1. Introduction
Certain classes of linguistic expressions are characterized by distinctive syntactic distributions.Footnote 1 Interrogative phrases are a canonical example: in a survey of content questions in 902 languages, Dryer (Reference Dryer, Matthew S. Dryer and Zenodo2013) finds that approximately a third of the world’s languages require interrogative phrases to appear at the left periphery. Other kinds of expressions that have been noted to trigger phrasal displacement include (i) focus sensitive items (see e.g. Branan and Erlewine Reference Branan and Erlewine2023), (ii) certain kinds of definite expressions (e.g., Germanic languages: Holmberg Reference Holmberg1986, Diesing and Jelinek Reference Diesing and Jelinek1995; Ch’ol: Coon Reference Coon2010, Little Reference Little2020a), and (iii) certain types of quantifiers (e.g., Hungarian: Kiss Reference Kiss and C. T1991, Szabolcsi Reference Szabolcsi and Anna Szabolcsi1997; Chamorro: Chung Reference Chung1998; Garifuna: Barchas-Lichtenstein Reference Barchas-Lichtenstein, Keenan and Paperno2012; Mixtec: Ostrove Reference Ostrove2018).
This article adds to this literature by providing a novel description and analysis of quantificational expressions in Chuj (Mayan), with a focus on the syntactic position of the arguments that host them. As briefly noted in previous work (Coon et al. Reference Coon, Baier and Levin2021; Alonso-Ovalle and Royer Reference Alonso-Ovalle, Royer, Nicole Dreier, Kwon, Darnell and Starr2022, Reference Alonso-Ovalle and Royer2024), as well as for the related language Q’anjob’al (O’Flynn Reference O’Flynn, Paperno and Keenan2017), a subset of Chuj quantificational expressions appear at the clause’s left periphery. For example, in (1) and (2), the bolded quantificational arguments are prohibited from appearing after the verb. Moreover, Chuj’s morphology indicates that the quantified agent in the (a) examples has undergone A
$'$-extraction: Agent Focus (af) morphology (-an) arises on the verb whenever a transitive subject is A
$'$-extracted (Buenrostro Reference Buenrostro2004, Reference Buenrostro2013, Reference Buenrostro2021; Coon et al. Reference Coon, Pedro and Preminger2014; Royer Reference Royer2025).
From a Mayan perspective, the unacceptability of quantified DPs in postverbal positions is unexpected. Mayan languages (England Reference England1991, Aissen Reference Aissen1992, Clemens and Coon Reference Clemens and Coon2018, Little Reference Little2020b), Chuj included, exhibit verb-initial word order in discourse neutral contexts (3):
Against these empirical facts, this article aims to answer the following questions about the nature of Chuj quantificational expressions, and their effects on the position of nominal arguments:
Why must some quantifiers appear in a sentence-initial position?
If any, what kind of movement operations are involved to derive the initial position of the quantifiers in sentences like (1a) and (2a)?
How are sentence-initial quantifiers formally distinguished from other expressions, including other quantifiers, that do not have to occur in this position?
After providing relevant background on Chuj in section 2, the rest of the article advances three main proposals, shedding light on all three questions above. In doing so, this article also contributes to the overall understanding of Chuj’s quantifier inventory, a topic that remains heavily understudied in Mayan linguistics as a whole (Henderson Reference Henderson2016).
In section 3, we first establish that despite their surface similarity, the sentences in (1) and (2) involve two different sub-classes of Chuj quantifiers. In particular, a number of syntactic diagnostics lead us to make the following empirical claims:
In other words, despite the fact that sentences containing tzijtum and masanil look identical on the surface – as shown in (1) and (2) – we argue that they exhibit an entirely different syntax. Our use of “A-quantification” versus “D-quantification” is based on terminology used in previous work, including Partee (Reference Partee, Emmon Bach, Jelinek and Partee1995), Keenan and Paperno (Reference Keenan and Paperno2012), and Davis and Matthewson (Reference Davis, Matthewson, Daniel Siddiqi, Barrie, Gillon, Haugen and Mathieu2019). Specifically, we use “D-quantifiers” to refer to quantificational items that are syntactically contained within the extended nominal domain, and “A-quantifiers” to refer to those that fall outside the nominal domain (e.g., verbs, predicates, auxiliaries, adverbs). Importantly, both D- and A-quantifiers can semantically quantify over entities, the type of quantification we focus on here.
In section 4, we turn to a novel syntactic analysis of quantifiers like tzijtum, taking into account the observation from section 3 that this item belongs to a class of nonverbal predicates. Our analysis is based on Mateo Toledo’s (Reference Toledo2012) and Coon’s (Reference Coon, Hsin-Lun Huang, Poole and Rysling2014) approaches to nonverbal predication in other Mayan languages, as well as on Coon et al.’s (Reference Coon, Pedro and Preminger2014) and Coon et al.’s (Reference Coon, Baier and Levin2021) analysis of Mayan extraction and relativization. A simplified syntax for (1), paraphrased with English, is provided in (5). Quantifiers like tzijtum ‘many’ are nonverbal predicates that require a DP argument. When this argument is a transitive subject heading a relative clause, as in (1), Agent Focus is triggered on the verb (Bielig Reference Bielig2015, Royer Reference Royer2025):
In proposing a predicative analysis of this quantificational expression, our analysis diverges from a previous analysis of this same item in Kotek and Erlewine (Reference Kotek and Erlewine2019), which identified tzijtum as part of a DP. It also diverges from O’Flynn’s (Reference O’Flynn, Paperno and Keenan2017) treatment of xiwil ‘many’ in Q’anjob’al, which likewise analyzes it as a D-quantifier. On the other hand, our analysis aligns with Mateo Toledo’s (Reference Toledo2012) approach to xiwil in Q’anjob’al, which also treats it as a nonverbal predicate.
Section 5 then turns to a syntactic analysis of masanil ‘all’ in sentences like (2). We first show that the behaviour of quantifiers like masanil is distinct from that of other D-quantifiers in the language, which typically appear in regular postverbal argument positions. Building on previous work on related phenomena (Horvath Reference Horvath, Simin Karimi, Samiian and Wilkins2007, Cable Reference Cable2010, Hedding Reference Hedding2025, Branan and Erlewine Reference Branan and Erlewine2023), we argue that the subset of D-quantifiers that show a strong preference for appearing in sentence-initial positions do so because these quantifiers are sensitive to focus alternatives. We implement this by adopting an analysis of masanil as lexically associated with an A
$'$-feature, which we analyze as a general [Q] feature in the context of Cable (Reference Cable2010). As such, masanil DPs are eligible goals for the Q-probe on Foc in the left periphery, forcing movement to [Spec, FocP]:
As we will argue, only a select subset of D-quantifiers carry the [Q] feature, and as such, most D-quantifiers in the language do not have to appear in a sentence-initial position.
Finally, section 6 concludes with a discussion of important topics for future work.
2. Chuj clausal syntax and the left periphery
This section establishes some background on Chuj. We start in section 2.1 with basic information about the language of study in this work, San Mateo Ixtatán Chuj, and the methodology we employed in the collection of data. We then provide a brief description of Chuj clausal syntax in section 2.2, and of its left periphery in section 2.3.
2.1 Chuj data and methodology
The main language of study in this paper is Chuj, a language belonging to the Q’anjob’alan sub-branch of Mayan languages (Kaufman Reference Kaufman1974, Law Reference Law2014).Footnote 2 Chuj currently has roughly 95,000 speakers, primarily located in the Department of Huehuetenango in Guatemala and in the State of Chiapas in Mexico, but also in diaspora communities across North America (Maxwell Reference Maxwell, David Levinson, Dow and Van Kemper1993, Hopkins Reference Hopkins2021, Kaplan, Reference Kaplan2021, Royer et al. to appear). There are two principal dialects of Chuj: San Mateo Ixtatán and San Sebastián Coatán.
All of the data we present in this paper come from the San Mateo Ixtatán dialect, under two principle sources. First, some of the data come from original fieldwork conducted by two of the authors, Cristina Buenrostro and Justin Royer, using a theoretically-driven fieldwork methodology (Matthewson Reference Matthewson2004, Bowern Reference Bowern2008, Tonhauser and Matthewson Reference Tonhauser and Matthewson2016, Bochnak and Matthewson Reference Bochnak and Matthewson2020). Second, other data come from segments of Mateo Pedro and Coon’s (Reference Mateo and Jessica2018) collection of Chuj narratives, available on the Archive of the Indigenous Languages of Latin America (AILLA; ailla.utexas.org). Examples taken from texts are indicated with “txt”.
2.2 Basic clausal syntax and (non-)verbal predication
Like other Mayan languages (England Reference England2001, Coon Reference Coon2016a, Aissen et al. Reference Aissen, England and Maldonado2017), Chuj is an ergative-absolutive, head-marking language, and exhibits predicate-initial word order in discourse neutral contexts. An example showing these properties is provided below:
We follow Mayanist tradition in referring to ergative morphemes as “Set A” (which also track possessors in the nominal domain), and absolutive morphemes as “Set B”. Notice that Set B is not represented in (7), because third person Set B (singular and plural) has no overt phonological manifestation; examples with overt Set B are provided below:
Note that both verb stems in (8) bear “status suffixes” (-a’ and -i in the above examples), which indicate information about transitivity and mood in Mayan (Coon Reference Coon2016a, Aissen et al. Reference Aissen, England and Maldonado2017). As apparent in (7), some status suffixes disappear in certain environments; these only appear when they are at the right edge of intonational phrases or when their absence would lead to illicit consonant clusters (Royer Reference Royer2022b).
Nonverbal predicates (NVP), which contrast with verbal predicates in systematically lacking tense-aspect morphology (Grinevald and Peake Reference Grinevald, Peake, Gilles Authier and Haude2012, Coon Reference Coon2016b), also appear at the beginning of the sentence. This is the case regardless of whether the predicate is nominal (9a), adjectival (9b), stative (9c), or existential (9d). Chuj does not have an overt copula.
2.3 Topic and focus
While Chuj exhibits predicate-initial word order in discourse neutral contexts, preverbal arguments are very common. This fact will prove relevant for the purposes of the current article, since, as will be shown later, a subset of Chuj quantificational expressions must appear in sentence-initial position in Chuj. An understanding of the Chuj left periphery is thus necessary to properly diagnose the syntax of its quantifiers.
As has been widely noted for many languages of the Mayan family (see e.g. Aissen Reference Aissen1992, Reference Aissen, Aissen, England and Zavala Maldonado2017b, Reference Aissen2023; Coon Reference Coon2016a), the left periphery is generally used to signal information structural statuses of the elements of the clause. In Chuj, there are two main types of preverbal arguments: foci and topics. We discuss both in turn below.
Topics in Chuj show the following properties: (i) occurrence at the left periphery, (ii) obligatory coindexation with a postverbal resumptive classifier pronoun (as is the case in other Q’anjob’alan languages; see e.g. Craig Reference Craig1977, Datz Reference Datz1980, Aissen Reference Aissen, Andrew Carnie and Guilfoyle2000), and (iii) the presence of a significant prosodic break immediately after the topic (indicated in examples with a comma; see Royer Reference Royer2022b on the prosody of topics in Chuj). Examples of subject and object topics are provided below:
As further shown in (10), topics must occur with the marker ha, which we follow Bielig (Reference Bielig2015) in glossing “pv” for ‘preverbal DP marker’.Footnote 3 We use this term because foci, which we turn to now, also require this morpheme.
While foci also appear at the left periphery with ha, two morphosyntactic facts make it possible to distinguish them from topics. First, unlike topics, foci never trigger resumption. This is shown for both subject and object foci in (11). Second, focused transitive subjects, as in (11a), require a special verb form called “Agent Focus” (see e.g. Coon et al. Reference Coon, Pedro and Preminger2014; Aissen Reference Aissen, Jessica Coon, Massam and Travis2017a, Reference Aissen, Aissen, England and Zavala Maldonado2017b). While such verbs always take two semantic arguments, they are syntactically intransitive. In particular, they (i) lack Set A (ergative) agreement, (ii) show Agent Focus voice morphology on the verb stem (-an), and (iii) appear with an intransitive status suffix (11a) (visible only when the verb is final in the intonational phrase; Royer Reference Royer2022b):
In discussing foci, it is important to distinguish between cases of ‘contrastive focus’ versus ‘new information focus’. As has been observed for a range of other Mayan languages (see e.g. Aissen Reference Aissen, Aissen, England and Zavala Maldonado2017b and citations therein), contrastively focused arguments must appear at the left periphery. In other words, they cannot be focused in situ. As shown below, Chuj is no exception. Contrastively focused DPs cannot felicitously be interpreted in their in situ postverbal position:
While DP arguments must appear preverbally, there is more flexibility with regards to the position of non-arguments. PPs, for example, can be interpreted with contrastive focus when occurring postverbally:
The same is also true of possessor DPs in possessive phrases:
Previous work on focus in Mayan distinguishes contrastive foci from new information foci, as they commonly show different syntactic profiles. In particular, this work shows that, contrary to arguments with contrastive foci, in situ arguments can be interpreted (to varying degrees) with new information focus. Some languages, such as Tsotsil (Aissen Reference Aissen, Aissen, England and Zavala Maldonado2017b), freely allow in situ arguments to have new information focus; others, such as K’iche’ (Velleman Reference Velleman2014) and Yucatec Maya (Verhoeven and Skopeteas, Reference Verhoeven and Skopeteas2015), allow all types of arguments to have new information focus in situ, except external arguments, which must be in the same preverbal focus position as contrastively-focused expressions.
Interestingly, Chuj seems to show a third possibility: as far the morphosyntax is concerned, all types of foci are treated equally. This is shown in the following question-answer pairs, a canonical diagnostic for new information focus. For both external (16) and internal (17) arguments, we see that new information foci must appear pre-verbally:
Note that the parentheses in (16b) and (17b) indicate optionality. In other words, even when providing a fragment answer, speakers require the use of ha, a marker which essentially signals occurrence at the left periphery. This suggests that new information focus in Chuj, at least for arguments, must always be expressed by means of a focus construction.
As with contrastive foci, new information foci can sometimes remain in situ, specifically when the focused element is not a main argument of the verb. Again, this is the case for PPs (18) and possessors (19). In the case of possessors, displacement of the possessor triggers pied-piping of the entire possessive phrase (19c).
It is also worth noting that ix Xuwan, without ha, can serve as a fragment answer to the question in (19), offering further support that the possessor does not need to be moved.
In sum, the distribution of focused DPs in Chuj – regardless of whether the focus is interpreted as contrastive or new information – can be summarized as follows:
Finally, it is important to note that the Chuj focus position also hosts wh-expressions in interrogative clauses (21a), nominal expressions modified by the focus-sensitive item nhej ‘only’ (21b), and the head of relative clauses (21c). As in other languages, wh-expressions must be in focus position in order to trigger interrogative meaning, even in echo questions (see also AnderBois Reference AnderBois, Aissen, England and Maldonado2017 and Caponigro et al. Reference Caponigro, Torrence and Maldonado2021 on other Mayan languages).Footnote 5
Turning now briefly to the relative syntactic positions of topics and foci, we note that their order is not free. As shown below, topics are always ordered before foci (Bielig Reference Bielig2015).
Building on Aissen (Reference Aissen1992) and Bielig (Reference Bielig2015), and based on the data seen above, we adopt the extended syntax in (23). Specifically, we assume that topics are base-generated in a peripheral position to the left.Footnote 6 Foci, on the other hand, are A
$'$-extracted into the focus position (see Coon et al. Reference Coon, Pedro and Preminger2014, Reference Coon, Baier and Levin2021 and Aissen Reference Aissen, Aissen, England and Zavala Maldonado2017b on this analysis of Mayan focus):
Having provided a description of Chuj clause structure, we are now in a position to provide a syntactic description of its quantifiers. As highlighted in the introduction to this article, we will show that distinct quantifiers must appear sentence-initially for different reasons: while some are sentence-initial because they are nonverbal predicates, others must appear sentence-initially due to a requirement that they appear in focus position.
3. Sentence-initial quantifiers: D-quantifiers vs A-quantifiers
We now turn to the main topic of this paper: the fact that a subset of Chuj quantificational items seem, at least on the surface, to require their host argument to appear at the clause’s left periphery.Footnote 7
Recall the first two examples from the introduction, repeated below.
While the data in (24) and (25) show quantified subjects, we also note here that the exact same facts hold of quantified objects. Data comparable to those in (24) and (25), but with tzijtum and masanil now quantifying over the object, are provided below:
Again, the fact that these quantificational expressions cannot appear in postverbal position is unexpected. As seen in the previous section, Chuj is a verb-initial language. Moreover, notice the presence of the Agent Focus morpheme -an on the verb in both (24) and (25). Keeping to our analysis of the Chuj focus construction in section 2, this means that the agent DP – i.e., the one being quantified over – must have been A
$'$-extracted. What, then, explains the fact that these expressions must be sentence-initial?
It is important to mention at this point that while the utterances like (24b) or (26b) (with the quantifier tzijtum) are consistently judged as unacceptable by our collaborators, this is not clearly the case for the sentence in (25b) and (27b) (with the quantifier masanil). We have chosen to mark the utterances in (25b) and (27b) as “strongly degraded” for the following reasons. First, speakers do not themselves offer postverbal arguments modified by masanil, even when discourse neutrality is controlled for. Second, when prompted with sentences like (25b) or (27b), speakers vary considerably in their judgements. For example, when asked to judge the acceptability of a sentence like (25b) on a scale of 1 to 5 (1 = clearly unacceptable, 5 = perfectly acceptable), most judge the utterance as a “2” or “3”. However, depending on the speaker, judgements have varied from “1” to “5”. We have also not been able to identify contexts that would make the use of postverbal masanil more pragmatically acceptable. The judgements are therefore much weaker than with tzijtum, which consistently receives a judgement of “1” when appearing postverbally (as in (24b) or (26b) above).Footnote 8
Corpus findings also support the idea that masanil is strongly preferred sentence-initially. The results in Table 1 were extracted from a collection of 17 narratives from different speakers of the San Mateo Ixtatán dialect of Chuj (Mateo Pedro and Coon Reference Mateo and Jessica2018). As the table shows, when masanil modifies a DP subject or object, it is consistently in a preverbal position. When modifying oblique arguments, on the other hand, both preverbal and postverbal positions are attested. As we will see, this will be an important point of difference between masanil ‘all’ and tzijtum ‘many’, since the two behave differently with regards to oblique arguments.
Position of masanil within corpora of San Mateo Ixtatán Chuj
We are now ready to tackle the question of why tzijtum ‘many’ and masanil ‘all’ must appear sentence-initially. Given the Chuj background in section 2, we can entertain at least two analytical possibilities. On the one hand, it could be that (24) and (25) contain A-quantifiers, specifically expressed via nonverbal predication. Quantification expressed via nonverbal predication is common across the world’s languages, especially for Indigenous languages of North America (see e.g. Davis and Matthewson Reference Davis, Matthewson, Daniel Siddiqi, Barrie, Gillon, Haugen and Mathieu2019 and references therein). Under such a theory, the sentence-initiality of the quantifier would fall out immediately from the fact that Chuj is a predicate-initial language. The presence of Agent Focus morphology on the verb stem would result from agent relativization:
The second option would be one in which the quantifier is a D-quantifier, and thus part of a DP. Under this account, such a DP would, for some reason or other, have to be in focus position. The behaviour of the quantifier in this case would be reminiscent of the behaviour of wh-items, which also trigger phrasal displacement:
In the rest of this section, we argue that despite their superficial similarities, the quantifiers in (24) and (25) do not in fact belong to a homogeneous class. Based on several syntactic diagnostics, we argue that both options in (28) and (29) are needed, but for different quantificational expressions. Our main proposal is the following:
The discussion is structured around the predictions made by each analytical possibility, starting with predictions of predicative status in section 3.1, and turning to those of DP status in section 3.2.
3.1 Quantification and nonverbal predicate status
Analysis 1 in (28) predicts that the quantifier should behave as a nonverbal predicate. Therefore, a first diagnostic to consider is whether the respective quantificational expressions can serve as the main predicate of a simple predicative clause in Chuj. Having proposed that only tzijtum, but not masanil, is a nonverbal predicate, we make the following prediction:
Here, we discuss three diagnostics to identify nonverbal predication in Chuj. We show that in each case, tzijtum checks these diagnostics, whereas masanil does not.
A first diagnostic to consider is the ability to appear as the main predicate of a simple NVP clause. Recall from section 2 that Chuj does not have an overt copula, meaning that NVPs can combine with a DP to form a complete sentence. Two examples, with nominal (32a) and adjectival (32b) predicates, are repeated below for illustration:
Now, if tzijtum is a NVP, it should also be able to co-occur alone with a DP to form a complete sentence. As shown in (33), tzijtum can be used as such. In both occurrences in (33), the quantifier combines alone with a nominal expression to yield an utterance that literally translates as ‘they were many’.
The quantificational expression masanil, on the other hand, cannot be used alone with a DP to form a truth-conditional sentence. Indeed, the string of words in (34a) cannot be used to express something along the lines of ‘all of them are here’. Our collaborators converge on the intuition that (34a) is somehow ‘incomplete’. We assume this is because (34a) is not a full sentence. Indeed, to form a full sentence, an existential predicate must be used separately (34b).
A second sentence type to consider are possessive sentences. As in other Mayan languages (Coon Reference Coon2016a), possessive existential constructions in Chuj (the equivalent of possessive ‘have’ in English) are derived by combining an existential predicate with a possessed nominal, as shown in (35) (see also Freeze Reference Freeze1992 on this strategy more generally).
Now, consider the data in (36). The quantifier tzijtum can replace the existential predicate to establish the possessive ‘have’ relation. This again suggests that it itself can be a NVP.
On the other hand, while the string of words with masanil in (37a) is not in and of itself ungrammatical, it does not form a complete sentence. Something is missing: our consultants consistently translate this string of words as ‘all of your sons’. To express a possessive ‘have’ relation, an existential predicate is required (37b):
Finally, there is a third, Chuj-specific, prediction of a predicational analysis of quantifiers to be considered. Mayan languages of the Q’anjob’alan sub-branch feature complex secondary predication constructions (see e.g. Mateo Toledo Reference Toledo2012). Such constructions involve complex clauses, which combine a secondary nonverbal predicate with an aspectless clause. An example from Q’anjob’al, a language closely related to Chuj, is provided below:
(Q’anjob’al, Mateo Toledo Reference Toledo2012: 152)
Chuj also makes extensive use of such constructions. In (39a), we see that the adjective junk’olal ‘content’ can be used as a secondary predicate. The same kind of template can also be used with tzijtum ‘many’ (39b), as expected if tzijtum is a NVP:
The quantifier masanil, on the other hand, cannot serve as the NVP of secondary predicate constructions. The following string of words is simply ungrammatical in Chuj (no context can make the sentence acceptable):
In sum, while tzijtum checks all diagnostics for predicatehood, masanil fails them all.Footnote 9 This supports the idea that while tzijtum is a NVP, and therefore an A-quantifier, masanil is not. We now turn to diagnostics of DP status, showing that masanil, but not tzijtum, shows all signs of being part of a DP, further supporting our proposal in (30).
3.2 Quantification and DP status
Here, we discuss evidence in support of “DP status” in Chuj, showing that while masanil is syntactically part of the DP that it quantifies over, tzijtum is clearly external to it. We discuss three predictions: (i) the (in)ability to appear within the complement of a preposition, (ii) the (in)ability to appear as part of the possessor of a possessive DP, and (iii) the (in)ability to be a topic.
One way to identify DP status in Chuj is to check whether an expression can serve as the complement of a preposition. Prepositions in Chuj can only select for nominal expressions (DPs or NPs), and not for verbal and nonverbal predicates. If tzijtum is consistently a NVP, we therefore predict that it should not be able to appear within the complement of a preposition. On the other hand, if masanil is a D-quantifier, it should be able to appear in this position. This prediction is borne out. Let us first consider an example with masanil:
The above examples show that masanil can appear within the complement of the preposition t’a. Moreover, notice that the PP can be either preverbal or postverbal, a fact that aligns with the corpus findings we report in Table 1 above. This optionality will be relevant when sketching a syntactic analysis of masanil in section 5.
Turning now to tzijtum, we find that this quantifier cannot appear within the complement of a preposition, regardless of the position of the PP in the clause.
To convey the intended meaning, a construction containing a relative pronoun must instead be used. Crucially, this relative pronoun can only be used in cases where an oblique DP is being relativized. This suggests that the nominal expression chonhab’ has been relativized in (43), yielding a literal meaning along the lines of ‘the villages where we went are many’.
Again, these data follow if tzijtum is a NVP. Here, its sole argument is simply a relativized oblique argument. While masanil can appear in a similar string of words, it again does not convey a complete sentence (44a). To form a complete sentence with this string of words, a predicate is needed (44b). (Note that the predicate is in final position, because the masanil constituent is focused here).
A second way to identify DP status in Chuj is to test whether the relevant expression can be found within the possessor of a possessive phrase, which is restricted to DPs (and not verbs/NVPs). Given our proposal, we predict here that only masanil should be possible within a DP possessor, and not tzijtum. As shown below, this prediction is borne out for masanil. (Again, notice that when inside a possessor, the argument containing masanil can remain in its in situ postverbal position).
The prediction is also borne out for tzijtum, which cannot appear within the possessor of a possessive phrase (46a). Instead, to convey the intended meaning, tzijtum must appear sentence-initially (46b). Here, it appears to select for a relativized possessive DP; the literal translation being ‘the women that are such that I know their partner are many’:
A third and final way to identify whether a quantifier is internal or external to the extended nominal domain is to test whether it can appear as a topic. Indeed, as far we know, only DPs may serve as topics in Chuj. Therefore, if (30) is on the right track, only masanil should be able to occur as a topic. Again, this prediction is borne out. As shown in (47), masanil can felicitously occur as part of a DP that occupies a topic position.
The example in (47) shows most of the properties of topichood in Chuj discussed in section 2: (i) the topic appears at the left, (ii) it is immediately followed by a considerable prosodic break (indicated with a comma here), and (iii) it is coindexed with a postverbal resumptive pronoun, in this case heb’ ix ‘they’. The only relevant exception is that there is no preverbal marker ha. While we come back to this fact in section 5 below, we assume that masanil is indeed part of a topicalized DP in (47).
The quantifier tzijtum, on the other hand, is not possible as a topic. This is shown in (48a). To convey the intended meaning of (48a), the quantifier must occur after the topicalized DP, as in (48b). In this example, the resumptive pronoun is in the position of a relativized DP; in other words, the sentence literally translates as ‘As for the girls
$_1$, those
$_1$ who read the Popol Wuj were many’.
Again, the above data follow from the predicative analysis of tzijtum: if the quantifier is a NVP, it is expected that (i) it should not be able to be topicalized with the DP it quantifies over and (ii) it should be able to itself serve as the main predicate of a sentence in which a DP is topicalized, as is the case in (48b).
3.3 Summary
This article started by showing a surface similarity between the quantifiers tzijtum ‘many’ and masanil ‘all’, insofar as both tend to be found in sentence-initial positions, at least when they quantify over main arguments of the verb. In this section, we have shown that this similarity is only apparent: while tzijtum is an A-quantifier instantiated as a nonverbal predicate, masanil is a D-quantifier. The diagnostics we used to draw this distinction are summarized in Table 2 below.
Testing for NVP vs DP status
Having established that there are two types of “sentence-initial” quantifiers in Chuj, the next sections sketch and motivate syntactic analyses of each. As we will show, both types of quantifiers require an understanding of the left periphery, since either relativization, focus movement, or topicalization is required to derive their distribution. Section 4 starts with an analysis of A-quantifiers. Section 5 then turns to an analysis of D-quantifiers.
The next sections will also make two additional empirical contributions. First, we will show that neither tzijtum nor masanil are unique in their category, by identifying additional quantifiers of each type. Second, section 5 will identify a second category of D-quantifier, which contrary to masanil, is perfectly acceptable in postverbal position.
4. A-quantifiers as nonverbal predicates
This section sketches an analysis of A-quantifiers as nonverbal predicate (NVPs) in Chuj. The appendix provides an inventory of NVP quantifiers, which we claim all have the syntax discussed here.
While the verbal domain in Mayan has been thoroughly studied, there are relatively few analyses of nonverbal predication (though see Mateo Toledo Reference Toledo2012, Coon Reference Coon, Hsin-Lun Huang, Poole and Rysling2014, Armstrong Reference Armstrong2017, and Coon and Martinović Reference Coon, Martinović, Suet-Ying Lam and Ozaki2023). Here, we adopt Coon’s (Reference Coon, Hsin-Lun Huang, Poole and Rysling2014) analysis of NVPs in a closely related language, Ch’ol. Coon builds on parallel data in Austronesian languages to propose that the subjects of NVPs in Ch’ol are always instantiated as internal arguments (building on Sabbagh Reference Sabbagh2011), rather than external arguments (contra analyses of English-type predication in work such as Bowers Reference Bowers1993, Baker Reference Baker2003, and den Dikken Reference den Dikken2006). This proposal is schematized below, with only the relevant syntactic pieces shown:Footnote 10
While we refer readers to Coon (Reference Coon, Hsin-Lun Huang, Poole and Rysling2014) for extensive argumentation, we note here that the subjects of NVPs behave syntactically like internal arguments, rather than external arguments. In Ch’ol, sub-extraction out of possessive DPs is possible when the possessive DP is an unaccusative subject or transitive object, but not when it is a transitive subject:Footnote 11
(Ch’ol; examples from Coon Reference Coon, Hsin-Lun Huang, Poole and Rysling2014: 80-81)
Coon shows that the subjects of NVPs pattern with the unaccusative subjects and transitive objects of the verbal predicates in (50) – and therefore internal arguments. This supports the view that they occupy the internal argument (theme) position, as in (49).
(Ch’ol; examples from Coon Reference Coon, Hsin-Lun Huang, Poole and Rysling2014: 81)
Extending this syntax for NVPs to Chuj, we now sketch an analysis of A-quantifiers. A syntax for a basic NVP clause with tzijtum is provided below. Here, tzijtum is instantiated as a NVP which takes the pronoun heb’ ix ‘they (feminine)’ as its internal argument.
As for sentence (1), repeated below, which drew our attention at the beginning of this article, we propose that the subject of the NVP in this case is a relativized agent. Recall from section 2 that relativization of transitive subjects in Chuj gives rise to Agent Focus morphology on the verb stem. This analysis therefore derives the obligatory use of an Agent Focus construction whenever a quantifier like tzijtum quantifies over the agent:
Since this analysis relies on relativization, it predicts that whenever tzijtum quantifies over a transitive subject, an Agent Focus construction will be required (Coon et al. Reference Coon, Pedro and Preminger2014, Reference Coon, Baier and Levin2021). As illustrated below, this is indeed the case: a regular transitive verb without Agent Focus morphology cannot be used as an alternative to (53a).
This concludes our analysis of A-quantifiers as NVPs in Chuj. Importantly, we have been able to answer some of our initial questions in this article. One was why the relevant quantifier needed to appear in a sentence-initial position. The answer is straightforward: tzijtum is generally sentence-initial, because Chuj is a predicate-initial language. Another question was what kinds of movement operations are involved in the derivation of sentences containing quantifiers like tzijtum. Here, we argued that more complex sentences with tzijtum, such as the one in (1)/(53a), involve a process of relativization. This also explains why Agent Focus morphology is required whenever tzijtum quantifies over a transitive subject. As we will see in the next section, this is crucially different from sentences with masanil, which do not necessarily trigger the Agent Focus construction.
Before moving on to the next section, however, we note that a list of quantifiers which we have identified as being part of the same category as tzijtum is provided in the appendix. Interestingly, all of these quantifiers belong to the semantic class of “vague” (Partee Reference Partee, Joyce Powers and de Jong1989) or “value judgement quantifiers” (Keenan and Paperno Reference Keenan and Paperno2012) such as ‘many’ or ‘few’, whose interpretation is dependent on a contextually-determined standard of comparison. While we do not provide explicit examples with each quantifier, any of these quantifiers (except the “mass” quantifiers, which require a mass noun) could be inserted instead of tzijtum in the previous examples.
5. Analyzing two types of D-quantifiers
In this section, we turn to an analysis of Chuj quantifiers of the masanil type, the second class of quantifier that generally appears in a sentence-initial position in Chuj (as in (2)). In section 3.2 we established that masanil is a D-quantifier, occurring in the extended projection of DPs. This finding leads to the question of why DPs containing masanil occur in a sentence-initial position when they are main arguments of the verb.
Before tackling this question, we first establish a crucial empirical fact, namely that not all D-quantifiers in the language are subject to this requirement. That is, section 5.1 first provides a description of a third type of Chuj quantifier: D-quantifiers that do not have any effect on the syntactic position of their host argument, which we will call “Basic D-quantifiers”. We then turn in section 5.2 to an analysis of quantifiers like masanil ‘all’, which we call “Focus D-quantifiers”, arguing that these are differentiated from the basic ones insofar as they introduce a focus [A
$'$]-feature, generally targeted by a probe on Foc
$^0$.
5.1 Basic D-quantifiers
While masanil generally prefers its host argument to appear at the left periphery, not all D-quantifiers in Chuj show such a requirement. Consider, for instance, the different uses of jantak in (55). As indicated in translations, jantak can mean ‘many’ or ‘all’, depending on the context:Footnote 12
As shown above, jantak can be postverbal when modifying the subject (55a), object (55b), or a DP contained within an oblique phrase (55c). This is strikingly different from quantifiers like masanil ‘all’, which generally force DP arguments to appear at the left periphery.
A unifying property of Basic D-quantifiers in Chuj is that their arguments may serve as the sole, postverbal argument of a NVP. Minimal pairs with pitzan ‘awake’ are provided below:
This, of course, is not an environment in which a NVP quantifier, such as tzijtum, may be found (57). That is, since tzijtum is itself a NVP, it cannot co-occur with ptizan ‘awake’:
As for the quantifier masanil, it is again preferred in pre-predicate position:
Finally, we note that arguments with Basic D-quantifiers pattern with those without quantifiers in being able to serve as foci or topics. In such cases, they show the same morphology generally found on focused and topicalized nominal expressions, discussed in section 2.3. We highlight here the presence of the preverbal marker ha, crucially absent from sentences in which masanil DPs are preverbal. (We return to this point shortly in the next subsection.)
For completeness, we extend the analysis of indefinite quantifiers in Royer (Reference Royer2022a) to all quantifiers in (56), which proposes that the indefinite quantifiers jun and juntzanh in Chuj are D heads:
Having established that Chuj possesses D-quantifiers that do not exhibit any constraints on appearing at the left periphery, we now turn to an analysis of masanil DPs.
5.2. Focus D-quantifiers
In section 3.1, we presented several empirical arguments to demonstrate that masanil ‘all’, contrary to other quantifiers like tzijtum ‘many’, is a D-quantifier: a quantifier that arises internal to the extended nominal domain. Crucially, however, masanil sometimes requires the expressions it quantifies over to appear at the left periphery. In section 5.1, we showed that Chuj possesses D-quantifiers that do not exhibit this requirement. Here, we sketch an analysis of masanil DPs, which accounts for this point of variation among D-quantifiers.
Recall the main distributional conditions on masanil ‘all’:
The facts in (61) should be reminiscent of the distribution of foci in Chuj, whose distributional conditions are repeated below from (20):
We therefore propose that masanil, at least when used as a D-quantifier (see footnote 7 above), is a focus-sensitive item. Building on previous work (see e.g. Cable Reference Cable2010; Coon et al. Reference Coon, Baier and Levin2021, Hedding Reference Hedding2022, and Branan and Erlewine Reference Branan and Erlewine2023 for different formalizations of this particular feature), we tie this to a feature in the syntax of the DP hosting the quantifier, which we refer to as “[q]” here, after Cable’s (Reference Cable2010) approach to question particles:
While providing a formal semantic account of masanil falls outside the scope of this article, we note here that the proposal that universal quantifiers like masanil should be associated with a focus-sensitive feature has precedent. Brisson (Reference Brisson2003), for instance, analyzes all in English as an item that requires all relevant individuals in the domain of quantification to be considered, and therefore introduces domain alternatives. Another precedent is Zeijlstra (Reference Zeijlstra2017), which, building on the analysis of PPIs in Chierchia (Reference Chierchia2013), argues that some universals introduce alternatives. Zeijlstra specifically argues that universal quantifiers introduce alternatives when they are positive polarity items that must take wide-scope over negative operators. Our preliminary data suggest that masanil must also take scope over negation. The data in (64) are from elicitation. Example (65) was taken from a text, in which the preceding dialogue makes it clear that there is nothing that the speaker can buy (and hence the universal takes wide scope over negation):
We refer readers to Zeijlstra (Reference Zeijlstra2017) for a detailed formal implementation of this analysis of universal quantification. However, as far as we can tell, this analysis could be extended to account for the semantics of masanil.
On the morphosyntactic side, recall that masanil is different from other D-quantifiers insofar as it does not combine with the “preverbal marker” ha.Footnote 14 Our corpora confirm this fact: we have not found a single sentence in which ha and masanil co-occur.
We hypothesize that the particle ha occupies the highest position in the extended nominal domain, which, following Cable (Reference Cable2010), we label as “QP”. This idea, schematized in (67), builds on previous work that proposes that A
$'$-movement always results from movement of a phrase containing a particle (see e.g. Cable Reference Cable2010; Hedding Reference Hedding2025). This particle could be a wh-particle (Cable Reference Cable2010), or a specific focus-sensitive particle like ha (cf. Branan and Erlewine Reference Branan and Erlewine2023), but we assume that it could also be lexicalized as part of specific focus-sensitive expressions, such as wh-words or quantificational expressions (Hedding Reference Hedding2025). In other words, we suggest that masanil, as an alternative-sensitive item, occupies the same syntactic position as the particle ha (68), explaining why the two do not co-occur. This is schematized below. (Building on the analysis of maximal definites in Royer Reference Royer2022a, we assume that the head of DP in this case is the operator ι ‘iota’.)
Turning to an analysis of Focus D-quantifiers inside clauses, we propose that these consistently enter into Agree with a Q:foc-probe on Foc
$^0$, by virtue of carrying a [Q:foc] feature.Footnote 15
In short, then, masanil as a Focused D-quantifier bears an inherent [A
$'$]-feature, modeled as the Q-feature of Cable (Reference Cable2010), which requires it to move to the left periphery of clauses in Chuj.
With our analysis of Focus D-quantifiers sketched above, we conclude this section by highlighting three final points about the distribution of masanil DPs and their implications for our syntactic analysis. A first point to be addressed is the availability of Focus D-quantifiers to occur in topic position. That is, we saw in section 3.2 that masanil DPs can be used as topics in Chuj. Assuming, as proposed in section 2.3, that topics are base-generated in a high peripheral position, we suggest that masanil has a non-focused counterpart which occurs in such contexts but still bears a non-focus Q feature. One simple argument for this conclusion is the observation that the particle ha is also used to introduce topics; this particle now can be seen as a general-purpose Q head. In discussing this possibility of Q features in topicalization structures, Cable (Reference Cable2010: 232–3) notes that such Q features would necessarily lack the alternative-sensitive semantics otherwise associated with Q heads. One way to make sense of the finding that ha (and masanil) can occur in both topic and focus position would be to analyze them as only morphologically specified to realize [Q] features in Chuj; while moved focus phrases possess foc or alt in the syntax, the topic phrases clearly do not. Nevertheless, the vocabulary items associated with ha and Focus D-quantifiers such as masanil are compatible with different subtypes of Q, as they realize only the Q feature itself.
Second, something must be said about those cases in which a Focus D-quantifier does not have to move to the focus position. That is, we saw in section 3.2 that when masanil DPs arise in oblique or possessor position, they can remain in situ. We suggest that in such cases, the relevant DPs are simply unavailable to the [A
$'$]-probe. For possessors, for instance, this seems to be mostly the case in the language: sub-extraction out of DPs is highly constrained in Chuj, at least when it comes to the left-peripheral focus position (this is contrary to other Mayan languages, such as Ch’ol; see also discussion in Royer Reference Royer2025):Footnote 16
This restriction can be explained by simply positing that the possessive DP is a phase, and thus that the posessor cannot be extracted from out of it.
For PPs, we could again allude to the fact that PPs are phases, and therefore a QP cannot be targeted by Foc
$^0$. Alternatively, it could be that PPs are simply not targetable by the Probe on Foc
$^0$. For instance, Coon et al. (Reference Coon, Baier and Levin2021) propose that the focus [A
$'$]-probe in Mayan is relativized to probe specifically for DPs (which would translate here as “QPs”). In other words, PPs could therefore simply not be of the right syntactic category to be targeted by the probe in (69). The fact that they can optionally appear to the left of the predicate would then have to be derived without resorting to movement to the specifier of Foc
$^0$. This is independently argued by Royer (Reference Royer2025) for Chuj to account for patterns of syntactic binding: arguments coindexed with main arguments of the verb that are contained within oblique phrases do not reconstruct for binding (see Royer Reference Royer2025: section 3.4).
A final case to consider is what happens when two foci co-occur within the same sentence. This is particularly relevant, as it has been claimed that multiple foci are not possible within a same sentence across several Mayan languages, a fact which correlates with the unavailability of multiple wh-questions (Aissen Reference Aissen1996, Curiel Ramírez del Prado Reference Curiel Ramírez del Prado, Aissen, England and Maldonado2017, AnderBois and Chan Dzul Reference AnderBois, Dzul, Ivano Caponigro, Torrence and Maldonado2021, Coon et al. Reference Coon, Baier and Levin2021; Mateo Toledo Reference Toledo, Ivano Caponigro, Torrence and Maldonado2021; Polian and Aissen Reference Polian, Aissen, Ivano Caponigro, Torrence and Maldonado2021; Royer Reference Royer, Ivano Caponigro, Torrence and Zavala2021; Vázquez Álvarez and Coon Reference Álvarez, Jesús, Coon, Ivano Caponigro, Torrence and Maldonado2021). Chuj also bans multiple wh-questions:
If masanil DPs in argument position must generally move to the specifier of Foc
$^0$, we might expect ineffability to arise when they co-occur with wh-items. This expectation is not clearly borne out, however. Speakers vary in whether they are willing to accept the following kind of sentence:
Nevertheless, all speakers we have consulted indicate a preference for the following type of construction to express the intended meaning in (72). In fact, when prompted to translate this target sentence, this is the construction that is generally given:
As noted in footnote 7 above, we follow Little (Reference Little, John R. Starr, Kim and Oney2022) in assuming that the quantifier in this case is nominalized, and adjoined as a possessive phrase whose possessor is a null pronoun coindexed with one of the main arguments of the verb:
If these uses of masanil are in fact of category N, as they can be possessed, then they are either inaccessible to the Q-probe on Foc or they lack a Q feature all together, and may instead simply have an exhaustifying lexical meaning such as ‘entirety, totality’.
As for the speakers for whom non-fronted uses of masanil in (72) are acceptable, we have already seen evidence from their ability to occur in topic sentences that some uses of masanil do not seem to require the presence of a foc-specified Q head. If such non-focused Q-variants of masanil are generally available when it occurs as a topic, then we can assume that speakers may generally resort to these non-focus variants of masanil when they occur in argument positions. In fact, this claim can help make sense of one of the original observations made with respect to masanil, which is that their non-fronted variants, as in (2), are not fully ungrammatical but degraded. Perhaps this is due to speakers’ ability to accept the non-focused variant of masanil, albeit reluctantly, in argument positions.
5.3 Summary
This section identified and provided an analysis of two types of D-quantifiers in Chuj: Basic D-quantifiers, which have no effect on the syntactic distribution of nominal arguments, and Focus D-quantifiers, which require arguments to appear in a preverbal position. We proposed that while D-quantifiers that show no left-peripheral restriction are instantiated in the head of DP, D-quantifiers instantiate the head of QP: a phrase usually containing focus sensitive particle that enters into Agree with focus operators at the periphery of the CP domain.
While we only focused on masanil DPs here, we note that one other quantifier is like masanil by being a Focus D-quantifier. This expression is built from the lexical item yalnhej and any wh-expression except tas yuj ‘why’. Yalnhej+wh quantifiers are discussed at length in Kotek and Erlewine (Reference Kotek and Erlewine2019) and Alonso-Ovalle and Royer (Reference Alonso-Ovalle, Royer, Nicole Dreier, Kwon, Darnell and Starr2022, Reference Alonso-Ovalle and Royer2024). This quantifier is used as a modal indefinite, conveying speaker indifference or ignorance. As discussed in Alonso-Ovalle and Royer (Reference Alonso-Ovalle and Royer2024), arguments modified by yalnhej+wh quantifiers are strongly preferred in preverbal position:
Yalnhej-wh quantifiers check all of the same syntactic diagnostics as masanil: (i) they can co-occur with prepositions, (ii) they can serve as the possessors of DPs, and (iii) they can be topicalized. Critically, they also do not co-occur with ha. Being modal indefinites, it is unsurprising that yalnhej-DPs should be associated with a domain alternative feature: based on an in-depth semantic analysis of these quantifiers, Alonso-Ovalle and Royer (Reference Alonso-Ovalle, Royer, Nicole Dreier, Kwon, Darnell and Starr2022, Reference Alonso-Ovalle and Royer2024) specifically argue that they evoke alternatives (see also, e.g., Kratzer and Shimoyama (Reference Kratzer, Shimoyama and Otsu2002) and Alonso-Ovalle and Menéndez-Benito (Reference Alonso-Ovalle and Menéndez-Benito2018) on analyses of similar modal indefinites).
6. Conclusion
This paper has identified, described, and analyzed different classes of quantificational expressions in Chuj, an understudied Mayan language. In doing so, we have provided answers to the main questions posed at the beginning of the paper, which were:
Q1 Why must some quantifiers appear in a sentence-initial position, but not others?
Q2 If any, what kind of movement operations are involved to derive the initial position of the quantifiers in sentences like (1a) and (2a)?
Q3 How are these quantifiers formally distinguished from other expressions that do not have to be sentence-initial?
Starting with Q1, we argued that despite initial appearances, the Chuj quantifiers that appear in sentence-initial position are not part of a homogeneous class. On the one hand, some quantifiers, such as tzijtum ‘many’, are “A-quantifiers”, syntactically manifested as nonverbal predicates. Since Chuj is predicate-initial language, it immediately followed that such quantifiers should generally be found in sentence-initial positions. On the other hand, we argued that other quantifiers, such as masanil all, were “Focus D-quantifiers”: D-quantifiers that carry a focus [Q] feature (Cable Reference Cable2010), and as such are generally targeted by a left peripheral [A
$'$]-Probe. An important empirical finding in the case of Focus D-quantifiers was that their distribution paralleled that of focused expressions in general: only main arguments must be displaced to the focus position.
This brings us to Q2, which is again relevant to both A-quantifiers and Focus D-quantifiers. In particular, while Focus D-quantifiers are targeted for A
$'$-movement whenever they form a part of one of the main arguments of the predicate, the derivation of sentences with NVP A-quantifiers also often required a step of A
$'$-movement. That is, to explain sentences like (1), we argued that NVPs often select for relativized arguments. Being “head-raising relative clauses” (Coon et al. Reference Coon, Pedro and Preminger2014, Reference Coon, Baier and Levin2021), such sentences involve A
$'$-movement of a relativized DP. More generally, the identification of a large class of A-quantifiers in Chuj brings to mind similar observations in other Indigenous languages of the Americas (see e.g. Davis and Matthewson Reference Davis, Matthewson, Daniel Siddiqi, Barrie, Gillon, Haugen and Mathieu2019 and references therein). As for Focus D-quantifiers, this type of quantifier adds to the growing body of work that has shown that certain kinds of quantifier can, much like wh-phrases in more widely-studied languages, trigger consistent phrasal movement of certain arguments (see e.g.
Kiss Reference Kiss and C. T1991; Szabolcsi Reference Szabolcsi and Anna Szabolcsi1997; Chung Reference Chung1998, Reference Chung and Lisa Matthewson2008; Barchas-Lichtenstein Reference Barchas-Lichtenstein, Keenan and Paperno2012; Ostrove Reference Ostrove2018).
Finally, with regards to Q3, we argued that Focus D-quantifiers should be formally distinguished from other DPs that do not have to appear in sentence-initial position. As we showed in section 5, this includes a large class of D-quantifiers, which are most often found in in situ postverbal positions. Building on Cable (Reference Cable2010), Branan and Erlewine (Reference Branan and Erlewine2023), and Hedding (Reference Hedding2025), we proposed that the main difference among these kinds of D-quantifiers amounted to the presence or absence of the focus feature, which only Focus D-quantifiers bear. This allowed us to identify a third class of Chuj quantifier, thereby contributing to the current limited understanding of quantification in Mayan (Henderson Reference Henderson2016).
While there remains much potential for future work, we end this article by highlighting two pressing issues. First, while we have mostly ignored wh-words – which also have quantificational force and appear in sentence-initial position – these have been treated like DP arguments in previous work on Chuj (see e.g. Kotek and Erlewine (Reference Kotek and Erlewine2019) and Royer (Reference Royer, Ivano Caponigro, Torrence and Zavala2021), and Coon et al. (Reference Coon, Baier and Levin2021) on Mayan more generally). As such, they could receive an analysis similar to the one proposed for masanil DPs. In fact, the analysis on which we based our account of Focus D-quantifiers (Cable Reference Cable2010) was originally designed to account for the behaviour of wh-words in questions. However, it is important to highlight that wh-quantifiers seem to exhibit a hybrid behaviour, at least with regards to the diagnostics we established in section 3. For instance, wh-words can serve as the main predicate of NVP clauses (76a). At the same time, they can also occur in oblique phrases (76b) (with pied-piping with inversion; see Aissen Reference Aissen1996). Future work should therefore address this hybrid behaviour.
A second pressing issue regards the pragmatics and semantics of Chuj foci and quantifiers, which we also mostly ignored in this paper. With regards to pragmatics, we showed in section 2.3 that Chuj differs from other Mayan languages in treating contrastive and new information foci alike (see e.g. Aissen (Reference Aissen, Aissen, England and Zavala Maldonado2017b) on other Mayan languages). On the semantics side, there are also important topics to be addressed. For example, our syntactic analysis of quantifiers like tzijtum as NVPs might be taken to predict that these should only have “cardinal” interpretations, as opposed to “proportional” interpretations (see, e.g., the discussion in Keenan and Paperno (Reference Keenan and Paperno2012); Davis and Matthewson (Reference Davis, Matthewson, Daniel Siddiqi, Barrie, Gillon, Haugen and Mathieu2019)). While we leave a discussion of these data for future work, we note here that upon preliminary investigation, this prediction does not seem to be borne out: while syntactically NVPs, A-quantifiers in Chuj seem to be compatible with both cardinal and proportional interpretations.
Acknowledgements
We are very grateful to the Chuj collaborators on this project: Petul Federico Felipe Gómez, Matin Pablo, Xuwan García, Kaxin Paíz, Matal Torres, Xapin Torres and Elsa Velasco. ¡Yuj wal yos t’ayex! For feedback on previous versions of this paper, we thank Jessica Coon, Carol Rose Little, Gilles Polian, Rodrigo Ranero and Roberto Zavala. We are also grateful to the editor of this volume, Éric Mathieu, as well as to two anonymous reviewers. For funding, we thank the Banting Fellowships program of the Social Sciences Research Council of Canada, the Instituto de Investigaciones Antropológicas of the Universidad Nacional Autónoma de México, as well as the UC Berkeley Oswalt Endangered Language Grant.
Appendix A. Inventory of Chuj quantifiers
Inventory of quantifiers over entities in San Mateo Ixtatán Chuj
Open access
