Grammatical and Lexical Dialectal Variation in Spanish: The Case of deísmo

Abstract

Deísmo is a non-standard dialectal phenomenon consisting of the insertion of a non-required preposition de ‘of’ before a non-finite clause: Me apetece (de) salir ‘I want to go out’. In most papers, de is analyzed as a defective complementizer that does not change the meaning of the sentence. However, deísmo has also been associated with a prospective meaning with some verbs, and de has been considered as a marker of evidentiality with visual perception verbs. In this paper, we provide a formal analysis for deísmo constructions, in which de is located in a projection below that occupied by de in dequeísmo constructions). Secondly, we will show the results of a questionnaire whose objective is to figure out if there is an evidential meaning associated with deísmo. For the questionnaire, we made a preliminary search in Corpus Oral y Sonoro del Español Rural (COSER) and in Spanish Web Corpus 2018 (Sketch Engine). From this, we selected the most frequent verbs with deísmo in Castilla-La Mancha (Spain). An examination of the results revealed that, on the one hand, deísmo is lexically associated with certain verbs, but not necessarily with all of the same semantic class; and on the other hand, that there is not an evidential meaning associated with deísmo.

1. Introduction

Deísmo is a vernacular phenomenon of variation in the Spanish complementizer system that has received little attention in the linguistic literature (only in recent years, starting with Di Tullio (2011, 2012) and the seminal work of Camus (2013)), especially when compared to a partly parallel phenomenon such as dequeísmo.

As we will see in Section 2, deísmo consists of introducing the preposition de ‘of’ before infinitival complements of certain verbs (see example (1)). On the other hand, dequeísmo is the introduction of the preposition de ‘of’ before the sentential complement—with a finite verb and complementizer que ‘that’—of a different group of verbs (see example (2)). Although sometimes it has been said that both phenomena are basically the same, we will maintain that the particle de occupies different positions in the two constructions:

(1)	a.	La vida me hace de llorar [CORPES]
		The life CL.DAT-make of cry.INF
		‘Life makes me cry’.
	b.	Resulta de que ahora el presidente es otro [CORPES]
		Turns out of that now the president is other
		‘It turns out that the president is now a different one’.

The understanding of this construction, however, is very relevant in the framework of studies that in the last twenty years, starting with Rizzi (1997), have shown that the area traditionally assigned to conjunctions and wh-words, in the left sentence periphery, is more complex than initially thought.

The main goal of this paper is to find out whether the particle de ‘of’ that appears in the deísmo is a marker of indirect evidentiality, as has been proposed for equivalent constructions in other languages (van in Brabant Dutch) or for the form de that appears in dequeísmo. For this purpose, we use data from Castilla-La Mancha, a region where deísmo is a relatively widespread phenomenon in the spoken language. Section 2 offers a review of the syntactic characteristics of the construction with deísmo and proposes an analysis in which de is considered a complementizer that occupies a low position in the left sentence periphery, the same position that Italian di ‘of’ or Catalan de ‘of’ occupies before non-finite subordinate clauses. Section 3 reviews the studies on deísmo and the various proposals of the meaning of the form de: from being a complementizer without meaning, as most papers maintain, to having an evidential meaning (as proposed in Carrasco and Peinado (2022) for Spanish or in other languages for similar constructions). Finally, Section 4 presents the empirical study carried out mainly through two surveys, in which verbs of various lexical classes were presented in contexts of direct and indirect evidentiality. To choose the verbs for the first survey and ensure that they presented realizations with deísmo, we made an initial search in the oral interviews from the Corpus Oral y Sonoro del Español Rural (COSER), which allowed us to look for regional data. Secondly, based on the results of this questionnaire, we designed a second one. For this, we also made an additional search in another corpus that allows searching for regional data, in the newspapers of Castilla-La Mancha: Spanish Web Corpus 2018 (ketch Engine) (2018). Although the number of speakers consulted is low, with 20 informants between the two surveys, the results show that the form de is not associated with any evidential meaning.

2. What Is Deísmo?

In this section1 we establish the main properties of deísmo. In Section 2.1 we define this construction, which is a phenomenon of variation in the Spanish complementizer system; in Section 2.2 we explain what type of verbs appear more frequently with de ‘of’ in non-finite subordinate sentences. We will use this classification to prepare our questionnaire, which is presented in Section 4. Section 2.3 is dedicated to the syntactic nature of the particle de, and Section 2.4 to its position in the left periphery.

2.1. Definition of Deísmo

Deísmo is the apparition of the preposition de ‘of’ before non-finite sentences that are the object (2a) or the subject (2b) of a verb2 that does not lexically require the preposition:

(2)	a.	Le propuse de encontrarnos en casa
		her.CL-propose of meet.INF at home
		‘I proposed her to meet at home’.
	b.	En el fondo le gusta de acompañar a Paco [CORPES]
		deep down CL.DAT-like of accompany.INF to Paco
		‘Deep down he likes to accompany Paco’.

The preposition de is the standard way to introduce non-finite sentences in Italian, French or Catalan. Rizzi (1991, p. 516) claims that the preposition does not head a prepositional phrase in those examples, but is the complementizer of the non-finite subordinate, which is why the sentence can be pronominalized using an accusative pronoun (3b):3

(3)	a.	Credo di voler andare (it.)
		believe of want.INF go.INF
		‘I think I want to go’.
	b.	Lo credo (it.)
		CL.ACC believe
		‘I believe it’.

2.2. Contexts of Deísmo. Types of Verbs

In this section we will see which type of verbs are more commonly found with deísmo. However, as Camus (2013, p. 24) indicates, there is a random split between deísmo and non-deísmo contexts, in the sense that not all the verbs that belong to one of the classes we will see in this section are constructed with de and thus there seems to be a kind of lexical selection among those semantic classes. For instance, inside the group of psychological verbs of affection, some of the verbs which are very frequent, such as gustar ‘to like’, are not frequently seen with deísmo in Castilla-La Mancha, but this same verb can appear with de in Andalucía, as the Corpus Oral y Sonoro del Español Rural (COSER) (Fernández-Ordóñez 2005) data show (see

Table 1. Occurrences of examples of deísmo with eleven verb classes in COSER.

Predicates	Castilla-La Mancha	Other Provinces	Total
Psychological predicates
Apetecer ‘to feel like’	0	1 Huesca, 1 Huelva	2
Costar ‘to have a hard time’	0	1 Alicante, 1 Barcelona, 1 Valencia	3
Doler ‘to hurt’	0	0
Gustar ‘to like’	1 Albacete, 2 Cuenca, 2 Toledo	1 Badajoz, 1 Córdoba, 1 El Hierro, 2 Granada, 1 Huelva, 1 Huesca, 1 Jaén, 1 Lérida, 1 Sevilla, 1 Teruel, 1 Valencia	16
Interesar ‘to be interested’, temer ‘to be afraid’	0	0
Predicates of occurrence
Ocurrir ‘to happen’, suceder ‘to happen’	0	0
Tocar ‘to be one’s turn’	1 Cuenca, 1 Toledo	1 Ávila, 1 El Hierro, 1 Huesca, 2 La Coruña, 1 Palencia, 1 Teruel, 1 Zaragoza	10
Predicates of opportunity or convenience
Convenir ‘to be in one’s best interest’	0	0
Predicates of influence
Aconsejar ‘to advise’, amenazar ‘to threaten’, permitir ‘to permit’, proponer ‘to propose’, recomendar ‘to recommend’, rogar ‘to beg’, solicitar ‘to request’, sugerir ‘to suggest’	0	0
Decir ‘to tell’		2 Alicante, 1 Cantabria, 1 Huesca, 1 La Rioja, 1 Soria, 1 Valladolid	7
Mandar ‘to tell’	5 Toledo	1 León	6
Ordenar ‘to order’	0	2 Zamora, 1 Lugo	3
Pedir ‘to ask’		1 Almería, 1 Murcia	2
Prohibir ‘to forbid’	0	1 Tenerife
Pseudoimpersonal predicates
Hacer falta ‘need’	0	1 Badajoz, 1 Málaga	2
Ser costumbre ‘to be customary’	0	1 Guipúzcoa, 2 Madrid, 3 Navarra, 1 Palencia, 1 Segovia, 1 Zamora	9
Ser {necesario/una lástima} ‘to be {necessary/a pity}’	0	0
Predicates of intention
Intentar ‘to try’		1 Badajoz, 1 La Rioja, 1 Lugo, 1 Salamanca	4
Probar4 ‘to try’		1 Lérida	1
Propositional attitude predicates
Considerar ‘to consider’, creer ‘to believe’, imaginar ‘to imagine’, saber ‘to know’	0	0
Pensar5 ‘to think’		1 Huelva, 1 Huesca, 1 Palencia, 1 Pontevedra, 1 Santa Cruz de Tenerife	5
Predicates of will
Aceptar ‘to accept’, esperar ‘to hope’, preferir ‘to prefer’	0	0
Estar deseando ‘to look forward to’	2 Albacete	1 Álava, 1 Burgos, 1 Córdoba, 1 Guipúzcoa, 1 Jaén, 1 Madrid, 1 Palencia, 2 Sevilla, 1 Valladolid, 1 Zamora	13
Necesitar ‘to need’	0	1 Huelva, 1 Zaragoza	2
Querer ‘to want’	1 Toledo	1, Huelva, 1 Lanzarote, 1 Málaga, 1 Palencia	5
Causative predicates
Hacer ‘to make’	1 Ciudad Real, 1 Toledo	1 Álava, 1 Cáceres, 1 Menorca, 1 Tenerife, 1 Zaragoza	7
Perception predicates6
Ver ‘to see’	1 Toledo	2 Badajoz, 1 Huelva, 1 Málaga, 2 Sevilla	7
Oír ‘to hear’	0	1 Badajoz, 1 Barcelona
Escuchar ‘to listen’	0	0
Auxiliary verbs7
Poder ‘can’	0	1 Córdoba, 1 Lérida	2
Soler ‘used to’	0	1 Burgos, 2 Madrid, 1 Sevilla	4

). We will see in the results of our questionnaires that there is also a lot of individual variation, because even among speakers from the same village that clearly have deísmo, not all of them use the preposition de with the same verbs of a semantic class (see Section 4).

Among all the classes below, the Standard Spanish structure without de is more common than the one with the preposition. The first five classes are the ones that have been described to appear more frequently with deísmo (Di Tullio 2011, p. 180; Camus 2013, sct. 3.2; De Benito and Pato 2015, p. 35; Gutiérrez-Rodríguez 2019, p. 96; Carrasco and Peinado 2022):

(a)

Psychological predicates with an infinitive clause as subject: apetecer ‘to feel like’, costar ‘to have a hard time’, doler ‘to hurt, to regret’, gustar ‘to like’, interesar ‘to be interested’, etc.

(b)

Predicates of occurrence with a subject infinitive clause: ocurrir ‘to happen’ suceder ‘to happen’, tocar ‘to be one’s turn’, etc.

(c)

Predicates of influence or conative verbs: decir8 ‘to ask’, mandar ‘to order’, pedir ‘to ask’, proponer ‘to suggest’, prohibir ‘to forbid’, etc.

(d)

Pseudoimpersonal constructions with copulative ser ‘to be’: ser costumbre ‘to be customary’; ser necesario ‘to be necessary’, ser una lástima ‘to be a pity’, etc. Many of these constructions have a modal meaning, such as ser necesario ‘to be necessary’, ser fácil ‘to be easy’, etc.

(e)

Predicates of intention: intentar ‘to try’, probar ‘to try’.

(f)

Propositional attitude predicates: considerar ‘to consider’, creer ‘to believe’, pensar ‘to think’, etc.

(g)

Predicates of will. The verb desear ‘to wish’ accepts of only in the periphrasis estar deseando lit. ‘to be wishing’.

(h)

Causative predicates: the causative verb hacer ‘to make’ can take the preposition of, as in Me hizo de cantar ‘he made me sing’.

(i)

Perception verbs: ver ‘to see’, escuchar ‘to listen’, etc.

(j)

Auxiliary verbs: poder ‘can’, soler ‘to use to’.

In view of these classes, there does not seem to be a common meaning in the verb that explains the preposition de in deísmo, as we have seen that almost all classes of verbs that can take a non-finite sentence, can also take de (with all the lexical and individual variation due to the non-standard nature of the phenomenon).

2.3. The Particle de ‘of’ as a Complementizer

In this section, we will review the arguments that have been used to show that the segment de cannot be considered a subcategorized preposition that heads a prepositional phrase. Camus (2013) has already shown that de cannot be considered a preposition in deísmo constructions; in this section we will review his arguments.

If we compare the de from deísmo and the preposition de subcategorized using a verb, as in olvidarse ‘to forget’, we see that the syntactic behaviour of both elements is entirely different. The complement of the “real” preposition de can be a non-finite clause (4a), a finite clause headed by que ‘that’ (4b) or a noun phrase (4c) (Di Tullio 2012, p. 268; Camus 2013):

(4)	a.	Me olvidé de comprar el pan
		CL.1SG-forgot.PAST.1SG of buy.INF the bread
		‘I forgot to buy the bread’.
	b.	Me olvidé de que venía
		CL.1SG-forget.PAST.1SG of that come.PAST.3SG
		‘I forgot that you were coming’.
	c.	Me olvidé de tu cumpleaños
		CL.1SG-forget.PAST.1SG of your birthday
		‘I forgot your birthday’.

In the case of deísmo, only the non-finite clause can appear with de. A neuter pronoun such as eso ‘that’ is impossible (5a), and a finite clause could be possible (5b) but it would be considered an example of dequeísmo (Di Tullio 2012, p. 268):

(5)	a.	*Nos pidió de eso
		CL.1PL.DAT-ask.PAST of that
		‘He asked that’.
	b.	Nos pidió de que vieras a su familia
		CL.1PL.DAT-ask.PAST of that you see his family
		‘He asked to see his family’.

Furthermore, the non-finite sentence can be cliticized by the accusative clitic lo ‘it’, as any other sentential direct object:

(6)	a.	¿Mirarás de averiguar esos datos? (Camus 2013, p. 16)
		have a look.FUT.2SG of find out.INF those figures
		‘Will you try to find out those figures?’
	b.	—Sí, lo miraré
		yes CL.3SG.N.AC-have a look.FUT.1SG
		‘—Yes, I will try’.

In Spanish prepositional complements can never be cliticized using the accusative pronoun lo: Depende de Juan (Lit. depends of Juan)/*Lo depende.

Another piece of evidence against the prepositional nature of de in deísmo structures is offered by its behaviour in pseudocleft sentences (Camus 2013, p. 18). The fronted phrase must include de if it refers to a subcategorized prepositional phrase (7a). However, if the de is not a real preposition, it does not have to appear in the fronted phrase (7b):

(7)	a.	De lo primero que me alegro es de hacer lo correcto (Camus 2013, p. 19)
		of the first that CL.1sg-be happy.PRES.1SG is of do.INF that correct
		‘The first thing that makes me happy is doing the right thing’.
	b.	Lo primero que no permito a mis hijos es de llegar tarde
		The first that not allow.PRES.1SG to my children is of arrive.INF late
		‘The first thing I do not allow my children is to be late’.

In addition, the non-finite clause can be the subject with so-called pseudoimpersonal verbs of psychological affection such as apetecer ‘to feel like’, pesar ‘to regret’, among many others. All these verbs have a dative experimentizer and a subject that can be realized by a determiner phrase in agreement with the main verb, as in (8a), or as a clause (8b):

(8)	a.	Me duelen sus desplantes (Camus 2013, p. 19)
		CL.1SG-hurt.PRES.3PL their rudeness.PL
		‘Their rudeness hurts me’.
	b.	Me duele de no haberlo dicho a tiempo
		CL.1SG-hurt.PRES.3SG of not have.INF-CL.3.N.AC say.PP on time
		‘I regret not having said it before’.

In (8a), sus desplantes agrees with the verb duelen because it is the subject of the clause; the non-finite clause in (8b) is also the subject. As there are not prepositional phrases in the subject position, we can affirm that de no haberlo dicho a tiempo in (8b) is not a prepositional phrase.

Finally, as we expect from a complementizer, de moves with the clause in the case of topicalization (9a), contrastive focus (9b) or right dislocation (9c) (examples from Camus 2013, p. 20):

(9)	a.	¿Verdad que no les permites a tus hijos de llegar tarde?
		true that not CL.3.PL.DAT-allow.PRES.2SG to your children of arrive.INF late
		‘You do not allow them to be late, do you?
		—Eso es, de llegar tarde no se lo permito.
		that is, of arrive.INF late not 3.CL.DAT-CL.3.N.AC allow.PRES.1SG
		—That’s true, that I do not allow it to them.’
	b.	De gastar ese dinero le da vergüenza, no de tenerlo.
		of spend.INF that money CL.3.DAT-give.PRES.3SG shame not of have.INF.CL.3SG.M.AC
		‘It’s spending that money that embarrasses him, not to have it.’
	c.	Esto mismo no permito, de llegar tarde.
		this self not allow.PRES.1SG of arrive.INF late
		‘This I do not allow, to be late.’

To sum up, all this evidence shows that, in deísmo clauses, de is not a real preposition that heads a prepositional phrase; instead, it can be considered a complementizer, as we explain in the next paragraphs.

2.4. The Complementizer de ‘of’ and Its Position in the Left Periphery

We have seen that the sequence <de + infinitive clause> can be replaced by the accusative pronoun lo, as a that clause does. The more extended analysis is that the element de is not a real preposition but a prepositional complementizer (Rizzi 1997). Hence, it does not head its own prepositional phrase. The segment de can be considered, as Di Tullio (2011, 2012), Camus (2013) and Gutiérrez-Rodríguez (2019) proposed, an overt realization of the COMP node that heads the embedded infinitive clause. We detail below which position in the left periphery of the sentence the complementizer can occupy.

Rizzi (1997) and subsequent work assume that the left periphery is complex and contains different functional heads and their projections: two fixed components, with the heads of Force and Finiteness; and two components that are activated only if needed, the Topic Phrase and the Focus Phrase. The Force Phrase is responsible for the distinction among types of clauses: declaratives, exclamatives, interrogatives, etc., while Finiteness contains the specification for finite and non-finite clauses:

(10)	Force P > TopP > FocP > FinP > TP

For Italian, Rizzi (1997, 2001) proposed that the complementizer che ‘that’ occupies the head of ForceP, while di ‘of’, which takes a non-finite clause, heads FinP. In Spanish, the complementizer que ‘that’ heads ForceP and thus precedes topics, as we can see in the examples in (11) from Brucart and Hernanz (2015, p. 92), where que precedes the topic (la Traviata):

(11)	a.	El inspector cree que la Traviata, la Castafiore la cantó en París
		The policeman thinks that la Traviata, la Castafiori CL.FEM.SG-sing.PAST in Paris.
	b.	*El inspector cree la Traviata que la Castafiore la cantó en París
		The policeman thinks la Traviata that la Castafiori CL.FEM.SG-sing.PAST in Paris.

Rizzi claims that di in Italian occupies a lower position and heads FinP, as its position with respect to the topic (il tuo libro in the example below) shows:

(12)	a.	Credo, il tuo libro, di apprezarlo molto (Rizzi 1997, p. 288)
		Think, the your book, of liking-CL.AC.M a lot
		‘I think that I like your book a lot’.
	b.	*Credo di, il tuo libro, apprezarlo molto
		Think of, the your book, liking-CL.AC.M a lot.

Brucart and Hernanz (2015, p. 92) showed that the complementizer de in Catalan occupies the same position, lower than que ‘that’ in the complementizer system:

(13)	a.	A la soprano li fa il.lusió, [Top La Traviata], de cantar-la al Liceu
		To the soprano CL.-do anticipation, la Traviata, of sing-CL.AC.FEM to the Liceu.
		‘The soprano is looking forward to singing la Traviata in the Liceu’
	b.	*A la soprano li fa il.lusió de, [Top La Traviata], cantar-la al Liceu
		To the soprano CL.-do anticipation of, la Traviata, sing-CL.AC.FEM to the Liceu.

Following Gutiérrez-Rodríguez (forthcoming), we propose that, in deísmo sentences, the segment de is a prepositional complementizer which occupies the lower functional head Fin. Although the data revealed some degree of inconsistency among speakers, probably due to normative pressure, most of the speakers agreed that topics precede de in deísmo sentences:9

(14)	a.	Propuso, la película, de verla mañana
		Propose, the film.F.SG, of see.INF-CL.AC.F.SG tomorrow
		‘He proposed to see the film tomorrow’.
	b.	??/* Propuso de, la película, verla mañana.
(15)	a.	Permitió, el libro, de leerlo más tarde
	b.	??/* Permitió de, el libro, leerlo más tarde.

To sum up, we have seen that the segment of in deísmo is not a preposition that heads a prepositional phrase, but a complementizer in Fin, a head below the one where that originates, in the left periphery.

3. The Meaning of the Particle de ‘of’

Our aim in this section is to investigate whether the segment de ‘of’ is a meaningless subordination marker in constructions with deísmo (as Gómez Torrego 1999, Camus 2013 and De Benito and Pato 2015 have argued), or whether it provides some kind of meaning (at least for some class of verbs). For example, it has been said that de has a prospective value with verbs of will, intention or influence (Di Tullio 2011): Me propuso de ir al cine ‘He suggested to me to go to the movies’. It has also been proposed that it is an evidential marker when combined with verbs of visual perception (Carrasco and Peinado 2022): La vimos de bailar ‘We saw her dance’. On the other hand, the question of whether the particle de contributes any meaning to the construction with deísmo has been related to its value in the construction with dequeísmo (the use of de before finite clauses, as in Creo de que hay tiempo suficiente ‘I believe there is enough time’). In the case of dequeísmo, these values range from stylistic factors, such as the desire to be more expressive or emphatic (e.g., in Náñez 1984 and De Mello 1995), to nuances related to epistemic modality, such as decreasing the speaker’s commitment to the utterance (Bentivoglio and D’Introno 1977; Schwenter 1999; and Demonte and Soriano 2005, among others). Since our research focuses on finding out whether de contributes evidential value in deísmo, we will only refer in this section to proposals in which de is attributed a modal value in a broad sense as a marker of doubt, hearsay or, in general, weak speaker commitment, weak involvement or the introduction of a speaker personal perspective.

There are at least three reasons to conduct an investigation into whether de has evidential meaning in the construction with deísmo, that is, whether it encodes the source of knowledge of the information conveyed by the speaker. First, while deísmo and dequeísmo need not be coincident phenomena, the two have been related in the literature and feature an unselected particle de heading a subordinate sentence. Second, it has been proposed that some complementizers are indirect evidentiality markers (see Rooryck 2001): for example, that in English or van in Brabant Dutch (Van Craenenbroeck 2004); or certain uses of que in root sentences in Spanish (Demonte and Soriano 2014 and Sánchez López 2020, among others). Given that in Section 2 we have shown that the particle de is a complementizer, it makes sense to ask whether it encodes evidential or modal values. Third, as we will see in Section 3.2, Carrasco and Peinado (2022) have argued that the particle de is a morpheme of indirect evidentiality with verbs of visual perception, so it is plausible to investigate whether this value extends to other verb classes. In the following sections we will examine the relationship between evidentiality and complementizers in general (Section 3.1), and the relationship between evidentiality and the complementizer de in Spanish (Section 3.2).

3.1. Evidentiality and Variation in the Complementizer Phrase

According to Rooryck (2001), Aikhenvald (2018) and Sánchez López (2020), among the expressions (or grammatical strategies) that encode the source of information are complementizers. As we will see, at least that in English, van in Brabant Dutch and de in Spanish have been studied as complementizers encoding evidential values. Before studying these complementizers in Section 3.1.2 and Section 3.2, let us digress briefly to see what is considered evidentiality and what the limits with the epistemic modality are.

3.1.1. Some Brief Notes on Evidentiality and Epistemic Modality

Evidentiality is a semantic-functional notion that encodes the source of knowledge of the information. In the sentence I saw John arrive late, the verb of perception encodes that the information comes from the speaker’s visual experience, while in the example Logically, John will be late, the adverb presents the information as a reasoning; if we add a parenthetical expression, According to my brother, John will be late, the information is attributed to a person other than the speaker. Although the evidential meaning of these examples is indisputable, it is not clear that evidentiality is a grammatical category in languages such as English or Spanish, as tense or person are.10 These languages lack markers that express evidentiality in a systematic way. On the other hand, it is also not clear that evidentiality and epistemic modality are distinct notions.

For the purposes of this paper and for the design of the questionnaires (see Appendix A and Appendix B), we will basically follow Aikhenvald’s proposal on evidential values. As we will see in Section 4, for our research we create contexts of direct and indirect evidentiality. In contexts of direct evidentiality (16a,b), the speaker has first-hand information because he or she has seen it, heard it or participated in some way in the situation. In (16c,d) the information has been obtained indirectly, but it is still personal because it comes from the speaker’s own reasoning, which either deduces it from observed facts or assumes it using logical reasoning (e.g., based on cause-effect), general knowledge, etc. In (16e,f) the access to information is indirect and not personal; it is accessed by the words of other people; it is second-hand information.

(16)		Evidential values (adapted from Aikhenvald 2018, p. 13)
		First-hand information (direct and personal)
	(a)	Visual: information acquired using sight.
	(b)	Non-visual sensory: obtained using other senses.11
		Indirect and personal information
	(c)	Inference: based on visible or tangible evidence or results.
	(d)	Assumption: information other than that obtained by tangible results; includes logical reasoning, assumptions or general knowledge.
		Second-hand information (indirect and not personal)
	(e)	Reported: information heard from another person without reference to the person who said it.
	(f)	Quotative: information heard from another person with reference to the authorship of the quoted source.

Regarding the limits between evidentiality and epistemic modality, there are authors who understand that they are distinct categories, despite the proximity between both notions. The evidentiality specifies the source of the information and epistemic modality, and the epistemic modality is reserved for grammatically encoding the degree of the speaker’s commitment to the truth of the proposition (Surely John knows vs. It is possible that John knows). The issue, however, is not straightforward because the source of the information has effects for the speaker’s commitment to the truth of the utterance. Let us illustrate this divergence with the examples in (17), taken from Fábregas (2020, p. 169):

(17)	a.	Evidentemente, han entrado en casa cuando no estábamos
		‘Evidently, they entered the house when we were not home’.
	b.	Según mi hermano, nos vas a llamar hoy
		‘According to my brother, you are going to call us today’.

Both sentences have declarative modality, but they do not show the same degree of commitment of the speaker to the propositional content. With evidentemente ‘evidently’, the speaker considers that he/she has first-hand evidence to take the content of the proposition at face value. The expression según mi hermano ‘according to my brother’ indicates that the evidence comes from a secondary source and blames someone else for the commitment to the truth of the propositional content. In this last case, the speaker may not take the truth of the proposition for granted, as it shows the possibility of adding the expression pero yo no lo creo ‘but I do not believe it’ in (18a). Notice, however, that it is not possible to make the same claim in (18b), since it is paradoxical to deny what we believe to be self-evident.

(18)	a.	Según mi hermano, nos vas a llamar hoy, pero yo no lo creo
		‘According to my brother, you are going to call us today, but I do not believe it’.
	b.	#Evidentemente, han entrado en casa cuando no estábamos, pero yo no lo creo
		‘Evidently, they entered when we were not home, but I do not believe it’.

What these data show is that evidentiality influences the degree of the speaker’s commitment to the truthfulness of what is asserted in the declarative modality. As we will see in the next sections, many studies understand evidentiality in a broad sense that includes the speaker’s commitment to what is asserted: the more direct the source of the information, the greater the speaker’s commitment. For the purposes of our research in Section 4, we will understand evidentiality as the encoding of the speaker’s source of information, but without losing sight of the close boundaries with epistemic modality.

3.1.2. Complementizers as Marks of Evidentiality

Let us recall that one of the purposes of our research is to find out whether the particle de ‘of’ in the construction with deísmo provides an evidential meaning. Since we have shown that de is a complementizer (see Section 2.2), it makes sense to ask whether other complementizers have been described from this perspective in other languages. We will analyze two cases: that in English and van in Brabant Dutch.

According to Van Gelderen (1999; taken from Rooryck (2001, p. 63)), the complementizer that has evidential value with verbs of language. In the example of (19), the presence of that implies general knowledge, i.e., that all the attendees knew that the wine was corked; however, the absence of the complementizer implies that the content of the subordinate sentence is John’s opinion. For this reason, in the example of (19), it would be ungrammatical to dispense with that because the context attributes the information to the general knowledge of all the attendees:

(19)	They all looked at each other across the table, but none of them dared to say anything. There was an uncomfortable silence. Finally, John said *(that) the wine was corked.

Van Craenenbroeck (2004, p. 43) notes that the presence of van in front of non-finite clauses in the Brabant Dutch varieties is related to evidential nuances with different classes of verbs. With raising and epistemic verbs, the absence of van implies direct evidentiality, while its presence means indirect evidentiality. In (20) van denotes that the speaker has direct evidence of the truth of the proposition; for example, because he is seeing how Marie enjoys swimming. If van is elided the speaker claims that Marie enjoys swimming because the speaker has been told so or infers it somehow, but he or she could be wrong. It follows from this asymmetry that in (21) van cannot appear with an expression meaning direct evidence of the type in mijn ogen ‘in my eyes’.

(20)	Marie schijnt (van) zwemmen plezant te vinden (Van Craenenbroeck 2004, p. 43)
	Mary seems (of) swim (INF) agreeable to find
	‘Mary seems to enjoy swimming’.
(21)	In mijn ogen scheen hij (*van) zich gisteren te amuseren
	in my eyes seemed he (of) himself yesterday to enjoy
	‘In my view he seemed to enjoy himself yesterday’.

With epistemic verbs (22), the presence of van also affects the meaning of the sentence. According to Van Craenenbroeck (2004, p. 44), when van is not present in the sentence, the speaker “is convinced that he can speak for everyone; he is convinced of the truth of the proposition expressed in the complement clause”. The presence of van, on the contrary, introduces doubts about the truth expressed using the non-finite clause.12

(22)	Ik meen (van) voor iedereen te kunnen spreken (Van Craenenbroeck 2004, p. 44)
	I believe (of) for everybody to can speak
	‘I believe I can speak for everyone.’

A final fact that demonstrates the evidential value of van in these varieties of Dutch is that this particle cannot appear with factive verbs, since they presuppose the truth of the complement. Note the difference implied by the change of the factive verb to regret for a non-factive verb such as to think: John {regrets/thinks} that the door is open. Only with the factive verb can we be certain that the door was open. This fact correctly predicts that in (23) the factive verb blijken ‘to turn out’ cannot be combined with the particle van, which is an indirect evidential marker of uncertainty.

(23)	Hij blijkt (*van) er niet bij betrokken te zijn (Van Craenenbroeck 2004, p. 48)
	he turns.out (of) there not with involved to be
	‘It turns out he is not involved in it’.

Finally, it is interesting to draw attention to Van Craenenbroeck’s hypothesis about the progressive loss of meaning of the particle van. According to this author, in Brabant Dutch the particle van has gone from being a spatial marker in the nominal domain to an aspectual or evidential marker in the verbal domain. The use as a spatial preposition still exists in examples such as Jan komt van Brussel ‘John comes from Brussels’ and the use as an evidential marker is the one we have reviewed in this section in the non-finite constructions. We will not go into the details of this, but both usages share what Van Craenenbroeck calls a dissociation reading that is in the original meaning of van ‘away from’.

There is a third phase of loss of meaning of complementizer van. In the non-Brabant part of the southern Dutch area, the use of complementizer van is completely optional and this lexical item has neither evidential nor aspectual meaning. The summary of his proposal on van is as follows: “At first it was a mere preposition, then it started functioning as a complementizer but still retained its semantic features and finally, it lost those features and became a mere grammatical element” (Van Craenenbroeck 2004, p. 61).

Finally, the author extends the hypothesis to the complementizers of other languages and proposes to analyze French de and Italian di as grammaticalized variants (24). In this sense, the complementizer de in the constructions with deísmo in Spanish (25) would coincide with the mentioned languages, since it does not contribute any meaning, as we will demonstrate in Section 4.

(24)	a.	Jean a essayé de chanter (Van Craenenbroeck 2004, p. 63)
		John has tried of sing (INF)
		‘John has tried to sing’.
	b.	Gianni ha tentato di cantare
		John has tried of sing (INF)
		‘John has tried to sing’.
(25)		Juan ha intentado de cantar
		Juan has tried of sing (INF)
		‘Juan ha intentado cantar’.

In short, some complementizers are not simple subordinating elements, but evidential marks in a broad sense: for example, in English that with the verbs of language implies general knowledge; and in Brabant Dutch van introduces evidential (information that is not first-hand) or epistemic (uncertainty) values. We will see in the following section that evidential values have also been attributed to the complement de ‘of’ in Spanish.

3.2. The Complementizer de ‘of’ as a Evidentiality Marker

As we advanced at the beginning of Section 3, dequeísmo has been explained as the encoding of an attenuating modal/evidential meaning (Bentivoglio and D’Introno 1977; García 1986; Schwenter 1999; Demonte and Soriano 2005; Guirado 2006; Del Moral 2008; RAE-ASALE 2009; and Gómez Molina 2011, among others). According to these authors, the presence of the particle de before a finite sentence adds a low commitment to the propositional content expressed using the subordinate clause headed with the complementizer que ‘that’. In a similar direction as Van Craenenbroeck (2004) for van (see Section 3.1.2), some authors have related the original locative/source meaning of the preposition de to an interpretation of separation or distance with respect to the content of the subordinate clause in dequeísmo (García 1986, p. 50; Schwenter 1999, p. 76). In this respect, Del Moral (2008, p. 183) stated that there has been a “categorial reanalysis of Spanish preposition de as having extended its concrete prepositional use of spatial marker to an abstract modal marker in dequeísta clauses”. Demonte and Soriano (2005) go a step further in the formal analysis of the complementizer de and place it as the head of an evidential phrase (see Section 2).

In Silva-Villar and Gutiérrez-Rexach (2012), some of the arguments put forward to defend the evidential character of the complement de in the construction with dequeísmo are questioned. Both the presumed frequency of non-first person subjects and the tendency to use past tenses (as markers of distance or less engagement) have been challenged in a survey conducted by these authors. First, they did not find that the percentage of third-person-only dequeístas speakers was significant. Second, certain parameters, such as the discourse type or mode, can determine the choice of grammatical subject. Finally, dequeísmo in Modern Spanish occurs increasingly with verbs of cognition, and particularly when the subject of the main clause is the first person, according to the study carried out by Del Moral (2008, p. 191). The example of (26) is, in fact, a typical case of dequeísmo.

(26)	Yo creo de que […] hay un tiempo para todo [CREA, oral]
	I believe of that […] there.is a time for everything
	‘I believe there is a time for everything’.

Finally, Silva-Villar and Gutiérrez-Rexach (2012) called into question other arguments that Demonte and Soriano (2005) have adduced to defend the attenuation evidential value of dequeísmo. Among other evidence, the incompatibility of dequeísmo with factive verbs or with verbs showing commitment by the speaker such as garantizar ‘guarantee’ and jurar ‘swear’. However, the following data show that dequeísmo is possible with these verbs: lo que hizo fue lamentar de que el presidente… ‘What he did was to regret that the president…; Te garantizo de que cuando lo hayas hecho, ocurrirá algo mágico ‘I give you my word that when you have done it, something magic will happen’ (Silva-Villar and Gutiérrez-Rexach 2012, p. 22).13 Finally, if de checks the evidentiality feature, it is predicted that it will be incompatible with evidential adverbs occupying the same position. However, this prediction is not borne out, as shown in the following example: Evidentemente de que las cifras que ellos hablan me parece que es imposible captar esa posibilidad para tanto recurso… ‘Evidently the numbers that they are talking about look to me as if it impossible to capture the possibility for so many resources…’ (Silva-Villar and Gutiérrez-Rexach 2012, p. 20).

These brief notes on dequeísmo have served to question its evidential value and will also serve to focus the debate on whether the complementizer de has any meaning in the construction with deísmo. Camus (2013) points out that dequeísmo and deísmo are different phenomena and proof of this is precisely that the complementizer de in deísmo has no evidential meaning, unlike what some authors have proposed for dequeísmo. According to Camus (2013), neither the geographical areas in which both phenomena are observed nor the kinds of verbs that appear in both constructions coincide.14 Although not all authors agree on whether the particle de has evidential value in dequeísmo, studying whether deísmo has this meaning may shed light on the relationship between both phenomena. For this reason, we will carry out our experiment in Section 4, in which we do not find a correlation between deísmo and evidential meaning. As we saw in Section 2, the preposition de could be considered a meaningless marker that introduces non-finite sentences, parallel to the one in the other Romance languages. In the remainder of this section, we will review what has been said about the meaning of de in deísmo constructions.

With regard to deísmo, it has been said that it is an expletive element without meaning (Camus 2013). According to De Benito and Pato (2015, p. 41), “de would have originally conveyed some kind of source meaning […], but that, later on, the rapid increase in the use of de was possible because its very broad meaning favored its weakening and it eventually became a barely significant introductory particle to the infinitive, rather than a real contributor to the meaning of the sentence”. As can be seen in this quote, these authors support a thesis that has also been defended by Van Craenenbroeck (2004) for van (see Section 3.1.2) and by Del Moral (2008) for dequeísmo (Section 3.2). In these two papers, the evidential meaning of van and de was related to the original locative/source meaning of these particles. Unlike these studies, De Benito and Pato (2015) did not see clearly that the particle de brings a general meaning to the present construction with deísmo.

Di Tullio (2012, 2022) attributed a prospective value to deísmo with volitional verbs (verbs of desire, influence, intention or proposal). The particle de brings out the future-oriented feature that these verbs usually express:

(27)	María propuso a su hermana de salir
	María suggested to her sister of go.out.INF
	‘María suggested to her sister to go out’.

With verbs that have more than one meaning, such as decir ‘to tell’, pensar ‘to think’ and probar ‘to try’, deísmo forces the prospective reading, especially when it combines with process and actions infinitives (see notes 15, 16 and 21 for the interpretation of these verbs). When de is combined with decir, pensar and probar, these verbs change from being verbs of language and thought to volitional verbs. The verb decir comes to mean ‘to propose’, the verb pensar becomes synonymous with ‘to plan’ and the verb probar does not mean ‘to demonstrate’ but ‘to try’. With verbs that are not volitional, the particle de does not seem to contribute any meaning, so that it cannot be asserted that deísmo inherently has a prospective meaning:

(28)	a.	Dijo de ir juntos
		said.PAST.3SG of go.INF together
		‘He proposed to go together’.
	b.	Estaba pensando de llegar pronto
		was.PAST.1SG thinking of arrive.INF early
		‘I was planning to arrive early’.
	c.	Probé de ir en coche
		tried.PAST.1SG of go.INF in car
		‘I tried going by car’.

Finally, Carrasco and Peinado (2022) studied deísmo with verbs of visual perception. For these constructions, they proposed that the segment de is a morpheme that heads an evidential phrase. The authors related the type of complement that the verb ver ‘to see’ selectsto different meanings. When ver selects a nominal phrase or a non-finite verb (29a,b), the complements represent the objects of perception. The speaker reports what he/she has seen (direct perception or non-epistemic, according to the authors). In semantic terms, the complement in (29a) would be a first-order entity, a person, and in (29b) it is a second-order entity, an event (Carrasco and Peinado 2022, p. 85). However, in (29c), if the verb selects a finite clause, the speaker reports what he knows and conveys knowledge that has been acquired by sight (indirect or epistemic perception). The complement in (29c) is a third-order entity, i.e., a proposition, which reproduces a content of consciousness. The question is what is meant by the construction with deísmo: Juan ha visto de llover (Lit. Juan has seen of rain.INF). Formally, this example would seem to be closer to (29b); however, the authors claim that meaning is closer to (29c).

(29)	a.	Juan ha visto a María
		Juan has seen to María
		‘Juan has seen María’.
	b.	Juan ha visto llover
		Juan has seen rain.INF
		‘Juan has seen rain’.
	c.	Juan ha visto que llovía
		Juan has seen that rained
		‘Juan saw that it was raining’.

A proof to defend their proposal was based on the contrasts in (30), which are examples adapted from Carrasco and Peinado (2022, p. 101). As we have just seen, the construction with deísmo (Juan ha visto de llover) and the construction with the finite clause (Juan ha visto que llovía) refer to states of consciousness, so that we could not attribute contradictory knowledge to the subject. Precisely for this reason, the example without deísmo (30a) is natural, but not the examples with deísmo and the finite clause (30b), according to the judgments of the speakers consulted by the authors. In (30b), we could not say that Juan knows that it is raining, because he has seen it, but he thinks that it is not raining.

(30)	a.	Juan ha visto llover, pero ha pensado que el vecino regaba sus plantas
		‘Juan saw it raining, but thought the neighbor was watering his plants’.
	b.	#Juan ha visto {que llovía/de llover}, pero ha pensado que el vecino regaba sus plantas
		‘Juan saw that it was raining, but he thought that the neighbor was watering his plants’.

The authors referred to “indirect perception” (with a finite clause or with deísmo), when the speaker conveys knowledge acquired by sight and not when the object of perception is presented (with a non-finite clause without deísmo). In the construction with deísmo, the visual experience is always direct, not inferred;15 the indirect nature refers to the fact that what is presented is knowledge and not the object of perception. In their hypothesis, de is a morpheme that heads an evidential phrase and encodes this meaning of indirect perception. It is not known, however, to what extent this interpretation extends to other verbs that admit deísmo and are not verbs of perception. This is a question to be investigated.

The aim of Section 3, which we now close, has been to review what has been said about the meaning of deísmo in Spanish. For this purpose, we have reviewed in Section 3.1.2 a construction similar to deísmo in Brabant Dutch. In Section 3.2, we reviewed the possible meaning of the particle de in deísmo and dequeísmo. From Section 3, we can deduce that it is plausible to think that deísmo could convey some indirect evidential value, but the data provided in the various studies dealing with this question have been called into question. In the next section we will precisely carry out a survey to show whether deísmo has some evidential meaning.

4. Deísmo and Evidentiality: The Data

The aim of our study was to find out whether deísmo, a frequent phenomenon in the non-standard language of Castilla-La Mancha, carries an evidential meaning analogous to that which has been described, that is, on the one hand for phenomena similar to deísmo in other languages and, on the other hand, for dequeísmo or deísmo in Spanish (see Section 3).

The method we followed for our research was the following: First, starting from the list of frequent verbs with deísmo collected in the literature (see Section 2.2), we made an initial search in the Corpus Oral y Sonoro del Español Rural (COSER), which allowed us to look for regional data. Having observed the small number of realizations with de in interviews as long as those in COSER, and realizing the difficulty of obtaining the data in the right evidential contexts, we decided to use a questionnaire to obtain our data. We are aware of the limitations of using this type of data instead of examples of real use.

With the verbs chosen, we designed a preliminary questionnaire that was carried out by our contacts with eight informants from Castilla-La Mancha who have deísmo in their dialects. Secondly, based on the results of this questionnaire, we designed a second one. For this, we also made an additional search in another corpus that allows searching of regional data, in newspapers of Castilla-La Mancha: Spanish Web Corpus 2018 (Sketch Engine) (2018). On this occasion, we went to two towns in Castilla-La Mancha (Pozuelo de Calatrava and Pedro Muñoz) to conduct the questionnaire in person with twelve other informants. Since we were seeking to understand the meaning, if any, associated with the complementizer de, we needed, in addition to obtaining the speakers’ responses to the questionnaire, to ask for clarification about the nuances of meaning associated with the construction.

As we will see in the following sections, obtaining data on deísmo entails some difficulties: First of all, it is mostly oral data that only occur in certain areas. This forced us to limit our searches to corpora including information on the region of origin of the speakers. Secondly, deísmo is a substandard phenomenon on which there is not much normative pressure and in which there is much variation among speakers, who alternate between standard constructions and constructions with deísmo. Third, there is no uniform geographical distribution of the verbs that may appear in the construction with deísmo. We know from the COSER study by De Benito and Pato (2015) of some villages16 where speakers have been recorded, but this does not mean that deísmo is used by the majority of people or that there is a uniform solution in those localities. In our case, the way to make sure that our informants were speakers with deísmo was to contact them through fellow linguists who knew them. This fact limited the number of informants, but we believe that the data collected serve to confirm that the complementizer is not an evidential marker in the construction.

4.1. Deísmo Data in Corpus Oral y Sonoro del Español Rural (COSER)

The first reason to look for examples of deísmo in COSER is that deísmo is a rural oral phenomenon and, therefore, examples would be expected to appear in this corpus. COSER is made up of recordings of the language spoken in rural areas of the Iberian Peninsula. It consists of long interviews that were obtained with the aim of offering a representative sample of the dialectal variety. The second reason is that, as opposed to other linguistic corpora, such as the Corpus de Referencia del Español Actual (CREA), the Corpus del Español del Siglo XXI (REAL ACADEMIA ESPAÑOLA: Banco de Datos CORPES n.d.) and the Corpus del Español de Mark Davies (Davies 2016) in the COSER data can be selected by province.

The nature of deísmo, which is an oral, colloquial phenomenon belonging to a vernacular variety, makes it difficult to find data on the phenomenon in the corpora. This explains why, both in this article and in most of the previous works, the amount of data used is small and, therefore, there is not much quantitative analysis. As for the source of the data of the papers on deísmo, they either come from corpora (COSER and REAL ACADEMIA ESPAÑOLA: Banco de Datos CORPES n.d.) or were elicited through surveys of native speakers of the deísmo variety. That is the case of Camus (2013) and Carrasco and Peinado (2022, which used grammaticality judgments from the introspection of native speakers of the variety. Neither article provided data on the number of speakers or examples consulted, nor performed quantitative analysis of the data.)

Based on the classification of Gutiérrez-Rodríguez (forthcoming), the verbs we searched for in COSER are the verbs shown in the Table 1, grouped by lexical class. As shown in Table 1, we found no examples of some verbs in constructions with deísmo anywhere in Spain. With other verbs, there were examples, but not from Castilla-La Mancha. Finally, for some of these verbs we did obtain examples in this region; we marked the occurrences in brackets: gustar ‘to like’ (5), tocar ‘to be one’s turn’ (2), mandar ‘to order’ (5), estar deseando ‘to look forward to’ (2), querer ‘to want’ (1), hacer ‘to make’ (2) and ver ‘to see’ (1).17 When there were no constructions with deísmo with any of the verbs, it is indicated with the number 0. When there were examples, these are indicated with the exact number next to the province.18 The last column shows the total number of constructions.

4.2. The First Questionnnaire

4.2.1. Design

We designed a first questionnaire with twenty questions, of which four were control questions (see Appendix A).19 We passed this questionnaire to eight informants through our contacts. Since we were seeking to find out whether the particle de ‘of’ has evidential value, we created two contexts with the most frequent verb classes in the construction with deísmo (psychological, influence, intention and will). The verbs chosen were aceptar ‘to accept’, apetecer ‘to feel like’, convenir ‘to be in one’s best interest’, decir ‘to tell’, gustar ‘to like’, intentar ‘to try’ and proponer ‘to propose’, all of which are very frequent verbs in everyday speech. In the survey, the seven verbs appeared in contexts of direct evidentiality (you have heard, you are seeing, you see, etc.), and indirect evidentiality (you have been told, apparently, you think that), as in the following example with the verb intentar ‘to try’:

(31)	Question 8 (direct evidentiality) and question 3 (indirect evidentiality) of the first questionnaire:
	8. Imagínate que estás en el salón con tu hijo y estás viendo cómo monta una estantería de IKEA. Dirías…
	‘Imagine that you are in the living room with your son and you are watching him assemble an IKEA shelf. Would you say…’
		Mi hijo está intentando de montar la estantería.
		‘My son is trying to assemble the bookshelf’
		Sí / No / No estoy seguro/a
		‘Yes / No / I am not sure’.
	3. Te han dicho que tu hijo lleva toda la tarde arreglando la lavadora. Dirías…
	You have been told that your son has been fixing the washing machine all afternoon. Would you say…’
		Mi hijo está intentando de arreglar la lavadora.
		‘My son is trying to fix the washing machine.’
		Sí / No / No estoy seguro/a
		‘Yes / No / I am not sure’.

In this first questionnaire with a small number of informants, we wanted to find out if there was deísmo with some verbs and if there was evidentiality in the construction. Our speakers, who are identified in the table with the letter H (Hablante ‘speaker’) and a number from 1 to 8, were consistent with their answers to the control questions, but we could not find patterns in the evidentiality values, nor was there much uniformity in the acceptance of the sentences with deísmo, as we will see when discussing the results in Section 4.2.3.

4.2.2. Results

Table 2. Results of the first questionnaire ordered by verb in its evidential context.

Questions from the First Questionnaire Ordered by Verb	Answers According to Evidentiality
	Yes	No	NS
1. A mi hijo le gusta de nadar (directa) ‘My child likes to swim’	H2	H1 H3 H4 H5 H6 H7 H8
16. A mi vecino le gusta de ir a los toros (indirecta) ‘My neighbor likes to go to the bullfights’	H1	H2 H3 H4 H5 H6 H7 H8
20. Le gusta de pasear todas las tardes (directa) ‘He likes to go for a walk every afternoon’	H2 H3 H5 H6	H1 H4 H7 H8
12. Le gusta de ir al Retiro todos los días (indirecta) ‘He likes to go to the Retiro every day’	H1 H2 H5 H6	H3 H4 H7 H8
8. Mi hijo está intentando de montar la estantería (directa) ‘My son is trying to assemble the bookshelf’	H1 H3 H4 H5 H6	H7 H8
3. Mi hijo está intentando de arreglar la lavadora (indirecta) ‘My son is trying to fix the washing machine’	H1 H2 H5 H6 H8	H3 H4 H7
18. Mi amigo ha aceptado de ir al cine (directa) ‘My friend has agreed to go to the movies’	H1 H2 H3 H4 H6	H5 H7 H8
4. Mi amigo ha aceptado de venir a mi boda (indirecta) ‘My friend has agreed to come to my wedding’	H1 H4 H6 H7 H8	H2 H3 H5
7. Le apetece de tirarse del tobogán (directa) ‘He feels like going down the slide’	H1 H2 H4 H6	H3 H5 H7 H8
19. Le apetece de irse a casa (indirecta) ‘He wants to go home’	H1 H2 H3	H5 H6 H7 H8	H4
5. Es obvio que le conviene de ponerse a la sombra (directa) ‘It’s obvious that it’s good for you to go to the shade’	H2 H4 H8	H1 H3 H5 H6 H7
11. Por lo visto, conviene de vacunarse (indirecta) ‘It seems to be a good idea to get vaccinated’	H2 H4 H3	H1 H5 H6 H7 H8
10. Me propuso de ir al cine (directa) ‘He suggested to me to go to the movies’	H3 H4 H5 H7	H1 H2 H6 H8
15. Le propuso de ir al cine (indirecta) ‘He suggested to her to go to the movies’	H1 H2 H3 H4 H5 H6 H7 H8
2. Le ha dicho de ir con ella a hacer la compra (directa) ‘He told her to go shopping with her’	H2 H3 H4 H5 H6 H7 H8	H1
17. Le ha dicho de ir con ella a hacer la compra (indirecta) ‘He told her to go shopping with her’	H1 H2 H3 H4 H5 H7 H8	H6
13. Se preocupa de cuidar el jardín ‘He takes care of the garden’	H1 H2 H3 H4 H5 H6
14. Finalmente, se libró de examinarse otra vez ‘Finally, he got out of taking another exam’	H1 H2 H3 H4 H5 H6
6. Mi madre me pidió que fuera con ella ‘My mother asked me to go with her’	H1 H2 H3 H4 H5 H6
9. Ese libro, Juan no le ha leído todavía ‘That book, John has not read it yet’	H1 H2 H3 H4 H5x H6

summarizes the results of the first questionnaire. The first column shows the verbs with their two contexts (direct and indirect evidentiality) and the number of the corresponding question in the questionnaire (see Appendix A). The other columns correspond to the possible answers: Yes, I would say it (Yes), No, I would not say it (No) and Not sure (NS). The answers of the eight speakers are coded H1 (Hablante 1 ‘speaker 1′), H2 (Hablante 2 ‘speaker 2′), H3 (Hablante 3 ‘speaker 3′), etc.

Table 3. Results of the first survey: number of realizations per verb and per speaker.

	H1 (59 age)	H2 (56)	H3 (25)	H4 (26)	H5 (26)	H6 (56)	H7 (14)	H8 (64)	Total per Verb
Gustar20 ‘to like’	1 ind	1 dir	0	0	0	0	0	0	2
Gustar ‘to like’	1 ind	2	1 dir	0	2	2	0	0	5
Intentar ‘to try’	2	2	1 dir	1 dir	2	2	0	1 ind	7
Aceptar ‘to accept’	2	1 dir	1 dir	2	0	2	1 ind	1 ind	7
Apetecer ‘to feel like’	2	2	1 ind	1 dir	0	1 dir	0	0	5
Convenir ‘to be in one’s best interest’	0	2	1 ind	2	0	0	0	1 dir	4
Proponer ‘to propose’	1 ind	1 ind	2	2	2	1 ind	2	1 ind	8
Decir ‘to tell’	1 ind	2	2	2	2	1 dir	2	2	8
Total per speaker	10	13	9	10	8	9	5	6

shows the results of the first questionnaire ordered by verb and speaker. In this table, the speaker appears with his/her age (in parentheses) in the first row. When a 0 appears in one of the boxes, it means that the speaker did not accept the preposition de ‘of’ with that verb. If 1 appears, it means that the speaker only accepted the preposition de in one of the contexts, marked as ind (indirect evidentiality) or dir (direct evidentiality). Finally, 2 means that the speaker accepted the preposition de in both contexts. The table also includes the total number of realizations of the verb with deísmo (in the last column on the right) and the total number of realizations with deísmo of each speaker (in the last row).

To facilitate the discussion in Section 4.2.3, we have summarized in

Table 4. Summary of Table 2: verbs and evidential values.

	Direct Evidential Context		Indirect Evidential Context
	Yes	No	Yes	No
Gustar	5	11	5	11
Intentar	5	2	5	3
Aceptar	5	3	5	3
Apetecer	4	4	3	4
Convenir	3	5	3	5
Proponer	4	4	8	0
Decir	7	1	7	1

the results of Table 2 on the relationship between the verb and its evidential value (leaving aside the details about the speaker).

For the verb gustar, only five speakers accepted de in the context of direct evidentiality compared to 11. This is to be expected if de is a marker for indirect evidentiality. However, the same results were obtained in the context of indirect evidentiality, when we would expect more affirmative answers with de. Analyzing the rest of the verbs in the same way, we observed that in no case were more examples with de accepted in indirect evidentiality contexts, except with the verb proponer (and even with this verb, half of the speakers also accepted de in direct evidentiality contexts).

Regarding the age of the informants, although the low number of speakers makes it difficult to draw conclusions regarding this variable, data grouped by age are presented in

Table 5. Results of the first survey by age.

AGE	Examples Accepted with de	Direct Evidentiality Contexts	Indirect Evidentiality Contexts
14–26 (H3, H4, H5, H7)	32	17	15
56–64 (H1, H2, H6, H8)	38	17	21
Total	70	34	36

below.

In terms of age, there were eight speakers in this first survey, which can be grouped into two age bands. The younger speakers accepted 32 realizations with de, of which there was a majority in contexts of direct evidentiality, although the difference with contexts of indirect evidentiality was small (17/15). In the case of older speakers, they accepted more realizations with de, which was to be expected, but again the difference with younger speakers in the total number of sentences with de was small (38/32). In this case, there were more examples of indirect evidentiality but, again, we believe that no consequences can be drawn from the small difference: 17 of direct evidentiality versus 21 of indirect evidentiality.

4.2.3. Discussion

As shown in the last column of Table 3, all the verbs chosen present deísmo. The verb that presented it to a lesser extent was convenir ‘to be in one’s best interest’, which speakers accepted in very few sentences with de, and is why is it was suppressed in the second questionnaire (see Appendix A and Appendix B). The most frequently accepted verbs with de were proponer ‘to propose’ (included in the second questionnaire) and decir21 ‘to tell’ (excluded from the second questionnaire), which all speakers accepted with de ‘of’. We did not include decir de ‘to tell’ as a verb of influence (‘to propose’) in the second questionnaire because there was no version without de ‘of’ with this meaning: Me dijo *(de) ir al cine ‘He told me to go to the cinema’. The structure with de is similar to the value it acquires when combined with the subjunctive (Me dijo que fuéramos juntos ‘He told me to go together’). The other two verbs we kept in the second questionnaire were gustar ‘to like’ and intentar ‘to try’, both of which had sufficient realizations and are representative of the psychological and intention verb classes. On the other hand, we decided to exclude aceptar ‘to accept’ and apetecer ‘to feel like’, the former because we already had another verb of the same lexical class with more realizations (proponer ‘to propose’), and the latter because we left gustar ‘to like’ as the representative of the class.

As for the speakers (see Table 3), all of them accepted sentences with deísmo but the youngest (H7) accepted fewer (5 out of a total of 16 examples), which was not surprising because younger age is associated with greater use of the standard and, therefore, less use of vernacular varieties. Even so, we find it remarkable that a 14-year-old speaker accepted sentences with deísmo, and also that it was accepted by the three speakers aged 25 and 26 (H3, H4 and H5), which shows a certain vitality of the phenomenon. Moreover, there was no correlation between the age of the speakers and the number of realizations with deísmo (see Table 5 and its discussion above).

Regarding evidentiality (see Table 2 and Table 3), we believe that de ‘of’ can be considered a marker of evidentiality in a speaker if this preposition is accepted consistently either in contexts of direct evidentiality or indirect evidentiality. However, if the speaker accepted the preposition de in direct evidentiality contexts with some verbs and in indirect evidentiality contexts with others, de was not considered a marker of evidentiality. If the speaker accepted sentences with the preposition de in both contexts, we also deduced that it has no evidential value. The results regarding evidentiality were inconsistent in all cases, except in speaker 1 (and only for the verbs decir ‘to tell’, gustar ‘to like’ and proponer ‘to propose’), who accepted the examples with de only in the contexts of indirect evidentiality (see Table 3). However, this same speaker accepted the preposition with three other verbs (aceptar ‘accept’, apetecer ‘to feel like’ and intentar ‘to try’) in both direct and indirect evidentiality contexts. In Table 3, the total count of realizations with deísmo offered 12 results with de with indirect evidentiality value and 10 with direct evidentiality value. Table 4 shows that none of the verbs, except perhaps proponer, had more realizations of de in contexts of indirect evidentiality. Our conclusion, from this first limited sample, was that the particle de cannot be considered a marker of evidentiality, direct or indirect.

In summary, from the first questionnaire we concluded that, although there is deísmo, there is also an enormous variation and instability in the judgments regarding the presence of the particle de with a verb. This questionnaire also served us to confirm that there was deísmo in Castilla-La Mancha, even with verbs discarded in the literature, such as gustar ‘to like’, intentar ‘to try’ and convenir ‘to be in one’s best interest’ (Camus 2013, p. 24).

4.3. Study of Deísmo in Spanish Web Corpus 2018 (Sketch Engine) (2018)

To complete the data obtained in COSER, we performed the same search for verbs in the Spanish Web Corpus 2018 (Sketch Engine) (2018), which enables the creation of regional subcorpora using written press. The data appeared in the following newspapers of Castilla-La Mancha: latribunadetoledo.es; lanzadigital.com; latribunadealbacete.es (accessed on 1 August 2023). We only obtained examples of the following verbs: gustar ‘to like’ (2), hacer ‘to make’ (12), intentar ‘to try’ (7), pensar ‘to think’ (1), permitir ‘to allow’ (1), proponer ‘to propose’ (2), querer ‘to want’ (2), tocar ‘to be one’s turn’ (2) and ver ‘to see’ (4). The fact that they appear in written language makes us think that they are frequent in spoken language and that is why we decided to carry out the second questionnaire including some of these verbs (see Appendix B). Of the verbs that we initially searched for in COSER, we did not find any examples of mandar ‘to order’, convenir ‘to be in one’s interest’, hacer falta ‘to need’, ser costumbre ‘be costumary’, prohibir ‘to forbid’, pedir ‘to ask’, ordenar ‘to order’, oír ‘to hear’ or escuchar ‘to listen’.

4.4. The Second Questionnnaire

4.4.1. Design

For the second survey we decided to go to two La Mancha towns where we knew there was deísmo (Pedro Muñoz and Pozuelo de Calatrava). In both of them we had contact persons who previously looked for the twelve speakers with deísmo. The interviews took place in an informal setting in the homes of the informants, so we were able to note many additional data and realizations during our conversation.22

Given that on this second occasion we were in charge of reading the questions to the informants and writing down their answers, we were sure that they were not answering randomly or unreflectively. In addition, by conducting the interviews in person, we were able to better control the responses of the speakers, from whom we could ask for clarifications. This allowed us to take notes of their comments and to record other cases of deísmo while we were conversing.23

We designed this twenty-question questionnaire with ten different verbs, each in a direct and indirect evidentiality context, just as we did in the first survey (see Appendix B). We chose one verb from each lexical class (see Section 2.1).24 The lexical classes and predicates chosen were the following: psychological verbs, gustar ‘to like’; verbs of occurrence, tocar ‘to be one’s turn’; verbs of intention, intentar ‘to try’; verbs of thought, pensar ‘to think’; verbs of influence, proponer ‘to propose’ and mandar ‘to order’; verbs of will, querer ‘to want’; and causative verbs, hacer ‘to make’. In addition, we included two semantically impersonal predicates, ser costumbre ‘to be customary’ and hacer falta ‘to need’.

In all cases, we chose either the most frequent verb with deísmo in the data we obtained, or the verb that had examples of deísmo in Castilla-La Mancha. From the first questionnaire we kept gustar ‘to like’, intentar ‘to try’ and proponer ‘to propose’. All three verbs had given results with deísmo in the first questionnaire and in Sketch Engine. We eliminated decir because it was not clear whether it forms a construction with deísmo (see note 21) and convenir ‘to be advisable’ because it is the verb that fewer speakers accepted with de ‘of’ in the first survey and did not give results in Sketch Engine. In return, we introduced verbs to cover more lexical classes with deísmo, so that we significantly increased the type of predicates with respect to the first questionnaire.

In the questionnaire, we added a fourth option for negative answers: I would not say it, but I have heard it (see Appendix B). The first objective of this question was to understand the degree of dialectal awareness of the informant. The second objective was to examine whether in this environment it really is an unknown construction, because if the construction is used, it could be the case that the informant says he does not use it because he considers that the standard form is better.

4.4.2. Results

Following the same criteria we used for Table 1 of the first questionnaire,

Table 6. Results of the second questionnaire ordered by verb in its evidential context.

Questions from the Second Questionnaire Sorted by Verb	Answers According to Evidentiality
	Yes	No	Heard	NS
1. A mi hijo le gusta de nadar (directa) ‘My son likes swimming’	H4	H1 H2 H3 H5 H6 H7 H8 H9 H10 H11 H12	H1 H2 H3 H6 H7
12. Aunque nunca me lo he encontrado en el parque, me han dicho que le gusta de ir allí todos los días (indirecta) ‘Although I have never met him in the park, I have been told that he likes to go there every day’	H1 H2 H4 H5 H7	H3 H6 H8 H9 H10 H11 H12	H6
2. Mi madre le ha mandado de hacer los deberes (directa) ‘My mother told him to do his homework’	H1	H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12	H2 H4 H5 H6
17. A los niños les han mandado de ir con uniforme (indirecta) ‘The children have been instructed to wear uniforms’	H1 H3 H8 H9 H10 H11	H2 H4 H5 H6 H7 H12		H4
8. Mi hijo está intentando de montar la estantería (directa) ‘My son is trying to assemble the shelf’	H2 H7	H1 H3 H4 H5 H6 H8 H9 H10 H11 H12	H1 H6
3. Mi hijo está intentando de arreglar la lavadora (indirecta) ‘My son is trying to fix the washing machine’	H5 H7 H9 H11 H12	H1 H2 H3 H4 H6 H8 H10	H1 H2 H6
6. Quiero de alquilar mi apartamento (directa) ‘I want to rent my apartment’		H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12	H5 H6 H7
4. Mi amigo quiere de vender el coche (indirecta) ‘My friend wants to sell the car’		H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12	H6
5. Me ha tocado de ir a Burgos (directa) ‘I had to go to Burgos’	H1	H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12	H2 H3 H5 H6
7. Me ha tocado de sacar al perro (indirecta) ‘I had to take the dog out’	H2 H6	H1 H3 H4 H5 H7 H8 H9 H10 H11 H12	H3 H5 H12
19. Estoy pensando de ir al norte este verano (directa) ‘I am thinking of going north this summer’	H1 H3 H8 H10 H11	H2 H4 H6 H7 H9 H12	H7 H9	H5
9. Mi hijo está pensando de casarse pronto (indirecta) ‘My son is thinking of getting married soon’	H1 H3 H4 H5 H6 H7 H11	H2 H9 H10 H12	H2 H8 H10	H8
10. Luis me ha propuesto de ir al cine (directa) ‘Luis asked me to go to the movies’	H1 H2 H3 H6 H7 H8 H9 H10 H11 H12	H4 H5
15. Pedro le propuso de ir al cine a mi hija (indirecta) ‘Pedro asked my daughter to go to the movies’	H1 H2 H3 H4 H7 H9 H10 H11	H5 H6 H8 H12	H8
11. Hace falta de barrer el suelo (directa) ‘Sweeping the floor is necessary’	H4 H6 H7	H1 H2 H3 H5 H8 H9 H10 H11 H12	H5
13. Hace falta de regar las plantas (indirecta) ‘Watering the plants is necessary’	H7 H9 H11	H2 H3 H4 H5 H6 H8 H10 H12	H1	H1
18. En mi casa era costumbre de comer pavo en Nochebuena (directa) ‘In my house it was customary to eat turkey on Christmas Eve’	H1 H2 H4 H9 H11 H12	H3 H5 H6 H7 H8 H10	H5
14. Antes era costumbre de regalar el vestido a la novia (indirecta) ‘It used to be customary to give the dress to the bride as a gift’	H1 H2 H3 H9 H10 H11 H12	H4 H5 H6 H7 H8	H5 H6
16. Ahora nos hacen de fichar todos los días (directa) ‘Now they make us clock in every day’	H1 H6 H7 H9 H10 H11	H2 H3 H4 H5 H8 H12
20. Nos van a hacer de entrar a las ocho todos los días (indirecta) ‘They are going to make us come in at eight o’clock every day’	H1 H3 H4 H5 H6 H8 H9 H10 H11	H2 H7 H12	H12

summarizes the results of the second survey. The first column shows the verbs with their two contexts (direct and indirect evidentiality) and the number of the corresponding question in the questionnaire (see Appendix B). The other columns correspond to the possible answers: Yes, I would say it (Yes); No, I would not say it (No); I would not say it, but I have heard it (Heard); and I am not sure (NS). The responses of the twelve speakers are coded with H1 (Hablante 1 ‘speaker 1’), H2 (Hablante 2 ‘speaker 2’), H3 (Hablante 3 ‘speaker 3’), etc.

Table 7. Results of the second questionnaire: realizations per verb and per speaker.

	H1 (57 age)	H2 (53)	H3 (26)	H4 (60)	H5 (73)	H6 (74)	H7 (64)	H8 (61)	H9 (63)	H10 (52)	H11 (51)	H12 (35)	Total per verb
Gustar ‘to like’	1 ind	1 ind	0	2	1 dir	0	1 ind	0	0	0	0	0	5
Mandar ‘to order’	2	0	1 ind	0	0	0	0	1 ind	1 ind	1 ind	1 ind	0	6
Intentar ‘to try’	0	1 dir	0	0	1 ind	0	2	0	1 ind	0	1 ind	1 ind	6
Querer ‘to want’	0	0	0	0	0	0	0	0	0	0	0	0	0
Tocar ‘to be one’s turn’	1 dir	1 ind	0	0	0	1 ind	0	0	0	0	0	0	3
Pensar ‘to think’	2	0	2	1 ind	1 ind	1 ind	1 ind	1 dir	0	1 dir	2	0	9
Proponer ‘to propose’	2	2	2	1 ind	0	1 dir	2	1 dir	2	2	2	1 dir	11
Hacer falta ‘to be necessary’	0	0	0	1 dir	0	1 dir	2	0	1 dir	0	1 ind	0	5
Ser costumbre ‘to be customary’	2	2	1 ind	1 dir	0	0	0	0	2	1 ind	2	2	8
Hacer ‘to make’	2	0	1 ind	1 ind	1 ind	2	1 dir	1 ind	2	2	2	0	10
Total per speaker	12	7	7	7	4	6	9	4	9	7	11	4

shows the results of the second questionnaire ordered by verb and by speaker (H1, H2, H3…). In this table, the speaker appears with his/her age in the first row. When a 0 appears in one of the boxes, it means that the speaker did not accept de with that verb. If 1 appears, it means that the speaker only accepted the particle de in one of the contexts, indicated as ind (indirect evidentiality) or dir (direct evidentiality). Finally, 2 means that the speaker accepted de in both contexts. The table also includes the total number of realizations of the verb with deísmo (in the last column on the right) and the total number of realizations with deísmo for each speaker (in the last row).

To facilitate the discussion in Section 4.4.3, we have summarized in

Table 8. Summary of Table 6: verbs and evidential values.

	Direct Evidential Context		Indirect Evidential Context
	Yes	No	Yes	No
Gustar	1	11	5	7
Mandar	1	11	6	6
Intentar	2	10	5	7
Querer	0	12	0	12
Tocar	1	11	2	10
Pensar	5	6	7	4
Proponer	10	2	8	4
Hacer falta	3	8	6	6
Ser costumbre	6	6	7	5
Hacer	6	6	9	3
Total	35	83	55	64

the results of Table 6 on the relationship between the verb and its evidential value (leaving aside the details about the speaker).

The results of Table 8 indicate that with four verbs (pensar, proponer, ser costumbre and hacer) more examples were accepted with de in contexts of indirect evidentiality. However, those same verbs were also accepted with de quite a few times in contexts of direct evidentiality, so the results did not clearly show that de is a marker of indirect evidentiality even with these verbs. In particular, the verb proponer had more results with de in direct evidential contexts. For the rest of the verbs, all of them received less acceptance of de in an indirect evidential context, which we would not expect if de were a marker of indirect evidentiality.

Nevertheless, there was a difference in the total values, insofar as in the contexts of direct evidentiality there were 83 realizations without de versus 35 with de, which was to be expected if de is a marker of indirect evidentiality. On the other hand, we would expect that in the contexts of indirect evidentiality the numbers would be reversed, so that there would be more examples with de than without it, but this was not the case, since while 55 examples were considered acceptable with de, there were 64 results with de considered unacceptable. Thus, even if the presence of de was slightly higher in indirect evidentiality contexts, from those numbers we cannot deduce that de is a marker of this type of evidentiality. We will expand on this in the following discussion. Before discussing the results of all these tables in detail in the next section, we conclude with a summary by age, as we did for the first survey. However, in the second questionnaire, there were only two young informants, so these data should be taken with caution.

The speakers who accepted de the most were those between 50 and 60 years old, and although in the three age groups the preposition was accepted a little more in contexts of indirect evidentiality, the difference in acceptance was not large: there were 49 accepted cases in total for indirect evidentiality and 38 for direct evidentiality. On the other hand, the two youngest speakers showed less acceptance of sentences with the preposition de than the others, which is to be expected with a dialectal phenomenon, as discussed before.

4.4.3. Discussion of the Second Survey

With respect to the verbs that already appeared in the first survey (gustar ‘to like’, intentar ‘to try’ and proponer ‘to propose’), it was confirmed in this second questionnaire (see Table 6 and Table 7) that there is deísmo with all three verbs and that proponer is the one which is more accepted in sentences with deísmo. The verb querer ‘to want’ was the only one of all the verbs studied with zero results (see Table 7). In this second survey we included this verb firstly because of its frequency in general Spanish and secondly because it had the most realizations of the verbs of will in COSER (see Table 1), and it also had results in Sketch Engine (see Section 4.3). The final result with zero realizations in our second questionnaire seemed to confirm what Camus (2013, p. 24) pointed out for this verb in Castilla-La Mancha, in spite of the data we had previously obtained in the corpora25 (Table 1 and Section 4.3). All the other verbs presented five or more realizations of deísmo in Table 7. Only the verb tocar ‘to be one’s turn’ had three, but this verb was the one for which more speakers indicated the answer I would not say it, but I heard it, which makes us think that it is also a verb used in the deísmo construction. It was also confirmed that deísmo is frequent in the causative construction in Castilla-La Mancha (10 results in Table 7), as noted by Camus (2013).26

As for the informants (all of them speakers with deísmo according to our contacts), we confirmed that they all accept sentences with deísmo. Although the sample of informants was small (12), there seemed to be no relationship between deísmo and age (see Results by age in

Table 9. Results of the second survey by age..

AGE	Examples Accepted with de	Direct Evidentiality Contexts	Indirect Evidentiality Contexts
26–35 (H3, H12)	11	4	7
50–60 (H1, H2, H4, H10, H11 H8)	44	20	24
61–75 (H5, H6, H7, H8, H9)	32	14	18
Total	87	38	49

), although the two younger speakers accepted the presence of the preposition de on fewer occasions.

Regarding evidentiality (see Table 6 and Table 7), as in the first questionnaire, we considered de to be a marker of evidentiality only if this preposition was accepted either in contexts of direct evidentiality or indirect evidentiality. However, if the speaker accepted de in contexts of direct evidentiality with some verbs and in contexts of indirect evidentiality with others, de was not considered a marker of evidentiality. Finally, if the speaker accepted the particle in both contexts, it was not considered a marker of evidentiality either. The judgments of our speakers were not consistent, although there were two speakers (H3 and H11 in Table 7) who, when they choose only one option, it was the one of indirect evidentiality. However, these same speakers accepted the presence of de in both direct and indirect evidentiality contexts with other verbs.

One conclusion from our study is that the variation seems to be related to individual choices associated with specific verbs (rather than lexical classes) because there was a lot of variation among speakers, except with the verb proponer ‘to propose’ and the causative hacer ‘to make’. What we do find remarkable is that the total count of the particle de in contexts of indirect evidentiality was double that of direct evidentiality (see Table 7). Despite this result, we believe that the conclusions of the first questionnaire, that there is no relationship between the presence of the preposition and evidentiality, can be confirmed.

5. Conclusions

In this paper we have analyzed in detail the construction with deísmo in Castilla-La Mancha. We proposed that de is a complementizer located in the head Fin in the left sentence periphery. The analysis of the data collected in the two surveys carried out showed that, firstly, there is great variability in the construction, since the speakers of the same town do not coincide neither in the verbs nor in the lexical classes in which they accept the particle de before the non-finite subordinate clause. Secondly, the data also showed that the complementizer de cannot be associated with an evidential meaning. Although the results of the two questionnaires indicate that the preposition de is not a marker of evidentiality, more data, preferably from real use, seem necessary to strengthen these conclusions.