TPs or (at least in some cases) CPs. There is abundant evidence for this difference (see e.g. Fernández Lagunilla 1999 for Spanish, Lobo 2002, Lobo 2006, and Lobo 2013; Brito 2003 for EP): only external modifiers can have a modal or auxiliary verb (cf. (80) with (60)) and are temporally independent ((81) vs. (61)), while internal modifiers behave like PGs. Moreover, only external modifiers can have an independent – always postverbal – subject, to which they assign nominative case (82):

(80)
Pudiendoquedarseencasa,Pedrodecidióacompañarme. (Sp.)
can.gerstayathomePedrodecidedgo.with=me

‘Although he could have stayed at home, Pedro decided to go with me.’

(Fernández Lagunilla 1999: 3446)

(81)
Empezandohoyeltrabajo,acabaríamosmañana.
startingtodaytheworkwe.would.finishtomorrow

‘If we started this job today, we would finish tomorrow’

(Fernández Lagunilla 1999: 3458)

(82)
VindoeleaoPorto,falaremosnisso.(EP)
cominghe.nomto-thePortowe.will.walkin.that

‘If he comes to Porto, we’ll talk about it/that.’

(Brito 1984: 432)

I suggest that this difference between PGs and external gerunds can be explained if we adopt Gallego’s (2010) proposal that the latter have the P merged in T, thus later than in PGs. However, incorporation can take place because here the verb has to move independently to TP or CP, due to the sentential nature of external gerunds – thus it incorporates the P even if it is higher in the structure. Note that the different merging site of the P correlates with the fact that gerunds used as external modifiers are much more flexible in their temporal and aspectual interpretation than PGs, since they can indicate, among other things, anteriority or posteriority, a result or a progression.

5 Conclusions

In this article I have proposed a new approach to the analysis of both PGs and PIs. The proposal is based on the detailed observation and comparison of data from several Romance languages, mainly Spanish and Portuguese. My investigation reveals that the two constructions share more than just their syntactic distribution: theoretical considerations, their diachronic evolution and the synchronic evidence all point to the fact that PGs and PIs should be analysed as having the same structure. On the other hand, both perceptive constructions with bare infinitives and English gerunds have a different structure, and must be accounted for in a different way.

The crucial difference between PGs and PIs is that in PGs the preposition is merged in a low AspP, while its merge has been procrastinated in PIs, thus the P is higher. This process has resulted in a more expanded structure for PIs, which means that they can host more elements than can PGs, such as negation and some types of adverb. A crucial point in favour of this analysis is that when Spanish cannot use a PG for syntactic reasons (i.e. when the event is negated) it also uses a PI headed by the preposition sin.

This study offers a detailed comparison of PGs and PIs, which confirms the interpretation of gerunds as PPs (first proposed by Mateu 2002). A further (and new) piece of evidence discussed in this paper concerns the analysis of the diachronic evolution that has led some varieties to switch from PGs to PIs. The preposition is responsible for the aspectual reading: the data discussed here lead me to suggest that the preposition sits in an AspP and is not the head of the SC.

Another issue tackled by this paper is the merging site of the antecedent: I discuss new evidence, gained from the reconstruction test, which shows that it must be externally merged in Spec,vP, from where it moves to the Spec of the SC. This finding has important consequences for the analysis of SCs: SCs are usually seen as basic configurations of a subject and a predicate (see Williams 1975; Stowell 1983; Cardinaletti and Guasti 1995; Moro 1997, Moro 2000; Dikken 2006, a.o.). The fact that in some cases their subject is merged within the predicate should lead to a reconsideration of the concept of SCs itself.

Finally, the proposed analysis offers a neat and detailed picture of PGs and PIs, in which we find a great deal of structural overlapping, since the only difference is the position of the preposition. However, the relationship between gerunds and PIs becomes more complex if we include other types of gerunds: the alternation between these and PIs cannot be reduced simply to a difference in the position of the preposition, other factors are also at play, such as the movement of the gerund to CP, when it is used adverbially. Although the present analysis sheds some light on these aspects, more research is needed to fully understand the overall syntax of Romance gerunds.

Literature

Sources of the examples
ASIt = Atlante Sintattico d’Italia, http://asit.maldura.unipd.it (7/2/2018).
Bonv. S IIBonvesin de la Riva. Opere Volgari. Edited by Gianfranco Contini. 1941. Roma: Società Filologica Romana.
Maggi,Maggi, Carlo Maria. Le rime milanesi. Edited by Dante Isella. 1994. Milano: Fondazione Pietro Bembo.
ProverbiaProverbia que dicuntur supra natura feminarum. Edited by Cesare Segre & Mario Marti. 1959. La Prosa del Duecento. Milano-Napoli: Ricciardi.
Reyes, La cenaReyes, Alfonso (1920). La cena. In Alfonso Reyes. El plano oblicuo. Cuentos y diálogos. Madrid: Europa.
Trav. Orl. Fur.Anonymous author. Travestimento del I e di parte del II canto dell’Orlando Furioso. Edited by Vincenzo Joppi. 1878. Testi inediti friulani dei secoli XIV al XIX. Archivio Glottologico Italiano 4. 233–253.
Trad. Orl. Fur.Paolo Fistulario. Traduzione del Canto quarto e quinto dell’Orlando Furioso. Edited by Vincenzo Joppi. 1878. Testi inediti friulani dei secoli XIV al XIX. Archivio Glottologico Italiano 4. 253–267.

Acknowledgements

I am grateful to Celina Agostinho, Paola Benincà, Elena Ciutescu, Patrizia Cordin, Roberta D’Alessandro, Marina Fernández Lagunilla, Alexandra Fiéis, Ángel Gallego, M. Lluïsa Hernanz, Maria Lobo, Cecilia Poletto, Avel·lina Suñer, Silvia Terenghi, two anonymous reviewers and the audiences of the seminars at the universities of Barcelona, Cambridge, Girona, Göttingen, Padova and Utrecht for their comments and suggestions; and Rachel Murphy for proof-reading the English. The research leading to these results has received funding from the European Union’s 7th Framework Programme for research, technological development and demonstration (grant agreement 613465 AThEME), and from European Union’s Horizon 2020 research and innovation programme (grant agreement 681959 Microcontact).

References

Araus, Gutiérrez & Maria Luz. 1992. Sobre el gerundio en función adjetival. In María Vaquero & Amparo Morales (eds.), Homenaje a Humberto López Morales, 205–220. Madrid: Arco/Libros.Search in Google Scholar

Barbosa, Afrânio G. 1999. Para uma história do português colonial: aspectos lingüísticos em cartas de comércio. Rio de Janeiro: Universidade Federal de Rio de Janeiro dissertation.Search in Google Scholar

Barbosa, Pilar & Fátima Cochofel. 2004. O infinitivo preposicionado em PE. In Actas do XX Encontro Nacional da Associação Portuguesa de Linguística, 387–400. Lisboa: Colibri.Search in Google Scholar

Belletti, Adriana. 1990. Generalized verb movement. Aspects of verb syntax. Torino: Rosenberg & Sellier.Search in Google Scholar

Belletti, Adriana. 2008. The CP of clefts. CISCL Working Papers 2. 7–17.Search in Google Scholar

Belletti, Adriana & Luigi Rizzi. 1988. Psych verbs and θ-theory. Natural Language and Linguistic Theory 6(3). 291–352.10.1007/BF00133902Search in Google Scholar

Belvin, Robert & Marcel Den Dikken. 1997. There, happens, to, be, have. Lingua 101. 151–183.10.1016/S0024-3841(96)00049-6Search in Google Scholar

Bolinger, Dwight. 1971. The nominal in the progressive. Linguistic Inquiry 2. 246–250.Search in Google Scholar

Borgonovo, Claudia. 1996. Gerunds and perception verbs. Langues et Linguistique 22. 1–19.Search in Google Scholar

Brito, Ana Maria. 1984. Sobre as noções de sujeito e argumento externo: Semelhanças entre a estrutura de F e a estrutura de SN em portugués. Boletim de Filologia 29. 421–478.Search in Google Scholar

Brito, Ana Maria. 1995. Sobre algumas construções pseudorelativas em português. Línguas e Literaturas 12. 25–54.Search in Google Scholar

Brito, Ana Maria. 2003. Subordinaçâo adverbial. In Maria Helena Mira Mateus, Ana Maria Brito, Inês Duarte & Isabel Hub Faria (eds.), Gramática da Língua Portuguesa, 695–728. Lisboa: Caminho.Search in Google Scholar

Brito, Ana Maria, Eduardo Raposo. 2013. Complementos, modificadores e adjuntos no sintagma nominal. In Eduardo Raposo et al. (ed.), Gramática do Português, 1043–1113. Lisboa: FCG.Search in Google Scholar

Cardinaletti, Anna & Maria Teresa Guasti. 1995. Small Clauses. New York: Academic Press.10.1163/9780585492209Search in Google Scholar

Carrilho, Ernestina & Sandra Pereira. 2011. Sobre a distribução geográfica de construções sintácticas não-padrão em português europeu. In Armanda Costa, Isabel Falé & Pilar Barbosa (eds.), Textos seleccionados do XXVI Encontro Nacional da Associação Portuguesa de Linguística, 125–139. Lisboa: APL.Search in Google Scholar

Casalicchio, Jan. 2013. Pseudorelative, gerundi e infiniti nelle varietà romanze: Affinità (solo) superficiali e corrispondenze strutturali. München: Lincom Europa.Search in Google Scholar

Casalicchio, Jan. 2014. Predicative gerunds in Spanish and Catalan. Quaderni di lavoro ASIt 18. 103–126.Search in Google Scholar

Casalicchio, Jan. 2016a. Pseudo-relatives and their left-periphery: A unified account. In Ernestina Carrilho, Alexandra Fiéis, Maria Lobo & Sandra Pereira (eds.), Romance languages and linguistic theory. selected papers from “Going Romance” 28, Lisbon, 23–42. Amsterdam: John Benjamins.10.1075/rllt.10.02casSearch in Google Scholar

Casalicchio, Jan. 2016b. The use of gerunds and infinitives in perceptive constructions: The effects of a threefold parametric variation in some Romance varieties. In Ermenegildo Bidese, Federica Cognola & Manuela Caterina Moroni (eds.), Theoretical approaches to linguistic variation, 53–87. Amsterdam: John Benjamins.10.1075/la.234.03casSearch in Google Scholar

Casalicchio, Jan. 2016c. Ricostruire la diacronia della sintassi ladino-dolomitica con l’aiuto di Joppi. Il caso dei costrutti percettivi. In Federico Vicario (ed.), Ad limina Alpium. VI Colloquium Retoromanistich, Cormons, dai 2 ai 4 di Otubar dal 2014, 97–126. Udine: Società Filologica Friulana.Search in Google Scholar

Casalicchio, Jan & Patrizia Cordin. Forthcoming. A grammar of central Trentino. Leiden/Boston: Brill.10.1163/9789004430976Search in Google Scholar

Casalicchio, Jan & Laura Migliori. 2018. Progressive and predicative constructions with gerund in Romance. A contrastive analysis. Revue Roumaine de Linguistique 63(3). 253–270.Search in Google Scholar

Cernecca, Domenico. 1976. Modi infinitivi del verbo nell’istrioto di Valle d’Istria. In Centro di Studi per la Dialettologia italiana (ed.), Problemi di morfosintassi dialettale, 227–238. Pisa: Pacini.Search in Google Scholar

Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press.Search in Google Scholar

Cinque, Guglielmo. 1992. The pseudo-relative and acc-ing constructions after verbs of perception. Working Papers in Linguistics – University of Venice 2. 1–31.10.1017/CBO9780511554261.009Search in Google Scholar

Cinque, Guglielmo. 1999. Adverbs and functional heads. A cross-linguistic perspective. Oxford/New York: Oxford University Press.Search in Google Scholar

Cinque, Guglielmo. 2006. Restructuring and functional heads. The cartography of syntactic structures. Oxford/New York: Oxford University Press.Search in Google Scholar

Ciutescu, Elena. 2013. Remarks on the infinitival subject of perception verb complements: Evidence for two syntactic configurations. Revue Roumaine de Linguistique 58(3). 292–312.Search in Google Scholar

Cunha, Celso. 1986. Conservação e inovação no português do Brasil. O Eixo e a Roda 5. 199–232.10.17851/2358-9787.5.0.199-230Search in Google Scholar

Demirdache, Hamida & Myriam Uribe-Etxebarria. 2000. The primitives of temporal relation. In Roger Martin, David Michaels, Juan Uriagereka & Samuel Jay Keyser (eds.), Step by step. Essays on minimalist syntax in honor of Howard Lasnik, 157–186. Cambridge, MA: MIT Press.Search in Google Scholar

Di Tullio, Ángela. 1998. Complementos no flexivos de verbos de percepción física en español. Verba 25. 197–221.Search in Google Scholar

Dikken, Marcel den. 2006. Relators and linkers. The syntax of predication, predicate inversion and Copulas. Cambridge, Mass.: MIT Press.10.7551/mitpress/5873.001.0001Search in Google Scholar

Duarte, Inês. 1992. Complementos Infinitivos Preposicionados e outras Construções Temporalmente Defectivas em Português Europeu. In Actas do VIII Encontro da Associação Portuguesa de Linguística, 145–158. Lisboa: APL.Search in Google Scholar

Duarte, Inês & Anabela Gonçalves. 2002. Construções de subordinação funcionalmente defectivas: O caso das construções perceptivas em Português Europeu e em Português Brasileiro. In Actas do XVII Encontro Nacional da APL (Lisboa, Out. 2001), 161–173. Lisboa: APL.Search in Google Scholar

Fábregas, Antonio. 2008. Categorías híbridas en Morfología Distribuida. El caso del gerundio. In María José Rodríguez-Espiñeira & Jesús Pena (eds.), Categorías híbridas y límites intercategoriales, 57–87. Santiago: Universidad de Santiago.Search in Google Scholar

Fábregas, Antonio & Ángel Jiménez-Fernández. 2016. Extraction from gerunds and the internal structure of verbs. Linguistics 54(6). 1307–1354.10.1515/ling-2016-0029Search in Google Scholar

Felser, Claudia. 1999. Verbal complement clauses. A minimalist study of direct perception constructions. Amsterdam: John Benjamins.10.1075/la.25Search in Google Scholar

Fernández Lagunilla, Marina. 1999. Las construcciones de gerundio. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española, Vol. 2. 3443–3500. Madrid: Espasa.Search in Google Scholar

Fernández Lagunilla, Marina. 2011a. Restricciones de la negación con el gerundio adjunto modal. In M. Victoria Escandell Vidal, Manuel Leonetti & Cristina Sánchez López (eds.), 60 problemas de gramática. Dedicados a Ignacio Bosque, 247–252. Madrid: Akal.Search in Google Scholar

Fernández Lagunilla, Marina. 2011b. El gerundio en función de adjetivo y la oración de relativo. In José J. de Bustos Tovar, Rafael Cano-Aguilar, Elena Méndez García de Paredes & Araceli López Serena (eds.), Sintaxis y análisis del discurso hablado en español. Homenaje a Antonio Narbona, 763–778. Sevilla: Universidad de Sevilla.Search in Google Scholar

Fiéis, Alexandra & Maria Lobo. 2009. Aspectos da sintaxe das orações gerundivas no Português Medieval e no Português Europeu Contemporâneo. In Ana Maria Brito et al. (ed.), Textos seleccionados do XXV Encontro Nacional da Associação Portuguesa de Lingüística, 419–434. Porto: Associação Portuguesa de Lingüística.Search in Google Scholar

Fiéis, Alexandra & Maria Lobo. 2011. Propriedades de gerúndios e de infinitivos em português antigo e em variedades não standard do português europeu contemporâneo. In Armanda Costa, Pilar Barbosa & Isabel Falé (eds.), Textos Seleccionados do XXVI Encontro Nacional da Associação Portuguesa de Linguística, 256–265. Lisboa: APL.Search in Google Scholar

Franco, Ludovico & M. Rita Manzini. 2017. Instrumental prepositions and case: Contexts of occurrence and alternations with datives. Glossa 2(1). 8.10.5334/gjgl.111Search in Google Scholar

Gallego, Ángel & M. Lluïsa Hernanz. 2012. Tipos de tiempo defectivo. In Emilio Ridruejo Alonso (ed.), Tradición y progreso en la lingüística general, 197–217. Valladolid: Universidad de Valladolid.Search in Google Scholar

Gallego, Ángel. 2010. On the prepositional nature of non-finite verbs. Catalan Journal of Linguistics 9. 79–102.10.5565/rev/catjl.95Search in Google Scholar

Giorgi, Alessandra & Fabio Pianesi. 1996. Verb movement in Italian and syncretic categories. Probus 8. 137–160.10.1515/prbs.1996.8.2.137Search in Google Scholar

Grillo, Nino & Keir Moulton. 2015. Sorting out Pseudo-Relatives ling.auf.net/lingbuzz/002738/current.pdf (accessed 7 February 2018).Search in Google Scholar

Guasti, Maria Teresa. 1992. Pseudorelatives and prepositional infinitives. A unified account. Geneva Generative Papers 1. 53–65.Search in Google Scholar

Hale, Kenneth. 1986. Notes on world view and semantic categories. Some Warlpiri examples. In Pieter Muysken & Henk van Riemsdijk (eds.), Features and projections, 233–254. Dordrecht: Foris.10.1515/9783110871661-009Search in Google Scholar

Hernanz, M. Lluïsa. 1994. Concordancia, rección y aspecto: Las contrucciones absolutas en español. In Beatriz Garza Cuarón, José A. Pascual & Alegría Alonso González (eds.), II encuentro de lingüistas y filólogos de España y México, 367–402. Salamanca: Universidad de Salamanca.Search in Google Scholar

Hornstein, Norbert. 1999. Movement and control. Linguistic Inquiry 30(1). 69–96.10.1162/002438999553968Search in Google Scholar

Iliescu, Maria & Louis Mourin. 1991. Typologie de la morphologie verbale romane. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.Search in Google Scholar

Landau, Idan. 2000. Elements of control: Structure and meaning in infinitival constructions. Dordrecht: Kluwer.10.1007/978-94-011-3943-4Search in Google Scholar

Liver, Ricarda. 1991. Manuel pratique de romanche, sursilvan vallader. Précis de grammaire suivi d’un choix de textes. Chur: Lia Rumantscha.Search in Google Scholar

Lobo, Maria. 2002. Aspectos da sintaxe das orações gerundivas adjuntas do português. In Actas do XVII Encontro Nacional da APL, 247–265. Lisboa: APL.Search in Google Scholar

Lobo, Maria. 2006. Dependências temporais: A sintaxe das orações subordinadas gerundivas do português. Veredas 10(1–2). 1–26.Search in Google Scholar

Lobo, Maria. 2013. Subordinação adverbial. In Eduardo Raposo, Maria F. Bacelar Do Nascimento, Maria A. C. Mota, Luísa Segura & Amália Mendes (eds.), Gramática do Português, 1981–2057. Lisboa: Fundação Calouste Gulbenkian.Search in Google Scholar

Manzini, M. Rita & Ludovico Franco. 2016. Goal and DOM datives. Natural Language & Linguistic Theory 34. 197–240.10.1007/s11049-015-9303-ySearch in Google Scholar

Manzini, M. Rita, Leonardo Savoia & Ludovico Franco. To appear. DOM and Dative in Italo-Romance. In Andras Barany & Laura Kalin (eds.), Case, agreement, and their interactions: New perspectives on differential object marking, Berlin/New York: Mouton de Gruyter.Search in Google Scholar

Masullo, Pascual. 2008. The syntax-lexical interface. Prepositionalizing motion verbs in Spanish. Ms., University of Pittsburgh.Search in Google Scholar

Mateu, Jaume. 2002. Argument structure: Relational construal at the syntax-semantics interface. Barcelona: Universitat Autònoma de Barcelona dissertation.Search in Google Scholar

Móia, Telmo & Evani Viotti. 2004. Differences and similarities between European and Brazilian Portuguese in the use of the “gerúndio”. Journal of Portuguese Linguistics 3. 111–139.10.5334/jpl.21Search in Google Scholar

Moro, Andrea. 1997. The raising of predicates. Cambridge: Cambridge University Press.10.1017/CBO9780511519956Search in Google Scholar

Moro, Andrea. 2000. Dynamic antisymmetry. Cambridge, MA: MIT Press.Search in Google Scholar

Oliveira, Fátima, Luís F. Cunha & Gonçalves Anabela. 2004. Aspectual verbs in European and Brazilian Portuguese. Journal of Portuguese Linguistics 3. 141–173.10.5334/jpl.22Search in Google Scholar

Panagiotidis, E. Phoevos. 2010. Nonargumental Mixed Projections. Syntax 13(2). 165–182.10.1111/j.1467-9612.2009.00134.xSearch in Google Scholar

Pereira, Sílvia Afonso. 2015. Predicative constructions with gerunds in European Portuguese dialects. Dialectologia. Special Issue 5. 351–371.Search in Google Scholar

Pesetsky, David & Esther Torrego. 2004. Tense, case, and the nature of syntactic categories. In Jacqueline Guéron & Alexander Lecarme (eds.), The syntax of time, 495–537. Cambridge, MA: MIT Press.Search in Google Scholar

Radford, Andrew. 1975. Pseudorelatives and the unity of subject raising. Archivum Linguisticum 6. 32–64.Search in Google Scholar

Rafel, Joan. 2000. Complex Small Clauses. Barcelona: Universitat Autònoma de Barcelona dissertation.Search in Google Scholar

Raposo, Eduardo. 1989. Prepositional infinitival constructions in European Portuguese. In Osvaldo Jaeggli & Kenneth J. Safir (eds.), The null subject parameter, 277–305. Dordrecht: Kluwer.10.1007/978-94-009-2540-3_10Search in Google Scholar

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of Grammar, 281–337. Dordrecht: Kluwer.10.1007/978-94-011-5420-8_7Search in Google Scholar

Schifano, Norma. 2018. Verb movement in Romance. A comparative study. Oxford: Oxford University Press.10.1093/oso/9780198804642.001.0001Search in Google Scholar

Stowell, Tim. 1983. Subjects across categories. The Linguistic Review 2. 285–312.Search in Google Scholar

Tekavčić, Pavao. 1972. Grammatica storica dell’italiano, Vol. 2. Morfosintassi. Bologna: il Mulino.Search in Google Scholar

Tissot, Livio. 1976. Dizionario primierotto: Parole, frasi, modi di dire, proverbi del dialetto di Primiero. Trento: Provincia autonoma di Trento.Search in Google Scholar

Torrego, Esther. 2010. Variability and the case patterns of causative formation in Romance and its implications. Linguistic Inquiry 41. 445–470.10.1162/LING_a_00004Search in Google Scholar

Williams, Edwin. 1975. Small Clauses in English. In John P. Kimball (ed.), Syntax and semantics, Vol. 4. 249–273. New York: Academic Press.10.1163/9789004368828_010Search in Google Scholar

Zanuttini, Raffaella. 1997. Negation and clausal structure. A comparative study of Romance languages. Oxford/New York: Oxford University Press.Search in Google Scholar

Published Online: 2019-05-25
Published in Print: 2019-05-27

© 2019 Casalicchio, published by De Gruyter

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Downloaded on 28.3.2024 from https://www.degruyter.com/document/doi/10.1515/probus-2019-0003/html
Scroll to top button