Dizque in Andean Spanish and Beyond

Abstract

This paper examines the reportative evidential dizque in Andean Spanish as spoken in Ecuador and Peru. Taking, as a starting point, the synchronic and diachronic syntactic analyses of this type of markers in Romance that have been discussed in the literature, I propose an analysis that makes explicit how their syntax is mapped into semantics, and provide a semantics that captures the evidential and lack of certainty implications of dizque. I argue that expressions with dizque must be uttered when reportative evidence is available to the speaker, and, building on previous literature, that the lack of certainty flavor that these expressions have is a not-at-issue entailment. I show a number of consequences that follow from this kind of approach. I further point out how my proposal can capture the cross-Romance variation that is found in this domain, and discuss some key differences between dizque and other reportative evidentials cross-linguistically in connection to the expression of lack of certainty.

1. Introduction

Several Romance languages have developed a reportative evidential due to a grammaticalization process involving the verb say and a complementizer (Aikhenvald, 2004; Cruschina & Remberger, 2008; Alcázar, 2018). Examples of these markers are American Spanish dizque, Brazilian Portuguese diz que, Romanian ciča, Sicilian dicica, among others. Broadly speaking, when somebody utters a sentence with these markers, they convey the meaning that they got the information that they are reporting from somebody else.

For instance, (1), with dizque in American Spanish (henceforth, Spanish), is normally understood as conveying the meaning that María went to the cinema and that the speaker got this information through report—in fact, the reportative evidence available to the speaker may be second- or third-hand, as well as folklore, as the extensive literature on the topic indicates (see Kany, 1944; Magaña, 2005; Company, 2006; Travis, 2006; Olbertz, 2005, 2007; Babel, 2009; De la Mora & Maldonado, 2015; Demonte & Fernández-Soriano, 2013, 2014; Grajales, 2017; Alcázar, 2018; Saito, 2019, 2021; Sanromán Vilas, 2020; Martínez Vera, 2023; Martínez Vera & Camacho, 2025). That there is reportative evidence is captured here in the gloss by means of ‘they say’.

(1)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

In contrast to many of its Romance counterparts, dizque also involves a lack of certainty implication, which has been argued for by Travis (2006), Cruschina and Remberger (2008), De la Mora and Maldonado (2015), Alcázar (2018), Martínez Vera (2023), and Martínez Vera and Camacho (2025) (see also Diewald, 2011; Heine, 2013; Squartini, 2014; Heine et al., 2016; Sanromán Vilas, 2020). In fact, as argued for by De la Mora and Maldonado (2015), such a function is always present in constructions with dizque in Mexican Spanish; a similar claim has been made by Martínez Vera and Camacho (2025) for Colombian Spanish. Thus, cases such as (1) further convey that the speaker has reservations over truth of the expression, i.e., that the utterer lack certainty with regard to whether María went to the cinema.

The properties of these markers in Romance have been studied from different points of view. For instance, there has been a focus on their syntactic properties, both from a synchronic and a diachronic point of view (Cruschina & Remberger, 2008; Demonte & Fernández-Soriano, 2013; Cruschina, 2015; Saito, 2019, 2021; Martínez Vera & Camacho, 2025). There have also been different attempts to single out the semantic and/or pragmatic properties of these markers based on corpora and fieldwork (Travis, 2006; Demonte & Fernández-Soriano, 2013; De la Mora & Maldonado, 2015; Alcázar, 2018; Sanromán Vilas, 2020; Martínez Vera, 2023; Martínez Vera & Camacho, 2025). Finally, there has been interest in these markers from narrower and broader perspectives, i.e., there has been a focus on specific markers—for instance, on Spanish dizque (see Alcázar, 2018 for an overview) or Sicilian dicica (Cruschina, 2015)—, as well as an interest in these markers from a cross-Romance perspective (Cruschina & Remberger, 2008; Alcázar, 2018; Sanromán Vilas, 2020) and from a broader typological perspective (Aikhenvald, 2004, 2018).

One of the tasks that is yet to be done in more detail is to establish how the syntax of these markers is mapped into their semantics, and to provide an explicit analysis as to how the different semantic properties of these markers that have been discussed in the literature are to be captured (see Martínez Vera & Camacho, 2025 for an initial attempt to do this). The aim of this paper is to fill these gaps. Concentrating on dizque in a specific variety, namely, Andean Spanish as spoken in Ecuador and Peru—a variety that has been in contact with both Quechuan and Aymaran languages—, I propose an analysis that addresses how the mapping from syntax to semantics of matrix declarative sentences with this marker takes place, by focusing on the diz- ‘say’ and the C que components.1 I further provide a semantics for dizque that makes explicit that the reportative and lack of certainty implications are to be understood as not-at-issue entailments—I also capture why the speaker need not be committed to the at-issue proposition that is uttered. 2 I further point out how my proposal can capture the cross-Romance variation that is found in this domain. I also discuss how dizque is different from other reportative evidentials cross-linguistically regarding the possibility of indicating lack of certainty, in particular, the expression of lack of certainty in the presence of a reportative is not usually entailed (see, e.g., Anderbois, 2014). By doing so, I point out different options in connection to the effect of uttering an expression with a reportative evidential in discourse within a cross-linguistic perspective (see Faller, 2002, 2019; Pancheva & Rudin, 2019).

The paper is organized as follows: Section 2 discusses the syntax of dizque, including relevant synchronic and diachronic aspects. Section 3 introduces the proposed analysis of Andean Spanish dizque. Section 4 discusses a number of consequences that the analysis makes, including discussion regarding the kind of evidence that the speaker must have to utter a sentence with dizque, as well as explicit argumentation as to why the lack of certainty implication of this marker is to be understood as a not-at-issue entailment. Section 5 discusses some extensions of my proposal in connection to similar markers in Romance, as well as in connection to the broader cross-linguistic picture. Section 6 is the conclusion.

2. Syntax of Dizque

As has been widely documented in the literature (see, e.g., Aikhenvald, 2004, 2018), markers such as dizque (and its counterparts in other Romance languages, as well as more generally) are formed by some form of the verb say, decir in Spanish, and a complementizer C, que in Spanish. In particular, dizque is a reduced form of dice que ‘s/he says’ (Kany, 1944; Travis, 2006; among others); as Kany (1944) further points out, dizque is further reduced in some varieties (e.g., in Venezuela). Dizque is an indivisible element that does not convey grammatical information such as person, number, tense, or mood, as a regular conjugated form of the verb decir ‘say’ would (Travis, 2006). In this regard, dizque is a fixed form.

In this section, I focus on the syntax of dizque and its connections to its mapping to semantics, specifically, I adopt previous proposals with regard to the nature of the elements that are present in dizque. Emphasizing a diachronic perspective, Saito (2019, 2021) argues that dizque is formed by a say element and a C element; say and C undergo fusion (see Halle & Marantz, 1993). The say component preserves its key semantic properties, in particular, (i) it conveys a report (a proposition here) and (ii) the source of the report is preserved somehow (e.g., in the semantics).3 As for (i), Saito argues that say takes a sentential complement. Of relevance here is that such a complement captures that there is a proposition that is said. As for (ii), Saito indicates that, in contrast to the full verb, dizque lacks a v layer, which plays a key role in grammaticalization processes involving speech verbs (Klamer, 2000; I. Roberts & Roussou, 2003). Under the assumption that v introduces an overt external argument (Harley, 1995; Marantz, 1997), Saito argues that the absence of v means that there is no overt external argument—for Saito, the absence of v also means that the say component is decategorized, i.e., it is no longer verbal. Of relevance here is that, while there may be a source for some report, such a source cannot be syntactically instantiated (e.g., there is no possible syntactic counterpart of a by-phrase, as in the passive voice, where a passive v would be present; see Saito, 2019, 2021 for further discussion). Importantly, this does not preclude it from being semantically instantiated, as Saito argues. Section 3 provides an analysis of the properties in (i)–(ii) from a semantic perspective. Section 4.2 discusses the different possible report sources in more detail.

While Saito’s proposal suggests that dizque is located high in the structure, which is based on the fact that such an element incorporates a C element, a question arises as to where exactly it would be located. Adopting a cartographic approach, whereby the CP is split into several projections (Cinque, 1999; Rizzi, 2001), Cruschina and Remberger (2008) (see also Demonte & Fernández-Soriano, 2013; Martínez Vera & Camacho, 2025) argue that markers such as dizque are adverbial in nature (in particular, high adverbs) and sit in an evidential projection. This is the position that I adopt in this paper. With regard to the adverbial nature of dizque, they indicate that this kind of element cannot be modified by, e.g., negation or adverbs, which contrasts with what happens with the verb decir ‘say’. Thus, while sequences with the verb, e.g., no dice que… ‘(s/he) doesn’t say that…’ and siempre dice que ‘(s/he) always say that…’ are possible, sequences with dizque, e.g., *no dizque… ‘not dizque…’ and *siempre dizque ‘always dizque…’ are ungrammatical (see Travis, 2006; Cruschina & Remberger, 2008 for additional arguments).4 Under the view that this kind of elements (in particular, dizque) are high adverbs, Cruschina and Remberger (2008) (see also Martínez Vera, 2023) indicate that there is a default order whereby these elements appear in a sentence initial position, as in (2) (=(1)).

(2)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

Nonetheless, as high adverbs, it should also be possible for these elements to appear, for instance, after a topicalized element, such as the subject, as exemplified in (3). The sentential complement, however, is the same in (2)–(3) (see Cruschina & Remberger, 2008; Martínez Vera, 2023 for additional discussion with regard to the use of these elements as parentheticals).

(3)	María	dizque	fue	al	cine.
	María	dizque	went	to.the	cinema
	‘María went to the cinema, they say.’

With regard to the evidential projection where dizque sits, Cruschina and Remberger (2008) adopt Cinque’s (1999) (rigid) hierarchy of adverbs, whereby evaluative adverbs precede evidential adverbs, which in turn precede epistemic adverbs. What this means is that an evaluative adverb, e.g., desafortunadamente ‘unfortunately’, may precede dizque, but not the other way around (4); dizque should not be able to co-occur with other evidential adverbs, e.g., aparentemente ‘apparently’ (5); and dizque may precede an epistemic adverb, e.g., posiblemente ‘possibly’, but not the other way around (6). As shown below, these predictions are borne out (see also Martínez Vera, 2023).

(4)	a.	Desafortunadamente	dizque	María	fue	al	cine.
		unfortunately	dizque	María	went	to.the	cinema
	b.	?*Dizque	desafortunadamente	María	fue	al	cine.
		dizque	unfortunately	María	went	to.the	cinema
		‘Unfortunately, María went to the cinema, they say.’

(5)	?*Dizque	aparentemente	/Aparentemente	dizque	María	fue	al	cine.
	dizque	apparently	apparently	dizque	María	went	to.the	cinema
	‘Apparently, María went to the cinema, they say.’

(6)	a.	Dizque	posiblemente	María	fue	al	cine.
		dizque	possibly	María	went	to.the	cinema
	b.	?*Posiblemente	dizque	María	fue	al	cine.
		possibly	dizque	María	went	to.the	cinema
		‘María possibly went to the cinema, they say.’

Based on the discussion, I assume that sentences with dizque have the structure in (7), where dizque heads an evidential projection with a sentential complement α.5

(7)	[EvidP dizque α]

3. Semantics of Dizque

This section proposes a semantics for the structure in (7). Below I will propose a lexical entry for dizque that takes a proposition as argument—in what follows, I assume that the sentential complement α in (7) denotes a proposition. The semantics proposed below addresses the following issues: (i) what is the status of the proffered content, i.e., α in (7)?, (ii) what is the status of the say-like component in dizque?, and (iii) what is the status of the lack of certainty implication that the sentences with dizque have? After making my theoretical proposal explicit, in the next section I turn to the discussion of how such an account accommodates a number of cases with dizque focusing on each component, namely, the proffered content, the evidential meaning (i.e., the say-like component) and the lack of certainty implication.

My proposal appears in (8). I adopt an approach where the proffered content is at-issue, and the evidential contribution and the lack of certainty implication are not-at-issue (see, among others, Potts, 2005; Murray, 2010, 2014, 2017; Korotkova, 2016, 2017; Koev, 2018; Faller, 2019; Pancheva & Rudin, 2019; Martínez Vera, 2024). In particular, I propose that both the reportative meaning and the lack of certainty implication are not-at-issue entailments. While both the evidential and the lack of certainty implications are not-at-issue, I state the evidential and doubt meanings in two different lines.6 The interpretation function incorporates a context parameter c, whose key components in what follows include the speaker sp and addressee add, i.e., ⟦⋅⟧^c.

Specifically, ⟦dizque⟧^c applied to proposition p denotes the (very same) proposition p at the at-issue level (see Pancheva & Rudin, 2019); the evidential (not-at-issue) meaning is the proposition that speaker sp has reportative evidence for p (see Murray, 2017); and the lack of certainty (not-at-issue) meaning is that the proposition that sp believes that not p is at least as likely than p (i.e., sp is not certain that p is the case). Below I provide further details with regard to each meaning component—for current purposes, it suffices that the w variables in (8) are interpreted as actual world w_@ (e.g., the belief worlds of sp would be determined in connection to what is accessible from w_@).

(8)	For any proposition p ∈ Dst, ⟦dizque⟧c(p) =
	at-issue:	λw[p(w)]
	evidential:	λw[Rep(sp, p, w)]
	lack of certainty:	λw[¬p ≥ Dox-sp(w) p]

The at-issue meaning basically includes the proposition under consideration. While the approach I adopt here is static, I would like to indicate some connections to what it would mean to utter an expression with dizque in an exchange (see Section 5.2 for additional discussion). The approach that I adopt here is inspired by Stalnaker (1978, 2002), where the Common Ground is formed by common beliefs, i.e., beliefs that are shared among the participants in a conversation.7 While I do not sketch an analysis that goes into detail with regard to exchanges,8 what is relevant here is to make explicit that, by uttering an expression of the form dizque α, where α denotes proposition p, speaker sp presents p to addressee add, i.e., speaker sp makes p salient towards addressee add (see Faller, 2002, 2019; Martínez Vera, 2024; Martínez Vera & Camacho, 2025). In this sense, I adopt the principle in (9), which suffices for current purposes—(9) merely attempts to make somewhat explicit the illocutionary force (and the purpose) of uttering an expression with dizque in an exchange (see more discussion in this regard below).

(9)	Uttering an expression with dizque
	By uttering an expression of the form dizque α, where α denotes proposition p,
	speaker sp presents p to the addressee add, i.e., sp makes p salient towards add.

It is worth emphasizing that, by uttering an expression with dizque, speaker sp is not asserting the relevant proposition. Importantly, under the view of the Common Ground indicated above, assertions are to be understood in connection to what sp believes and would like somebody else to believe, i.e., what sp puts out there in an exchange is what she believes such that sp intends that her belief becomes shared belief among participants. In contrast, when sp presents a proposition, she does not indicate anything with regard to her belief of that proposition; she simply puts it out there in the open (because, e.g., it is relevant for the topic under consideration in an exchange).

Turning to the reportative meaning, (8) captures such a contribution as not-at-issue, in particular, the speaker has reportative evidence for the at-issue proposition. The presence of reportative evidence implies that there is a source for the report, but such a source remains unspecified;9 this is how the properties indicated by Saito (2019, 2021) discussed in Section 2 are captured. In connection to how I implement the presence of reportative evidence, I assume that the possibility of including this kind of meaning when uttering an expression with dizque means that the speaker had access to the relevant report even if only indirectly; otherwise, she would have not been able to indicate that there is reportative evidence. Such a proposal, where the source of the report remains unspecified, builds on the literature (see, e.g., Travis, 2006; Cruschina & Remberger, 2008; Martínez Vera, 2023; Martínez Vera & Camacho, 2025), which indicates that dizque is compatible with second- and third-hand evidence, as well as folklore (see Section 4.2 for examples in this regard). In connection to this, that the source remains unspecified and that uttering a sentence with dizque makes no reference to assertive force further mean that such a source may or may not believe that the proposition under consideration is true or not, which is, in principle, compatible with a speaker reporting a proposition, where the source of the report has reportative evidence for that proposition as well; this is precisely what the literature on dizque suggests.10

Recall at this point that the literature also points out that expressions with dizque convey lack of certainty. Under my approach, this meaning is captured as not-at-issue in (8) as well, where it is made explicit that the speaker has not settled her beliefs with regard to some proposition, i.e., her beliefs are divided in this regard in that only a preference is indicated. In particular, the speaker remains uncertain over proposition p being the case by indicating that she believes that not p is at least as likely as p. Building on how the notion of doubt is captured in the literature (Bolinger, 1968; Stalnaker, 1984; Farkas, 1985; Heim, 1992; Villalta, 2000, 2008; Anand & Hacquard, 2013; see Villalta, 2008 in particular for an approach focusing on Spanish where the alternatives considered are not restricted to some proposition and its negated counterpart), two key notions are relevant here in connection to lack of certainty. First, as is standard in the formal semantics literature (see Hintikka, 1962; Yalcin, 2007; Hacquard, 2010; among many others), I assume an accessibility relation R, where R ⊆ D_s × D_s (s is the type of worlds) and function f is related to R. Specifically, f is that function from D_s to ℘(D_s) such that, for any world w, f(w) = {w′ | wRw′}. The relevant function here is doxastic, which applies to a world relative to some individual, and give as output the set of worlds that conforms to what that individual believes in that world. This is defined in (10).

(10)	Doxastic alternatives
	For any world w ∈ Ds, individual x ∈ De, Dox-x(w) = {w′ | w′ conforms to what x believes in w}.

The notion of doubt builds on belief, in particular, the presence of a preference in connection to doubt is made explicit by means of the at least as-likely-as relation in (11), which is closely tied to the presence of doxastic alternatives. This is how the lack of certainty not-at-issue entailment in (8) is to be understood here.

(11)	At least as-likely-as relation
	a.	For any worlds w, w′, w″ ∈ Ds, individual x ∈ De, w′ ≥ Dox-x(w) w″ iff w′, w″ ∈ Dox-x(w) and w′ is closer to w than w″.
	b.	For any world w ∈ Ds, individual x ∈ De, proposition p ⊆ Ds, ¬p ≥ Dox-x(w) p iff for any worlds w′, w″ ∈ Dox-x(w) such that w′ ∈ p and w″ ∈ ¬p, w″ ≥ Dox-x(w) w′.

For illustration, consider (12)a (=(1)). Its syntactic structure appears in (12)b and its denotation appears in (12)c. (12)b conveys the at-issue proposition that María went to the cinema. Due to (9), speaker sp makes salient the proposition that María went to the cinema to addressee add. (12)b conveys the not-at-issue evidential meaning that sp has reportative evidence for the proposition that María went to the cinema. (12)c conveys the not-at-issue doubt meaning that sp casts doubt over the proposition that María went to the cinema (i.e., sp believes that it is more likely that María did not go to the cinema)—recall that, for simplicity, the world of evaluation could be thought of as actual world @ (e.g., the belief worlds of sp would be determined in connection to what is accessible from @).11

(12)	a.	Dizque		María		fue		al	cine.
		dizque		María		went		to.	the cinema
		‘María went to the cinema, they say.’
	b.	[EvidP dizque [α María went to the cinema]]
	c.	⟦(12)b⟧c =
			at-issue:		λw[María went to the cinema in w]
			evidential:		λw[Rep(sp, λw′[María went to the cinema in w′], w)]
			lack of certainty:		λw[¬λw′[María went to the cinema in w′] ≥ Dox-sp(w)
							λw′[María went to the cinema in w′]]

4. Applying the Analysis

Having indicated how my proposal captures the meaning of sentences with dizque at the outset, I make explicit a number of applications and consequences in connection to the different meanings involved in expressions with such a marker. Section 4.1 focuses on the at-issue meaning. Section 4.2 concentrates on the evidential meaning. Section 4.3 focuses on the lack of certainty implication. In what follows, I show that these applications and consequences naturally follow from the analysis in Section 3. In addition, in Section 4.4, I outline a comparison between dizque and the Spanish verb dudar ‘doubt’, which tries to establish a direct comparison with an element that is closest to lack of certainty, and, importantly, has been extensively studied in Spanish.

4.1. At-Issue

With regard to the at-issue meaning in (8), in combination with (9), recall that my proposal states that the speaker makes the at-issue proposition under consideration salient towards the addressee. My proposal makes (at least) three predictions in this regard, as made explicit in (13).12

(13)	At-issue meaning in combination with (9)
	a.	Since the speaker only presents the at-issue proposition, she may indicate that she does not believe that proposition.
	b.	Since the speaker only presents the at-issue proposition (so she does not indicate whether she believes such a proposition), it is infelicitous (for, e.g., the addressee) to challenge that the speaker believes that proposition.
	c.	Since the at-issue proposition is made salient towards the addressee, she may challenge it, specifically, she may indicate that she disagrees with it.

These issues can be examined via a battery of tests, in particular, via the commitment test and the direct challengeability test (see Murray, 2017 for a survey of the tests).13 (13)a targets the commitment of the speaker with regard to a proposition based on whether she believes such a proposition or not, specifically, the speaker need not be committed to (believing) that proposition. In particular, when the speaker utters a sentence of the form dizque α, where α denotes proposition p, her commitment to p can be overtly challenged by means of a follow-up that indicates that she does not believe p, such as pero yo no lo creo ‘but I don’t believe it’ (Martínez Vera, 2023). If (13)a is the case, i.e., if the speaker does not (necessarily) believe such a proposition (because she presents but does not assert it), then such a follow-up should be felicitous. This is borne out, as shown in (14), where the speaker’s commitment to (1) is challenged.

(14)	Dizque	María	fue	al	cine,	pero	yo	no	lo	creo.
	dizque	María	went	to.the	cinema	but	I	not	cl	believe
	‘María went to the cinema, they say, but I don’t believe it.’

Turning to (13)b,c, the test that I use here is the direct challengeability test. This test involves a follow-up to an utterance of the form dizque α, where α denotes proposition p. The follow-up is made by an individual different from the speaker, who questions what is indicated, and then proposes a correction indicating that not p (Faller, 2002, 2019; Matthewson et al., 2007; Murray, 2010, 2014, 2017; Korotkova, 2016, 2017). Since the setup in Section 3 is stated in terms of beliefs, a direct challenge is to be understood in terms of an individual’s beliefs (towards some proposition). (15) (=(1)) will be directly challenged, i.e., the proposition that María went to the cinema will be challenged.

(15)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

Two follow-ups have been proposed prior to stating the correction. With regard to the first one, Korotkova (2016, 2017) proposes a general follow-up of the form you’re mistaken (here I assume that the speaker is female, so the follow-up in Spanish would be estás equivocada), which indicates that the speaker is wrong in some regard, in particular, in connection to the belief of p here. Importantly, the prediction in (13)b states that it should be infelicitous to question the speaker’s belief of p. As (16) shows, following up (15) with estás equivocada ‘you’re wrong’ is infelicitous, which is as expected: if the speaker does not believe that p is true, then it does not make sense to state that she is wrong with regard to their belief that p.

(16)	#Estás	equivocada.	María	no	fue	al	cine.
	you’re	wrong	María	not	went	to.the	cinema
	You’re wrong. María didn’t go to the cinema.’

The second follow-up involves a more direct challenge of p by, e.g., the addressee by means of the follow-up that’s not true (in Spanish, eso no es cierto) (see, e.g., Faller, 2002; Murray, 2017). (13)c states that it should be possible to disagree with p, since p is made salient towards the addressee. As shown in (17), this is possible, which is as expected, as (13)c indicates.

(17)	Eso	no	es	cierto	(para	mí).	María	no	fue	al	cine.
	that	not	is	true	for	me	María	not	went	to.the	cinema
	‘That’s not true (for me). María didn’t go to the cinema.’

4.2. Evidential Meaning

With regard to the evidential meaning, my proposal in (8) states that there is reportative evidence, to which the speaker had access. Such a meaning is not-at-issue. My proposal makes (at least) five predictions in this regard, as made explicit in (18).

(18)	Evidential meaning
	a.	The evidence must be reportative, i.e., it should be felicitous to utter a sentence with dizque in the presence of this kind of evidence, and it should be infelicitous to utter a sentence with dizque in the presence of other kinds of evidence (e.g., direct or inferential evidence).
	b.	Since there is no specification as to what kind of reportative evidence there must be, it is possible that such an evidence be (i) second- or (ii) third-hand, as well as (iii) folklore.
	c.	Since the speaker had access to reportative evidence, she cannot indicate that she does not have this kind of evidence.
	d.	Since the reportative evidence is not-at-issue, it is infelicitous (for, e.g., the addressee) to directly challenge it.
	e.	Since the reportative evidence is not-at-issue, it is felicitous (for, e.g., the addressee) to indirectly challenge it.

The issue in (18)a indicates that the evidence must be reportative. This means that it should be possible to utter (19) (=1) against the context indicated below, where the speaker heard about María going to the cinema (see Martínez Vera, 2023 for additional examples).

Context targeting reportative evidence: The speaker heard (from Rosa) that María would be going to the cinema. The speaker tells (19) to somebody else.

(19)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

However, uttering a sentence with dizque should not be possible in the presence of direct or inferential evidence. Focusing on direct evidence first, this means that uttering, e.g., (19) against a context where the speaker saw María going into the cinema is infelicitous—this is in fact the case, as expected. However, it is worth pointing out that such a straightforward scenario may not be conclusive in this regard for two reasons. First, in my proposal, the speaker only presents the proposition under consideration (the speaker does not assert it, by means of which her belief of such a proposition would be overtly indicated). Second, sentences with dizque involve lack of certainty, i.e., it is conceivable that uttering (19) against this simple context is infelicitous not because there is direct evidence, but, instead, due to the fact that it is unlikely that the speaker will want to convey lack of certainty if she has such a strong evidence. For these reasons, below I present another context, which is more nuanced: the speaker has direct evidence, but there is room for lack of certainty on the speaker’s side. Importantly, uttering (20) against the context below is still infelicitous. This is precisely what my account predicts in (18)a.

Context targeting direct evidence: The speaker normally wears glasses. When she does not wear them, her sight is very blurry. She is not wearing them. She sees something on the white floor in the kitchen; it is reddish and looks like a stain (it is very likely that it is a stain), but she cannot tell for sure. The speaker’s roommate goes into the kitchen. The speaker tells (20) to this roommate.

(20)	#Dizque	hay	una	mancha	en	el	piso.
	dizque	there.is	a	stain	on	the	floor
	‘There is a stain on the floor, they say.’

The context below and the example in (21) target inferential evidence. The same considerations are taken into account in connection to the context that follows, specifically, there is room for lack of certainty. Importantly, as my account predicts, uttering (21) against this context involving inferential evidence is infelicitous—it is also infelicitous to utter (21) against a simpler context where María only goes to the gym by bike (i.e., if María’s bike is not at home, it is safe to infer that she went to the gym); the sentence in such a case would be Dizque María fue al gimnasio ‘María went to the gym, they say’.

Context targeting inferential evidence: The speaker lives with two roommates, María and Susana. María has a bike; when her bike is not at home, it means that she went to the gym or to the park. The speaker arrives home and María is not there; her bike is not there either. Susana arrives a little later than the speaker. The speaker tells (21) to Susana.

(21)	#Dizque	María	fue	al	gimnasio	/ al	parque.
	dizque	María	went	to.the	gym	to.the	park
	‘María went to the gym/park, they say.’

Turning to (18)b, my proposal can readily accommodate instances of second- and third-hand evidence, as well as folklore (Travis, 2006; Cruschina & Remberger, 2008). Second-hand evidence is normally understood in terms of the presence of a report such that the source of the report has direct evidence for the relevant proposition. Third-hand evidence involves the presence of a report, but the source of that report is not a direct witness. Folklore refers to established stories in a community, which are shared among speakers (see Willett, 1988 for an overview as to how to characterize different kinds of evidence). This is the case because the source remains underspecified; the only requirement is that there is reportative evidence, which means that there is a source. For this reason, uttering (22) (=(1)) and (23) against the proposed contexts is felicitous, which follows from my account.

Context targeting second-hand evidence: Rosa told the speaker that María went to the cinema. Rosa claims that she saw María going into the cinema. Nonetheless, the speaker has doubts about Rosa’s report, because Rosa tends to lie. The speaker tells (22) to Susana.

Context targeting third-hand evidence: Rosa told Julia that María went to the cinema. Julia told this to the speaker. The speaker tells (22) to Susana.

(22)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

Context targeting folklore: The speaker heard the story about the origin of the Tahuantinsuyo at school in her history classes. Talking to a friend of hers that has never heard about this traditional story, the speaker utters (23).

(23)	Dizque	Manco	Cápac	emergió	del	lago	Titicaca.
	dizque	Manco	Cápac	emerged	of.the	lake	Titicaca
	‘Manco Cápac emerged from the Titicaca Lake, they say.’

The prediction in (18)c indicates that the speaker had access to reportative evidence, so that she cannot indicate that she does not have it. This can be tested by means of the so-called commitment test applied to the evidential proposition (Murray, 2017). This test involves a follow-up where the speaker indicates that she does not have the relevant evidence. In this case, it would be a follow-up of the form but nobody told me that (in Spanish, pero nadie me dijo eso). Applying this test to (1) in a context where Rosa told the speaker that María went to the cinema is infelicitous, which is as indicated in (18)c. This is shown in (24).

(24)	#Dizque	María	fue	al	cine,	pero	nadie	me	dijo	eso.
	dizque	María	went	to.the	cinema	but	nobody	cl	said	that
	‘María went to the cinema, they say, but nobody told me that.’

The prediction in (18)d indicates that, since the reportative evidence is not-at-issue, it is infelicitous (for, e.g., the addressee) to directly challenge it. The direct challengeability test used here is the one introduced in Section 4.1. The speaker utters a sentence and somebody else (e.g., the addressee) challenges what is said. Here the correction proposed indicates that the speaker does not have the relevant evidence, e.g., that nobody told the speaker what she indicates. As expected under my analysis, such a challenge is infelicitous, because the evidential contribution is not-at-issue. This is shown in (25), where (25)b challenges (25)a (=(1))—the relevant context here is one in which Rosa told the speaker that María went to the cinema.

(25)	a.	Dizque	María	fue	al	cine.
		dizque	María	went	to.the	cinema
		‘María went to the cinema, they say.’
	b.	#Eso	no	es	cierto. /	Estás	equivocada.	Nadie	te	dijo	eso.
		that	not	is	true	you’re	wrong	nobody	cl	said	that
		‘That’s not true/You’re wrong. Nobody told you that.’

The last prediction discussed here is (18)e, which states that, since the reportative evidence is not-at-issue, it is felicitous (for, e.g., the addressee) to indirectly challenge it. As the literature points out (see, e.g., Faller, 2002; Murray, 2017), while it is infelicitous to directly challenge the evidential contribution, as shown in (25), it is possible to indirectly challenge it. An indirect challenge involves uttering a follow-up that asks the speaker to be (more) explicit about the evidence that she has. In this sense, this challenge does not state that the speaker does not have the relevant evidence (which is what a direct challenge does); instead, it asks for clarification. This is the case when, e.g., making a question asking for clarification with regard to the evidence that the speaker has, such as and who told you that? (in Spanish, ¿y quién te dijo eso?) when challenging a sentence with a reportative evidential. The idea behind such a challenge is that the information is present (even though it is not-at-issue) and, as such, is recoverable. As expected under my account, such a challenge is possible, as shown in (26)—the relevant context here is one in which Rosa told the speaker that María went to the cinema.

(26)	a.	Dizque	María	fue	al	cine.
		dizque	María	went	to.the	cinema
		‘María went to the cinema, they say.’
	b.	¿Y	quién	te	dijo	eso?
		and	who	cl	said	that
		‘And who told you that?’

4.3. Lack of Certainty Implication

As indicated in (8), the lack of certainty implication is a not-at-issue entailment. In this sense, the predictions in connection to the lack of certainty implication are, to some extent, similar to those discussed in Section 4.2 with regard to the evidential meaning. In particular, in this section I focus on the predictions in (27).

(27)	Lack of certainty implication
	a.	Since the speaker lacks certainty with regard to the proposition in the denotation of an expression with dizque, she cannot indicate that she has no doubts about it.
	b.	Since the speaker lacks certainty with regard to the proposition in the denotation of the expression with dizque, she cannot utter such an expression if she has no doubts about that proposition (i.e., she believes that p is true).
	c.	Since the the lack of certainty implication is not-at-issue, it is infelicitous (for, e.g., the addressee) to directly challenge it.
	d.	Since the lack of certainty implication is not-at-issue, it is felicitous (for, e.g., the addressee) to indirectly challenge it.

The predictions in (27)a,b target whether there must always be lack of certainty in the presence of an expression with dizque, i.e., they target whether the doubt meaning is an entailment. The prediction in (27)a indicates that the speaker expresses doubt about the proposition in the denotation of an expression with dizque, so that she cannot indicate that she has no doubts about it. This can be tested by means of the so-called commitment test applied to not-at-issue meanings (Murray, 2017). Here, this test involves a follow-up where the speaker indicates that she has no doubts, i.e., pero yo no tengo dudas al respecto ‘but I don’t have (any) doubts in this regard’ (Martínez Vera, 2023). Applying this test to (1) is infelicitous, which is as indicated in (27)a. This is shown in (28).

(28)

#Dizque

María

fue

al

cine,

pero

yo

no

tengo

dudas

al

respecto.

dizque

María

went

to.the

cinema

but

I

not

have

doubts

to.the

respect

‘María went to the cinema, they say, but I don’t have (any) doubts in this regard.’

In a similar vein, (27)b targets whether the lack of certainty component is entailed. Specifically, if the lack of certainty implication is entailed when uttering an expression with dizque, then if the speaker has no doubts about some proposition (i.e., she believes that p is true), she should not be able to utter such an expression. This is borne out, as (29) (=(1)) shows. In this case, the speaker has no reason to express lack of certainty towards the proposition under consideration, since the source of the information is reliable. In such a context, it is infelicitous to utter an expression with dizque.14

Context: The speaker talks on the phone with Rosa. Rosa is a reliable source, and the speaker trusts her completely. Rosa tells the speaker that María went to the cinema. The speaker has no reason to doubt Rosa’s report. She hangs up the phone and tells (29) to her roommate.

(29)	#Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

Turning to (27)c, this prediction indicates that, since the lack of certainty implication is not-at-issue, it is infelicitous (for, e.g., the addressee) to directly challenge it. The direct challengeability test used here is the one introduced in Section 4.1. The speaker utters a sentence and somebody else (e.g., the addressee) challenges what is said. Here the correction proposed indicates that the speaker does not have any doubts about the relevant proposition. As expected in my analysis, such a challenge is infelicitous, because the lack of certainty implication is not-at-issue. This is shown in (30); (30)b challenges (30)a (=(1)).

(30)

a.

Dizque

María

fue

al

cine.

dizque

María

went

to.the

cinema

‘María went to the cinema, they say.’

b.

#Eso

no

es

cierto. /

Estás

equivocada.

Tú

no

tienes

dudas

al

respecto.

that

not

is

true

you’re

wrong

you

not

have

doubts

to.the

respect

‘That’s not true/You’re wrong. You don’t have (any) doubts in this regard.’

The last issue is (27)d, which states that, since the lack of certainty entailment is not-at-issue, it is felicitous (for, e.g., the addressee) to indirectly challenge it. Building on the discussion on Section 4.2, I make use of an indirect challenge of the form, ¿y qué dudas tienes? ‘and what are your doubts?’, which asks for clarification regarding the speaker’s doubts. As indicated in the previous section, the idea behind such a challenge is that the information is present (even though it is not-at-issue) and, as such, is recoverable. As expected under my account, such a challenge is possible, as shown in (31).

(31)	a.	Dizque	María	fue	al	cine.
		dizque	María	went	to.the	cinema
		‘María went to the cinema, they say.’
	b.	¿Y	qué	dudas	tienes?
		and	what	doubts	have
		‘And what are your doubts?’

4.4. Dizque vs. Dudar ‘Doubt’

As discussed throughout this paper, dizque has a lack of certainty meaning, which has been explicitly captured in (8) as a not-at-issue entailment. In fact, there are some authors, for instance, De la Mora and Maldonado (2015) (see also Travis, 2006; Cruschina & Remberger, 2008; Alcázar, 2018; Sanromán Vilas, 2020), who, focusing on Mexican Spanish, crucially claim that the epistemic meaning of doubt is actually the distinctive feature of dizque (see also Martínez Vera & Camacho, 2025 for a somewhat different take on this issue).15 While dizque may well include a lack of certainty meaning, a question arises as to what exactly the nature of this meaning is. As argued for in this paper, the lack of certainty implication should be treated as not-at-issue. Building on this, I compare dizque and the verb dudar ‘doubt’ in some respects. Such a comparison is motivated by the fact that, given that dizque and dudar ‘doubt’ would be similar, at least in the intuitive sense that they both indicate lack of certainty, a question arises as to how they should be distinguished (if at all). In what follows, I make explicit what some of the similarities and differences between these expressions are.

In this regard, it is worth pointing out what the common core of these lexical items would be in the current approach. As discussed in Section 3, the meaning of lack of certainty is captured here in terms of a likelihood relation where some proposition is at least as likely as some other proposition—as mentioned there, this meaning is actually based on proposals for the characterization of the lexical item dudar ‘doubt’, as in Villalta (2008). Specifically, in this paper, lack of certainty is captured in terms of a likelihood relation whereby a speaker indicates that, relative to her belief worlds, the negation of proposition p in the denotation of the expression that is uttered is at least as likely as p (see (11) for the formal details). Importantly, both dizque and dudar ‘doubt’ would incorporate a meaning along these lines.

In what follows, I argue that dizque and dudar ‘doubt’ can actually be told apart. Here I focus on two key elements to argue for such a distinction, namely, (i) where the doubt/lack of certainty meaning is incorporated (i.e., whether it is at-issue or not-at-issue), and (ii) what is the relation of the speaker and proposition p in utterances of the form dizque α and dudar α (where α denotes p). It is worth pointing out that I have made a methodological choice to keep the parallels between both kinds of expressions as strict as possible. Since cases with dizque involve an expression of lack of certainty on behalf of the speaker, I am using cases with dudar ‘doubt’ where the subject of this predicate is in the first person singular for the most part, so that the doubt that is considered is tied to the speaker (at the end of this section, I discuss third person subjects with dudar ‘doubt’ to some extent).

With regard to (i), i.e., whether the doubt/lack of certainty meaning is at-issue or not-at-issue, I discuss two issues here. The first one regards the issue of direct challengeability; the second one is tied to the presence of the indicative or subjunctive mood in the complement of dizque and dudar ‘doubt’. With regard to the direct challengeability test, recall that the lack of certainty meaning in expressions with dizque cannot be challenged (see Section 4.3). The example in (32) (=(30)) illustrates this. As indicated there, this suggests that the lack of certainty meaning of dizque is not-at-issue.

(32)

a.

#Dizque

María

fue

al

cine.

dizque

María

went

to.the

cinema

‘María went to the cinema, they say.’

b.

#Eso

no

es

cierto. /

Estás

equivocada.

Tú

no

tienes

dudas

al

respecto.

that

not

is

true

you’re

wrong

you

not

have

doubts

to.the

respect

‘That’s not true/You’re wrong. You don’t have (any) doubts in this regard.’

In contrast, the direct challengeability test applied to expressions with dudar ‘doubt’ suggests that the doubt meaning is at-issue (this is actually the standard approach to this kind of predicate; see Anand & Hacquard, 2013). This is shown in (33): the speaker expresses that she doubts that María went to the cinema (33)a, and, importantly, challenging her doubt in this regard is possible (33)b. As expected, this indicates that the doubt meaning of dudar is at-issue.

(33)

a.

Dudo

que

María

fuera

al

cine.

doubt

that

María

went

to.the

cinema

‘I doubt that María went to the cinema.’

b.

Eso

no

es

cierto. /

Estás

equivocada.

Tú

no

tienes

dudas

al

respecto.

that

not

is

true

you’re

wrong

you

not

have

doubts

to.the

respect

‘That’s not true/You’re wrong. You don’t have (any) doubts in this regard.’

The second issue, as mentioned, is tied to the presence of the indicative or subjunctive mood in the complement of dizque and dudar ‘doubt’. As Villalta (2008) shows (see also Anand & Hacquard, 2013), attitude predicates selecting the subjunctive mood in Spanish involve a semantics where a comparison between a proposition and some alternative(s) is established. This is illustrated in (34) with querer ‘want’. Expressions with this predicate involve the comparison of alternatives with regard to the desires of the speaker.

(34)	Quiero	que	María	*va /	vaya	al	cine.
	want	that	María	go.ind	go.subj	to.the	cinema
	‘I want María to go to the cinema.’

In contrast, the indicative appears in plain declaratives (i.e., declarative sentences without, e.g., an attitude verb or a verb of saying), as well as in the embedded clause in sentences with epistemic predicates, predicates of communication, among others (these predicates select the indicative). Importantly, in all these cases, the semantics involved does not include a notion of preference.16 In (35), I illustrate this with a plain declarative (35)a, as well as with a sentence with creer ‘believe’ (35)b; in all these cases, the relevant verb appears in the indicative.

(35)	a.	María	fue /	*fuera	al	cine.
		María	went.ind /	*went.subj	to.the	cinema
		‘María went to the cinema.’
	b.	Creo	que	María	fue /	*fuera	al	cine.
		believe	that	María	went.ind /	*went.subj	to.the	cinema
		‘I believe that María went to the cinema.’

In this regard, there is an interesting contrast between expressions with dizque and dudar ‘doubt’, as shown in (36)–(37). While both elements incorporate the doubt/lack of certainty meaning made explicit by means of the comparison of alternatives, there is a contrast with regard to the selection of the subjunctive mood in the complement. In particular, while dudar ‘doubt’ behaves as Villalta (2008) indicates, i.e., the subjunctive is selected (in the embedded clause), as in (37), this is not the case when dizque is used, where the indicative is present, as in (36).

(36)	Dizque	María	fue /	*fuera	al	cine.
	dizque	María	went.ind	went.subj	to.the	cinema
	‘María went to the cinema, they say.’

(37)	Dudo	que	María	*fue /	fuera	al	cine.
	doubt	that	María	went.ind	went.subj	to.the	cinema
	‘I doubt that María went to the cinema.’

Building on the comparison between dizque and dudar ‘doubt’ in (36)–(37), I suggest here that the selection of the indicative or the subjunctive mood can only be sensitive to the semantics at the at-issue level, specifically, the key in the selection of the subjunctive is the presence of a semantics of preference, which has to be at-issue. This tells apart dizque and dudar ‘doubt’ in this regard, since the semantics of preference is at-issue only in the case of dudar ‘doubt’.

With regard to (ii), i.e., what is the relation of the speaker and proposition p in utterances of the form dizque α and dudar α (where α denotes proposition p), the literature on dizque points out that the speaker may know17 that the proposition under consideration is false (see De la Mora & Maldonado, 2015; Alcázar, 2018; Sanromán Vilas, 2020; Martínez Vera, 2023; Martínez Vera & Camacho, 2025). Thus, assuming that, in an exchange, it is relevant to utter (38) (=(1)) (for instance, this would be the case when it is pertinent to indicate that a third party thinks that María went to the cinema), it is possible that the speaker utters such an expression when she knows that María didn’t go to the cinema. Thus, uttering (38) against the context below is possible.

Context where the speaker knows that the relevant proposition is false: The speaker’s sister was supposed to go to the cinema with María, but María canceled because she was sick, so she didn’t go to the cinema; the speaker was told about this. In parallel, a friend of the speaker told her that María went to the cinema. The speaker tells (38) to somebody else.

(38)	Dizque	María	fue	al	cine.
	dizque	María	went	to.the	cinema
	‘María went to the cinema, they say.’

In contrast, it is infelicitous to utter dizque’s counterpart with dudar ‘doubt’ when the speaker knows that the relevant proposition is false. Thus, it is infelicitous to utter (39), with dudar ‘doubt’, against the context below (which basically repeat the context used for (38)).

Context where the speaker knows that the relevant proposition is false: The speaker’s sister was supposed to go to the cinema with María, but María canceled because she was sick, so she didn’t go to the cinema; the speaker was told about this. The speaker tells (39) to somebody else.

(39)	#Dudo	que	María	fuera	al	cine.
	doubt	that	María	went	to.the	cinema
	‘I doubt that María went to the cinema.’

The question that needs to be answered is why this contrast arises. Here I suggest that the key to understand this contrast lies in what is the role of uttering expressions such as (38)–(39) (in an exchange). As discussed in Section 3 in connection to (9), when uttering an expression of the form dizque α, where α denotes proposition p, a proposition is made salient towards the addressee. In contrast, when a sentence with dudar ‘doubt’ is uttered, an assertion is made, i.e., the speaker indicates that she believes the at-issue proposition and intends that such a belief becomes a shared belief; such a role is absent when uttering expressions with dizque. Crucially, in (38), with dizque, while the speaker may lack certainty with regard to whether p is true, she is not trying to make such a meaning a part of the Common Ground—that meaning is a not-at-issue entailment only. This is where the difference with dudar α, where α denotes proposition p, lies. In (39) for instance, the speaker would be trying to make her belief towards p a common belief—her doubt of p is the at-issue meaning. The problem is that, if the speaker utters (39), she would not be making a truthful contribution (i.e., Grice’s maxim of quality would be violated), because she would be stating that she doubts p when she is actually in a position of saying something stronger (that she believes that not p in this case).18

Note that the discussion indicates that the crucial difference between an expression with dizque and an expression with dudar ‘doubt’ does not lie in whether the speaker expresses doubt (she does that in both cases). The difference lies in whether the speaker is making an attempt to make her beliefs common beliefs, which is crucially tied to the distinction between presenting vs. asserting an at-issue meaning: only with dudar ‘doubt’ is the doubt meaning asserted, not with dizque, where such the lack of certainty meaning is not-at-issue. In fact, that this is the case, i.e., that this is where the key difference lies, can be tested if the expression with dudar ‘doubt’ removes the beliefs of the speaker towards the relevant proposition, by, e.g., changing the subject of the sentence, so that the doubt that is expressed is not that of the speaker—by doing so, the speaker’s beliefs towards that proposition become irrelevant, since what she is asserting that somebody else has doubts with regard to the proposition (this case is thus closer to dizque in that it would be relevant to indicate what is somebody else’s take on the issue under consideration, as mentioned above).

Context where the speaker knows that the relevant proposition is false: The speaker’s sister was supposed to go to the cinema with María, but María canceled because she was sick, so she didn’t go to the cinema; the speaker was told about this. Susana, however, is still unsure in this regard, i.e., she thinks that María didn’t go to the cinema, but remains unsure. The speaker tells (40) to somebody else.

(40)	Susana	duda	que	María	fuera	al	cine.
	Susana	doubt	that	María	went	to.the	cinema
	‘Susana doubts that María went to the cinema.’

5. Extensions

This section discusses additional cases in connection to reportative evidentials taking the discussion about dizque as a starting point. In Section 5.1, I provide initial discussion with regard to the variation that is found in Romance in connection to dizque’s counterparts, and suggest that the proposal discussed in this paper can be naturally extended to such cases. In Section 5.2, I turn to the more general cross-linguistic variation that is found in connection to reportative evidentials by focusing on the possibility of expressing lack of certainty.

5.1. Reportative Evidentials in Romance

This section turns to the variation found in connection to reportative evidentials that come from the verb say in Romance, i.e., I will briefly discuss more extended variation in Romance languages by considering the dizque’s counterparts that are present in these languages. In particular, I focus on the presence of reportative and lack of certainty implications, and suggest that my analysis of Spanish dizque can be naturally extended to similar evidentials in other Romance languages (although acknowledge that more detailed discussion is needed for some of the varieties discussed below).

Cruschina and Remberger (2008) (see also Cruschina, 2015), for instance, comparatively discuss Spanish dizque, Sicilian dicica, Galician disque, Sardinian nachi, and Romanian ciča. Sanromán Vilas (2020) comparatively discusses Spanish dizque, Brazilian Portuguese diz que and Galician disque (see also Martínez Vera, 2023). Examples with the relevant markers appear below (the markers are glossed as sayC, following Cruschina and Remberger’s, 2008 convention; (41)a–d are taken from Cruschina & Remberger, 2008, pp. 110–111; (41)e was provided to me by Renato Lacerda).

(41)	a.	Dicica	Maria	jè	malata.	(Sicilian)
	b.	Disque	María	estaba	enferma.	(Galician)
	c.	Nachi	Juanne	est	malàidu.	(Sardinian)
	d.	Cică	Ion	e	bolnav.	(Romanian)
	e.	Diz que a	Maria	está	doente.	(Brazilian Portuguese)
		sayC	the Mary/John	is	sick
		‘Mary/John is ill, they say.’

It appears to be the case that there are basically two different types of markers. On the one hand, there are markers which convey the meaning that the speaker has reportative evidence. On the other hand, there are markers which convey the meaning that, in addition to the reportative evidence, the speaker expresses lack of certainty with regard to the information conveyed. The first type of markers would be Sicilian dicica, Sardinian nachi, and Romanian ciča (see Cruschina & Remberger, 2008). The second type of markers would be Spanish dizque, Galician disque (see Sanromán Vilas, 2020) and Brazilian Portuguese diz que (see Alcázar, 2018; see also Casseb & Cristina, 2001; Martínez Vera, 2023). For illustration, consider the example in (42) with Brazilian Portuguese diz que (Renato Lacerda, p.c.), which is very similar to (1). In this case, the speaker has reportative evidence and expresses lack of certainty with regard to the proposition that María has rheumatism.

(42)	Diz	que	a	Maria	tem	reumatismo.	(Brazilian Portuguese)
	diz	que	the	Maria	has	rheumatism
	‘Maria has rheumatism, they say.’

Nonetheless, it is worth pointing out that there is discrepancy with regard to Brazilian Portuguese diz que and Galician disque. Thus, in contrast to Sanromán Vilas (2020), for Cruschina and Remberger (2008), Galician disque belongs to the first group, i.e., it only conveys that the speaker has reportative evidence. In contrast to Alcázar (2018), for Sanromán Vilas (2020), Brazilian Portuguese diz que belongs to the first group, i.e., it only conveys that the speaker has reportative evidence. These discrepancies may well be due to variation within these languages.

Importantly, my proposal can readily capture the variation (as indicated, however, additional, more detailed discussion is needed for some of the markers).19 The first type of markers only incorporates the not-at-issue reportative meaning. The second type of markers incorporates both the not-at-issue reportative meaning and the not-at-issue lack of certainty meaning. A final note is in order with regard to the reportative meaning. Cruschina and Remberger (2008) indicate that all the markers that they discuss, i.e., Spanish dizque, Sicilian dicica, Galician disque, Sardinian nachi, and Romanian ciča, are compatible with second-hand evidence. With regard to third-hand evidence, in particular, with regard to folklore, they indicate that Spanish dizque, Galician disque, Sardinian nachi and Romanian ciča are compatible with it (Brazilian Portuguese also seems compatible with second- and third-hand evidence, as well as folklore; see Sanromán Vilas, 2020; Martínez Vera, 2023). This indicates that the approach taken here, where the source remains largely underspecified, seems appropriate. There is, however, some discrepancy with regard to Sicilian dicica in that, as Cruschina and Remberger (2008) indicate, it may not so readily accommodate cases involving folklore. According to Cruschina (2015), Sardinian dice che is the form that more readily appears in these cases, which is perhaps indicative of some form of competition between dicica and dice che in this regard.

5.2. Lack of Certainty Meaning and Cross-Linguistic Variation

The discussion about Andean Spanish dizque has emphasized that the lack of certainty meaning is a not-at-issue entailment (see Section 3). In the previous section, I suggested that at least some varieties of Brazilian Portuguese and Galician would have a reportative evidential that would basically be the same as dizque. It is worth pointing out that the property that a reportative evidential may be used if the speaker is skeptical with regard to some proposition has been reported before (not only in the literature about Romance languages). For instance, (43), with the reportative schijnen in Dutch (taken from Koring, 2013, p. 49), can be used by the speaker when she wants to express doubt with regard to the relevant proposition.

(43)	Anneloes	schijnt	thuis	te	zijn,	maar	ik	geloof	er	niets	van.
	Anneloes	seems	home	to	be	but	I	believe	there	nothing	of
	‘I’ve heard that Anneloes is at home, but I don’t believe it’

A similar claim has been made in connection to other reportative evidentials. For instance, Kaufmann argues in Schwager (2010) that the Tagalog reportative daw indicates doubt with regard to the prejacent when the evidential is stressed. In a similar vein, Valenzuela (2003) argues that the Shipibo-Konibo reportative -ronki allows “sarcastic” denials with “marked intonation” and other means (see also Mortelmans, 2000; Faller, 2002, 2019; Anderbois, 2014; among others).

In this regard, I would like to briefly mention two proposals in connection to reportative evidentials and the expression of disagreement; these are discussed by Faller (2002, 2019) and Pancheva and Rudin (2019).20 Both proposals are implemented in Farkas and Bruce’s (2010) dynamic model (see also Gunlogson, 2001, 2008; C. Roberts, 2012), where exchanges are to be understood in connection to public commitments made by discourse participants.21 What is key in the current discussion is that these proposals further assume a distinction between an animator and a principal; the animator is the individual that is actually talking, whereas the principal is the individual whose commitment is being described. Sentences with a reportative have the property that the animator and the principal are two different individuals (which are also different from the addressee). This contrasts with plain assertions, where the animator and the principal are the same individual.

What is of relevance in the current discussion is the link to the possibility of the expression of lack of certainty towards the relevant proposition. For instance, for Faller (2002) (see also Anderbois, 2014), the expression of lack of certainty is not the default case with reportative evidentials cross-linguistically. Specifically, the expression of lack of certainty must be overtly indicated, since, by default, by uttering a proposition with a reportative, the animator endorses such a proposition. Faller (2019) implements this by means of the illocutionary force of presentation (see Section 3) in combination with a pragmatic Collaborative Principle, which indicates that “Discourse participants must provide evidence of a detected discrepancy in commitment as soon as possible” (Faller, 2019, p. 25; see Walker, 1996, p. 269). Pancheva and Rudin (2019) adopt a different approach. Contra Faller (2019), they argue that there is no bias in that agreement with what is stated is adopted by default. In their view, the animator is committed to indicating that somebody else (in particular, the principal, but see their discussion for more complicated cases) is committed to the relevant proposition. Importantly, if the participants in an exchange think that some reported proposition p is credible, they may endorse it (in their words, there is a positive bias towards p); if the report is not credible, they may not endorse it (in their words, there is a negative bias towards p).

Having briefly sketched the cross-linguistic picture in this regard, as well as these two approaches to reportative evidentials, the key questions are: (i) what is the locus of dizque within a cross-linguistic perspective? and (ii) how does dizque fit within the analyses that have been sketched above within this broader perspective? With regard to (i), as indicated, the reportative evidentials that have been studied in the formal semantics literature, may convey lack of certainty with regard to the relevant proposition only in a limited number of cases, specifically, when there are additional means, such as stress on the evidential marker (e.g., daw in Tagalog), or when overt disagreement is indicated (e.g., si in Cuzco Quechua). Importantly, there is a key contrast between the cases discussed above and dizque: dizque is the first reportative evidential for which the lack of certainty meaning is to be understood as entailed. In other words, while most reportative evidentials may convey lack of certainty in some limited cases, dizque necessarily conveys lack of certainty. Pending additional research of the cross-linguistic picture, the reportative in Spanish would be, in this sense, unique cross-linguistically.

The question in (ii), i.e., how dizque fits within the analyses sketched above, the key element is how to capture that the lack of certainty meaning of dizque is an entailment. As mentioned, my intention is not to provide a detailed implementation of this within these approaches; my goal is more limited in that I merely want to point to possible research directions. Specifically, under Faller’s (2019) approach, that the lack of certainty meaning of dizque is an entailment means that the Collaborative Principle is not needed in this case (or that it is redundant), since dizque already brings in a potential disagreement in that lack of certainty is expressed towards the relevant proposition. In this sense, the lack of certainty meaning of dizque would override a pragmatic mechanism such as the Collaborative Principle. Under Pancheva and Rudin’s (2019) approach, the lack of certainty meaning of dizque would simply mean that there is always a negative bias, i.e., the animator will always be of the opinion that the relevant proposition is not credible. Such a possibility is already available under this approach; the lexical semantics of dizque would only make this explicit.

6. Conclusions

This paper has provided a unified analysis of dizque in Andean Spanish. Taking, as a starting point, the synchronic and diachronic syntactic analyses of this type of markers in Romance that have been proposed in the literature, I proposed an analysis that makes explicit how the syntax of dizque is mapped into semantics, and provide a semantics that captures the evidential and epistemic modality properties of this element. In particular, I showed that expressions with dizque must be uttered when reportative evidence is available to the speaker, and that the lack of certainty implication that these expressions involve is to be understood as a not-at-issue entailment. I discussed a number of consequences in connection to the analysis, which serve as a baseline to test several properties of this kind of element across Romance. I further pointed out how my proposal can capture the cross-Romance variation that is found in this domain by means of the analysis adopted in this paper for Spanish dizque, and discussed some key differences between dizque and other reportative evidentials cross-linguistically, where crucially, Spanish dizque differs from other reportative evidentials in that only in the former is the lack of certainty meaning entailed.