Wayana language

Wayana's morphology, like that of other Cariban languages, is very rich, and makes extensive use of affixes.[10] There are five major speech classes in Wayana: nouns, verbs, postpositions, adverbs, and particles. Each class can be distinguished by the presence (or absence) of specific affixes. Particles take no affixes, and are not morphologically bound. They "expres[s] many different meanings such as evidentiality, negation … [and] emotivity …".[19] Barring particles, all members of each speech class are clearly recognizable by the affixes that they can take. In her "Grammar of Wayana", Tavares addresses the major aspects of each speech class in-depth. In regards to nouns, she discusses possessive, personal, and number affixes, Wayana pronouns, and special cases such as sound symbolic words. For verbs, she outlines the two "sets" of verbs found in Wayana ("Set I" –on which tense, aspect, and mood are marked through affixes, and t-V-(he) verbs – so-called due to their characteristic ambifix). She also addresses valence-changing morphemes, and gerundive verb forms. For postpositions, Tavares examines the personal prefixes, and spatial suffixes, as well as semantic classes of postpositions (including spatial postpositions). Derivational morphology is addressed for all classes, as class change (e.g. nominalization, verbalization) through affixes occurs often. Derivation is especially important for adverbs, as most adverbs are derived from other speech classes. Tavares presents the few underived adverbs that exist, as well the processes of adverbialization.[8]

Possession in Wayana occurs in only one way. It consists of three elements - the possessed noun, the possessor, and the possessive suffix. The possessor comes before the possessed noun as either a free noun or free pronoun, or a personal prefix, and the possessed noun is necessarily inflected by one of four suffixes.[8] For example, we have:

Data from Tavares (2005), pp. 121

Note: -n indicates the possessive suffix.

As we can see from 1c., the possessor cannot be expressed as both a personal prefix and a noun/pronoun, or else it is ungrammatical. The personal prefixes differ based on the succeeding environment. They are summarized in the table below:

Modified from Tavares (2005), pp. 121

As mentioned previously, there are four allomorphs of the possessive suffix that inflect the possessed noun: -n(u), -(lï), -t(ï), -ø. The full form of the suffixes are realized only in some phonological environments. -n(u) is "the most productive" (Tavares, 2005, pp. 127) allomorph, as it is applied to loanwords (2c.). -t(ï-) can only be applied to three nouns (ëmi, ëlek, ëpi – face, boil, medicine). Both -(lï) and -ø occur after specific nominalizing affixes; each occurs after a different nominalization affix (for example the ones in 3b., 5b, though there are several others).–ø also occurs after some inherently possessed nouns (5a.), and nouns with the devaluative suffix (-tpë, -tpï(lï)) (5c.).[8]

Data from Tavares (2005) pp. 126 – 136

It is important to note that nouns in Wayana fall under three classes, in regards to possessibility: unpossessable, optionally possessable, and inherently possessed. Unpossessable nouns do not take possessive prefixes or suffixes. Optionally possessed nouns are semantically diverse, and can occur in either possessed or unpossessed form (most examples given use optionally possessed nouns). Inherently possessed nouns "denote entities that stand in a stable, intrinsic relationship with another entity" (Tavares, 2005, pp. 136), and must always be possessed. Kinship terms (such as in 5a.) and some body parts fall under this category. Inherently possessed nouns often, but not always, take –ø as the possessive suffix.[8]

Valency refers to the number of participants that a verb takes. In Wayana, the subject of an intransitive sentence (S) is distinguished from subject of a transitive sentence (A) and the object of a transitive sentence (O). The valency on a verb stem can be decreased or increased by using certain affixes. Valence can be decreased by only one affix – the detransitivizing prefix – but can be increased by four different suffixes. Causational morphemes (such as the Wayana –po) are not considered as valence-increasing, since the participants they introduce are optional.[8]

The detransitivizing prefix is realized as e-, ëh-, ët-, depending on the succeeding phonological environment. The prefix can be taken by a transitive verb stem to detransitivize it, so that only one argument can be taken; the verb is then "marked as an SA intransitive verb" (Tavares, 2005, pp. 253), meaning that it only takes one argument, which is the subject and agent. For example, we see, how, in 6a. and b., a verb stem that usually requires another participant, can be satisfied by just one acting as an agent, in the presence of the detransitivizing prefix.[8]

Data from Tavares (2005), pp. 254

Conversely, intransitive verbs can be made to take new participants by using transtivizing suffixes: -ka, -nëp(ka), -nïp(ka), ma, and lë. These suffixes only "occur on SO intransitive stems", "adding a new nuclear participant to the event described by the verb" (Tavares 2005, pp. 254). The new participant becomes the agent and subject of the transitive sentence, and the old S, the object. The suffixes do not occur on SA intransitive verbs (like the ones formed by the detransitivizing suffix), as they cannot be transitivized.

The distribution of these suffixes is not well described, and seems to be mostly based on phonology of the surrounding environment. As can be seen below, the different suffixes behave very similarly.

Data from Tavares (2005), pp. 255–256

The root "elemi", seen in 7c. can also take the transitivizing suffix –ka, as well as –nëp(i), but this causes a different meaning, for an unknown reason.[8]

Wayana's case system presents "an … unprecedented type of split ergativ[ity]" (Tavares, 2005, pp. 412),[8] where there are two verb types – Set I and t-V-(h)e. The former presents an unclassified mixed system (sometimes analyzed as active-stative, or inverse) while the latter presents ergative case. In both verb sets, the subject of an intransitive sentence (S) is distinguished from subject of a transitive sentence (A) and the object of a transitive sentence (O), as mentioned previously. There is no grammatical property that seems to trigger the use of one set over the other. Both systems can "occur independent of tense, with all persons … in main clauses, and with all verb stems" (Tavares, 2005, pp. 234),[8] and the choice of verb set is perhaps dependent on discourse factors.[8]

Set I verbs in Wayana take a variety of personal prefixes, marking first, second, dual (first and second), and third person. The prefix used differs based on the valency of the verb in question, and person(s) being marked.[8]

For transitive verbs, one prefix marks both subject (A), and object (O). The different prefixes, based on which person A and O are marking, are below.

Modified from Tavares (2005), pp. 206

When one of the arguments of the transitive verb is a speech act participant (SAP) - either first person, second person or dual person (both first and second person) - and the other argument is a third person, the personal prefix differs based on whether the speech act participant is acting on the third person (is A), or is being acted on by the third person (is O). Often, first situation is termed direct, and the latter inverse. We can see this demonstrated by the differing prefixes for 1A3O (direct) and 3A1O (inverse), as well as 2A3O (direct) and 3A2O (inverse).[8]

This distinction between direct and inverse prefixes is crucial when examining intransitive verbs. While each prefix marks the subject (S), as the chart below shows, there are two distinct sets of personal prefixes for each SAP person. One set matches the direct prefixes in transitive verbs, and the other set matches inverse prefixes. Thus, the S is split, and referred to either as SA or SO, depending on which transitive prefix set the intransitive prefix resembles. SA corresponds to the direct prefixes, and SO, the inverse. For example, 1SA prefix "w-" is the same as the 1A3O direct transitive prefix.[8]

To illustrate, below are two example of sentences using Set I transitive (8) and intransitive verbs, with SA (9a) and SO arguments (9b) :

Data from Tavares (2005), pp. 208 & 426

There have been a few analyses proposed to explain Wayana's person marking, including active-stative. If one focuses on the first and second person prefixes, that might be reasonable, as S is marked either as A or O, a characteristic of active-stative languages. Yet this analysis holds for first and second person only – the other persons (i.e. dual person) "do not present such a binary opposite" (Tavares, 2005, pp. 208); as well, the "semantic basis is lacking" (Tavares, 2005, pp. 208). Other analyses include: inverse (Gildea, 1998, as cited by Tavares) and portmanteau prefixes (Hoff, 1995, as cited by Tavares). Instead, it is likely that the set I verbs in Wayana are of a mixed system that is not fully encompassed by one single analysis.[8]

Verbs in this set, characterized by the ambifix t- -(h)e around the verb root, are less complicated than set I verbs in regards to case. They are not inflected for person, tense, evidentiality, or number, and, as mentioned, present ergative case marking. The intransitive subject, S, and transitive object, O are both unmarked (having absolutive case), while the transitive object is marked for ergative case with the ergative morpheme, ja. Two examples follow:

Data from Tavares (2005), pp. 231

In the transitive sentence, 10., the A, which is the pasina fish, is marked as such by the ja ergative morpheme. The O, the speaker's fishhook pole, is unmarked. In the intransitive sentences, 11. and 12., the S (the speaker, and "a huge rain", respectively) are completely unmarked, just as the O is in 10..

Wayana thus shows split ergativity, with one set of verbs presenting ergative case, and the other, a mixed system. Wayana is a special case in split ergativity, since it seems that the two sets are not grammatically or semantically triggered, and occur freely in discourse. For example, both 13. and 14. use the verb root ("ënepï"/"bring"), yet 13. is marked as a t- -(h)e verb, and 14. is marked as a set I.

Data from Tavares (2005), pp. 201, 202

Even another Cariban language, Tiriyó, with split ergativity (and similar person-marking), "restrict[s]" t-V-(h)e verbs (the Tiriyo cognate) to the remote past" (Tavares, 2005, pp. 234), making Wayana's case system quite unique, even amongst Cariban languages.[8]

Wayana explicitly marks tense only on set I verbs, where tense morphemes are bound with aspect, modality, and number. Broadly, there are two categories of tenses – non-past, and past, which includes recent past, distant past, and habitual past; all tenses are marked with affixes. Aspect is not marked on Wayana verbs, but is "impl[ied] to different degrees" (Tavares, 2005, pp. 211) in each tense, or else given by context or other morphemes. Below is a chart of the suffixes for each tense.

Modified from Tavares (2005), pp. 210

The non-past suffix, -ja can refer to future events, facts about the world, and non-past sentences can encode aspect, such as imperfective habitual or continuous aspect. In the examples below, we see how habitual aspect is marked through another morpheme (15.), how near or distant future is marked through context (saying "in a minute" versus "next year") (16a., 16b.). We also see in 17. that, since perfective aspect is not explicitly marked, the non-past can be understood as either imperfect and continuous ("I am making bread"), or perfect ("I will make bread"), based on the context of speech.

Data from Tavares (2005), pp. 212–213

The recent past suffix, -ø, refers to "events that took place in the past twenty -four hours", including "events that have just occurred", "events that have happened hours ago, and events that happened in the previous day" (Tavares 2005, pp. 215). 11) and 12) provide examples for this tense:

Data from Tavares (2005), pp. 215

The distant past (referring to an event that occurred more than 24 hours before speech) is marked by two different affixes, -ne and kun-. Their distribution is based on the person of the verb's arguments. –ne occurs when one or both of the verb's arguments is an SAP, or when there is a free (pro)noun object before the verb in a 3A3O configuration. Kun- "occurs elsewhere" and marks third person as well as distant past. 20. gives an example of the distant past, using the –ne suffix (as there is a preverbal free noun O), and 21 gives an example of kun-.

Data from Tavares (2005), pp. 216

All past affixes only "imply perfective meaning" (Tavares 2005, pp. 212) on their own. However, two habitual suffixes can mark habitual aspect in the distant past. They are -(j)(ë)mëhneja and -(h)e. The former takes the same personal prefixes as non-past forms do, while the latter takes no personal prefixes, though they seem to be identical, semantically – marking a repeated event in the past.

22. and 23. provide examples of -(j)(ë)mëhneja and -(h)e, respectively.

Data from Tavares (2005) pp. 218, 237

While set I verbs mark tense, t-V-(h)e verbs do not have morphology to do so. They do, however, refer to events in the past, present, and future, with different aspects. Tavares does not make it clear in her grammar how those events are referred to, but likely it is through context, and other morphemes, such as the habitual morpheme hepï (from the data presented on Tavares 2005, pp. 232).[8]

[1] This can be a free pro(noun) that is the O on a transitive verb. See (1b) for an example. Also note that the differing realizations of the prefixes are phonologically conditioned or based on tense (in the case of 3A3O)