5.1. Noun classification

Kamsá exhibits a grammatical system that -typologically- falls somewhere between noun class and noun classifiers, fitting neither definition neatly as shown in this section. However, here I choose the term «class» to refer to the system in Kamsá, because the markers are an obligatory part of the noun and they trigger agreement.

There are at least eight class markers that exist in the language, as well as a subset of nouns in which the noun itself is a suffix (these are often called «repeaters» in other South American languages). On nouns, the class markers are not separable morphemes, but rather they form a phonological word with their host and cannot be removed or changed. Demonstratives, adjectives, and numerals that agree with a noun, however, sometimes display a matching suffix, which is indeed a separable morpheme (whether or not these modifiers receive such marking depends on their position in relation to the noun, which is discussed below). A few classes are signaled by both a prefix and a suffix, but no classes are signaled by a prefix without a suffix. As is common cross-linguistically within gender/class systems, nouns of a given class in Kamsá exhibit shared semantic features.

Peculiarly, the majority of nouns in Kamsá do not belong to any noun class, and there are no classes encoding animate referents, whether human or non-human, or biological gender. Nouns that exhibit class markers represent a very small subset of the nouns in the language. Noun class markers have been described in several works, including ^{McDowell (1994)}, ^{Howard (1967)}, and ^{Juajibioy Chindoy & Wheeler (1974)}, all providing slightly different sets of classes. ^{Sandoval Camargo (2017)} also analyzes these affixes as having characteristics of both noun classes and classifiers. Here, I define a class as a set of nouns requiring a particular agreement marker from a modifying adjective. Based on this requirement, I present eight classes. Table 5 shows the morphological form of each class, along with a semantic description and an example of one of the nouns in the class. The numbering of the classes is arbitrary. In addition to class markers, the language also has repeaters (discussed below).

Agreement is obligatory for adjectives that follow the noun as well as for modifiers (determiners, adjectives, and numerals) that are being used anaphorically without the noun. However, when the adjective precedes the noun, it does not exhibit the class marker, as seen in example 5a. Only when an adjective follows a noun, as seen in 5b, it does receive the class marker to signal agreement.

The agreement marker is also present when a following adjective is being used predicatively, as in example 5c.

More research is needed to determine why there is this distinction in adjective marking; thus far, based on my corpus, there is no compelling evidence to explain the difference in marking other than surface word order.

Adjectives are the only modifiers that can either precede or follow nouns. Other modifiers of NPs (such as numerals and determiners) always precede their noun. In examples 6a-c, the modifiers precede the noun, and do not agree with the noun in class.

A modifier as a numeral or determiner can follow a noun only when it is being used anaphorically. When this happens, it is marked for class agreement, as seen in example 7, where the numeral «unga» ‘three’ has the class marker «-be» to refer to «mashakbe» ‘lulo’.

Three of the eight classes require the agreeing adjectives to have the prefix «wa-» in addition to their respective suffixes. Not all nouns of each of these classes have the prefix «wa-» (although many do) but the adjectives always do. Examples 8a-c show one example of each of these classes: «-sha» (for hairy things), «-jwa» (for cloth), and «-ya» (for clothes). In example 8a, the adjective «botaman» ‘beautiful’ has both the suffix «-sha» and the prefix «wa-» to agree with «wa-jajonë-sha» ‘bird’s nest’. In example 8b the adjective «bwangan» ‘red’ has the prefix «wa-» and the suffix «-jwa» and in example 8c «bwangan» ‘red’ has the prefix «wa-» and the suffix «-ya».

Example 8d shows an example from the «-ya» class in which the noun only has the class marker «-ya» (no prefix) while the adjective has both the prefix and the suffix.

Classifier and gender systems are common in Amazonian languages but not in Andean languages (^{van Gijn, 2014}). Regardless of whether Kamsá’s system should be considered a classifier system or class/gender system, it patterns more closely with Amazonian languages than Andean languages.

5.2. Repeaters

5.3. Evaluatives

Kamsá has at least three evaluative nominal suffixes-that is, a morphological means of assigning a value (such as big, small, good, or bad) to particular referents, by affixing to a noun. These may also occur on adjectives as agreement markers but are optional. The three evaluatives are: a diminutive «-tem», a pejorative «-ëfja» (with allomorphs «-ëja» and «-ëfa»), and an affective (or commiserative, i.e., with the sense ‘poor X’) «-jema». Evaluatives are often seen to be derivational morphology or somewhere between derivational and inflectional. Table 6 illustrates the three Kamsá evaluatives, providing an example of each.

5.4. Oblique case markers

Kamsá exhibits nominal marking that lies somewhere along a spectrum connecting case suffixes and postpositions-not meeting all the criteria for either category. There is no core argument case marking in the language (i.e., no overt marking for nominative or accusative arguments). There is also no case marking for the semantic goal in a sentence; that is to say, what would be an indirect object in some languages does not have any special oblique marker in Kamsá. However, there are various oblique case markers, which are distinct from postpositions in that they are not separate words: they are bound morphemes that fulfill common semantic functions found in oblique case markers (e.g., signaling instrumental, comitative, possessive, or allative thematic relations). By «case» I do not mean to imply any theoretical definition of «Case», but rather just a description of bound oblique marking that signals various thematic relations. They could also possibly be analyzed as clitics rather than suffixes.

My main argument for calling these suffixes case markers (or oblique markers) and not postpositions is that they are phonologically part of the word. They cause phonological changes in noun stems. Elsewhere in the language, there are no phonological changes across word boundaries. This is not to say they are not clitics, only that they are not separate words.

All cases except the genitive are different for animate nouns and for inanimate nouns: namely, animate nouns have the suffix «-bi» preceding the case marker whereas inanimate nouns do not.

As stated above, in Kamsá, A, S, and O do not have any overt case marking. The noun stem in NPs fulfilling such grammatical roles appears alone without any affixes. Furthermore, in ditransitive constructions, neither the Theme nor the Goal argument has any special marking. In example 10a, the A argument «batá» ‘aunt’ has no suffix. In example 10b, this same noun serves as the O argument, and still has no suffix. In example 10c, «batá» ‘aunt’ is the S argument, yet again with no suffix. Finally, in example 10d, this noun is the recipient (Goal) of the ditransitive verb and has no case-marking suffix. Note that the other object, «sana» ‘food’, which is the gift (Theme) argument, also has no case marking.

Obliques, however, are marked with the oblique case marking suffixes. The oblique case markers are shown in Table 7. The form of the case is shown in the first column, and the label in the second. The fourth column has examples of two words with the case marker, using an animate and inanimate noun («batá» ‘aunt’ and «shknen» ‘plate’).

The following sentences exemplify each of the noun cases. Note that for possession, the dependent is marked, as seen in example 11a.

Genitive

Benefactive

Comitative

Locative

Allative

Ablative

Illative

6.1. Subject marking

For intransitive verbs, the subject marker can encode person, number, or both; that is to say, there is variation in how specific the person/number marker has to be, how much information it encodes. For example, first person plural argument can be indexed on the verb with «bsë-» (which encodes both first person and plural), or with «së-» (the first person marker, which does not encode number but is generally assumed to be singular unless there are contrary indication), or with «mo-» (the plural marker, which does not imply anything about person). Similarly, a second person plural argument can be indexed with the prefix that encodes both second person and plural «s̈mo-», or with the second person marker «ko-» (which is generally singular), or with the plural marker «mo-». There is no special marker that indexes both first person and dual number, so first person dual arguments can be indexed with the first person plural marker, the first person (generally singular) marker, or the (unmarked-for-person) dual marker.

However, second person dual arguments have a designated marker, «s̈o-», and so they can be marked as such, or with the second person plural marker «s̈mo-», or with the dual marker «bo-» which is unmarked for person⁵. Thus, plural referents can be indexed on the verb for just their person (first or second) or for their number (plural) or for both person and number (first person plural or second person plural); dual referents can be indexed with either dual or plural verbal morphology for the second person and with either dual or plural or unmarked (but generally taken to be singular) verbal morphology for the first person; singular referents are indexed by their person alone. All person/number combinations for speech-act participants have a unique agreement prefix, except first person dual (as there is no verbal morpheme that encodes both first person and dual number).

Table 8 shows the past tense conjugation of the verb «jaban» ‘come’ with the most specific verbal prefixes for each person/number.

6.2. Object marking

Verbs index objects when the object is first or second person. The object marker precedes the subject marker. When the object is second person and the subject is first person, the morpheme in the subject slot is dual subject as seen in example 12a. When the object is first person and the subject is second person, the two are marked accordingly, with the first person object marker followed by the second person subject marker as seen in example 12b. In example 12c, the object is first person and the subject is third person, so the object is marked with «s̈-» (the first person object marker), and the subject is marked with just «o-»⁶. Note that for first person the subject and object markers are different: «së-» for subjects and «s̈-» for objects whereas for the second person, the subject and object marker are the same: «k-».

6.3. Evidentiality «n-» and «j-» or «Ø-»

There is a two-way evidentiality system expressed through verbal prefixes: «n-» signals that information being conveyed is known by the speaker and «j-» or «Ø-» for information about which the speaker is not sure. It is not clear if the form of the evidential marker for unknown information is underlyingly «j-» or «Ø-», because it may be occurring with the verbalizer «j-» and thus could be underlyingly /j-j-/. Because there is a prohibition against geminates in Kamsá, this form would be realized as [j-]. The evidential marker comes after the person markers and before the aspect markers. The marker for known information is «n-». It is often used in the recent past, and never used in the distant past. It is almost always used in first person singular, perhaps because if someone is talking about himself/herself, he or she presumably knows the information.

^{Jamioy Muchavisoy (1999)} describes the evidential system as being «n-» for ‘close’ and «j-» for removed. According to Jamioy Muchavisoy, in example 13a, the «j-» shows that the speaker has distance from responsibility, while in example 13b the speaker is closer. More research is needed to determine the uses and meanings of these two evidential markers. This system could be analyzed as engagement rather than evidentiality, referring to a grammaticalized system that can express the «speaker’s assumptions about the degree to which their attention or knowledge is shared (or not shared) by the addressee» (^{Evans, Bergqvist & San Roque, 2018}).

6.4. Imperatives

There are two ways of forming commands in Kamsá: either through the verbal imperative marker, or by using a second person future form of the verb. In example 14a, the verb «jatbana» ‘collect’ is marked with the second singular marker «ko-» followed by the future prefix «ch-». The sentence in example 14b shows both strategies of command: the morphological imperative «mo-lempia» ‘Clean!’ and the future-as-imperative with «kochjase» ‘Eat!’. 14c shows the morphological imperative with a native Kamsá word, «jasan» ‘eat’.

8.1. Word order

Transitive clauses have very free word order, so much so that some have asserted that Kamsá has no dominant word order (^{Fabre, 2001, p. 5}). In his typological sketch of Kamsá, Fabre says that out of one hundred instances selected from his corpus he found a slight predominance of VO (56%) over OV (44%). Most AGENT NPs (presumably in transitive sentences) appear first, but the inverse order is not rare either (^{Fabre, 2001}). The following examples show various options for word order. In example 16a and example 16b the word order is AVO and in example 16c and example 16d the word order is SOV. All six possible orders of S, O, and V are possible in Kamsá.

8.2. Causatives

Causatives are formed morphologically. A causative meaning is formed by putting an object-marking prefix on the verb to represent the causee. This is the case for both transitive and intransitive verbs. A verb such as «j-os̈achn» ‘cry’, which has no object, will have an object marking prefix to index the causee. In example 17a, the verb «j-os̈achn» ‘cry’ is marked with «to-» to show past tense and index a third singular subject (Alena). In example 17b, «j-os̈achn» ‘cry’ is a causative and is marked with the direct object pronominal prefix «s̈o-». In example 17c, the verb is marked with the dual «bo-». In example 17d, only the first person subject is marked with «te-», an alternate past form for the first person singular. The function of «j-» is unclear. In example 17c the «bo-» shows that there is an animate subject and an animate object, while in example 17d the verbal prefix just marks first person singular subject and past tense. In examples 17e-f, a causative with «jasan» ‘eat’ and an object is shown. As in example 13b, the verb is marked with «s̈o-» the first person object marker, to show that the causee is first person singular. The object «bay-ëng» ‘worms’, the things being eaten, do not have any special marking.

8.3. «-am» on a complement

To show purpose, the suffix «-am» is added to an uninflected verb. This is perhaps the same suffix as the benefactive case marker on nouns and it seems to be a nominalization process. The suffix «-am» must be added to the uninflected verb, not any other constituents (e.g. not the object). Example 18a shows «-am» on the uninflected verb ‘buy’ and example 18b shows «-am» on a noun, Carmen, for comparison.

8.4. Subordinate clauses with «=as», «ora» and «kaus»

Subordinate clauses are often formed with clause-final clitics as «=as» or «=ora». Clauses with «=ora» are always temporal whereas classes with «=as» can encode many different subordinate relations. Subordinate clauses may either precede or follow matrix clauses. A common strategy for marking subordinate clauses employs the clitic «=as», which comes at the end of the subordinate clause. In my data, there are no sentences with «=as» in which the clause ends in an S or A. The clitic «=as» can encode a variety of subordinate meanings, including causal, concessive, temporal.

Temporal clauses may use the subordinator «ora» ‘when’, which may derive from Spanish «hora» ‘hour, time, term’. The subordinator «ora» comes at the end of its clause (the temporal subordinate clause). It has the alternate forms «or», «gor», and «orna». In example 20, «gor» comes at the end of the first clause.

The subordinator «kaus», which also shows the forms «kausa» or «kausna» («kaus» + topic marker «=na»), may be used to show the cause of an event or state. In example 21, «kausa» comes at the very end of the clause, after the verb. The verb is fully inflected for person and number.

8.5. Comparatives and superlatives

Comparative constructions are formed with «mas» (from Spanish «más» ‘more’) followed by the adjective. The object of comparison is marked with the benefactive «-am» for inanimate nouns and «-biam» for animate nouns. There are no irregular comparatives; all comparatives are formed with «mas» + adjective. Superlatives are also formed with «mas» and the adjective (i.e., no distinction is made between ‘more’ and ‘most’).

In example 22a, the adjective «bna» ‘tall’ is preceded by «mas» and «tobias̈» ‘girl’ is marked with the benefactive «bi-am», while «bobonts» ‘young man’ is not marked. In example 22b, the superlative is shown with «mas» ‘more’ and «ts̈abá» ‘good’.

8.6. WH questions

There is wh-movement for questions: question words always come at the beginning of the clause. In example 23a, the word order is SOV, with the object «wameshnen» ‘mote’ coming before the verb. In example 23b, the question word «ndayá» ‘what’ is at the very beginning of the sentence. Similarly, in example 23c, the place the girl went, the market, follows the subject and in example 23d, the allative question word «ndmoy» ‘to where’ is at the beginning of the sentence. As discussed above, declarative sentences can have any word order. Questions, however, must have the question word first in the sentence.