Grammatical Object Passives in Yucatec Spanish

Abstract

Yucatec Spanish displays a type of sentence that appears to mix elements of an active impersonal and a passive. For example, “te castigaron por mi tío” may be interpreted as “you were punished by my uncle”, where a by-phrase headed by the preposition por introduces an agent rather than a cause or reason. The verb has active morphology—it is always third-person plural, and accusative clitics (e.g., te) and DOM-marked objects are possible. This type of sentence, which I descriptively label an active–passive (A-P) hybrid, has been mentioned in previous literature on contact varieties in Mayan-speaking regions of Mexico and Guatemala, but it has not been precisely described or analyzed formally. I argue that A-P hybrid constructions are instances of grammatical object passives. Grammatical object passives have certain active properties—accusative case is assigned to a theme argument and the morphology of the verb is active, but like passives, they require that the expression of the agent be a by-phrase rather than a grammatical subject. I claim that this is possible in this variety of Spanish due to the emergence of a null pronoun, absent in other varieties of Spanish, that can merge in the specifier of Voice and restrict, rather than saturate, an agent argument, permitting the subsequent addition of a third-person by-phrase. I demonstrate that this analysis is able to explain its hybrid properties as well as other person restrictions on the by-phrases that express the agent. Finally, I describe avenues of future research that will help discern the role that language contact may have played in the emergence of A-P hybrids.

1. Introduction

Yucatec Spanish (YS) is the variety spoken throughout the Yucatán peninsula in southern Mexico primarily in the states of Campeche, Yucatán and Quintana Roo. Yucatec Maya is spoken by approximately 800,000 people in the peninsula, the vast majority of whom are bilingual (INEGI 2009; INALI 2012). Yucatec Spanish is known primarily as a contact variety whose phonetics, phonology, lexicon and morphosyntax have been influenced by Yucatec Maya (see Barrera Vázquez (1945), Michnowicz (2015), Guitiérrez Bravo et al. (2019), Bove (2020) for overviews and analyses of particular contact phenomena). In this paper, I describe and analyze a construction in Yucatec Spanish that has characteristics of both an active impersonal sentence and a passive sentence, which I descriptively label as active–passive (A-P) hybrid sentences. An example is shown in (1) The speaker is talking about when babies are thought to suffer from mal de ojo (“evil eye”) in this example—the clitic los refers to babies.1

(1)

Though A-P hybrids look superficially like active transitive sentences in which there is a null third-person subject and a by-phrase that describes a cause or reason, they are not interpreted in this way. What is interesting about (1) is that the nominal expression in the by-phrase is interpreted as the ‘looker’, not the cause or reason for some other people looking at babies. This is clear from the context; mal de ojo is thought to be given to infants when certain people, such as drunk men or pregnant women, look them in the eye, and also from elicitation. Consultants reject the interpretation of (1) in which there is some implicit plurality different from las mujeres embarazadas, which is the ‘looker’.

Aside from this important interpretative difference, the key characteristics of A-P hybrids are the following: (i) the verb is formally active and always inflected as a third-person plural, (ii) a theme argument surfaces with accusative case and (iii) an initiator/experiencer argument (normally the external argument of a transitive verb) surfaces as a by-phrase. The by-phrase may be plural as in (1) or singular as in (2). The context for this example is that a hunting dog accompanying two men on a deer hunt disappeared and was pulled into the underworld by a shape-shifting deer (ese venado encantado).

(2)

The context of this story clearly shows that the by-phrase introduces the initiator of the event rather than the cause or reason. Here, the initiator ese venado encantado is singular.

A-P hybrids are mentioned in every overview of the unique morphosyntactic properties of Yucatec Spanish (Barrera Vázquez 1945, p. 344; Michnowicz 2015, pp. 31–32; Guitiérrez Bravo et al. 2019, p. 280) and have been argued to arise from a semantic influence from Yucatec Maya in Lema (1991). They can also be found in the Spanish of bilingual speakers from other Mayan-speaking communities outside of Yucatán. The examples in (3) are cited in Craig (1977) as specific ways of rendering passive sentences from Popti’ (also known as Jakalteko) into Spanish by bilingual Popti’–Spanish speakers. Craig (1977) mentions that this could be a more general characteristic of Guatemalan Spanish.

(3)		(Craig 1977, p. 97)
		(Craig 1977, p. 98)

While I know of no in-depth studies of A-P hybrids in Guatemalan Spanish, the following datum appears to corroborate the existence of the construction in a broader range of bilingual speech communities. Example (4) is a quote from Guatemalan actor María Mercedes Coroy (also a native speaker of Kaqchikel) describing the terror experienced by Mayan communities during the Guatemalan civil war. Here, el ejército (“the army”) is clearly the agent rather than a cause or reason.

(4)

In spite of their documented existence, A-P hybrids have not been accurately described or systematically analyzed through corpus research or elicitation. Given this situation, I have two aims in this paper. The first is to sharpen previous descriptions of the properties of A-P hybrids using novel production data and elicited acceptability judgments from bilingual Spanish–Yucatec Mayan speakers. The second is to present a formal analysis of A-P hybrids based on work by Legate (2014).

The paper is organized into four sections. In Section 2, I provide the theoretical background on different kinds of active impersonal and passive sentences cross-linguistically Legate (2014), focusing on a type of construction that exhibits mixed properties of active and passive sentences called grammatical object passives. In Section 3, I outline the core descriptive properties of A-P hybrids, showing what they have in common with transitive active sentences and what they have in common with passives. I argue that, given their mixed properties, A-P hybrids are grammatical object passives. I then situate A-P hybrids within a general typology of active and passive constructions across different varieties of Spanish. In Section 4, I discuss avenues for future research regarding the role that language contact has played in the emergence of A-P hybrids. In Section 5, I provide a brief conclusion.

2. Theoretical Background on Passives and Active Impersonal Sentences

In order to properly integrate A-P hybrids into a theory of passives, it is instructive to start with the standard way of how the difference between active and passive sentences is treated in modern theoretical approaches to syntax. Consider the sentences in (5).

(5)

Active sentences have a Voice head that semantically introduces an initiator. I will represent the initiator argument with the label $\theta$ on Voice. A variable associated with the initiator is saturated by an external argument that is licensed syntactically as the specifier of Voice. I will represent the presence of a syntactic argument in the specifier of Voice as a “D” feature. Finally, a set of uninterpretable phi features license an internal argument with accusative case, which is represented as u$\varphi$. The external argument receives nominative case through an agree relation with finite T. The features of Voice in (5-a) are shown in (6).

(6)

Passive sentences have a Voice head that semantically introduces an initiator, but they lack both of the other characteristics of active Voice: there is no syntactic argument in the specifier of Voice to saturate a variable associated with the initiator (no “D” feature), and there is no uninterpretable set of phi features that license an internal argument with accusative case. The lack of phi features in Voice means that the internal argument is licensed through an agree relation with T, from which it receives nominative case. Finally, while initiators may remain implicit in passives, the presence of the initiator predicate in Voice may license a PP adjunct—a by-phrase—that is linked to this initiator (see Bruening (2013) for a detailed discussion). The representation of the passive sentence in (5-b) prior to any kind of movement is shown in (7) (I have not included the auxiliary in the representation for ease of exposition).

(7)

This analysis of the distinction between active and passive sentences is largely based on Burzio’s generalization (Burzio 1986). Essentially, the presence of a “D” feature on Voice that syntactically licenses an external argument correlates with the presence of uninterpretable phi features that license an internal argument through structural case assignment. In an insightful treatment of passive constructions cross-linguistically, Legate (2014) has shown that the notion “passive voice” should be understood in a more nuanced way than simply the absence of an external argument and accusative case. What can be conceived generally as passive constructions show a variety of morphosyntactic differences across languages, and these differences can be modeled in terms of features in Voice and the inventory of Voice morphemes and pronominal elements that merge in external argument positions.

Some languages, like Acehnese (an Austronesian language spoken primarily in Indonesia), have Voice morphemes that impose person restrictions on the initiator that is adjoined as a PP to VoiceP. In (8), the verbal prefix di- indicates that the initiator is third-person—it is not a subject agreement morpheme.

(8)

Acehnese (Legate 2014, p. 9)

Legate (2014) captures this phenomenon by proposing that certain languages have phi features that may merge with a Voice head and restrict, rather than saturate, the person features of the initiator that may be subsequently added as PP adjuncts (see Chung and Ladusaw (2004) for the difference between restriction and saturation). In (9), di- would be a manifestation of $\varphi$ and would impose a restriction on the phi features of the initiator adjunct.

(9)

Apart from this restriction imposed by the prefix, constructions like (9) behave like canonical passive sentences. Voice licenses a semantic initiator argument but has no “D” feature and cannot assign accusative case to the theme. As a result, the theme gets nominative case from finite T.

Other languages, such as Ukrainian and Icelandic, have sentences that exhibit a mixture of active and passive properties. Legate (2014) labels such sentences as “grammatical object passives”. Some examples can be observed in (10) below.

(10)	a.
		‘The church was built by Lesiv’	Ukrainian (Legate 2014, p. 94)
	b.
		‘The car was inspected by the mechanic’	Icelandic (Legate 2014, p. 89)

The verb in grammatical object passives exhibits a form of default agreement. In Ukrainian (10-a), the auxiliary verb appears with neuter morphology, while in Icelandic (10-b), it agrees with an expletive. The theme argument is marked with accusative case, and the initiator may be expressed as an oblique or PP adjunct. The most important difference between grammatical object passives and canonical passives is that the theme receives accusative case. Legate (2014) proposes that grammatical object passives have the following structure.

(11)

In grammatical object passives, the Voice head shares some properties of active Voice—it is endowed with uninterpretable phi features that can structurally license an internal argument with accusative case and it has a specifier. However, there is no “D” feature, and the external argument position is not saturated with a full DP. Instead, a deficient pronoun, labeled $\varphi$P in (11), merges in this position and restricts the predicate associated with the initiator such that only a by-phrase with a third-person initiator can modify the predicate. This permits the addition of a by-phrase or quantification of the initiator-linked variable through existential closure. In this sense, the external argument behaves as it does in a passive sentence.

Finally, Legate (2014) contrasts grammatical object passives with impersonal constructions. Impersonal constructions are sometimes translated as passives because the external argument appears to be implicit, but they behave like active sentences with a fully fledged active Voice head. The external argument position in impersonal constructions is usually a null pronoun that fully saturates the variable associated with the initiator. Because it functions like a full DP argument, fully referential by-phrases are not possible in impersonal constructions, as shown in (12).

(12)

Polish (Legate 2014, p. 96)

The structure for active impersonal sentences is shown in (13) below.

(13)

Due to the different kinds of sentences that exhibit mixed active and passive properties, Legate (2014) claims that the difference between passive and active should not be understood as a binary “all or nothing” distinction. On the contrary, it seems better to conceive it as a cline that can be modeled in terms of features in Voice and the nature of the external argument. With this cline now clearly outlined, we can turn to A-P hybrids in Yucatec Spanish and propose a precise analysis of them.

3. Description and Analysis of A-P Hybrids in Yucatec Spanish

In this section, I present the main descriptive properties of A-P hybrids, sharpening previous descriptions and introducing novel data. I then argue that A-P hybrids in Yucatec Spanish should be treated as a particular type of grammatical object passive as they exhibit all the primary characteristics of similar constructions in the languages discussed in Section 2.

3.1. Not a “SE Construction”

Since Barrera Vázquez (1945) first mentioned A-P hybrids as a unique characteristic of Yucatec Spanish, most cited examples, such as those in (14) below, contain the reflexive se clitic.

(14)	Yucatec Spanish
	a.
		‘It was taken from him by his father’	(Barrera Vázquez 1945, p. 344)
	b.
		‘It was told to him by Don Julio’	(Lema 1991, p. 1279)

Because of the prevalence of se in such examples, it seems to have been tacitly assumed that they are a type of “SE construction” in which se is obligatory (Barrera Vázquez 1945; Lema 1991).

There are two pieces of evidence against the idea that A-P hybrids are a type of “SE construction”. First, there are A-P hybrid sentences that lack se, as shown in (15).

(15)	Yucatec Spanish
	a.
		‘An email was sent by the principal’	recorded, July 2021
	b.
		‘It was put in the room by his father-in-law’	(Lema 1991, p. 1279)

Second, there are A-P hybrid sentences in which se is unacceptable, as shown in (16).

(16)

In conclusion, A-P hybrids are not “SE constructions” because they do not require the presence of the se clitic. When se is present, as in examples in (14), it is a surface manifestation of the third-person dative clitic that is linearly adjacent to a third-person accusative clitic (so-called spurious se).

3.2. The Verb Is Invariable

As can be appreciated from the data presented in Section 3.1, the verb in A-P hybrid sentences is invariable—it is always third-person plural.

(17)

When the verb is not third-person plural, the interpretation associated with A-P hybrid sentences is not possible. That is, the by-phrase cannot be interpreted as an agent, as shown in (18), where the verb is third-person singular.

(18)

One consultant communicated that (18) is possible only if my uncle is interpreted as the cause or reason for which someone else killed the deer. This shows that a third-person plural morphology is fundamental in licensing the agent reading associated with the by-phrase.

3.3. Themes Are Expressed as Grammatical Objects

The theme arguments of verbs in A-P hybrid constructions surface as grammatical objects. They may be accusative clitics (19-a) or a full DP. The latter may appear with (19-b) or without (19-c) the differential object marker a.

(19)

3.4. Initiators Are Expressed as By-Phrases

The property that distinguishes A-P hybrids from superficially similar constructions that are present in all varieties of Spanish is the interpretation assigned to the by-phrase. Unlike other varieties of Spanish, the by-phrase is interpreted as an initiator (agent, experiencer or direct causer) rather than an indirect cause or reason. For example, in (20) below, mi abuela ‘my grandmother’ and los vecinos ‘the neighbors’ are interpreted as the agents of the verbs moler ‘grind’ and invitar ‘invite’, respectively.

(20)

In other dialects of Spanish, these interpretations are not possible. Instead, the only possible reading that could be assigned to the by-phrases is one of cause or reason. For instance, (20-b) is possible only if it is interpreted as ‘They invited us because of the neighbors’, where the agent of the verb cannot be co-referential with the DP in the by-phrase.

There is an important restriction on the person features of the DP that may appear within the initiator by-phrase of A-P hybrid sentences: it can only be third-person. Some of the examples from previous sections are repeated below.

(21)	Yucatec Spanish
	a.
		‘We were invited by the neighbors’	elicited, July 2021
	b.
		‘It was told to him by Don Julio’	(Lema 1991, p. 1279)

As can be observed in (21), third-person initiators may be singular or plural and include both common and proper names.

First- and second-person pronouns, on the other hand, are unacceptable as initiators in A-P hybrid sentences, as shown in (22).

(22)

3.5. Analysis: A-P Hybrids in Yucatec Spanish as Grammatical Object Passives

I propose that A-P hybrid sentences in Yucatec Spanish are instances of grammatical object passives. Specifically, what Yucatec Spanish has that other varieties lack is a null subject pronoun, a $\varphi$P, that has the following characteristics.

(23)

$\varphi$P is formally third-person plural. This is what triggers third-person plural morphology on the verb. However, it is associated with an underspecified generic meaning, like impersonal pronouns more generally. As can be seen in (23-b), $\varphi$P combines with an event that has an unsaturated external argument (type <e, st>) and restricts the variable associated with the initiator (x) such that it cannot be a participant in the speech situation.

The representation of (15), repeated as (24) below, is shown in (25).

(24)

(25)

As is the case with other grammatical object passives, Voice has some features that are like an active construction—it assigns accusative case to the theme argument un correo and it has an external argument, a $\varphi$P. The formal features of $\varphi$P trigger third-person plural agreement on T, which is spelled out as -n. However, it also has some properties of a passive. Voice has no “D” feature and $\varphi$P does not saturate the argument variable associated with the initiator introduced in Voice, which permits the addition of a by-phrase. The semantic composition of the VoiceP is shown below, based on Bruening (2013) and Legate (2014).

(26)

When $\varphi$P is combined with Voice${}^{\prime }$, it yields an event with an unsaturated variable x but imposes a restriction on the type of argument that may saturate the variable—it must be third-person. Next, a by-phrase may link a DP that is embedded inside it with the initiator predicate’s argument variable so long as it is third-person and compatible with the restriction imposed by $\varphi$P. When VoiceP is merged with aspectual and temporal functional projections, existential closure applies to both the argument variable x and the event variable e (see Bruening (2013) and Legate (2014) for details).

Treating A-P hybrid sentences as grammatical object passives accounts for all the properties outlined in the above subsections. First, it explains why there is no se present. If we take impersonal and passive se to be a Voice morpheme that appears when there is a syntactic “D” feature that is not checked (MacDonald and Maddox (2018)), the absence of a syntactic “D” feature would account for why we do not see se in A-P hybrids. Second, the Voice head assigns accusative case to the theme argument. Third, the formal features of $\varphi$P trigger third-person plural agreement on the verb. Finally, $\varphi$P does not saturate the variable associated with an initiator predicate but restricts it such that it cannot be interpreted as first- or second-person. This permits the presence of a by-phrase so long as it is compatible with this restriction.

There are two additional pieces of evidence in favor of the grammatical object passive analysis of A-P hybrids. The first concerns depictive modification of the external argument. As discussed above, the $\varphi$P in the specifier of Voice in grammatical object passives is not a full DP argument. Depictive secondary predication is thought to only be possible when the null argument is a full DP. Legate (2014) shows that there is a difference between Polish impersonals and Icelandic grammatical object passives in this regard. Only the former license a depictive secondary predicate as a modifier of the external argument, as shown in (27-a).

(27)	a.
		‘They robbed Jana while (they were) drunk’ Polish impersonal (Legate 2014, p. 97)
	b.
		Intended: ‘Breakfast is eaten naked’ Icelandic GrObjPass (Legate 2014, p. 89)

A-P hybrids behave like grammatical object passives with respect to depictive modification, as shown in (28).

(28)

This provides evidence in favor of a $\varphi$P in Voice that lacks a “D” feature and thus does not license a secondary depictive adjective. The initiator in the by-phrase is too deeply embedded within a PP to serve as the subject of the secondary predicate, so the sentence is unacceptable.

A second piece of evidence in favor of the grammatical object passive analysis comes from binding. If a null pronoun is a full DP, it can bind an anaphor in object position but deficient pronominal elements may not. MacDonald and Maddox (2018) claim that impersonal and passive SE constructions have a pronoun with a “D” feature capable of saturating the external argument introduced by the Voice head. This pronoun may be co-referential with a possessor of a body part object.

(29)

In contrast to SE constructions, A-P hybrids do not permit the external argument to bind a possessor of a body part object, as shown in (30).

(30)

I claim that the unaccepability of (30) is due to the fact that $\varphi$P in the external argument position is not a full DP and cannot bind a possessor in the body part object. Likewise, the initiator in the by-phrase is too deeply embedded and does not c-command the body part object.

3.6. A-P Hybrids and Other Impersonal/Passive Constructions across Different Varieties of Spanish

In this section, I compare and contrast A-P hybrids with other types of impersonal and passive constructions in Spanish. A-P hybrids are superficially very similar to sentences with “arbitrary plural” pronominal subjects (Jaeggli 1986), such as (31).

(31)

General Spanish (Jaeggli 1986, p. 45)

The arbitrary null pronoun in the subject position of such sentences is formally plural but semantically underspecified. It only indicates that there is an unspecified third-person subject that may be singular or plural. This mismatch between formal features and semantic interpretation is shared with A-P hybrids. Where arbitrary plural sentences differ from A-P hybrids is in their inability to accept by-phrases.

(32)

The difference between arbitrary plural sentences and A-P hybrids can be captured in the features of Voice and the nature of the pronominal element in the specifier of Voice. Arbitrary plurals are transitive sentences in which a full DP saturates the argument of a Voice head with a “D” feature, while A-P hybrids are grammatical object passives in which a $\varphi$P restricts the argument variable of an initiator predicate introduced in Voice.

(33)

A-P hybrids are also similar to non-paradigmatic SE constructions (see MacDonald and Maddox (2018), Ormazabal and Romero (2019) and (Saab 2020) for recent overviews and analyses), which include what are traditionally known as impersonal and passive SE constructions.

(34)

MacDonald and Maddox (2018) have argued, based on data such as (29), that non-paradigmatic se constructions have a null pronoun with a “D” feature that saturates the external argument variable introduced in Voice. Moreover, it has been argued in both Ormazabal and Romero (2019) and (Saab 2020) that the variable agreement on the verbs in non-paradigmatic SE constructions is not really an indication of an active–passive distinction. Both sentences in (34) are active transitive sentences, and the apparent agreement between the verb and the object in (34-b) can be handled outside of the narrow syntax. Based on this work, I assume that non-paradigmatic SE constructions have the structure below. I follow MacDonald and Maddox (2018) in assuming that se is a Voice morpheme generated in the Voice head.

(35)

Given this treatment of non-paradigmatic se constructions, one would expect them to pattern like arbitrary plural sentences with respect to by-phrases. This is largely true as referential by-phrases that refer to individuals are unacceptable.

(36)

There are, on the other hand, some by-phrases with unspecified readings that either refer to institutions or generically to a type or class of people that are accepted by some speakers.

(37)	General Spanish
	a.
		(MacDonald and Maddox 2018, p. 10)
	b.
		‘We were summoned by the government’ (Ormazabal and Romero 2019, p. 66)

Pujalte (2013) and Ormazabal and Romero (2019) suggest that the peculiar restrictions on these by-phrases warrant a treatment that is distinct from the agentive by-phrases we see in canonical passives. They are adjuncts that modify an already saturated VoiceP and thus do not identify the initiator in the same way as in a canonical passive or, indeed, a grammatical object passive. Note that A-P hybrids permit a far wider range of by-phrases, suggesting that they are more like passives than non-paradigmatic SE constructions. In (38), repeated from Section 3, we see that by-phrases may refer to specific individuals in A-P hybrids.

(38)	Yucatec Spanish
	a.
		‘That deer, it was killed by my uncle’	elicited, July 2021
	b.
		‘It was told to him by Don Julio’	(Lema 1991, p. 1279)

Finally, let us consider analytic passives. Like A-P hybrids, these sentences have no “D” feature in Voice. However, unlike A-P hybrids, they have no specifier and no restriction on potential by-phrases. This means that they are compatible with first- and second-person by-phrases, as shown in (39).

(39)

The lack of restrictions on by-phrases are due to the fact that Voice has an unsaturated and unrestricted initiator role that can be freely modified by an adjunct.

(40)

In conclusion, the range of different impersonal/passive constructions that we observe across different varieties of Spanish can be differentiated based on whether Voice has a specifier and the nature of the pronoun in that specifier. A-P hybrids occupy a unique place in this typology that is slightly distinct from any other existing construction. The last issue that will be discussed concerns how they might have arisen and the potential role that language contact with Yucatec Maya played in causing their emergence.

4. The Role of Language Contact in the Emergence of A-P Hybrid Sentences

Given that the documentation of Spanish A-P hybrid sentences is limited to bilingual Mayan-speaking communities in Mexico and Guatemala, an important area of future research pertains to the role that contact with Mayan languages has played in the emergence of A-P hybrids. Rather than make a definitive proposal, my comments in this section will outline three ways of understanding how A-P hybrids could have emerged in Yucatec Spanish. For each scenario, I will comment on the potential influence of Yucatec Maya, leaving to future research more definitive answers as to the whether the cause of the emergence of A-P hybrids is due solely to language contact.

4.1. An Internal Change?

As with any linguistic innovation, perhaps the most important question about A-P hybrids is if they have developed due to an internal change or an external change (e.g., contact with another language). Even in situations where an externally induced change due to language contact may seem like an obvious explanation, it is often difficult to rule out internal causes as potential explanations of the same change (see Thomason (2020) and references therein). For instance, it is possible that A-P hybrids are the result of an internal change in the grammatical system of speakers of Yucatec Spanish (and other Mesoamerican varieties of Spanish). As described in Section 3.5 and Section 3.6, the difference between an active impersonal sentence and an A-P hybrid involves a small change in the features of Voice and the element in its specifier, which enable the expression of an initiator as a by-phrase. This kind of variation has been documented in the history of Spanish SE constructions and modeled in terms of a change in the features of the null pronoun in the specifier of Voice MacDonald and Maddox (2018). The fact that internal changes in the feature structure of null generic pronouns in SE constructions have been proposed, and lead to superficial differences in terms of the presence or absence of by-phrases, could be applied to impersonal third-person plural constructions as well. In order to explore this hypothesis in more detail, one would have to collect historical data on the presence of by-phrases that identify initiators in impersonal third-person plural constructions and search for potential examples of A-P hybrids in varieties of Spanish that are not in contact with Mayan languages. I lack any of the relevant data in order to further consider this hypothesis currently, so it cannot be ruled out as a potential factor in the emergence of A-P hybrids.

4.2. A Case of Contact-Induced Change?

Another possible explanation for the emergence of A-P hybrids is that they have developed due to contact with Yucatec Maya (or with Mayan languages more generally). In order to properly evaluate this potential explanation, we need to explore which characteristics of Mayan languages could potentially contribute to the development of A-P hybrids. Since nearly all the data in this paper come from bilingual Yucatec Maya–Spanish speakers, I will focus on certain morphosyntactic characteristics of Yucatec Maya that could have played a role in the development of A-P hybrids in Spanish.

Even though Yucatec Maya is a null subject and object language in which grammatical relations are head-marked on predicates, it has no dedicated null generic pronouns that give rise to an active impersonal construction similar to arbitrary plurals or non-paradigmatic SE constructions in Spanish. Instead, passives are used to suppress initiators and to express the full range of meanings of both impersonal and passive sentences in Spanish. Passives in Yucatec Maya are synthetic, formed either by glottalizing the vowel in a verb root or adding a glottalized suffix to a verbal stem Bricker et al. (1998).

(41)

In (41-a), the vowel in the verb root juch’ is glottalized to form the passive, while in (41-b), a passive suffix with a glottalized vowel -a’ab is added to the verbal stem kíin-s.

As can be observed in (41), theme arguments are marked differently depending on aspect. Yucatec Maya is a split ergative language. In the incompletive aspect, intranstive subjects are marked with the same set of bound pronouns as transitive subjects, a nominative-accusative pattern. This is illustrated in (42).

(42)

Note that both the intransitive theme subject in (42-a) and the transitive agent subject in (42-b) are marked with the same bound pronoun in. In the completive aspect, intransitive subjects are marked with the same set of bound pronouns as transitive objects, an ergative–absolutive pattern. This is illustrated in (43).

(43)

In these completive sentences, the intransitive theme subject in (43-a) and the transitive theme object in (43-b) are marked with the same bound pronoun -en.

The initiator in Yucatec Maya passives may either be implicit or expressed as the complement of the relational noun/preposition tumen ‘by.’ To my knowledge, there are no reported person restrictions on initiators in by-phrases in Yucatec Maya. However, the bilingual speakers consulted in this study judged first- and second-person initiators in passives as marked compared to the third-person initiators, as shown by the distinction between (44-a) and (44-b).

(44)

Now that some basic characteristics of passives in Yucatec Maya have been presented, we can see that there are two properties that are shared between them and A-P hybrids: (i) they are both synthetic, expressed without the need for an auxiliary and (ii) both have person restrictions on the initiator argument (at least for some speakers). Given these similarities, one possible proposal is that Yucatec Maya acts as a model for the transfer of a pattern into Spanish in which a set of morphosyntactic features correlate with a particular meaning (see Heine and Kuteva (2005), (Matras 2009), Grant (2020) for different ways of conceptualizing transfers of patterns). Further investigation of this idea should look into the frequency of A-P hybrids with respect to other ways of expressing impersonals and passives across Spanish varieties. If bilingual speakers use Spanish A-P hybrids in similar contexts and with similar frequencies to the passives used in Yucatec Maya, this may provide evidence that A-P hybrids have arisen due to contact-induced change.

4.3. A Case of Convergence?

The last potential explanation for the development of A-P hybrids that will be mentioned here is that of convergence. As discussed in Grant (2020), convergence is used in a variety of ways in the contact linguistics literature. I will follow one of the more narrow definitions of convergence, whereby an innovation in a particular language develops as a result of a combination of internal changes and contact-induced influence from another language (see Pato Maldonado (2002) for a detailed study of possessive constructions that have arisen through convergence in Guatemalan Spanish). In the case of A-P hybrids, an explanation based on convergence might claim that these arose due to an internal change in the feature specifications of Voice and the pronominal element in its specifier and that the use of A-P hybrids was further reinforced due to its superficial similarity with passive constructions in Yucatec Maya. The result is a construction that has similarities to existent impersonal constructions in many Spanish varieties and is also similar to Yucatec Maya passives in being synthetic and imposing person restrictions on the expression of the initiator.

5. Conclusions

In this paper, I have provided a novel, detailed description and analysis of a class of sentences in Yucatec Spanish that I descriptively labeled active–passive (A-P) hybrids. I argued that A-P hybrids are an instance of grammatical object passives Legate (2014). On the one hand, they are like active sentences in that there is a null pronominal element in the specifier of Voice and accusative case is assigned to the theme. On the other hand, they are like passive sentences in that the null pronominal element does not saturate the external argument variable, leaving open the possibility that it may be modified by a DP introduced as a by-phrase and then bound through existential closure. Finally, I reviewed some avenues of future research that should shed light on whether A-P hybrids have arisen due to internal changes, due to language contact with Mayan languages or due to a combination of both factors.