nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
1. Introduction
1.1. Presentation
This is a descriptive study of Cabecar postpositions which combines the synchronic and diachronic perspectives. The structure of the presentation is as follows: The rest of this introductory section provides some basic information on Cabecar. Section 2, the bulk of the paper, details the structure of the postpositional phrase, classifies the postpositions, and describes their function and structure. Section 3 assesses the degree of grammaticalization of Cabecar primary postpositions by a set of criteria defined beforehand. To some extent, this is an extract from the Cabecar grammar currently elaborated in cooperation with Guillermo González Campos from Universidad de Costa Rica (González & Lehmann 2023).
Cabecar postpositions are not particularly noteworthy in typological perspective; with few exceptions, they are of nominal origin. Nor is the theory their description is based on here a breakthrough, although two conceptions employed here are not widely used in the description of postpositions: First, it is expedient to distinguish between event-oriented and participant-oriented postpositions (§2.3.2). Second, the degree of grammaticalization of the class of primary postpositions is assessed on a cross-linguistically applicable scale (§3.1).
1.2. Background on the language
Cabecar is spoken by some 14 000 people in certain territories in the south-east of Costa Rica as shown in Diagram 1.
Cabecar is a Chibchan language of the Viceitic branch as shown in Diagram 2. Its dialects are subdivided into a northern and a southern group. Examples from the latter are marked by ‘s.d.’
Diagram 2 Genetic affiliation of Cabecar
| Northern Chibchan Paya | ||||
| Southern Chibchan | Votic | Rama | ||
| Guatuso = Maleku | ||||
| Huetar† | ||||
| Isthmic | Western | Viceitic | Cabecar | |
| Bribri | ||||
| Naso | Terraba† | |||
| Teribe | ||||
| Boruca† | ||||
| Doracic | Dorasque† | |||
| Changuena† | ||||
| Eastern | Guaimiic | Guaymi = Ngäbe | ||
| Bocotá = Buglere | ||||
| Cuna | ||||
| Magdalenian | … | |||
The following typological properties of Cabecar are prerequisites for the analysis: Beside usual word classes, the language has a class of verboids which includes positionals. The morphology comprises compounding, derivational and inflectional morphology in nouns, adjectives, and verbs. It is exclusively suffixing, mildly synthetic, and mostly agglutinative. Syntactic relations are characterized exclusively by dependent marking; heads are not indexed for dependents. Alignment of the actants of intransitive and transitive verbs is ergative in terms of case marking, but accusative for most phoric control. Alignment of the actants of a ditransitive verb is indirective. Word order is organized at two levels. The lower level is characterized by the inner complement relation visualized in Diagram 3: a bare NP directly precedes its head [K]X, where K may be a verb, a relational noun, or a postposition.
At this lower level, constituent order is rigidly left-branching. At the higher level, i.e., outside the inner-complement construction, constituent order is free.
2. Cabecar postpositions
2.1. Postpositional phrase
The postpositional phrase is an inner-complement construction, with the construction of Diagram 4 instantiating Diagram 3.
While the inner structure of P is analyzed comprehensively in the following subsections, it may be noted for N that it, too, may be complex up to the level of a clause (§3.2.1), converted into an adverbial clause by P functioning as a subordinator.
Example (1) illustrates a transitive clause with two postpositional phrases, the first of which is the ergative actant, while the second is an instrumental adjunct. All the three binary syntagms are inner complement constructions.
(1) Sá te kjalá wa ni̱ma̱ kuk-ë.
PL ERG fishing.net INSTR fish catch-IPFV
‘We catch fish with a net.’ (Duchi_9.1)
2.2. Kinds of postpositions
Two criteria are employed in the classification of postpositions:
1) Syntactic function of the postpositional phrase: Most postpositions signal whether the phrase headed by them functions as a complement or as an adjunct. This produces the classes of grammatical vs. lexical postpositions. There are a few postpositions, though, which are polysemous in either marking a complement or specifying the semantic role of an adjunct.
Inside lexical postpositions, a subclassification into event-oriented and participant-oriented postpositions obtains as spelled out in §2.3.2.
2) Morphological structure of the postposition: A monomorphemic postposition is primitive. Postpositions comprising more than one morpheme are complex. In the latter case, the categories of the morphological constituents and their relation constitute subclasses of postpositions as spelled out in §2.4.2.
Cross-classifying postpositions by these criteria produces Table 1:
Table 1 Classes of adpositions
| syntactic function morphological structure \ | grammatical | lexical |
| primitive | + | + |
| complex | - | + |
As may be read from Table 1, grammatical postpositions are primitive; complex postpositions are lexical.
2.3. Functions of postpositions
The criterion of syntactic function distinguishes between a postposition required by a rule of grammar and a postposition coding a specific semantic relation.
2.3.1. Grammatical postpositions
A grammatical (or “structural” or “functional”) postposition is one that may be governed by a verb or verboid or be otherwise required by a rule of grammar. The paradigm is enumerated in Table 2. In the semantic explanations, N represents the entity designated by the complement N of the postposition of Diagram 4.
Table 2 Grammatical postpositions
| form | label | meaning |
| ia | dative | to, for N |
| te/të | ergative | N is the actor in a situation also involving an undergoer |
| wa~kú | dispositive | by intervention / under control of N |
There is no known origin for the first two grammatical postpositions.1 While the criterion of the syntactic function generates the two classes of grammatical and lexical postpositions, these are not neatly disjunct. On the one hand, besides the postpositions of Table 2, a few lexical postpositions are governed by some verbs. On the other, the dative postposition may govern certain adjuncts, including a benefactive one (see example 25); and the ergative postposition is homonymous or cognate with the causal subordinator.
2.3.2. Lexical postpositions
A lexical postposition is one which codes a specific semantic relation between its complement and its head. As it is not governed by the latter, the postpositional phrase functions as a modifier. It modifies a verbal or verboidal clause core. Within the class of lexical postpositions, first the primitive ones will be reviewed. These subdivide by the semantic criterion of their orientation, yielding the classes of event-oriented and participant-oriented postpositions.
2.3.2.1. Event-oriented postpositions
An event-oriented postposition signals an oblique (non-direct) syntactic relation of its complement NP vis-à-vis its head - generally a verb or verboid. On a semantic basis, two kinds of event-oriented postpositions are distinguished: A participative postposition joins a peripheral participant to a situation. A situative postposition situates the modified situation according to a local, temporal or manner parameter; typical complements have a low position on the empathy hierarchy. Table 3 enumerates the participative postpositions.
Table 3 Participative postpositions
| form | label | meaning |
| da | comitative | (simultaneous) with N, at the time of N |
| wa | instrumental | with (tool) N, from (material) N |
| wä́së́(ka)/wö́ksë | conductive | moving actor takes N with him |
| yí̱ka | aversive | from, against N; before (time) N |
The first three items of Table 3 specify some sort of accompaniment. The last entry indicates an oppositive relation. N wa codes the role of an inanimate N as instrument used by the agent of the situation, as in (1) and (2). Wa INSTRUMENTAL is a grammaticalized variant of wa PERLATIVE (examples 16 and 29; cf. Span. por).
(2) ják wa
stone INSTR
‘with a stone’
N da may code the company of some concrete, typically human entity N and normally relates this to a participant of the situation, as in (3).
(3) yís da
1.SG COM
‘with me’
However, frequently N is some temporal or propositional entity, so N da (ra in (4)) means ‘at the time of N, when N, if N’.
(4) Yoko bukalëi te sá tk-á=wa ra,
[fire wasp:SPC ERG 1.PL cross-PFV=TOT] COM
sá dälä-w-é yökö käi.
1.PL burn-CAUS-IPFV fire MNR
‘If the fire wasp stung (us), it burns (us) like fire.’
This is an example of the principle that temporal concepts and relations are derived from more concrete ones.
[ N wä́së́ ] PoP codes the relation of the actor of the situation to an additional animate or inanimate participant N which the actor takes along in some movement, as in (5).
(5) Yaba tsíne shk-á=mi du wäsë.
child small walk-PFV=AND bird COND
‘The little boy went away with the bird (i.e. taking it with him).’
The conductive differs from the comitative and the proprietive (§2.3.2.2) in this feature of motion. It differs from the instrumental both in this feature and in being insensitive to N’s empathy.
As for situative postpositions, there is no dedicated manner postposition. All temporal postpositions are basically local postpositions. Therefore, the set of primitive situative postpositions is exhausted by Table 4.
Table 4 Primitive local postpositions
| form | label | meaning |
| (ja)mi | adessive/allative | at, by N [+ contact] |
| ka | lative | from N; at, to(wards) N; along (some place or time) N |
| kí̱ | superlocative | (up)on, over N [+ contact]; with respect to N |
| kjane | apudessive/-lative | near, by, beside N [- contact] |
| kjë́ | prosecutive | after N |
| na | inessive/illative | in(side), into N |
| ska | locative | in, on, at, to (place or time) N |
| wa | permeative | through(out) N, past N |
Unlike English prepositions such as at, to, from etc., the basic local postpositions are not regularly sensitive to the distinction between rest and motion. On the other hand, most local postpositions have non-local uses, too. By and large the language does not distinguish formally temporal concepts from local concepts. Beside such temporal uses, local postpositions have more abstract senses, too. For ki SUP, some are mentioned in Table 4, including a motivative use illustrated by (6).
(6) Yís kuta sibí-n-á=ká duláklëi kí.
1.SG sister jealous-D.MID-PFV=ASC boy SUP
‘My sister got jealous because of a/the boy.’
2.3.2.2. Participant-oriented postpositions
A participant-oriented postposition does not (like an event-oriented postposition) assign an oblique role to its complement vis-à-vis the predicate. Instead, its complement bears a predicative relation to some actant of the main predicate and shares its semantic role. Consequently, a participant-oriented postpositional phrase may constitute the predicate complement in a copula clause. In a verbal clause, it functions like a secondary predicate on the actant in question.
Again, by a semantic criterion, there are two subclasses of participant-oriented postpositions. Possessive postpositions relate a possessum to its possessor. The two possessive postpositions of Table 5 form a converse pair.
Table 5 Possessive postpositions
| form | label | meaning |
| cha | property | N’s (property), belonging to N |
| ta | proprietive | having N, provided with N |
[ N cha ]PoP means ‘N’s (property), belonging to N’. The main function of N cha is as the predicate complement of the copula in a predication of belonging (7).
(7) Bikö-le rä bá cha.
how.much-ever COP 2.SG PPY
‘All of them are yours / all of it is yours.’
Although this item has cognates in other Chibchan languages, there is no etymology for it; so, it must be old. It is some sort of relational dummy possessum with the original meaning ‘(someone’s) property’; which explains its participant-oriented syntax.
(8) Jé=jami sä te sájé cha wëik-ë
D.MED=AD 1.PL ERG 1.PL PPY harm-IPFV
r-á=mi tái ma itjä.
emerge-PROG=AND big wherever
‘Moreover, we constantly despise what is ours everywhere.’
What appears to be the substantivization of a postpositional phrase in (8) may reflect the original category of the formative, viz. a noun.
[ N ta ]PoP means ‘having, provided with N, characterized by N’. The main function of N ta is as the predicate complement of the copula in an ascription of possession, as in (9).
(9) Yís dä inaklä ta.
1.SG COP money PPV
‘I have money.’
The diachronic source of the postposition ta is a form of a positional meaning ‘be with N, have N’, which, however, is not extant in Cabecar. The semi-finite origin of positionals explains its participant-oriented syntax.
The other subclass of participant-oriented postpositions is equiparative postpositions. These function in the formation of comparative constructions of equality (superiority is marked by kí SUP, Table 7) and are, except for a variant of súta, enumerated in Table 6.
Table 6 Equiparative postpositions
| form | label | meaning |
| kä́i | equal | as/like N |
| sú̱ta | similative | like N, similar to N |
| yë́ | translative | in(to) the form of N, as N |
Table 7 Postpositions consisting of two primitive postpositions
| form | structure | meaning |
| ia=ra | dat=com | according to N |
| ka=na | lat=in | since/up to [point in time] N |
| kí̱=ka | sup=lat | on, over [+ contact] N; because of N; to experiencer N; (more) than N |
| kí̱= ta | sup=ppv | (more) than N |
| kí̱=wa | sup=per | throughout the surface of N |
| kja̱ne̱=ka | apud=lat | near N, towards N |
| na̱=ka | in=lat | into N; to N[’s senses] |
| ska=ra | loc=com | during, in [process] N |
| wä́së́=ka | cond=lat | moving subject takes N with him |
The postpositional phrase [ N kä́i ] (with its allomorphs) means ‘like N, the N way’, as in (4) and (21)f. [ N súta ]PoP means ‘like, similar to N’, as in (10). Manner constructions and comparative constructions on their model may use this postposition.
(10) yís súta
1.SG SIM
‘like me’
[ N yë́ ]PoP means ‘(turned/transformed) in(to) the form of N, (serving) as N, behaving like N’. In (11), it functions as a predicate complement of a verboid.
(11) jé rä koñó yë
D.MED COP rat TRL
‘that one was transformed into a rat’
The TRANSLATIVE is the dynamic counterpart of the SIMILATIVE, as it involves a transformation.
As noted, at least some of the postpositions of Table 5 and Table 6 originate in an (extinct) positional or a noun. A positional may head a participant-oriented adjunct; a noun may function as a displaced attribute. Several of the synchronic syntactic peculiarities of these postpositions may thus manifest persistence of their original category.
The two classes of event-oriented and participant-oriented postpositions have counterparts in other languages. Typically, the counterparts of participant-oriented postpositions either assign no case to their complement or they assign a different case. Generally, equiparative operators assign no case to their operand and are therefore not subsumed under the concept of postposition.2 This criterion is of no help in Cabecar since no postposition assigns a case to its complement (§3.2.2).
2.4. Morphological structure of postpositions
2.4.1. Primitive postpositions
Primitive (or simple) postpositions are monomorphemic and mostly monosyllabic. They form a closed set and are listed exhaustively in Table 2 - Table 6. Most postpositions are primitive and monosyllabic and lack an etymology. Some members of Table 3 - Table 6 may not have the look of a primitive postposition. However, none of them except súta is synchronically morphologically transparent. As will be seen in §3.2.4, primitive postpositions are structurally halfway between postpositions and suffixes.
2.4.2. Complex postpositions
A complex postposition is one consisting of more than one morpheme. It may be a combination of primitive postpositions (§2.4.2.1) or a denominal postposition (§2.4.2.2). There is currently no formation of deverbal postpositions.
All complex postpositions are lexical postpositions; their postpositional phrases are adjuncts rather than complements. Most complex postpositions have local or temporal functions.
2.4.2.1. Combination of two postpositions
Diagram 4 provides for only one position of P. However, sequences of two postpositions are not excluded. The systematic source for such complex postpositions is Diagram 5:
Here, P2 governs the PoP headed by P1. A subset of postpositions can take the position of P2 here, allowing either an NP or an adverbial as their complement, and the latter may be a PoP. This subset includes da comitative, ka LATIVE, na INESSIVE and ta PROPRIETIVE. Table 7 provides a representative sample of complex postpositions originated by the univerbation (Lehmann 2019, 2020) of P1 and P2 in Diagram 5.
Example (12) illustrates the univerbation of the dative with the comitative postposition to mark a judivative adjunct.
(12) Bá ia=ra, ¿mané sälírba ñá-la kjëi=ta ...?
1.SG DAT=COM which yam eat-S.POT taste=PPV
‘According to you, which yam is tastier ...?’
2.4.2.2. Denominal postpositions
Diagram 4 represents the postpositional phrase based on an unanalyzed (and possibly monomorphemic) postposition. For complex postpositions formed by rules of syntax, a dynamic view is more appropriate. These postpositions may be called denominal only in the sense that their lexical basis is a noun. However, they are not derived from it by any process of stem formation. Instead, the origin of a postpositional construction based on a complex postposition of Table 8 is a syntactic construction as shown in Diagram 6. Example (13) features its three components.
Table 8 Local postpositions based on region nouns
| postposition region noun \ | ka lat | (ja̱)mi̱ ad | kí̱ sup | na̱ in | s(a)ka loc | wa per | |
| bata | tip | on top of | at the end of | ||||
| kakirë́i | side | next to | beside | ||||
| kjä́/kö́ | edge | at the side of | beside, near | ||||
| ni̱kí̱ | inferior region | under | below | at the bottom of | |||
| ñá̱k | lower end | near the lower end of | at the other end of | along the other side of | |||
| shá̱ | middle | amidst | in the middle of, among | ||||
| tä́ | track | after (in a sequence) (15) | behind | after (16) | |||
| wä́ | face | in front of (17) | opposite (18) | ||||
(13) kal digí=na
tree bottom=IN
‘below the tree’ (s.d.)
In Diagram 6, Nd is typically a concrete noun phrase and Nd/h is typically, though not necessarily, a region noun which takes Nd as a possessive attribute. The complex NP thus formed depends on P, typically a primitive local postposition, like na in (13) and jami in (14).
2.4.2.2.1. Local denominal postpositions
If Nd/h of Diagram 6 is a region noun and P is a local postposition, they form a specific local postposition. For instance, in (14), ñala kjä=jami ‘by the roadside, next to the road’ describes the subject’s position more specifically than a mere ñala jami ‘by the road’. In this way, finer distinctions not made by the local postpositions themselves may be made by such complexes of Nd/h P.
(14) Pablo kaldu ñala kjä=jami.
Paul POS:stand path edge=AD
‘Paul is standing by the roadside.’
Table 8 visualizes the degree of productivity of this pattern by listing a representative set of combinations constituted by region nouns which combine with more than one local postposition and local postpositions that combine with more than one region noun.
Some of the formations listed in Table 8 are illustrated by examples (15-18).
(15) Jié d-ël=ju yís mína tö=ga.
1.SG emerge-s.PRF=AM 1.SG mother track=lat
‘He has arrived after my mother.’ (Margery 1989 s.v. töga, s.d.)
(16) Bá shk-ö yís tä=wa!
2.SG walk-SBJ 1.SG track=PER
‘Walk after me !’
(17) S’ j-ö =a min-a bá wä=ki !
1.SG let-SBJ=TOT go-SNR 2.SG face=SUP
‘Let me go in front of you!’
(18) Kjös káte öshkoro ju wä=saka
black.oak POS:stand chicken house face=LOC
‘The black oak stands opposite the chicken coop.’
As may be seen, the pattern is uniform, but not fully productive.
2.4.2.2.2. Non-local denominal postpositions
Beside the local postpositions, there is also a set of complex postpositions with a more abstract meaning. Table 9 lists some such postpositions. Examples (19) and (20) illustrate some of their uses.
Table 9 Non-local postpositions based on nouns
| form | structure | meaning |
| bala=na | interior=in | during N |
| kju̱é̱i̱=kí̱ | restriction=sup | because of N |
| tä́rí=wa | track=per | instead of N |
| tsá̱=ta | top=ppv | (more) than N |
| wä́=na | face=in | to N(‘s mind) |
(19) Yís m-á=ju tuwína bala=na.
1.SG go-PROG=AM night interior=IN
‘I am going during the night.’ (Margery 1989 s.v. balaná; s.d.)
(20) Yís yaba m-á yís tärí=wa.
1.SG child go-PROG 1.SG track=PER
‘My son is going in my stead.’
With the exception of ta, the postpositions taking the position of P in these formations are, again, the primitive local postpositions of Table 4. Likewise, except for kju é i, the nouns forming their erstwhile complements are, again, region nouns from Table 8f. As already observed for primitive postpositions, these complex postpositions undergo metaphorical extension of their basic local meaning.
As long as the combination of the noun with the postposition is a syntactic construction, it takes the regular form of Diagram 6. Now, both the set of nouns operative in the position of Nd/h and the subparadigm of primitive local postpositions P are limited. Consequently, the number of recurrent combinations is not large. Semantically, the sequence Nd/h P is perceived as a specific relation born by Nd to whatever the PoP of Diagram 6 depends on.
Consequently, in such a configuration, the sequence Nd/h P tends to univerbate (against the syntactic structure) and lexicalize as a complex postposition. The result is a construction according to Diagram 4, where the complex postposition takes the position of P. The transition from the combination Nd/h P of Diagram 6 to a complex postposition is fluid; writing with or without an intervening blank varies.
The conversion of complexes Nd/h P of Diagram 6 into postpositions has been operative in the language for a long time. Consequently, most current postpositions, whether transparent or opaque, are based on nouns. However, the set of postpositions is an open and heterogeneous one, containing members of different age and of different degrees of transparency. Some formations are not easily assignable to either Table 7 or Table 8.
Some postpositional phrases based on ta PROPRIETIVE univerbate and lexicalize, too. The similative postposition súta (10), listed as primitive in Table 6, is, in fact, one of these. Etymologically, it is denominal, as it is ta PPV governing some nominal form of su a ‘see’, thus ‘having the view (appearance) of N’.
3. Grammaticalization status of postpositions
3.1. Criteria
There is a gamut of increasing grammaticalization of case relators from the relational noun via the postposition to the case suffix. The operationalization of the general parameters of grammaticalization produces criteria for measuring the degree of grammaticality of postpositions (Lehmann 2015; Bisang et al. 2020). Table 10 gives a systematic survey of the criteria employed here. Some of them are tailored to the Cabecar system, which does not exclude their usefulness for other languages. Each criterion is a test that can yield a yes-no output. In their combination, however, they arrange the test items on a scale, for two reasons: First, each binary criterion divides the scale of grammaticality at a particular point. For instance, copying a case relator on subconstituents of an NP presupposes a rather high degree of grammaticalization, while direct government of its complement, i.e., without an intervening case, is found even with adpositions of a rather low degree of grammaticality. Second, even such criteria which relate to the same degree of grammaticality in one language need not generate the same result in another. There is so far no theory from which the systematic position of each criterion could be deduced.
Table 10 Relative degree of grammaticalization of adpositions
| grammaticalized criterion \ | less: like the major- category source | more: like an affix |
| Scope: complexity of complement | ||
| adposition takes clausal complement | yes | no |
| adposition takes coordinative complement | yes | no: adposition is repeated with each coordinate |
| adposition is copied on subconstituents of NP | no: there is one occurrence per NP | yes |
| Bondedness: tightness of bond between adposition and complement | ||
| adposition governs case | yes | no: complement is bare NP |
| adpositional syntagma can be split by intervening material | yes | no |
| complement can be left-dislocated and taken up by intercalated resumptive | yes | no |
| adpositions can be coordinated while sharing their complement | yes | no: complement is repeated for each adposition |
| Syntagmatic variability: | ||
| position of adposition vis-à-vis complement varies | yes | no |
3.2. Application to Cabecar postpositions
In the following, the criteria of Table 10 are taken as yes-no questions and answered for the Cabecar primitive postpositions.
3.2.1. Scope: complexity of complement
The complement of a postposition (N in Diagram 4) may, in general, be of any of the following categories:
a. pronoun (21)
b. lexical noun or NP (11)
c. clause (22)
(21) bá käi
2.SG MNR
‘like you’
(22) I w-o yís te i sh-ë=ké käi !
3 do-SBJ [1.SG ERG 3 say-IPFV=IPFV2] MNR
‘Do as I say!’
Adposition takes clausal complement? Yes. A postposition can govern a subordinate clause, as demonstrated for ra by (4), and for käi by (22).
Adposition takes coordinative complement? Yes. Mostly, a coordinative noun phrase is not governed by one postposition. Instead, the postposition is repeated after each coordinate component, like a suffix.
(23) I mu-á=wa i te bís tse-i wa
3 tie-PFV=TOT 3 ERG fique string-SPC INSTR
jémi tsini yagá wa
and rubber stuff INSTR
‘he tied it with a fique string and rubber matter’
However, leaving out the first occurrence of the postposition in such a construction is admissible (24).
(24) I kt-á=wa i te bák irä tebélí irä wa.
3 kill-PFV=EFF 3 ERG axe CRD knife CRD INSTR
‘He killed him with an axe or with a knife.’
Adposition is copied on subconstituents of NP? No. There is one occurrence of the postposition at the end of its phrase; it cannot be copied on subconstituents of an NP, as case suffixes may in some languages.
3.2.2. Bondedness: tightness of bond between adposition and complement
N in Diagram 4 cannot be zero; apart from a few lexicalized cases, postpositions have no alternate use as adverbs (like Engl. before, inside etc.). This is another feature in which the relation between a postposition and its complement is like the relation between a verb and its absolutive actant, justifying their subsumption under the concept of inner complement.
Adposition governs case? No postposition governs a case (there is no genitive, anyway); they all combine with a bare NP or an adverbial which may contain a semantically suitable postposition (as in §2.4.2.1).
Adpositional syntagma can be split by intervening material? Yes. The syntagma constituted by a postposition and its complement may be split up by intervening particles, in particular the negator ká (25).
(25) a. ká yís kuta te dí y-e tá i
NEG 1.SG male’s.sister ERG liquor drink-IPFV big
b. yís kuta ká te dí y-ë́ tá i
C. yís kuta te ká dí y-ë́ tá i
[#a - #c:] ‘my sister does not drink much corn drink.’
It can also be split up by some focus particles (26).
(26) Yís te yäk-ä-klä älí-w-á
1.SG ERG dine-SNR-FIN cook-CAUS-PFV
bás kjë́la=bä ia.
2.PL only=LIM DAT
‘I cooked food just only for you.’
Since these particles are clearly neither suffixes nor head of the complement of the postposition, the postposition cannot be a suffix.
Complement can be left-dislocated and taken up by intercalated resumptive? Yes. Optionally, the MEDIAL DEMONSTRATIVE jé anaphorically takes up the complement even in the inner complement construction (Diagram 3) (as elsewhere), as illustrated by the possessive attribute construction in (27).
(27) yëria (jé) kuta-wá
hunter D.MED male’s.sister-PL
‘hunter’s sisters’
Example (28) demonstrates application of this criterion to a dative postpositional phrase.
(28) jé rä sä yu-ä kié Guillermo jé ia
D.MED COP 1.PL form-VSN named William D.MED DAT
‘that is for the professor named William’
Adpositions can be coordinated while sharing their complement? No. Postpositions cannot be coordinated as in the translations of (29) f. Instead, the complement is repeated for each postposition, either anaphorically, as in (29), or literally, as in (30). This is the same for the primitive postpositions of (30) and the complex postpositions of (29), and for different coordinators.
(29) ják bata=ki jé=jami i niki=ka
stone tip=SUP D.MED=AD 3 bottom=lat
‘on and under the stone’
(30) Shára wa irä=le Shára ska irä=le
Grano.de.Oro PER CRD=DUB Grano.de.Oro LOC CRD=DUB
‘near or in Grano de Oro [toponym]’
Suppressing the second occurrence of the complement from the constructions of (29)f renders them ungrammatical. By the criterion of (non-)coordinability, the Cabecar case relators thus behave more like suffixes than like postpositions. This behavior of postpositions vis-à-vis their complement is inherited from the impossibility of having a coordinate head for a single possessive attribute.
3.3. Balance
Table 11 is a copy of Table 10, filled in for Cabecar primitive postpositions.
Table 11 Relative degree of grammaticalization of Cabecar primitive postpositions
| grammaticalized criterion\ | less | more |
| Scope: complexity of complement | ||
| adposition takes clausal complement | yes | |
| adposition takes coordinative complement | yes | |
| adposition is copied on subconstituents of NP | no | |
| Bondedness: tightness of bond between adposition and complement | ||
| adposition governs case | no | |
| adpositional syntagma can be split by intervening material | yes | |
| complement can be left-dislocated and taken up by intercalated resumptive | yes | |
| adpositions can be coordinated while sharing their complement | no | |
| Syntagmatic variability: | ||
| position of adposition vis-à-vis complement varies | no |
As it jumps into the eyes, a few more criteria indicate a lesser than a higher degree of grammaticalization. This induces us to say that Cabecar primitive postpositions are still postpositions, but on their way to suffixes.
4. Result
Cabecar postpositions behave differently on the various subparameters of grammaticalization. Not even the criteria spelling out one and the same parameter yield the same result, as announced before. The theoretical consequences of this empirical result remain to be investigated. No doubt the criteria have to be related to a unified theory which provides for the connection of each criterion with a particular position on a general scale of grammaticalization. The theory would also be furthered if there were pairs of criteria whose members were connected by an implicational generalization.
Abstracting from the differences, it may be generalized that Cabecar primitive postpositions are possibly enclitic, but not (yet) suffixes. Complex postpositions are free, but still perceptibly grammaticalized.
