Again on the Existence of Causative Periphrases in Spanish: The Case of “enviar/mandar a + Infinitive”

Abstract

The concept of verbal periphrasis has historically been a controversial one in Romance linguistics, especially in the Hispanic context, where there has been disagreement as to what multiverbal constructions should be considered periphrastic. One of the points of contention has been the class of infinitive causatives. This article revisits the controversy by focusing on Spanish “enviar/mandar a + infinitive” structures and drawing on historical corpus data. The analysis of various examples leads to the conclusion that strictly periphrastic instances of this constructional class are present across all main stages of the history of Spanish. Additionally, a series of quantitative analyses reveals what appear to be two distinct grammaticalization processes and a degrammaticalization process. These findings are discussed in connection with broader themes in the field, such as syntactic ambiguity and the concept of analyzability.

1. Introduction

Having been traditionally conceived as constructions1 that, despite involving the use of two verb forms, exhibit some kind of grammatical unity or cohesion as a result of grammaticalization, verbal periphrases (e.g., Span. “tener que + infinitive”: tengo que estudiar ‘I have to study’) are of special interest for the study and conceptualization of morphosyntactic analyzability. Indeed, besides exhibiting the seemingly paradoxical property of behaving like verbal heads2 despite their multiverbal nature, verbal periphrases, as defined above, are a prime example of how language users can recast lexical signs as grammatical instruments (grammaticalization in the narrow sense) thereby changing the underlying syntactic arrangement of certain schemata (syntactic reanalysis above the word level). Moreover, because the emergence of verbal periphrases from other (non-periphrastic) constructions typically leads to cases of syntactic ambiguity, linguists studying the former are forced to reflect on what it means for a structure to be “analyzable” in a certain way.

Although verbal periphrases constitute a large portion of the grammatical inventory of many languages (including mainly inflectional ones such as those of the Romance family), and terms such as verbal periphrasis and periphrastic conjugation have been commonplace in grammatical description for over a century (see Campión, 1884, p. 313 and passim; Donovan, 1922, pp. 125–163; Lanchetas, 1897, p. 21 and passim; Menéndez Pidal, 1908, p. 342; Zumpt, 1818/1850, pp. 163–166), there is disagreement as to what multiverbal structures ought to be regarded as periphrastic. This is, no doubt, partially due to the fact that terms such as the aforementioned have been used in different ways by different authors (see Haspelmath, 2000, pp. 654–655). However, not all disagreement can be attributed to terminological differences, since authors who appear to operate with essentially the same concept of verbal periphrasis—or at least with compatible ones—nevertheless assign different extensions to the class. When it comes to the Romance languages, one of the points of contention, especially in the Hispanic context, has been the class of infinitive causative, or factitive, structures (see Aissen, 1979, pp. 35–89; Alfonso Vega, 1998, pp. 35–41; Fábregas, 2019, pp. 71–73; Fernández de Castro, 1999, pp. 26–29, 97–99; Gougenheim, 1929, pp. 310–377; Porto Dapena, 1977, pp. 37–39), illustrated in (1) and (2) with medieval and present-day Spanish examples, respectively:3

(1)

a.

Señor […],

el

infante

Roboán

te

embía

a

lord

the

prince

Roboán

you

send.prs.3sg

to

rogar

que […]

no

quieras

hazer

mal

en

el

beg.inf

that

not

want

do

wrong

in

the

reino

de

Pandulfa

kingdom

of

Pandulfa

‘Lord, Prince Roboán asks you to plan no wrong in the kingdom of Pandulfa’ (Anonymous, Libro del cavallero Cifar, 1300–1305 [last third of the 15th c.], Spain [RAE, 2015])

b.

&

fizo=les

matar

muchos

daquellos

monges

and

make.pret.3sg=them

kill.inf

many

of.those

monks

‘And he made them kill many of those monks’ (Alfonso X, Estoria de Espanna, ca. 1270 [last fourth of the 13th c.], Spain [RAE, 2015])

c.

Estonçe

el

rey

mandó

llamar

todos

los

grandes

then

the

king

order.pret.3sg

call.inf

all

the

great

honbres

de

su

corte

men

of

his

court

‘Then the king sent for all the great men in his court’ (Gutierre Díaz de Games, El Victorial, 1431–1449 [late 15th c. or early 16th c.], Spain [RAE, 2015])

d.

assi

como

te

leuantares

manda

a

thus

as

yourself

rise

order/send.imp.sg

to

limpiar

todas

las

plaças

&

las

calles

de

clean.inf

all

the

public.squares

and

the

streets

of

la

cibdad

the

city

‘as soon as you rise have all the public squares and streets of the city cleaned’ (Alfonso X, General Estoria. Cuarta parte, ca. 1280 [1280], Spain [RAE, 2015])

(2)

a.

Incluso

había

enviado

a

decir

a

sus

even

have.impf.3sg

send.ptcp.m.sg

to

say.inf

to

his

visitantes

que

la

perdonaran

visitors

that

her

forgive

‘She had even sent word to her visitors asking them to forgive her’ (Pablo Simonetti, La soberbia juventud, 2014, Chile [RAE, 2025])

b.

Este

frío

me

hace

dormir

abrazado

a

ti

this

cold

me

make.prs.3sg

sleep.inf

embraced

to

you

‘The cold makes me sleep hugging you’ (Vicente Molina Foix, El abrecartas, 2006, Spain [RAE, 2025])

c.

Uno

de

los

hijos

mandó

construir

un

one

of

the

children

order.pret.3sg

build.inf

a

mausoleo

pomposo

mausoleum

pompous

‘One of his children had a pompous mausoleum built’ (Daniel Sada, Casi nunca, 2008, Mexico [RAE, 2025])

d.

Dicen

que

mandó

a

matar

a

su

padre

say

that

order/send.pret.3sg

to

kill.inf

to

his

father

‘They say he had his father killed’ (Fernan Espinosa, Mi vida por un libro, 2001, Colombia [RAE, 2025])

In this article, I revisit the controversy focusing on “enviar/mandar a + infinitive” structures such as (1a), (1d), (2a), and (2d), considered in diachronic perspective. Such structures are deemed worthy of attention for two main reasons. First, despite insightful early work on them by Porto Dapena (1977; Cuervo & ICC, 1978/1998a, pp. 713–714), they have been largely overlooked in discussions about the existence (or lack thereof) of causative periphrases in Spanish. Second, given the verbal meaning at the historical base of these structures (‘to send’), they offer an opportunity to document a scarcely attested grammaticalization pathway.

The remainder of the text is structured as follows. In Section 2, I offer a working definition of verbal periphrasis and conduct a selective review of the literature on Spanish verbal periphrases and infinitive causatives. In Section 3, I revisit Porto Dapena’s data and arguments and derive diagnostics for identifying periphrastic instances of “enviar/mandar a + infinitive.” In Section 4, I use these diagnostics to investigate the presence of periphrases in question in two corpora—CDH (RAE, 2015) and CORPES XXI (RAE, 2025)—and I offer a general semantic characterization of the relevant structures. In Section 5, I further investigate the diachrony of “enviar/mandar a + infinitive” drawing on quantitative data from the aforementioned corpora. Finally, in Section 6, I conclude with a recapitulation and a few remarks connecting my findings with broader themes in the field, including the concept of analyzability.

2. Spanish Infinitive Causatives and the Concept of Verbal Periphrasis: Findings and Problems

As already pointed out, there is no consensus regarding the grammatical status of infinitive causatives such as those in (1–2), which are regarded as periphrastic by some scholars (e.g., Cano Aguilar, 1977a, p. 243 and passim; 1977b; Enghels & Comer, 2020; Mourelle de Lema, 1981, pp. 17–19) and as non-periphrastic by others (e.g., Fábregas, 2019, pp. 71–73; Fernández de Castro, 1999, pp. 26–29, 97–99; Olbertz, 1998, p. 38), while other authors simply omit them (e.g., García Fernández et al., 2006; Morera, 1991; Roca Pons, 1958). To add to the confusion, a third group of scholars, while not using terms such as periphrasis or periphrastic, characterizes some of the relevant structures as having properties traditionally associated with verbal periphrases or refers to the finite verb forms in question as auxiliaries (e.g., Bello, 1847/1891, pp. 319–320; Davies, 1995, pp. 71–74; Pottier, 1961, p. 329).

Since terms such as the aforementioned do not always carry exactly the same meaning in the literature, some terminological remarks are in order at this point. Although certain scholars would regard any structure involving a finite verb form and a non-finite one sharing some kind of syntactic link as periphrastic, especially if the second verb form immediately follows the first, others require that the two verb forms show some kind of grammatical unity or cohesion, as indicated in the introduction. This will be the case here, too. Although particular definitions and criteria vary, I will follow Coseriu (1976/1996, pp. 126–130) and van Pottelberge (2007), who view verbal periphrases as linguistic signs in essentially the Saussurean sense—a view I take to be compatible with those of most authors espousing a somewhat narrow concept of verbal periphrasis. More precisely, I shall define verbal periphrases, as found in the Romance languages, as partially schematic signs whose signifier (signifiant) consists of a grammaticalized verbal base, to be inflected (usually in finite form4) according to the syntactic context and the speaker’s communicative intention, and a slot for a verb indicating the nature of the event as conceptualized by the speaker, to be inflected in a specific non-finite form (potentially in addition to a linking element corresponding to a preposition or conjunction), and whose signified (signifié), on the other hand, can correspond to a variety of grammatical categories. Consistently with grammatical tradition, the first verb form in a structure instantiating a verbal periphrasis will be said to be that of the auxiliary verb, while the second will be said to be that of the main or auxiliated verb.

The above definition, which is consistent with the findings of construction grammar on schematicity (see Bybee, 2010, pp. 22–28), captures the nature of uncontroversial periphrases such as Spanish “tener que + infinitive” (e.g., tengo que estudiar ‘I have to study’; see Fernández de Castro, 1999, pp. 185–189; RAE & ASALE, 2009, pp. 2141–2145; Roca Pons, 1958, p. 72), where the auxiliary is materially (i.e., “formally”) identical, in all conjugations, to a full verb in the present-day language, whose meaning nevertheless survives only vestigially in the periphrasis. Additionally, this definition applies to constructions whose meaning, productivity, and degree of grammaticalization often make grammarians forget that they are, in fact, periphrases. Such is the case, for example, of Spanish “haber + participle” (e.g., he estudiado ‘I have studied,’ había estudiado ‘I had studied’; see Alarcos Llorach, 1994/1999, p. 264; Coseriu, 1976/1996, pp. 130–131; Lanchetas, 1897, pp. 89–90), which usually are only considered in relation to specific (simple) tenses, and where the auxiliary (unlike those of the corresponding constructions in most other Romance languages) mirrors an archaic verb (see Roca Pons, 1958, p. 95).5 By contrast, the proposed definition does not apply to constructions in which the two verbs, no matter their frequency of combination or their joint productivity, constitute distinct lexical signs, such as “desear + infinitive” (e.g., deseo comer ‘I want to eat’) or “estar + participle” (e.g., está cansada ‘she is tired’) (see Fernández de Castro, 1999, pp. 63–64, 92–94; Gómez Torrego, 1999, pp. 3336–3337; Lüdtke, 1990, pp. 203–204).

When it comes to infinitive causatives, those opposed to treating them as periphrastic have offered several arguments, some of which, taken at face value, would seem to prove their incompatibility with the definition provided here (see Alfonso Vega, 1998, pp. 35–41; Fábregas, 2019, pp. 71–73; Fernández de Castro, 1999, pp. 26–29, 97–99; Fernández Martín, 2018, pp. 56, 66–67; Gómez Torrego, 1999, pp. 3327–3328; Vivanco, 2019, pp. 50–51). These arguments can be divided into semantic and syntactic, though, of course, one should expect the semantic properties of a construction to be reflected at the syntactic level, and vice versa.

On the semantic side, scholars have pointed to evidence taken to indicate that the two verb forms in the relevant structures contribute their own lexical meaning and argument structure. For instance, in (3a), the lexical autonomy of hizo is evidenced by the fact that its first semantic argument is clearly distinct from the first semantic argument of condenar. Similarly, in (3b), the independent semantic contribution of obligaban is made patent by the adverb espresamente, which can only modify a verb referring to a speech act.

(3)

a.

Nos

hizo

condenar=te

us

make.pret.3sg

sentence.inf=you

‘He made us sentence you’ (Fernández de Castro, 1999, p. 28)

b.

reafirmar

las

pazes

y

confederaciones

antiguas

que

confirm

the

peaces

and

confederations

old

that

perpetuamente

fueron

juradas

por

los

reyes

pasados

de

perpetually

were

sworn

by

the

kings

previous

of

Francia […]

que

espresamente

obligauan

a

mantener

France

that

expressedly

obligate.impf.3pl

to

keep.inf

y

guardar

a

sus

subcesores

and

save.inf

to

their

successors

‘ratify the old pacts and confederations that were sworn by the former kings of France, which their successors were expressly obligated to uphold and preserve’ (Hernando del Pulgar, Crónica de los Reyes Católicos, prob. 1480–1484, Spain [Alfonso Vega, 1998, p. 40])

On the syntactic side, scholars have pointed to a number of properties taken to be characteristic of non-periphrastic multiverbal structures, such as different clitic positions for each verb form, as in (3a); absence of clitic promotion or “climbing,” as in (4a); non-contiguous verb forms, as in (4b); and the possibility of paraphrase with a complementizer phrase, as in (4c):

(4)

a.

Juan

le

hizo

escribir=la

Juan

him

make.pret.3sg

write.inf=it

‘Juan made him write it’ (Fábregas, 2019, p. 71)

b.

tienen

los

omnes

que

el

faze

a

ellos

consider

the

men

that

he

make.prs.3sg

to

them

despreciar=le

despise.inf=him

‘men hold that he makes them despise him’ (Pero López de Ayala, Crónica de Pedro I, ca. 1400, Spain [Alfonso Vega, 1998, p. 37])

c.

Macarena

hizo

{ jugar

a

Luis

/

que

Luis

Macarena

make. pret.3sg

play.inf

to

Luis

that

Luis

jugara }

play.3sg.impf.sbjv

‘Macarena made Luis play’ (Vivanco, 2019, p. 46)

There are, however, at least two problems with these lines of argument. First, traditional periphrasticity tests have proven fallible (see Fábregas, 2019, pp. 3–15; Fernández de Castro, 1999, pp. 38–57; RAE & ASALE, 2009, pp. 2116–2125). For example, the clitic promotion test, one of the oldest and most widely used in Romance linguistics (see Aissen, 1979, pp. 56–59, 63; Fábregas, 2019, pp. 49–55; Keniston, 1936, p. 163; Lenz, 1920, pp. 374–375; Myhill, 1988), does not lead to the automatic exclusion of uncontroversially non-periphrastic structures such as (5a), nor does it lead to the automatic inclusion of uncontroversially periphrastic structures such as (5b). Second, and most importantly, even if it were possible to analyze certain cases (or, in fact, all the relevant examples ever discussed in the specialized literature) as non-periphrastic, through the application of syntactic tests or semantic analysis, this would not constitute irrefutable evidence against the existence of causative periphrases in Spanish. Indeed, etymologically related and materially identical periphrastic and non-periphrastic constructions often coexist, giving rise to numerous cases of syntactic ambiguity that, nevertheless, may not compromise communication (see Coseriu, 1960/1962, p. 123; Davies, 2000, pp. 119–121; Juge, 2002, pp. 80–88; Porto Dapena, 1977; RAE & ASALE, 2009, p. 2108). Furthermore, as the scientific adage goes, absence of evidence is not evidence of absence.

(5)

a.

Este

disco

lo

deseaba

hacer

desde

hacía

diez

this

album

it

wish.impf.1sg

make.inf

since

made

ten

años

years

‘I had been wanting to make this album for ten years’ (La Vanguardia, 13/07/2021, Spain [RAE, 2025])

b.

Tenemos

que

hacer=lo

así

have.prs.1pl

that

do.inf=it

thus

‘We have to do it like this’ (Paloma Pedrero, El pasamanos, 2019, Spain [RAE, 2025])

Notwithstanding the above, the burden of proof lies with those claiming there to be true causative verbal periphrases in Spanish, and arguments such as those offered by Aissen (1979, pp. 35–89), Álvarez (2006), and Zubizarreta (1985, pp. 280–286), based on questionable grammaticality judgments and unreliable syntactic tests, clearly must be regarded as inadequate. But, as it turns out, there already exist studies that, drawing on various data, make a persuasive case for the existence of causative periphrases—or something like them—in the history of the language (e.g., Cuervo & ICC, 1978/1998a, pp. 713–714; Davies, 1995, 1996; Enghels & Comer, 2020; Porto Dapena, 1977; Soares da Silva, 2012). Unfortunately, however, they appear to have been completely ignored by those who explicitly reject the notion of a true causative periphrasis in this language.

The first studies to be discussed here, as they are among the oldest and most overlooked and deal specifically with the constructional class at the center of the present work, are a short article and a dictionary entry by Porto Dapena (1977; Cuervo & ICC, 1978/1998a), the latter being the Diccionario de construcción y régimen (DCR) entry on enviar. In these studies, the author examines structures such as the following, which feature the verb enviar ‘to send’ or mandar ‘to order, to send’ followed by the preposition a ‘to’ and an infinitive:

(6)

a.

El

general

mandó

a

un

emisario

a

the

general

order/send.pret.3sg

to

an

emissary

to

negociar

la

paz

negotiate.inf

the

peace

‘The general sent an emissary to negotiate peace’

b.

Óscar

mandó

(a

alguien)

a

pedir=le

Óscar

order/send.pret.3sg

to

someone

to

ask.inf=him

consejo

a

Joaquín

advice

to

Joaquín

‘Óscar sent for Joaquín’s advice’

c.

Miguel

envió

su

bicicleta

a

arreglar

Miguel

send.pret.3sg

his

bicycle

to

fix.inf

‘Miguel sent his bicycle to have it fixed’

(Porto Dapena, 1977, pp. 28, 35, 30)

(7)

a.

Andrés

envió

a

decir

esto

a

su

hijo

{ por

/

Andrés

send.pret.3sg

to

say.inf

this

to

his

son

by

con }

un

amigo

with

a

friend

‘Andrés had a friend say this to his son’

b.

Óscar

mandó

a

pedir=le

consejo

a

Óscar

order/send.pret.3sg

to

ask.inf=him

advice

to

Joaquín

Joaquín

‘Óscar sent for Joaquín’s advice’

c.

Mandé

a

llamar=lo

en

una

carta

order/send.pret.1sg

to

call.inf=him

in

a

letter

‘I summoned him by letter’

(Porto Dapena, 1977, pp. 35–36)

According to Porto Dapena, despite their material and semantic similarities, the sentences in (6) and (7) are quite different syntactically. Evidently, the examples in (6) cannot be periphrastic, since the finite verb forms have their own objects.6 However, according to the author, the situation is different in (7), and not merely because the finite verb forms lack an overt object or are contiguous with the preposition and the infinitive, but mainly because there are semantic and syntactic factors that make it difficult even to imagine an object for the finite verb forms without major semantic changes. Thus, in (7a), a separate object for envió would create a conflict with the prepositional phrase por/con un amigo, which already refers to the intermediary or causee of the action; in (7b), a parallel constituent, as in (6b), would make it unclear whether the sought advice for Óscar (the initiator or causer) or for his intermediary; and, in (7c), it would shift attention to an intermediary other than the letter itself. Thus, Porto Dapena concludes that examples such as those in (7) are instances of a periphrasis, with enviar and mandar functioning as modal auxiliaries. As for the productivity of the construction, the author remarks that it is only attested with verbs of speech in 20th-century usage, as in (7), though he points out that this was not always the case.

A rather different line of research that has led to interesting findings concerning Spanish causatives focuses on syntactic properties traditionally used as periphrasticity or syntactic reduction tests (e.g., clitic promotion and others discussed above) but considers them from a diachronic and quantitative perspective. This line of inquiry is perhaps best represented by Davies (1995, 1996),7 who studies various infinitive causatives in a historical corpus of Spanish and Portuguese. More specifically, this scholar—who, adopting a generativist term, speaks of reduced rather than periphrastic structures (see Goodall, 1987, p. 135 and passim; Pearce, 1990, p. 25 and passim; Zubizarreta, 1985, p. 281 and passim)—examines a series of historical alternations pertaining to “case” marking, clitic placement, reflexivization, and word order, respectively. These alternations are illustrated below:

(8)	a.	le	hicieron		comer		el		pastel
		3sg.dat	make.pret.3pl		eat.inf		the		cake
	b.	la		hicieron		comer		el	pastel
		3sg.f.acc		make.pret.3pl		eat.inf		the	cake
		‘they made her eat the cake’ (Davies, 1995, p. 59)

(9)	a.	me	lo	dejaron		comer
		me	it	let.pret.3pl		eat.inf
	b.	me	dejaron		comer=lo
		me	let.pret.3pl		eat.inf=it
		‘they let me eat it’ (Davies, 1995, p. 63)

(10)	a.	lo	dejaron	sentar
		him	let.pret.3pl	sit.inf
	b.	lo	dejaron	sentar=se
		him	let.pret.3pl	sit.inf=refl
		‘they let him sit’ (Davies, 1995, p. 59)

(11)	a.	le	hicieron	comer		el		pastel	a	Pedro
		to.him	make.pret.3pl	eat.inf		the		cake	to	Pedro
	b.	le	hicieron	a	Pedro		comer		el		pastel
		to.him	make.pret.3pl	to	Pedro		eat.inf		the		cake
		‘they made Pedro eat the cake’ (Davies, 1995, p. 59)

Davies takes the properties of (8b), (9b), (10b), and (11b) (accusative marking, lack of clitic promotion, reflexivization, and SV[O] word order) to be characteristic of non-reduced structures, which are assumed to feature a “lower” subject position. By contrast, he takes the properties of (8a), (9a), (10a), and (11a) (dative marking, clitic promotion, lack of reflexivization, and V[O]S word order) to be characteristic of reduced structures, which are assumed to lack a “lower” subject position. Importantly, however, in practice, Davies uses these properties not as mere tests to determine which examples are reduced structures and which are not, but rather as quantitative indicators of syntactic shift. Thus, from a methodological standpoint, Davies’s work can be grouped with variationist studies such as those by Laca (2002), Myhill (1988), Rosemeyer (2016), and Torres Cacoullos (1999), among many others.8 By examining the frequency of the properties exemplified in (8–11) in each period, the author identifies four parallel shifts between the Middle Ages and the present day in both Spanish and Portuguese—from dative marking, clitic promotion, lack of reflexivization, and V(O)S word order to accusative marking, lack of clitic promotion, reflexivization, and SV(O) word order—which he interprets as evidence of a more general shift from reduced to non-reduced causatives. Furthermore, he relates this shift—which, as he shows, is not unique to causatives (see Davies, 1995, p. 74 and passim)—to the loss of the distinction between active and passive infinitive forms in Late Latin (e.g., aedificāre, aedificāri > aedificāre), which caused the formerly active forms to be interpreted mainly passively (an effect that persisted in Old Romance).

Even if Davies does not use the terms periphrasis or periphrastic but instead chooses to speak of reduced structures, the two notions are clearly related, as the concept of verbal periphrasis implies a process of syntactic reduction, even if such a process is not exactly what generativists such as those cited by Davies mean by this term. Thus, it does not seem far-fetched to interpret Davies’s findings as pointing to the existence of productive causative periphrases in the medieval language, which nonetheless would seem to have declined or even disappeared in the course of subsequent centuries. In fact, this would be consistent with the author’s own interpretation of the shift from the perspective of grammaticalization (Davies, 2000, p. 112ff.), as well as with Soares da Silva’s (2012, p. 54ff.) explicit description of it as a process of degrammaticalization.

Despite the coexistence of approaches such as Porto Dapena’s and Davies’s, both of which have offered valuable insights for understanding Spanish causatives, I know of no studies taking what could be described as a mixed approach to argue for or against the existence of true verbal periphrases in this domain. Moreover, the allegedly periphrastic nature of certain cases of “enviar/mandar a + infinitive” has not been properly revisited since Porto Dapena’s work on the subject.9 In the remainder of the article, I will attempt to fill this gap. Specifically, after a critical examination of Porto Dapena’s data and arguments, I will conduct a diachronic investigation of this constructional class, presenting both qualitative and quantitative evidence pertaining to the nature and historical development of the relevant constructions.

3. Porto Dapena’s Data and Arguments Revisited

As we have seen, Porto Dapena believes that not all “enviar/mandar a + infinitive” structures are the same syntactically. Crucially, he argues this to also be true of materially identical structures such as (6b) (in the version with the omitted direct object) and (7b), repeated below as (12a) and (12b), respectively. According to the author, only the latter structure is truly periphrastic, since in the former mandar retains its lexical autonomy and therefore contributes its own argument structure, which includes a second semantic argument expressible as a direct object. Thus, the contrast in (12) serves as a reminder of the importance of taking into account syntactic ambiguity when attempting to distinguish periphrastic and non-periphrastic structures.

(12)	a.	Óscar	mandó		(a	alguien)	a		pedir=le
		Óscar	order/send.pret.3sg		to	someone	to		ask.inf=him
		consejo	a	Joaquín
		advice	to	Joaquín
	b.	Óscar	mandó		a	pedir=le		consejo		a
		Óscar	order/send.pret.3sg		to	ask.inf=him		advice		to
		Joaquín
		Joaquín
		‘Óscar sent for Joaquín’s advice’
		(Porto Dapena, 1977, p. 35)

The key to dealing with syntactic ambiguity is, of course, to devise reliable diagnostics that allow the relevant constructions to be teased apart. In his attempt to identify such diagnostics, Porto Dapena achieves mixed results, partly due to his occasional reliance on constructed examples, his own grammaticality judgments, and traditional periphrasticity tests. For example, the author attempts to demonstrate the periphrastic nature of (7a), repeated below as (13a), by transforming it into (13b) and (13c), where enviar introduces a demonstrative pronoun and a complementizer phrase, respectively:10

(13)

a.

Andrés

envió

a

decir

esto

a

su

hijo

{ por

/

Andrés

send.pret.3sg

to

say.inf

this

to

his

son

by

con }

un

amigo

with

a

friend

‘Andrés had a friend say this to his son’

b.

Andrés

envió

esto

a

decir

a

su

hijo

Andrés

send.pret.3sg

this

to

say.inf

to

his

son

c.

*Andrés

envió

que

viniera

a

decir

Andrés

send.pret.3sg

that

come.3sg.impf.sbjv

to

say.inf

a

su

hijo

to

his

son

(Porto Dapena, 1977, pp. 36–37)

According to the author, (13b) would only be possible as “un caso de hipérbaton bastante violento” [‘a rather violent case of hyperbaton’] (p. 36), while (13c) would be completely unacceptable. However, these claims are at odds with examples such as the following:

(14)

a.

E

maguer

que

don

Joan

le

enbiaua

esto

and

although

that

Don

Juan

to.him

send.IMPF.3sg

this

a

dezir

to

say.inf

‘And although Don Juan sent this to be said to him’ (Anonymous, Gran crónica de Alfonso XI, ca. 1348–1379 [15th/16th c.], Spain [RAE, 2015])

b.

A

la

hora

de

cenar

envió

que

to

the

time

of

dine

send.pret.3sg

that

viniesen

come.3pl.impf.sbjv

‘At dinnertime he sent for them to come’ (El Universal.com.mx, 20/12/2020, Mexico [RAE, 2025])

Such cases, though rare, suggest that the direct object of enviar as a full verb can refer to an abstract linguistic or psychological entity, such as the content of a message, request, or order. This should not come as a surprise, since both enviar and mandar can take objects headed by nouns such as mensaje ‘message,’ petición ‘request’, and orden ‘order,’ even if the message, request, or order in question is not written, as illustrated by the following example:

(15)

Mendieta

compareció

el

pasado

domingo

ante

los

medios

de

Mendieta

appeared

the

past

Sunday

before

the

media

of

comunicación

con

la

intención

de

enviarle

un

mensaje

communication

with

the

intention

of

send

a

message

al

Valencia

to.the

Valencia

‘Mendieta appeared before the media last Sunday with the intention of sending a message to Valencia’ (El País, 17/07/2001, Spain [RAE, 2025])

When it comes to (13a), this means that esto could well be the object of envió, with the infinitive having a passive meaning (decir ~ ser dicho ‘to be said’; see n. 6)—a situation that would explain the relegation of the intermediary to a prepositional phrase. Thus, cases of this type, whether they feature a verb of speech or another type of verb exhibiting similar complementation patterns, are not ideal examples to demonstrate the existence of a verbal periphrasis.11

Another limitation of Porto Dapena’s argumentation has to do with the meaning of enviar and mandar as full verbs. In this connection, the author assumes that mandar—whose original and still primary meaning is ‘to order, to command’12 (see Corominas & Pascual, 1980, s.v. mandar; RAE, 2014, s.v. mandar)—is always synonymous with enviar when followed by a + infinitive, at least in Peninsular Spanish. However, this is a debatable generalization. It is true that mandar is often interchangeable with enviar in the modern language, even if ‘to order’ remains its primary acceptation. Moreover, as Sanaphre Villanueva (2010, pp. 149–152) has shown, the schema “mandar a + infinitive” appears to have emerged and developed by analogy with “enviar a + infinitive,” which might explain the semantic convergence of the two verbs.13 Nevertheless, given the coexistence of both acceptations, ‘to order’ and ‘to send,’ mandar can be argued to still retain a feature of authority or control as part of its basic, invariant meaning.14

The semantic duality of mandar is important in the present context because it can affect the interpretation of certain examples, especially of constructed examples for which not much contextual information is provided. For instance, consider again (7c), repeated below as (16):

(16)	Mandé	a	llamar=lo	en	una	carta
	order/send.pret.1sg	to	call.inf=him	in	a	letter
	‘I summoned him by letter’
	(Porto Dapena, 1977, p. 36)

While it is not difficult to imagine the letter mentioned in this example as the only thing being sent anywhere (since modern postal services are not something over which one has much control), (16) could well be used to refer to a situation in which someone was ordered to write a letter, with the second argument of mandé (i.e., the person being ordered to write the letter) being left unexpressed.

Despite these shortcomings, Porto Dapena is able to identify at least two useful diagnostics for recognizing otherwise ambiguous periphrastic instances of “enviar/mandar a + infinitive.” These diagnostics, which were already illustrated in Section 3, are the presence of a prepositional phrase indicating the intermediary or instrument of the action signified by the infinitive and the presence of reflexive clitics referring to the initiator of said action. Most importantly, the author provides convincing examples from the DCR corpus using these diagnostics:

(17)

a.

Pasé

con

este

trabajo

hasta

la

tarde,

que

passed

with

this

trouble

until

the

evening

that

envió

el

Retor

de

la

Compañía

a

ver=me

send.pret.3sg

the

rector

of

the

society

to

see.inf=me

con

un

padre

que

me

animó

y

consoló

with

a

father

that

me

encouraged

and

consoled

mucho

much

‘I spent the entire day dealing with this trouble, until the rector of the society sent a father to visit me, and he encouraged and consoled me a great deal’ (Santa Teresa de Jesús, Libro de las fundaciones, 1573–1582, Spain [Cuervo & ICC, 1978/1998a, p. 714])

b.

Muchos

señores

con

embaxadas

se

enviaron

a

many

lords

with

missives

refl

send.pret.3pl

to

encomendar

en

sus

oraciones

commend.inf

in

his

prayers

‘Many lords commended themselves to his prayers by missive’ (Antonio de Fuenmayor, Vida y hechos de Pío V, 1595, Spain [Cuervo & ICC, 1978/1998a, p. 714; Porto Dapena, 1977, p. 38, fn. 9])

c.

Sabiendo

quién

era

el

escribano

de

la

causa,

knowing

who

was

the

clerk

of

the

cause

envié=le

a

llamar

con

un

picarillo

send.pret.1sg=him

to

call.inf

with

a

little.scoundrel

‘Knowing who the clerk of the cause was, I sent a little scoundrel to fetch him’ (Francisco de Quevedo y Villegas, La vida del Buscón llamado don Pablos, 1626, Spain [Cuervo & ICC, 1978/1998a, p. 714])

In all three cases, the person or thing sent is indicated by means of a prepositional phrase—con un padre in (17a), con embaxadas in (17b), and con un picarillo in (17c)—rather than a direct object as would be expected if enviar retained its lexical autonomy. Additionally, in (17b), while the missives in question must have been delivered by envoys, interpreting enviar as a full verb referring to the sending of the latter is ruled out by the reflexive clitic se, which clearly refers to no other than the lords who sent the missives.15 Thus, the above examples are strong evidence of the existence of a causative verbal periphrasis.

Two remarks on the cases in (17) are in order. First, all three examples are from Golden Age (1492–1681) texts. This is an interesting fact, for two reasons. On the one hand, it suggests that the “enviar a + infinitive” periphrasis may have declined or even disappeared in subsequent centuries—a process that could turn out to be a part of the shift hypothesized by Davies. On the other hand, it raises the question of whether the consolidation or maximum entrenchment of this grammatical unit occurred relatively late—something that would not fully align with Davies’s findings and hypotheses, as this scholar describes a series of unidirectional syntactic shifts taking place since the Middle Ages and attributes the emergence of syntactically reduced causatives to the loss of the distinction between active and passive infinitive forms. Second, all three examples feature enviar, which raises the question of whether comparable cases with mandar are attested in the history of Spanish.16

4. Searching for Periphrastic Cases of “enviar/mandar a + Infinitive” in the History of Spanish

The characteristics and distribution of Porto Dapena’s historical examples warrant further study of the use of “enviar/mandar a + infinitive” in the history of Spanish, especially in the periods before and after the Golden Age. As it turns out, CDH and CORPES XXI show that cases comparable to those in (17) are not unique to this period. For example, beginning the exploration in the Middle Ages, CDH yields several examples of the following types, from variably reliable manuscripts and editions (see Rodríguez Molina & Octavio de Toledo y Huerta, 2017) and featuring a variety of verbs in infinitive form:

(18)

a.

&

desque

vido

que

no

venía,

embio=la

a

and

after

saw

that

not

came

send.pret.3sg=her

to

llamar

con

un

portero

call.inf

with

a

guard

‘And after she realized that she was not coming, she sent a guard to fetch her’ (Anonymous, Libro del cavallero Cifar, 1300–1305 [last third of the 15th c.], Spain [RAE, 2015])

b.

Et

sabet

que

por

este

tuerto

tan

grand

que

el

and

know

that

for

this

wrong

so

great

that

the

Rey

me

ha

fecho

que

me

enbio

a

king

me

has

done

that

refl

send.prs.1sg

to

despedir

e

a

desnaturar

del

take.leave.inf

and

to

forswear.inf

of.him

‘And know that on account of so great a wrong that the king has done me I hereby take my leave of him and forswear him’ (Anonymous, “Cartas de Don Juan Manuel y vasallos suyos enviados al Rey de Granada y concejos de Murcia,” 1327, Spain [RAE, 2015])

c.

et

que

don

Iohan

no

y

auie

quesido

venir,

and

that

Don

Juan

not

there

had

wanted

come

enuiando=se

a

escusar

por

tales

razones

que

el

send.ger=refl

to

excuse.inf

for

such

reasons

that

he

et

los

de

su

cort

pudieron

bien

conoscer

que

no

and

the

of

his

court

could

well

know

that

not

auie

uolumtat

de

venir

a

su

seruicio

had

will

of

come

to

his

service

‘and that Don Juan had not wished to come, sending such excuses that he and everyone in his court could clearly recognize that he had no intention of coming in his service’ (Juan Fernández de Heredia, Gran crónica de España, III, 1376 – a. 1391, Spain [RAE, 2015])

d.

el

rrey,

por

sus

cartas,

enbió

a

llamar

a

the

king

by

his

letters

send.pret.3sg

to

call.inf

to

don

Alonso

Carrillo,

arçobispo

de

Toledo,

y

a

Don

Alonso

Carrillo

bishop

of

Toledo

and

to

don

Pero

Fernández

de

Velasco,

conde

de

Haro

Don

Pedro

Fernández

de

Velasco

count

of

Haro

‘the king, through his letters, summoned Don Alonso Carrillo, bishop of Toledo, and Don Pedro Fernández de Velasco, count of Haro’ (Diego Enríquez del Castillo, Crónica de Enrique IV, ca. 1481–1502, Spain [RAE, 2015])

Like the examples in (17), these cases feature no constituent that could be interpreted as the direct object of enviar as a full verb. Additionally, in (18a) and (18d) the intermediary or instrument of the action is introduced by a phrase whose function is not that of an object, while in (18b–c) there is a reflexive clitic referring to the initiator of the action. Furthermore, while in (18d) there must have been intermediaries besides the letters in question, an interpretation of enviar as referring to the sending of those individuals is ruled out by the use of the preposition por ‘by, through,’ which clearly expresses instrumentality.17 Thus, by the same logic as before, it seems possible to conclude that the “enviar a + infinitive” periphrasis already existed in Old Spanish. However, since no comparable cases appear to be attested before the 14th century, it would appear that this grammatical unit was a late medieval development.

The medieval period also furnishes various examples that lend themselves to periphrastic analyses but for which a disjoined interpretation of the two verb forms is not completely ruled out. These examples, which also represent variably reliable manuscripts and editions, include cases with verbs of speech in the narrow sense (e.g., decir ‘say,’ responder ‘answer’) as well as with other types of verbs. Relevant examples include (1a), repeated below as (19b), and the rest in (19):

(19)

a.

e

mandamos

que […]

le

recabden

e

nos

lo

and

order

that

him

detain

and

us

it

embíen

a

decir

send.prs.sbjv.3pl

to

say.inf

‘and we order that they detain him and send us word of it’ (Anonymous, “Alfonso X amplia y matiza exenciones y privilegios,” 1264, Spain [RAE, 2015])

b.

Señor […],

el

infante

Roboán

te

embía

a

lord

the

prince

Roboán

you

send.prs.3sg

to

rogar

que […]

no

quieras

hazer

mal

en

el

beg.inf

that

not

want

do

wrong

in

the

reino

de

Pandulfa

kingdom

of

Pandulfa

‘Lord, Prince Roboán asks you to plan no wrong in the kingdom of Pandulfa’ (Anonymous, Libro del cavallero Cifar, 1300–1305 [last third of the 15th c.], Spain [RAE, 2015])

c.

E

el

prior

e

el

conçejo

de

Çamora

and

the

prior

and

the

council

of

Zamora

enbiaron=le

a

rresponder

que

lo

fazian

por

su

send.pret.3pl=to.him

to

answer.inf

that

it

did

for

his

serviçio

service

‘And the prior and the council of Zamora sent him word in reply that they were doing it in his service’ (Anonymous, Gran crónica de Alfonso XI, ca. 1348–1379 [late 15th c. or early 16th c.], Spain [RAE, 2015])

d.

&

nabuchodonosor

rey

embio /

a

ayuntar

los

and

Nebuchadnezzar

king

send.pret.3sg

to

gather.inf

the

duques

&

los

condes

dukes

and

the

counts

‘And King Nebuchadnezzar ordered to gather the dukes and the counts’ (Anonymous, Biblia romanceada, ca. 1400, Spain [RAE, 2015])

As for “mandar a + infinitive,” although there are some medieval examples that lend themselves to periphrastic analyses, a disjoined reading (with mandar interpreted as either ‘to order’ or ‘to send’) is never completely ruled out. Furthermore, this schema, quite unlike its counterpart with enviar, is not attested with verbs of speech in the Middle Ages. Relevant examples include (1d), repeated below as (20b), and the rest in (20):

(20)

a.

&

si

al

juyz

fuere

prouado,

o

al

obispo,

and

if

to.the

judge

were

proven

or

to.the

bishop

que

julgó

tuerto

lo

que

mandó

a

that

judged

wrong

the

that

order/send.pret.3sg

to

tomar

a

aquel

a

que

lo

julgaron

take.inf

to

that

to

that

him

judged

‘And if it be proven to the judge, or to the bishop, that he judged wrongly what he ordered to be taken from him against whom the judgment was rendered’ (Anonymous, Fuero Juzgo, ca. 1250–1260, Spain [RAE, 2015])

b.

assi

como

te

leuantares

manda

a

thus

as

yourself

rise

order/send.imp.sg

to

limpiar

todas

las

plaças

&

las

calles

de

clean.inf

all

the

public.squares

and

the

streets

of

la

cibdad

the

city

‘as soon as you rise have all the public squares and streets of the city cleaned’ (Alfonso X, General Estoria. Cuarta parte, ca. 1280 [1280], Spain, [RAE, 2015])

c.

e

mando

a

juntar

alli

a

todos

los

and

order/send.pret.3sg

to

gather.inf

there

to

all

the

hombres

buenos

del

su

reyno

men

good

of.the

his

kingdom

‘and he ordered that all the worthy men of his kingdom be gathered there’ (Anonymous, Gran crónica de Alfonso XI, ca. 1348–1379 [late 15th c. or early 16th c.], Spain [RAE, 2015])

d.

E

en

aquell

caso

mandamos

a

poner

en

el

and

in

that

case

order/send.prs.1pl

to

put.inf

in

the

dito

officio

otra

persona

said

office

another

person

‘And in that case we order that another person be appointed to the said office’ (Anonymous, Ordinación dada a la ciudad de Zaragoza por el rey don Fernando I (el de Antequera), 1414, Spain [RAE, 2015])

The situation is similar when it comes to the Golden Age, for which CDH yields several examples like those in (17–20) but no clearly periphrastic cases of mandar. After the 17th century, however, we start to see cases like those in (17–18) with both auxiliaries and with a wide variety of verbs (including verbs of speech) in infinitive form, contrary to Porto Dapena’s claim in this last regard:

(21)

a.

y

la

noche

del

día

primero

le

and

the

night

of.the

day

first

him

mandó

a

llamar

con

su

Capitán

de

la

order/send.pret.3sg

to

call.inf

with

his

captain

of

the

Guardia

guard

‘and on the night of the first day, he had his captain of the guard summon him’ (Antonio Díaz del Real, “Declaración de Don Antonio Díaz del Real,” 1763, Philippines [RAE, 2015])

b.

ni

la

sumisión

con

que

se

envió

a

nor

the

submission

with

that

refl

send.pret.3sg

to

disculpar

de

la

imputación

que

se

le

hacía

excuse.inf

of

the

accusation

that

him

it

made

‘nor the submission with which he excused himself from the accusation brought against him’ (Manuel José Quintana, “Vidas de Vasco Núñez de Balboa, Francisco Pizarro, Álvaro de Luna y Bartolomé de las Casas,” 1832, Spain [RAE, 2015])

c.

Con

más

mensajeros

te

envío

a

llamar

with

more

messengers

you

send.prs.1sg

to

call.inf

‘I send for you with more messengers’ (Domingo Miras, Las brujas de Barahona, 1978, Spain [RAE, 2015])

d.

se

mandó

a

hacer

el

uniforme

refl

order/send.pret.3sg

to

make.inf

the

uniform

amarillo

que

lo

hizo

conocido

de

por

vida

yellow

ofthat

him

made

known

of

by

life

‘he commissioned the yellow uniform that made him famous for the rest of his life‘ (Ignacio Pérez Tuesta, Hombres de blanco, 2006, Chile [RAE, 2025])

In fact, most such cases attested between the 19th and 21st centuries in both CDH and CORPES XXI feature mandar instead of enviar, which is consistent with Porto Dapena’s (1977, p. 28) observation that the former verb has largely replaced the latter in the modern language. However, a closer examination of the data suggests that this is not merely a replacement of one verb or construction by another, but an instance of dialectal divergence, as most such cases are attested in the Americas, with a noticeable concentration in Chile and neighboring countries. Additional CORPES XXI examples from this region are provided below:

(22)

a.

y

entre

sus

excentricidades

se

manda

a

and

among

her

eccentricities

refl

order/send.prs.3sg

to

hacer

decenas

del

mismo

vestido

de

lino

con

make.inf

dozens

of.the

same

dress

of

linen

with

un

modista

de

Roma

a

dressmaker

of

Rome

‘and among her eccentricities she is commissioning dozens of the same linen dress to a dressmaker from Rome’ (Página/12, 23/03/2001, Argentina [RAE, 2025])

b.

y

fue

a

la

comisaría

para

cerciorarse

antes

and

went

to

the

police.station

to

make.sure

before

de

partir

de

que

su

padre

no

lo

hubiera

of

leave

of

that

his

father

not

him

have.3sg.impf.sbjv

mandado

a

buscar

con

Carabineros

order/send.ptcp.m.sg

to

search

with

Carabineros

‘And he went to the police station to make sure, before leaving, that his father had not sent local police looking for him’ (María José Poblete, El desvelo, 2012, Chile [RAE, 2025])

c.

le

fue

tan

bien

importando

telas

chinas

que

to.her

went

so

well

importing

fabrics

Chinese

that

se

mandó

a

hacer

un

cholet

refl

order/send.pret.3sg

to

make.inf

a

chalet

‘she did so well importing Chinese fabrics that she had a chalet built for herself’ (Liliana Colanzi, Ustedes brillan en lo oscuro, 2022, Bolivia [RAE, 2025])

d.

Matta

me

mandó

a

llamar

con

un

Matta

me

order/send.pret.3sg

to

call.inf

with

a

amigo

friend

‘Matta reached out to me through a friend’ (Gerardo Chávez, Antes del olvido, 2022, Peru [RAE, 2025])

A disjoined interpretation of the two verb forms is ruled out, once more, by the presence of a prepositional phrase already mentioning the intermediary of the action or by the presence of a reflexive clitic referring to the initiator of the action. Examples (22a) and (22c), additionally, illustrate a constructional variant attested only in the Americas, characterized by a ditransitive (potentially reflexive) structure and the use of hacer ‘make’ or a similar verb to refer to an act of commission. In these cases, the auxiliary function of mandar is further evidenced by the fact that the object is an object of result (since it refers to something that did not exist when the action was initiated) and thus must belong to an existential causative, such as hacer (see Lyons, 1968, p. 439; Porto Dapena, 1977, p. 32).18

In short, periphrastic cases of “enviar/mandar a + infinitive” seem to be attested in Old Spanish, Classical Spanish, and post-Classical Spanish, including the language of the present day. Several questions remain open, however, concerning the nature of the grammatical units instantiated in such cases. For instance, what is the basic, invariant meaning of the “enviar a + infinitive” and “mandar a + infinitive” periphrases? Do these periphrases, to the extent that they coexist, have the same basic meaning? Does the answer to these questions vary across historical periods or across coexisting varieties? Although I will not attempt to answer these questions here, a good starting point will be to determine the common semantic value deducible from the examples hitherto cited, abstracting away from any possible functional split within the class—whether synchronic or diachronic, syntopic or diatopic.19

Recall in this connection that Porto Dapena classifies enviar and mandar as modal auxiliaries. More specifically, he claims that their function is to indicate that “la acción se verifica indirectamente” [‘the action is verified/carried out indirectly’] (Porto Dapena, 1977, p. 38). The question is whether se verifica here means ‘is verified’ or ‘is carried out,’ both of which are common acceptations of this verb (see RAE, 2014, s.v. verificar). While under the former interpretation Porto Dapena’s characterization appears to be consistent with the data, under the latter it fails to account for some of the author’s own examples. Thus, it fails to account for (16) and (17b), both repeated in (23), where the action signified by the infinitive is presumably not verified at all, nor is it intended to be:

(23)

a.

Mandé

a

llamar=lo

en

una

carta

order/send.pret.1sg

to

call.inf=him

in

a

letter

‘I summoned him by letter’

(Porto Dapena, 1977, p. 36)

b.

Muchos

señores

con

embaxadas

se

enviaron

a

many

lords

with

missives

refl

send.pret.3pl

to

encomendar

en

sus

oraciones

commend.inf

in

his

prayers

‘Many lords commended themselves to his prayers by missive’ (Antonio de Fuenmayor, Vida y hechos de Pío V, 1595, Spain [Cuervo & ICC, 1978/1998a, p. 714; Porto Dapena, 1977, p. 38, fn. 9])

Therefore, the better approach seems to be to concentrate on the (potential) realization of the action rather than on its verification. From this point of view, taken together, the relevant examples can indeed be characterized as referring to the indirect realization of the action signified by the infinitive, by means of either an intermediary or some other kind of instrument. Crucially, indirect does not mean ‘in absentia.’ Thus, even if all the examples offered thus far refer to an action carried out without the presence of the initiator, the proposed characterization would also accommodate cases like the following:

(24)	Rodrigo	{ envió			/	mandó}		a	llamar	a	la
	Rodrigo	send.pret.3sg				order/send.pret.3sg		to	call.inf	to	the
	mesera	con	un	gesto			sutil
	waitress	with	a	gesture			subtle
	‘Rodrigo called the waitress over with a subtle gesture’

The above characterization can in fact also be extended to non-periphrastic cases such as those in (6), repeated below as (25). Naturally, this generalization involves a further abstraction, as it ignores the notion of caused movement or command present in cases such as those in (25). Nevertheless, such an abstraction serves to underscore the fact that, despite their differences, the two types of structures, periphrastic and non-periphrastic, share a functional space—a fact that will be a key consideration in Section 5.

(25)

a.

El

general

mandó

a

un

emisario

a

the

general

order/send.pret.3sg

to

an

emissary

to

negociar

la

paz

negotiate.inf

the

peace

‘The general sent an emissary to negotiate peace’

b.

Óscar

mandó

(a

alguien)

a

pedir=le

Óscar

order/send.pret.3sg

to

someone

to

ask.inf=him

consejo

a

Joaquín

advice

to

Joaquín

‘Óscar sent for Joaquín’s advice’

c.

Miguel

envió

su

bicicleta

a

arreglar

Miguel

send.pret.3sg

his

bicycle

to

fix.inf

‘Miguel sent his bicycle to have it fixed’

(Porto Dapena, 1977, pp. 28, 35, 30)

5. A Quantitative Diachronic Approach to “enviar/mandar a + Infinitive”

The historical and geographical distribution of the examples analyzed in Section 4 seems to confirm the hypothesis that “enviar a + infinitive” and “mandar a + infinitive” did not follow quite the same trajectory as other Spanish causative constructions. Indeed, the data examined thus far point to a late medieval consolidation of the “enviar a + infinitive” periphrasis and a subsequent constructional shift characterized by the development and entrenchment, at least in certain regions, of an analogue with “mandar a + infinitive” at the expense of the former construction. In what follows, I will take a more systematic look at this general hypothesis by focusing on a measure of entrenchment and grammaticalization widely discussed in the specialized literature: frequency of use (see Bybee, 2003, 2006, pp. 719–721; Haiman, 1994; Torres Cacoullos, 1999, pp. 154–157; Traugott, 2011, p. 28). Assuming a positive correlation between the frequency of use of a constructional class and the degree of entrenchment and grammaticalization within it, I predict that cases of “enviar a + infinitive” will exhibit an initial surge followed by a progressive decline, while “mandar a + infinitive” will show an increase concurrent with the latter process. These predictions will be tested with CDH and CORPES XXI data for the 12th–20th and 21st centuries, respectively,20 focusing strictly on cases in which enviar and mandar are immediately followed by the preposition and the infinitive.

For the present purposes, frequency of use will be operationalized as the number of cases of each schema per million words in each century, as calculated automatically in both corpora. Crucially, this will be determined based on queries expected to yield both periphrastic and non-periphrastic structures. In other words, it will not be assumed that all cases under study are periphrastic in nature. The rationale for this is twofold. On the one hand, frequency of use is intended to provide evidence of entrenchment and grammaticalization independently of the analyses offered in Section 3 and Section 4. On the other hand, the use of this measure is guided by the working hypothesis that periphrastic and non-periphrastic cases of “enviar/mandar a + infinitive” instantiate etymologically related constructions in weak complementarity, between which a clear line cannot always be drawn (see Davies, 2000, pp. 119–121; Myhill, 1988, p. 353; Sankoff & Thibault, 1981; Schwenter & Torres Cacoullos, 2010). Said working hypothesis has several methodological implications. On the one hand, operating on the basis of an a priori distinction between periphrastic and non-periphrastic cases becomes, of course, impossible, and focusing only on those cases deemed to be clearly periphrastic would in all likelihood lead to the exclusion of ambiguous cases that were intended as periphrastic by the speaker and/or might be interpreted as periphrastic by a number of language users. On the other hand, even if the researcher were able to reliably tease apart the two types of structures, non-periphrastic cases remain of interest due to the historical and associative21 link they share with their periphrastic counterparts.22

Cases were searched specifying the lemmas enviar and mandar as needed as well as the preposition and the infinitive, utilizing the grammatical, chronological, and distance criteria of each corpus. In CDH, where tense cannot be specified, infinitive forms were targeted with the search term “*r|*rme*|*rte*|*rnos*|*ros*|*rlo*|*rla*|*rle*|*rse*” in the first form field (for the node), so as to be able to find cases featuring different enclitic configurations as well as cases without enclitics, while avoiding, to the extent possible, irrelevant structures.23 In order to control for changes in the frequency of enviar and mandar as autonomous signs, the proportion of cases of each verb instantiating the relevant schema was also considered. This was determined by retrieving all occurrences of enviar and mandar in each century and dividing the number of “enviar a + infinitive” and “mandar a + infinitive” cases by the total number of cases of each verb.

The results of the general analysis are presented in

Table 1. Normalized frequency (per million words) of “enviar a + infinitive” and “mandar a + infinitive” and proportion of cases of enviar and mandar instantiating each schema by century.

Century	enviar a + Infinitive		mandar a + Infinitive
	Normalized Frequency	% of Cases of enviar	Normalized Frequency	% of Cases of mandar
12th	0	0	0	0
13th	8.13	1.04	0.93	0.04
14th	133.77	8.79	1.83	0.08
15th	69.67	8.76	2.59	0.13
16th	114.93	14.56	1.66	0.11
17th	62.30	12.04	0.95	0.10
18th	15.77	6.16	1.25	0.16
19th	7.72	4.03	3.02	0.63
20th	3.29	2.27	7.21	4.13
21st	1.53	1.19	8.82	9.25

and Figure 1, which confirm the above predictions. As the data show, “enviar a + infinitive” experienced a sharp increase between the 12th and 14th centuries, further suggesting that the corresponding periphrasis was a late medieval development. Thereafter, we observe a declining trend in favor of “mandar a + infinitive,” which starts to become more frequent around the 18th century. Furthermore, these changes in frequency are accompanied by broadly parallel changes in the proportion of cases of enviar and mandar instantiating each schema. This strongly suggests that the trajectories of “enviar a + infinitive” and “mandar a + infinitive” cannot be entirely attributed to changes in the frequency of enviar and mandar as autonomous signs.

Since, as pointed out in Section 4, a version of the “mandar a + infinitive” periphrasis appears to have become particularly common in Chile and neighboring countries, the same type of analysis was performed separately for this region (where Spanish was introduced in the 16th century) and for Spain, which was used as a baseline. The results of these analyses are shown in

Table 2. Normalized frequency (per million words) of “enviar a + infinitive” and “mandar a + infinitive” and proportion of cases of enviar and mandar instantiating each schema by century in Spain.

Century	enviar a + Infinitive		mandar a + Infinitive
	Normalized Frequency	% of Cases of enviar	Normalized Frequency	% of Cases of mandar
12th	0	0	0	0
13th	8.13	1.04	0.93	0.04
14th	133.77	8.79	1.83	0.08
15th	69.67	8.76	2.59	0.13
16th	113.82	14.90	1.67	0.11
17th	59.37	11.67	0.68	0.08
18th	14.66	6.37	0.56	0.08
19th	7.41	4.18	2.20	0.51
20th	3.65	2.54	2.15	1.23
21st	1.2	1.05	2.33	2.64

and

Table 3. Normalized frequency (per million words) of “enviar a + infinitive” and “mandar a + infinitive” and proportion of cases of enviar and mandar instantiating each schema by century in Argentina, Bolivia, Chile, and Peru.

Century	enviar a + Infinitive		mandar a + Infinitive
	Normalized Frequency	% of Cases of enviar	Normalized Frequency	% of Cases of mandar
16th	136.82	11.64	2.47	0.08
17th	89.64	12.75	1.14	0.08
18th	16.76	6.20	5.76	0.48
19th	11.80	3.39	3.19	0.36
20th	1.92	1.59	11.10	6.63
21st	1.53	1.28	11.78	11.20

and Figure 2 and Figure 3. When it comes to Spain, the trajectory of “enviar a + infinitive” is virtually identical to that shown in Table 1 and Figure 1, while in the case of “mandar a + infinitive” no upward trend is discernible (although between the 20th and 21st centuries there is a slight increase in both metrics). In Chile and neighboring countries, in contrast, while “enviar a + infinitive” exhibits a similar (but somewhat delayed) trajectory between the 16th and 21st centuries, “mandar a + infinitive” shows an even steeper increase than that in Table 1 and Figure 1 across both metrics.24

The preceding analyses point to two grammaticalization processes: one (corresponding to “enviar a + infinitive”) that reached its peak in the late medieval period and another (corresponding to “mandar a + infinitive”) that took place later in parts of the Americas and is probably still ongoing today. Given this situation, the Late Latin morphological merger discussed by Davies (1995, pp. 71–72) does not seem directly relevant here, a more plausible explanation being that these processes were the result of a series of analogical innovations taking place during the Middle Ages and beyond, which may have created a perfect storm for the emergence, diffusion, and grammaticalization of new causative constructions (see Sanaphre Villanueva, 2010, p. 149ff.).

On the other hand, the trajectory of “enviar a + infinitive” is consistent with a degrammaticalization process along the lines drawn by Davies (1995, 2000) and Soares da Silva (2012)—a process that could be explained as part of a broader shift in infinitival complementation in Ibero-Romance (see Davies, 1995, p. 74 and passim; Martins, 2006, p. 340ff.; Soares da Silva, 2012, p. 542). Importantly, to hypothesize a degrammaticalization process in this case is not to say that the “enviar a + infinitive” periphrasis has completely disappeared, but rather that it has lost ground to its non-periphrastic counterparts, including constructions in which enviar is not followed by a + infinitive. As for the role of “mandar a + infinitive” in this decline, it seems possible to conclude that it was limited at best, for two reasons. First, while the decline of “enviar a + infinitive” appears to have begun before the discovery of the Americas, the rise of “mandar a + infinitive” seems to be a chiefly Latin American phenomenon. Second, the two phenomena are neither parallel nor comparable in magnitude.

6. Conclusions

In this article, I have attempted to shed light on a controversial area of Spanish (and more broadly Romance) morphosyntax, focusing on an understudied constructional class. Despite claims to the contrary, I have argued that certain “enviar/mandar a + infinitive” structures instantiate true verbal periphrases, and, in fact, such structures can be found across all main stages of the history of Spanish. Furthermore, I have argued that “enviar a + infinitive” and “mandar a + infinitive,” as periphrastic constructions, are the product of two distinct grammaticalization processes, widely separated both chronologically and geographically, and that the former construction has also been involved in a degrammaticalization process.

These findings have implications beyond Hispanic and Romance linguistics. Methodologically, they illustrate the importance of considering syntactic ambiguity and of combining ad hoc diagnostics with quantitative evidence in the study of verbal periphrases and grammaticalization more generally (see Coseriu, 1960/1962, p. 123; Davies, 2000, pp. 119–121; Mair, 2004, p. 136ff.; Myhill, 1988, p. 353). Theoretically, they provide further evidence of the reversibility of grammaticalization, particularly if the phenomenon is understood as a process and not merely as an outcome (see Haspelmath, 2004; Norde, 2009; Willis, 2007). Typologically, they add to the documentation of the grammaticalization pathway ‘send’ > causative, which has been reported mainly in South Asian languages (see Kuteva et al., 2002/2019, p. 393) but also, to a lesser extent, within the Romance family (see Sousa Fernández, 2002, p. 296ff.).

Finally, the case of “enviar/mandar a + infinitive” calls for a few remarks on the concept of analyzability. Although linguists often use this term to refer to the “recognition of the contribution that each component [of a construction or structure] makes to the composite conceptualization,” according to Langacker’s (1987, p. 292) well-known definition,25 it is in fact possible to distinguish several types of analyzability from a morphosyntactic standpoint. First, a distinction can be drawn based on the type of knowledge at work (see Coseriu, 1988/1992, pp. 229–237; Lehmann, 2007, pp. 243–244, 250–252). According to this criterion, analyzability can be conceptualized either as a strictly linguistic property—that is, as the capacity of a construction or structure to be parsed in a given way in communication—or as a metalinguistic property—that is, as the capacity of a construction or structure to be analyzed in the usual sense, as a scholarly or otherwise reflective endeavor (for example, as undertaken by a linguist). Second, a distinction can be drawn according to the accuracy of the parse or analysis from the perspective of language users. Thus, one may use the term to refer either to the capacity of a construction or structure to be parsed or analyzed as intended by the speaker and/or as interpreted by the hearer or to the possibility of parsing or analyzing it according to any given materially compatible syntactic pattern, regardless of whether it matches the actual usage of the unit in question. When it comes to constructions such as those investigated in this article, their dismissal in studies on verbal periphrases seems to be largely motivated by the fact that, oftentimes, a disjoined analysis is available in principle. In other words, the exclusion has been based on analyzability as a metalinguistic property independent of usage and not as conceived by Langacker.