2.1. Vowels

Most Mayan languages have five vowels plus vowel length:

All of the Eastern languages except Kaqchikel, Mocho’ in the Western division, and all Yucatecan and Huastecan languages fit this pattern. Kaqchikel has converted what was possibly both a length and a quality difference in vowels to a quality difference only and now has either nine or six vowels differentiated on the basis of height or tense/lax distinctions, where the tense (higher) vowels correspond to long vowels in the other languages and the lax (lower) vowels correspond to the short vowels. This distinction is only found in final syllables. The difference between long and short vowels is also restricted to final syllables in some languages. The Western languages, except for Mocho’ and Classic Maya (the language of the hieroglyphs, e.g., Law, 2014, p. 38), have lost the distinction between long and short vowels entirely, and now have only five vowels (except Chol, which has a sixth high central vowel). Akatek has reintroduced vowel length distinctions. Other languages besides Chol that have a high central vowel include Mopan, Itzaj, and Lacandon, giving them a total of 11 vowels each. The extra vowel ([ə] or [ɨ]) is the reflex of original short (lax)/a/in final syllables, whereas the/a/comes from long (tense) /a:/ in this context and when non-final or before laryngeals. However, a three-way distinction is maintained (/a, aː, ɨ/), as is shown by the contrast of [k’ɨʃ] ‘tie,’ [k’aʃ] ‘knot,’ [k’aːʃ] ‘forrest’ in Itzaj (Hofling, 2000, p. 5).

Three Mayan languages have been analyzed as having distinctive (lexical) tone—Yucatec (e.g., Fisher, 1973; Frazier, 2009; Pike, 1946), (Southern) Lacandon (Bergqvist, 2008), and Uspantek (Bennett & Henderson, 2013; Campbell, 1977; Can Pixabaj, 2007). One of the tones in a tonal contrast generally developed from *VhC or *VɁC sequences, with some complicating factors in Uspantek, where tone also interacts with stress. It has been claimed that the Tsotsil of San Bartolo (Venustiano Carranza) has tone (Kaufman, 1972; Sarles, 1966), but Herrera Zendejas (2013) disputes this, and it has been claimed (Palosaari, 2011) that Mocho’ has low tone on long vowels coming from a *VɁC sequence. Martin (1984) showed that this is only true of Motozintlec and not of the related Tuzantec, where the corresponding words have a V́ɁV̀C sequence instead of V̀ːC (Palosaari) or V̂ːC (Martin). Palosaari (2011), while agreeing with Martin on the differences between Motozintlec and Tuzantec with regard to the reflexes of a *VɁV sequence, chose to treat these two varieties separately. Other languages or dialects of languages spoken along the southern Guatemalan–Mexican border (Mam and Teko) show phonetic pitch changes related to the realization of VɁ or VːɁ sequences. However, these are clearly phonetic and differ from dialect to dialect as to whether the pitch on a particular sequence is high or low and as to whether there is a copy vowel after the glottal stop or not.

It is customary in Yucatec studies to talk about “glottalized” or “broken” vowels for a V₁ʔV₁ sequence. It can be seen clearly in the variations in the realization of *VɁ or *VːɁ that the V₁ʔV₁ sequence is but one of the possible reflexes of these combinations and that it is usually accompanied by pitch changes between the first and second vowel.

2.2. Consonants

All Mayan languages have voiceless stops and affricates at minimally five points of articulation (bilabial, alveolar, and velar stops; alveolar and postalveolar affricates), plus the glottal stop ([Ɂ]). The Eastern and Q’anjob’alan languages have preserved from Proto-Mayan a uvular series of stops, while the other Western, the Yucatecan, and the Huastecan languages have not. It can be noted that there is another series of stops that is reconstructed for Proto-Mayan, *t^j, *t^j’, which none of the daughter languages has maintained. This series is usually assumed to consist of some sort of postalveolar or palatal stop because of its reflexes in the daughter languages ([t, ts, tʃ] plus the corresponding glottalized sounds). It cannot be reconstructed as any of these because each is a necessary reconstruction of some other correspondence. Mamean and Q’anjob’alan languages have innovated a series of retroflex affricates ([ʈʂ, ʈʂ’]) and a retroflex fricative ([ʂ]; note that this sound differs somewhat among the different languages, varying in exact place of articulation on the palate). Two Mamean dialects (Todos Santos Mam and Chajul Ixil) furthermore have a set of apico-postalveolar affricates and a fricative (something like [tsʃ, tsʃ’, ʃ̺]) in contrast with the alveolar, postalveolar, and retroflex series. Mamean languages have also innovated a palatal series of stops ([kʲ, kʲ’]); some K’ichean languages also have these palatal stops as allophones of the velar stops. Several languages (most Cholan, all Yucatecan, Tseltal, Tsotsil, Poqomam, some Poqomchi’) now have both a voiceless ejective bilabial stop as well as an implosive stop, giving a three-way series for the bilabial stops only ([p, p’, ɓ]); this can be shown to be an innovation. The implosive *ɓ is reconstructed for Proto-Mayan. In fact, in some languages it is not the only implosive. The uvular glottalized stop may be implosive in at least some dialects of the Eastern Mayan languages, and the alveolar glottalized stop has been found to be implosive in some dialects of Mam and Tz’utujil (e.g., Campbell, 1973; England, 1983; Pinkerton, 1986). Several changes are restricted to single sounds or individual languages. Poqomam has merged [ts] with/s/, the Q’anjob’alan languages are merging [q] with/x/. In Chol the alveolar stops and [n] have become palatalized ([tʲ, tʲ’, ɲ]); the alveolar [t] remains as an occasional manifestation of/ts/and [n] is found as an allophone of/ɲ/and in Spanish loans (Vázquez Álvarez, 2011).

As previously stated, those languages that have added retroflex stops also have a retroflex fricative ([ʂ]) in addition to the alveolar and postalveolar fricatives. While some languages (Poqomchi’, Poqomam, Q’eqchi’, Popti’, Chuj, some dialects of Tseltal and Tsotsil (Law, 2014, p. 41)) maintain a [χ‎/h] or [x/h] distinction from Proto-Mayan, many no longer have [h]. In some languages it has merged with [x], in others it has disappeared, become vowel length, and/or become a [j] or [w]. The correspondences are quite complicated. In Huastec *s has become [θ‎].

All Mayan languages have both a bilabial and alveolar nasal ([m, n]), but several (Popti’, Chuj, Mocho’) also maintain a velar nasal from Proto-Mayan (*ŋ). This sound merged with [n] in Yucatecan languages and all the Western languages where it was not maintained as a velar, but in the Eastern languages it changed to a fricative. This was presumably a velar fricative ([x]) initially in contrast to the already existing uvular fricative ([χ‎]), but it has by now merged with [χ‎]. As mentioned, the alveolar [n] became palatalized in Chol ([ɲ]), matching the palatalized alveolar stops. The remaining consonants are the approximants [l, w, j], which all Mayan languages have, and [r] (the pronunciation of this sound varies between a flap in intervocalic position and an assibilated r, usually voiceless, in other positions), which is characteristic of K’ichean languages and occurs in other languages in very restricted contexts (such as in affect words [ideophones] and borrowed words). Other instances of [r] from Proto-Mayan changed to [j] in most non-K’ichean languages but [t] in Mamean languages.

2.3. Laryngeals

The status of the laryngeals [Ɂ, h] has been debated. Both occur where other consonants are required. For instance, [Ɂ] occurs as either consonant in the roots of transitive verbs or positionals, which are restricted to CVC shapes in most Mayan languages. [h] occurs as the first or last consonant in a word. However, both also occur where most other consonants are prohibited, for instance in CVɁC or CVhC sequences (see also Section 2.1). This has led to some talk about so-called “glottalized” or “aspirated” vowels. In Proto-Mayan, the only consonants that could occur in this position were [Ɂ, h, s, ʃ, χ‎] (Kaufman, 1976). But note that all of the fricatives were permitted, not just [Ɂ, h]. These last can also be found as epenthetic elements to separate vowels that would otherwise occur together. But again, they are not the only epenthetic segments. The glide [j] is also used epenthetically in some languages. Individual languages may show other characteristics of the laryngeals that suggest that they behave, at least at times, like features of vowels as well as like consonants. The dual character of these sounds should be recognized.

2.4. Consonant and Vowel Inventories

Consonants:

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

Vowels:

1.

2.

3.

4.

5.

6.

2.5. Alphabet

The orthography for Mayan languages uses the following non-IPA symbols: b’ = [ɓ], tz = [ts], ch = [tʃ], tx = [ʈʂ], ky = [kʲ]. ty = [tʲ], kw = [k^w], ’ = [Ɂ], x = [ʃ] for languages that do not have retroflex sounds, x = [ʂ] for languages that have retroflex sounds and then xh = [ʃ], j = [x] or [χ‎], th = [θ‎], y = [j], nh = [ŋ], ñ = [ɲ], VV = [Vː], ï = [ɪ], ë = [ɛ], ü = [ʊ]. ö = [ɔ], ä = [ɨ ~ ə ~ ɜ]. Other symbols are equal to the IPA symbols. This alphabet will be used in the sections that follow.

3.1. Nouns and Adjectives

Nouns can be subcategorized according to changes they may undergo when possessed, or according to the structure of compounds (5). The largest class of nouns does not change when possessed. Others may lengthen or change a vowel when possessed; in disyllabic stems this is usually the last vowel, but it may be the first vowel in some languages. Other changes have to do with the type of possession. Inalienable nouns carry a suffix in their unpossessed forms. Some nouns, mostly parts, are always possessed, and some, mostly natural phenomena, are never (or not usually) possessed. Compound nouns can consist of two joined roots or of two separate roots that nevertheless form one lexeme. This lexeme, when possessed, usually has the possessive marker on the second noun. Compound nouns can also consist of two separate roots where the second is marked as the possessor of the first, or of two separate roots where both receive the mark of possession to form the compound.

tzutujil (García Ixmatá, 1997, pp. 117–122)

(5)

In addition, there are several subclasses of special-purpose nouns. The most important of these is a closed set of always possessed nouns that are used to indicate location (6), grammatical relations (7), or subordination (8), as well as comparatives (9), and possessives. These have been called relational nouns, they are used like prepositions except that they show the person and number of their complements through Set A markers. Most languages have one or two true prepositions as well, and some of the relational nouns show signs that they are grammaticalizing further as prepositions (losing their person marking). Some relational nouns in some languages are used in combination with prepositions (6).

k’iche’ (López Ixcoy, 1997, pp. 240, 234)

(6)

(7)

mam (Pérez & Jiménez, 1997, pp. 398, 227)

(8)

(9)

Numbers constitute another special class of nouns, at least morphologically. Ordinal numbers are derived from cardinal numbers (usually by a suffix) and are obligatorily marked with Set A markers. Thus they seem to be morphological possessed nouns. However, numbers behave syntactically like a separate class. They precede and modify other nouns in the noun phrase. Nouns cannot modify nouns in this way. Measures, numeral classifiers, and sortal classifiers are other subclasses of nouns with very special uses. They occur obligatorily with preceding numbers to count mass nouns or aggregated nouns, to classify nouns, or to specify characteristics, mostly physical, of counted nouns. The Q’anjob’alan languages and several of their neighbors (Mam, Ixil, Chuj) also have noun classifiers. These are nouns (they are often transparently related to common nouns) that accompany all concrete nouns in the noun phrase according to their semantic class and have pronominal functions as well (they are only used pronominally in some varieties of Mam). In Q’anjob’alan languages they establish classes on the basis of human (male, female, plus age/respect categories), non-human (animal, dog), or inanimate features (plant, tree, rock/metal, fiber, liquid, fire, etc.). Q’anjob’alan languages have the greatest number of different kinds of classifiers, with noun classifiers, numeral classifiers, sortal classifiers, and plural markers restricted to humans (Craig, 1986; Zavala, 2000) Example (10) shows all of these except the plural marker.

acatec (Zavala, 2000, p. 127)

(10)

Adjectives are a small class of roots (usually under 50) but a very large class of derived forms, given that at least one, and possibly multiple, adjectival forms can be derived from every verb and positional in a language. With the exception of plural marking in K’ichean languages, they take no inflection other than subject marking when they function as the heads of non-verbal predicates. They minimally have characteristic derivations to form abstract nouns and intransitive verbs (11–12).

poqomam (Santos Nicolás & Benito Pérez, 1998, pp. 165–166)

(11)

(12)

3.2. Verbs and Adverbs

In most Mayan languages transitive and intransitive verbs are clearly separate. With very few exceptions verbs are restricted to events. States are usually conveyed by non-verbal predicates, although a very few verbs in specific languages may be stative as well. Verbs take aspect (or (tense)-aspect-mood) marking while non-verbal predicates do not. Verbs differ with regard to person marking—intransitives mark subjects (Set B) while transitives mark both subjects (Set A) and objects (Set B). Some languages have a set of what have been called status markers. These are inflectional suffixes that indicate the category of the verb (intransitive or transitive, and sometimes they distinguish root transitive from derived transitive) and the broad type of structure in which the verb is used, usually distinguishing some sort of “dependent” structure (which may include imperative, hortative, or optative, and dependency to an auxiliary or within some sort of complex verb structure) from other verb structures. All Mayan languages indicate aspect on verbs through affixes, clitics, or particles; this may be combined with some temporal information in some languages (for instance, restricting the completive/perfective to the past) and may include some mood marking (such as an imperative). Examples in K’iche’ follow:

k’iche’ (López Ixcoy, 1997, pp. 187, 181, 192)

Intransitive Verb

(13)

Intransitive Verb in a “Dependent” Status

(14)

Transitive Verb

(15)

Transitive Verb in a “Dependent” Status

(16)

Adverbs take no inflection and are mostly particles, except for some adverbs for time in the past and time in the future, which are derived from numbers (alone for days in the past or future and in combination with the word for ‘year’ for years in the past or future). They include temporal adverbs, manner adverbs, and locatives.

3.3. Positionals and Affect Words

Positionals are a unique class of roots in Mayan languages that require derivation to be used as words; the most typical and defining derivations are those that create a positional predicate (non-verbal), an intransitive verb, or a transitive verb (17).

q’anjob’al (Mateo Toledo, 1999, pp. 110–112)

(17)

Positionals describe the position in which an object is found, often in combination with some of its physical characteristics. They occupy a semantic space that is often filled by intransitive verbs or adjectives in other languages. When positionals are discussed as a word class, what is meant is usually the positional predicate. Positionals have sometimes been called a kind of verb, especially in the older literature on Mayan languages. This is usually an error of analysis since positional predicates do not have the characteristics of verbs; that is, they do not take aspect marking. However, there are verbs derived from positionals, and these may exist in some languages in which the original positional is no longer used. Positional roots are very numerous in most Mayan languages, numbering between 250 and over 600 distinct roots. They make very specific and detailed reference to positions and forms or shapes or other characteristics and thus provide speakers with very precise ways to talk about states. They also figure productively in expressive language. Martin (1977) is a detailed study of positionals in Q’anjob’al, one of the languages with an exceptionally large positional inventory.

Mayan languages have another very large and important class of words or in some languages two classes of words—affect words and possibly affect verbs. These have also been called ideophones (in the case of affect words) and expressive predicates (in the case of affect verbs). They differ fairly substantially from language to language in terms of how they are used, whether more than one type can be distinguished, whether their phonology is more or less standard or unusual from a Mayan perspective, and whether they are more or less integrated grammatically. In general, affect words name actions and impressions, often sensory, associated with them, especially sounds, movements, and visual impressions. They are not necessarily onomatopoeic, although they can be and some of them are in each language where they occur. Some are roots, some are derived, and there are usually characteristic derivations that can be applied to affect words to derive other words, including affect verbs.

Affect verbs are usually formed from roots of almost any class through characteristic derivations, are numerous where they exist, and cover domains such as sounds, salient physical characteristics, distinctive body positions, motion, sudden events, and so on (Baronti, 2001). Affect verbs tend to be more grammatically integrated than affect words, where both exist. That is, they participate in a greater number of grammatical processes such as taking at least some aspect affixes, or negation, or taking oblique agent phrases with the agent relational noun, and so on. Affect words, on the other hand, tend to be restricted to occurrence with a very few light verbs (i.e., intransitive verbs of motion, quote verbs, or the verb ‘to do’) and to participate even less than affect verbs in other grammatical structures or processes.

Affect verbs have been described the best for Tseltalan languages (Kaufman, 1971; Maffi, 1990; Pérez González, 2009, 2011; Ringe, 1981) and K’iche’ (Baronti, 2001). Affect words have been compared to affect verbs in Tseltal (Pérez González, 2011) and have been described for Yucatec (Le Guen, 2011, 2012) and for Mam, where affect verbs are relatively marginal (England, 2004, 2006). It can be seen that in K’iche’ and the Tseltalan languages, affect verbs have somewhat similar characteristics and several of the suffixes that derive them are cognate, so they are presumably old structures in Mayan languages. Mam, on the other hand, seems to have almost lost the verbs or combined their functions and grammars with those of the affect words. Examples in Tseltal follow. There are many different suffixes that derive expressive predicates (18) and many different contexts for the use of ideophones (19); these examples only illustrate one of each.

tseltal (Pérez González, 2011, pp. 98, 154)

Expressive predicate:

(18)

Ideophone:

(19)

3.4. Function Words

Other classes of words in Mayan languages are closed and are primarily grammatical function words rather than content words. They include a very few prepositions, pronouns (usually based on person markers), demonstratives, articles, exclamations, interrogatives, negatives, conjunctions, and others.

4.1. Predicate Types

Predicates are either non-verbal (20), intransitive (21), or transitive (22–23) and are quite distinct in structure. Non-verbal predicates are formed on nouns, adjectives, numbers, or positionals or on locative, existential, or demonstrative bases, take Set B for subjects (they never take Set A) and do not take preposed aspect markers, although they may take aspectual suffixes in some languages. Intransitive verbs take Set B for subjects, and if the language has split ergativity, they will take Set A for subjects under the conditions that provoke a change in alignment (see Section 4.3 for additional types of alignment). Transitive verbs take Set A for their subjects and Set B for the object. Verbal predicates take aspect markers, preposed to the stem in first position (before the person markers). They may take suffixes for additional aspects such as perfect, and these are different for intransitive and transitive verbs. Verbs in some languages have “status markers”—suffixes that indicate the type of verb and its status with regard to independent or dependent (broadly). These are also distinct for intransitive and transitive verbs, as are derivations that produce verbs or that apply to verbs to produce other classes of words.

mam

Non-verbal predicate

(20)

Intransitive

(21)

Transitive with lexical subject and object

(22)

Transitive with audible Set B marker for object

(23)

Almost all Mayan languages require specific derivations to produce stems of one verb class from the other verb class; very few or no verb stems belong to both classes simultaneously. There are often a number of roots, however, that form positional predicates or transitive verbs with no derivation. An analysis made of Yucatec (Lois & Vapnarsky, 2003) holds that roots are not assigned to any category and thus there is no base distinction between transitive and intransitive verbs (or between verbs and nouns). It is not entirely convincing, however, (because there actually are regular differences between the transitive and intransitive forms) and certainly does not hold for other Mayan languages where the differences between the two kinds of verb roots are even more apparent. While Haviland (1994) shows that Tsotsil has a large number of roots (around 300) that cannot be easily assigned to intransitive, transitive, or positional classes because they mix properties of these classes, this analysis of polyvalence cannot be extended to most other languages beyond Ch’olan-Tseltalan and Yucatecan.

4.2 Order

Mayan languages were originally verb initial (England, 1991; Norman & Campbell, 1978). Most of them have VOA (Verb-Object-Agent, aka VOS) basic (syntactic) word order (24), but the Mamean and some Q’anjob’alan languages have VAO (aka VSO) basic word order (25), while Ch’orti’ is developing AVO (SVO) word order in independent clauses but maintains VOA (VOS) order in subordinate clauses (Quizar, 1979, 1994). Other Cholan languages and Tseltalan languages are basically VOA (VOS) with flexible order, however (Aissen, 1987; Polian, 2013a; Osorio May, 2005; Vázquez Álvarez, 2011) (26).

k’iche’ (flexible VOA)

(24)

q’anjob’al (rigid VAO) (Mateo Toledo, 2012a, p. 143)

(25)

chol (flexible VOA) (Vázquez Álvarez, 2011, p. 170)

(26)

There has been considerable discussion about basic word order in Mayan languages because of the flexibility that most of them show and because some are apparently moving to a situation in which pre-verbal occurrence of arguments, at one time reserved for topicalization or focus, is becoming less and less marked. This is the case of Ch’orti’, where the preverbal transitive subject, or A argument, is always the topic, according to Quizar. However, in these languages post-verbal A arguments also occur, sometimes alone, as in the few texts that Quizar examined, or sometimes, but rarely, also accompanied by a post-verbal O argument.

The reasons that examples of transitive clauses with two arguments are so rare are that person of both the subject and object is marked on the verb, independent pronouns are only used for some kind of highlighting (usually focus), no nominal form is required to express these arguments, and there are few occasions in natural discourse where both arguments will be expressed nominally. This is especially true since first and second person arguments are only rarely expressed by pronouns. Elicitation is not much help here except to show that two lexical arguments are possible, because speakers often try to supply a context for elicited sentences by, minimally, topicalizing one of the arguments, usually the subject. Yucatec and Huastec are other Mayan languages where AVO order is quite unmarked, both morphologically and syntactically, and K’ichean languages seem to be moving in that direction. Mamean and Q’anjob’alan languages, on the other hand, have highly marked AV(O) orders and in fact mark almost any order involving a preverbal nominal.

Durbin and Ojeda stated in their 1978 paper that they could not decide between AVO and VOA as the basic word order in Yucatec and said that they could find no differences in meaning or emphasis for clauses given in the two orders. Recent studies (Gutiérrez Bravo & Monforte y Madera, 2010; Skopeteas & Verhoeven, 2011) re-examine Yucatec and come to opposing conclusions. The only other strong case for AVO order, that made by Quizar for Ch’orti’, also merits some reconsideration. Although in her data Ch’orti’ has more examples of AVO than any other order with two arguments (10 out of 12 examples), it also has a number of VA examples (9 out of 28 examples; the rest are AV). This suggests that a verb initial order may still be the basic order, or there would not be any examples of VA. All of the AV or AVO orders are analyzed by Quizar as examples where the A is the topic. They are most probably examples of topicalization, and Ch’orti’ is probably moving toward a situation in which this kind of topicalization is no longer marked. The only thing that remains is for the VA orders to fall into disuse or to switch to a highly marked use for Ch’orti’ to in fact change completely to an AVO language. Yucatec may be in a similar situation, but it too still has V(O)A examples (not highly marked) and for that reason cannot be claimed to be a purely AVO language. The K’ichean languages show a somewhat earlier stage in the possible conversion from VOA to AVO languages—they still have both orders, but AVO is increasing in frequency and preverbal topicalization is not highly marked. Section 4.3 has examples of how structures with fronted agents can be highly marked. In addition, the fronting of arguments other than agents, and topicalization with fronted arguments may all be marked in some fashion, often by focus or topic particles, for instance. Example (27) shows the use of a particle marking topicalization through fronting in Tsotsil (Aissen, 1992, p. 49).

tsotsil

(27)

4.3. Alignment

Mayan languages are morphologically ergative, and many of them have at least some rules that show syntactic ergativity as well. The most usual of these is that the extraction of A arguments for purposes of focus (especially contrastive focus), or negation, relativization, or interrogation may be restricted or not allowed, while the extraction of S (intransitive Subject) or O arguments is not subject to the same constraints. Those languages that restrict the extraction of A arguments have three ways to do so. One is to convert the verb into an intransitive verb via an antipassive suffix (Mamean languages and partially in K’ichean languages). The resulting verb agrees with the extracted A argument, and the patient (original O argument) can only be expressed in an oblique phrase (28). Another way is to mark the verb with an agent focus suffix, -(o(o))n, which may be different from or homophonous with the antipassive (historically they were probably different, but they have shifted roles in some languages) (Q’anjob’alan proper languages, Chuj, and partially in K’ichean languages). The verb then agrees with the patient (original O argument) rather than the A argument, but the A argument is expressed directly rather than in an oblique phrase. Thus the verb is not quite transitive (it only marks one argument, and does so with the Set B markers), but it is not quite intransitive either (it clearly has two arguments, neither of which is oblique) (29). A third way, little used, is to retain a transitive verb but express the extracted agent in an oblique phrase introduced by a relational noun (Mam; England, 1983). Some Mayan languages show no restrictions whatever with regard to the extraction of A arguments (30). There may be other rules besides the extraction of A arguments that show syntactic ergativity. In Mam, for instance, control of the subject of an infinitive verb can only be by an absolutive argument, but any absolutive argument can show such control—an intransitive subject, a transitive object, or the subject of a non-verbal predicate. Or in K’ichee’ (Velleman, 2014) in situ focus is restricted to absolutive arguments.

mam: Antipassive for extraction of agents

(28)

q’anjob’al: Agent focus for extraction of agents (third person only) (Mateo Toledo, 2008, p. 76)

(29)

tseltal: Agent focus with a transitive verb (no change) (Polian, 2013, p. 776)

(30)

Most Mayan languages show split ergativity morphologically, but K’ichean languages (except Poqom) and Tseltalan languages do not. They are always ergative in their alignment. There are three different conditioning factors for split ergativity: aspect (lowland languages—Cholan, Yucatecan, Poqom) (31–32), occurrence in a dependent aspectless clause (Mamean and Q’anjob’alan languages) (33–34), and person (Mocho’) (35–36). Under the conditions that provoke a non-ergative pattern of person marking, intransitive verbs take Set A person markers for subject (S), and therefore are marked the same as transitive subjects (A). This then constitutes a nominative-accusative system, since both subjects (A and S) are marked the same while the transitive object (O) is marked differently. The exception to this is Mam, where under the conditions that provoke a change in marking, all arguments (A, S, O) are marked with the Set A markers, resulting in neutralization of person marking (but note that position of the markers on a transitive verb still distinguishes the object from the subject, in that the object always precedes the subject). Ch’orti’ is also an exception, since it has a third set (set C) that is used to mark the subjects of intransitive verbs in the imperfective aspect (Quizar, 1979). Mocho’ has a split on the basis of person—first and second persons take Set A markers for both subjects (A and S) and mark objects with Set B (nominative-accusative pattern), while third person follows an ergative pattern with Set A for A arguments and Set B for S and O arguments (Larsen & Norman, 1979, pp. 352–353).

chol (Vázquez Álvarez, 2011, pp. 25–26)

Ergative—Perfective Aspect

(31)

Accusative—Imperfective Aspect

(32)

q’anjob’al (Mateo Toledo, 2008, pp. 49–50)

Ergative

(33)

Accusative

(34)

mocho’ (Larsen & Norman, 1979, pp. 352–353)

Ergative—Third Person

(35)

Accusative—Local Person

(36)

A group of Mayan languages, probably starting with Chol and influenced by it, show patterns of agentive organization of intransitive verbs (on top of ergative alignment and split ergativity). In Chol intransitive verbs form three groups—agentive verbs, non-agentive verbs, and those which can pattern either agentively or non-agentively (Vázquez Álvarez, 2011). Agentive intransitive verbs cannot inflect directly for subjects but instead function with a transitive light verb cha’l ‘to do’ that inflects with set A for the subject (37). Non-agentive verbs, on the other hand, inflect directly for their subject with set B on the verb (38) and cannot occur in the light verb construction. Additionally, there are some verbs that can inflect for subject with either pattern (Set A on the light verb or Set B on the verb), depending on the volitionality of the action (39). Other languages such as Poqomchi’ and Q’eqchi’, presumably in contact with Chol or other Cholan languages in the past, make similar distinctions in the organization of intransitive verbs, and Mopan has a somewhat similar pattern, except that the marking of the two kinds of intransitive verbs is directly on the verb with no light verb intervening (40–41).

chol (Vázquez Álvarez, 2011, pp. 27, 265)

Agentive

(37)

Non-agentive

(38)

Labile

(39)

mopan (Danziger, 1996, pp. 392, 395)

Non-agentive (“Mutative”) with set B marking

(40)

Agentive (“Active”) with set A marking on the verb stem (nominalized)

(41)

Finally, in some Mayan languages obviation and inversion have been shown to be important organizing principles. The languages that have so far been analyzed as having obviation and/or inverse include Tsotsil (Aissen, 1997, 1999), Tseltal (Polian, 2013a), Huastec (Zavala Maldonado, 1994a, 2007), and Mam (Pérez Vail, 2014). In these languages the main way in which these principles affect the grammar is that a marked structure, such as passive, may be required when obviative participants act on proximate participants (obviation) or when third person participants act on speech act participants (inverse). Huastec is the only Mayan language that marks the inverse morphologically; the others make use of the difference between marked (e.g., passive) and unmarked (active) structures (42–43).

tsotsil (Aissen, 1997, pp. 725, 727, 728)

Obviative > Proximate

(42)

Proximate > Obviative

(43)

4.4. Voice

Mayan languages have, in many instances, multiple passives, an antipassive with several functions, an agent focus form, and an applicative with different functions in the languages where it occurs. The passives usually include one or more fairly straightforward syntactic passives and one or more lexical passives that add to or change the meaning of the verb in some way (44).

k’iche’ (Larsen, 1988, pp. 254–255)

(44)

Mam has the largest number of passives so far encountered in a Mayan language, with seven different passives in Cajolá. Pérez Vail analyzes these as not adding anything to the meaning (2 forms), involving temporal changes such as reiteration or duration (2 forms), adding achievement or accomplishment (1 form), or used only in contexts of movement plus purpose (2 forms) (Pérez Vail, 2014, pp. 114–115). As has been stated, passives are used obligatorily in some languages in contexts of obviation or inversion. This is probably more widespread than has so far been reported.

Antipassives are formed by either one or two suffixes in Mayan languages, -(V(V))n or –(o)w. There are several functions for the antipassive forms. They are used to express an unknown patient (absolutive antipassive, but note that the patient can be added in an oblique phrase) (45), for patient incorporation (incorporative antipassive) (46), and in some languages for lexical purposes (deriving new vocabulary, lexical antipassive). However, antipassive forms are also used in some languages for agent extraction or focus of the agent (focus antipassive) (47), while other languages use agent focus morphology, where the verb marks the patient rather than the agent, for these functions (48). Some languages, such as K’iche’, can use either the focus antipassive or the agent focus morphology, depending in the case of K’iche’ mostly on an argument hierarchy (1>2>3PL>3SG; the higher argument is marked).

q’anjob’al (Mateo Toledo, 2008, p. 72)

(45)

(46)

k’iche’ (López Ixcoy, 1997, pp. 368–369)

(47)

(48)

The two suffixes have been reconstructed to Proto-Mayan as originally having two different functions, that of antipassive (*-(V)n) and agent focus (*-w) (Dayley, 1983). The difference between the two is the following. Both suffixes detransitivize a transitive verb, which then takes only one person marker for the subject. The antipassive verb marks the agent or experiencer as subject (the original transitive subject). The other participant, the patient/theme, must be in an oblique phrase if expressed (introduced by a relational noun). The agent focus suffix, on the other hand, creates a verb that marks the patient or theme as subject (the original object). The agent, while not expressed on the verb, is not oblique, but is instead in a direct (unmarked) noun phrase, usually in front of the verb. Either kind of verb can be used, depending on the language, when the agent is focused (or negated, questioned, or relativized), and in some languages both are used.

Note that there has been terminological confusion with regard to these two verb forms. Dayley (1983) refers to both as the “focus antipassive.” Stiebels (2006) and others call both constructions “agent focus.” Here “antipassive” is reserved for an intransitivized verb that marks the agent (no matter what the function), “agent focus suffix/morphology” for the suffix that creates a verb that marks the patient, and “agent focus form” for the resulting verb. Both may figure in agent focus constructions. There has also been disagreement about whether these are in reality voices, especially with regard to the agent focus form. It is the case that both suffixes detransitivize the verb morphologically, such that the resulting verb marks its subject with Set B markers (or Set A markers under conditions of split ergativity) and usually takes other intransitive morphology. However, since the verb with the agent focus suffix still has an agent that is not in an oblique phrase, its arguments look more like arguments of an active verb in syntactic terms (both direct). The antipassive when used for agent focus as well as the agent focus form are semantically transitive, since two arguments are implied.

Furthermore, in some languages the specific suffixes have shifted functions. K’iche’, for instance, uses the –ow suffix for root transitives and –n for derived transitives either in the focus antipassive or for agent focus forms. In general, the verb marks the higher participant and so can mark the agent (antipassive) or the patient (agent focus form) depending on which one is higher. The K’iche’ absolutive antipassive, however, only uses the –n suffix. Mam only has the –n suffix now, and, except for in Todos Santos and Tacaná, the verb marked with this suffix agrees with the agent and not the patient. The patient if expressed is in a relational noun phrase. In Todos Santos and Tacaná, however, the verb marks the patient but the patient still appears in a relational noun phrase, producing a sort of hybrid structure with agent focus morphology but a patient structure that usually accompanies the antipassive (Dayley, 1983, pp. 45–46; England 1990, pp. 243–244). Todos Santos, which is the most northeastern Mam municipality, is clearly unlike other northern Mam dialects. Tacaná is a western Mam municipality; it is unknown whether other western Mam varieties are like Tacaná with regard to this feature. Southern Mam is like the north (Pérez & Jiménez, 1997).

In the Q’anjob’alan group agent focus is marked with –(o)n and is obligatory for agent extraction of third person agents. The antipassive is marked with –w (-waj, -wi) and is now very restricted in usage. It can only be used with certain verbs in the absolutive and incorporative functions. These languages maintain what was presumably the original situation with regard to the difference in the two suffixes. In Q’anjob’al the agent focus morphology is used to focus, question, or negate a third person agent (only); the verb agrees with the patient (Mateo Toledo, 2008, p. 76). Agents that are not third person can be focused with active forms. Other Mayan languages have made other responses to the extraction of agents, most of which do not involve either antipassive or agent focus morphology, or do so only in an attenuated form.

A final form that some Mayan languages have is the applicative. This is mostly a rearranging voice; while the valence of the verb is not increased or decreased, its argument structure is rearranged. There are two different patterns—that found in K’ichean languages (except Q’eqchi’) and Ixil and that found in Tseltalan languages, Chol, and Chontal. Both patterns use the same suffix (–b’e) on the verb. Huastec has the same functions as are found in both patterns, but with different suffixes.

In K’ichean languages and Ixil the suffix –b’e forms what has been called the instrumental voice. The verb form is used primarily when an instrument is in focus position in front of the verb. Whether the instrument becomes an argument of the verb or not varies in the different languages and even in different dialects of the same language; the more conservative pattern is that it is the argument marked by Set B on the verb, and the patient/theme becomes oblique in a relational noun phrase. In (49) the verb bears a third person set B marker in agreement with the fronted instrument, and the patient is indicated by the relational noun –eech.

k’iche’ (Norman, 1978, p. 462)

(49)

In other languages, however, such as Ixil and Kaqchikel, while a verb takes b’e with a focused instrument, the patient rather than the instrument continues to be marked on the verb with set B affixes. In Kaqchikel the fronted instrument also continues to be introduced by a relational noun (50). Norman analyzes this as a loss of instrumental voice.

kaqchikel (Norman, 1978, p. 465)

(50)

In the Tseltalan and Cholan languages, however, the same suffix is used for the promotion of recipients, benefactives, malefactives, and possessors of patients, which are then marked by set B on the verb. The patient/theme is no longer marked on the verb, but is also not in an oblique phrase. This pattern of agreement defines a classic primary object language (Dryer, 1986), in which the set B marking is used for the direct object in a monotransitive clause but for the recipient or benefactive (indirect object in Dryer’s terms) in a ditransitive clause. In Tsotsil, the patient (the secondary object) follows the verb and precedes the primary object. (51) shows that the secondary object precedes the primary object, while (52) shows that it is the primary object (first person) that is marked on the verb by the set B marker.

tsotsil (Aissen, 1983, p. 286)

(51)

(52)

K’ichean languages show some signs of once having used –b’e for marking recipients on the verb through set B markers, in some frozen lexical forms with –b’e that still exist. For instance, in Tz’utujil the examples in (53) show the difference between the verbs ‘speak,’ an antipassive verb from the noun tziij ‘speech, word,’ and ‘speak to (someone)’ with the applicative. Note that the noun phrase ja rixoqiil ‘his wife’ is a direct argument of the verb in (53b) while it is an oblique argument in (53a), introduced by the relational noun ruk’iin ‘with.’ Jar uleew ‘the land’ is oblique in both examples, in (53a) because the verb is antipassive and only marks the agent, and in (53b) because the verb has the applicative suffix and set B marks the recipient (ja rixoqiil).

tz’utujil (García Ixmatá, 1997, pp. 396–397, cited in England, 2001, p. 140)

(53)

Finally, while Huastec does not have –b’e, it has both its functions in two separate suffixes, -na’ for the instrumental and –ch(i)(nch) for the applicative (Dayley, 1983).

4.5. Other Topics

There are a number of other topics in syntax that are of interest but that will be only briefly summarized here. Mention has already been made of extraction of agents for focus. Just about any argument, oblique argument, or adverb can be focused in the position immediately before the verb. Some Eastern Mayan languages have particles or clitics that indicate focus of locatives and other adjuncts, such as wi in K’iche’. The applicative morpheme –b’e is used in K’ichean languages to indicate focus of an instrument. In addition, at least some kinds of arguments can be focused in situ (but not transitive subjects; Velleman, 2014). Topicalized arguments can also be indicated in a preverbal position, defined by Norman (1977) as the first position in the clause. The two pre-verbal positions—topic and focus—have been the basis of considerable work since Norman proposed them. Significant progress has been made on analyzing both topicalization and focus in various Mayan languages (especially K’iche’; see, e.g., Aissen, 1992, 1999; Baird, 2014; Can Pixabaj, 2004; Can Pixabaj & England, 2011; Larsen, 1988; Velleman, 2014).

Another topic that has received considerable attention is the structure of complement clauses (e.g., Can Pixabaj, 2015; Craig, 1977; Coon, 2010; England, 2013; Mateo Toledo, 2013; Polian, 2013b;Vázquez Álvarez, 2013). Mayan languages have complements that have finite verbs, non-finite verbs which lack aspect marking but do have person marking, or infinitives. There is a general prohibition of transitive infinitives such that in most languages transitive verbs must be detransitivized, usually through a passive or antipassive suffix, in order to derive an infinitive (but see Polian, 2013b).

Relatively little has been done on relative clauses, adverbial clauses, or conditional clauses, but the recent grammars of different Mayan languages usually have some description of these clause types. Relative clauses follow the head noun except in Chol, where the relative clause can precede or follow the head (Vázquez Álvarez, 2011). Temporal clauses provide a context for split ergativity in Q’anjob’alan and Mamean languages. In these languages they are aspectless and show either accusative marking or all-ergative marking (Mam). Mam furthermore has several different subordinators for temporal clauses that make a distinction between irrealis (potential or future actions) and realis (perfective, imperfective, proximate aspects). Negation in a number of languages may also differ according to aspect or mood, or according to type of constituent negated (verb, non-verbal predicate, etc.).

Several recent works have described secondary predicates in some Mayan languages (Can Pixabaj, 2010; Mateo Toledo, 2010, 2012a; Pascual, 2010; Polian & Sánchez Gómez, 2010; Vázquez Álvarez, 2010) and complex predicates (Mateo Toledo, 2008, 2012b). The study of complex predicates began with a series of analyses of movement verbs and directionals (Aissen, 1994; Craig, 1986; England, 1976; Mateo Toledo, 2008; Zavala Maldonado, 1993, 1994b) and has been extended to other kinds of complex structures by Mateo Toledo. In short, there is considerable room for further investigation of complex clauses and some characteristics of simple clauses as well. However, the study of syntactic structures in Mayan has been making steady progress. Phonetics and phonology and their integration with syntax have also been receiving new attention (e.g., Baird, 2014; Bennett & Henderson, 2013; Frazier 2009). Previous studies in this area were very scarce (but see Pinkerton, 1986).