Spike Gildea

This paper provides a first analysis of the tense-aspect system of Yawarana (yar), a Cariban language spoken in Amazonas State in Venezuela. The data analyzed stems from a documentation collection consisting of recordings of 13 of about 30 known conversational speakers of Yawarana. The inflectional morphology of Yawarana is relatively simple in comparison to nearby Cariban languages, with many fewer person prefixes, fewer inflectional suffixes, and no splits in alignment; in compensation, syntactic collocations with auxiliaries, clitics, and particles play a larger role in creating tense-aspect distinctions. Main clause verbs in Yawarana have a single suffix slot for inflectional tense-aspect morphology, the same slot that holds all category-changing derivational morphology. The inventory of inflectional suffixes in this slot includes three past tense suffixes, two that are identical to synchronic nominalizers (-sapë,-jpë) and one to an adverbializer (-se). This paper illustrates problems encountered in determining whether each of these forms primarily encodes tense or aspect. Crucial to answering this question is an examination of how the meaning of a given tense-aspect suffix combines with the inherent lexical aspect (especially telic vs. states) of different verbs. Examining all examples of these suffixes in our text corpus, we conclude that the suffix-se encodes past perfective,-jpë encodes past tense with no aspectual value, and-sapë is heterogeneous, with a perfect reading on lexical verbs and a simple past tense reading on the copula. Further, the two past tense forms of the copular auxiliary (one with-jpë, the other with-sapë) are specialized to occur in different constructions, chi-jpë with the progressive and wej-sapë with all other compound tense-aspects.

Most postpositions in modern Cariban languages are bipartite, composed of a stem that indicates properties of the ground with reference to which the object is located (e.g., liquid, open space, container, hand, back) plus a suffix that indicates the type of relation (e.g., static location, allative, ablative, perlative). While bipartite postpositions are ubiquitous in modern languages, surprisingly, very few are fully cognate: cognate stems often take a non-cognate suffix, or cognate suffixes often occur with noncognate stems. We conclude that the antecedent to modern bipartite postpositions was a syntactic phrasal construction in (Pre-)Proto-Cariban, in which modern stems were relational nouns, modern suffixes were cliticized postpositions, and different combinations lexicalized in different modern languages. We also track more recent innovations by which some older stems have become suffixes, some sequences of postpositions have become compound suffixes, and innovative stems have come from other parts of speech.

The main goal of this chapter is to discuss the value of the Construction Grammar framework to solving perceived problems with diachronic syntax. As such, one part of this chapter provides a condensed review of previous research in diachronic syntax, including a brief discussion of why many linguists have doubted the value of such work. While most of this early work did not emphasize the importance of constructions to our understanding of either synchronic or diachronic syntax, we do identify earlier examples of work for which the notion of construction was crucial, although not richly developed. The bulk of the chapter then proposes ways in which a constructional perspective/theory allows us to address some of these perceived problems with the study of diachronic syntax, hence providing a research context for the individual studies published in this volume.

This paper is essentially three papers rolled up into one: the first documents a typologically unusual case of tense-aspect-based split ergativity in the Akawaio language, one which is, in some ways, reminiscent of a voice distinction. The other two papers provide empirical tests of a hypothesized interaction between topicality and the selection of the ergative versus the nonergative clause types in Akawaio speech. The Fish Film study directly manipulates the attention of the speaker, and observes the effect of that manipulation on descriptions of animated scenes on the computer screen. Akawaio is the third language (of 20 tested so far) to show no effect of attention manipulation, either in the domain of the ergative split or otherwise. Givónian text counts measure the accessibility and importance (two aspects of topicality) of participants in traditional planned narrative discourse, and then seeks to correlate those measurements to grammar. Again, the results show no correlation between topicality and the ergative split. The final portion of the paper explains the unusual split as the transitional final stage in the replacement of one series of tense-aspect morphology with a
new series, with the default clause type now the robust innovative ergative, and the etymologically prior nonergative clause type now used only in stylistically marked contexts.

L'ergativité en Amazonie, v. 1, ed. by F. Queixalós.  Brasília: CNRS, IRD and the Laboratório de Línguas Indígenas, UnB.

The purpose of this paper is to offer a framework for organizing the three primary types of ergative main clauses in the Cariban language family, and to briefly characterize what is known about the morphological and syntactic properties of each. I will also discuss briefly the main point of typological interest, which is that the tense-aspect based ergative splits in individual northern Cariban languages do not conform with expected universal patterns (cf. §4). While I will offer brief examples from individual languages to illustrate the claims I make here, in most cases, many more examples and much more detailed argumentation can be found in prior publications (cf. specific citations in various subsections). I encourage the interested (and especially the skeptical) reader to use this brief synopsis as a guide to points of interest that can be found in those works.

Línguas Indígenas Brasileiras. Fonologia, Gramática e História. Atas do I Encontro Internacional do GTLI da ANPOLL, v. I, pp. 315-326, ed. by Ana Suelly Cabral and Aryon Rodrigues. Belem: Editoria Universitária U.F.P.A.

Arguing that the hierarchical verbal indexation system does not appear to be derived from the absolutive indexation system seen in Tupí-Guaranían subordinate clauses.

Lingüística Tupí-Guaraní/Caribe, ed. by Ignacio Prado Pastor, 163-77.  Revista Latinoamericana de Estudios Etnolingüísticos, v. VIII.  Lima, Peru.

Paper for the 1994 International Congress of Americanists, reconstructing the form and function of the object nominalizing prefix in Cariban and pointing out intriguing similarities between it and the object nominalizing prefix in TG.

... language consultants were Pragedes Salas and Rafael Moncala of Perro de Agua, Miguel Castillo of Santa F6, and Miguel Castro of Camana. ... In an intransitive clause, these two categories overlap, so that both markers agree with the intransitive subject, ken '3.Anim.Invisible'. ...

Scan of my paper in the Procedings of the Fourth Meeting of the Pacific Linguistics Conference, held May, 1989 at the University of Oregon.

How the Proto-Cariban nominalizer *-ne becomes the Panare 'Nonspec.T' (Nonspecific Aspect, Transitive) inflection. At the time I wrote this, I was unaware of the Construction Grammar model, but given that the "grammaticalization" in question in this example was the reanalysis of a derivational morpheme (an agent nominalizer) into an inflectional TA morpheme (the "nonspecific aspect Transitive" is much like the English Present Tense), I was forced to think about grammatical change as a constructional, rather than merely morphological, phenomenon. This included using changes in the syntax of the reanalyzed construction as evidence that verb form in question *in the construction in question* had been reanalyzed, in that it was no longer acting like a noun *uniquely in that construction*.

This paper presents a first survey of the morphology and syntax of verbal negation in the Cariban  language family. After a brief survey of the negative suffix(es) and a unique third person prefix on the negative verb (§1), we consider the argument structure of the “standard” negative construction (§2), some possible hypotheses for the etymologies of the non-cognate negative suffixes found in Kari’nja, Panare, Tiriyó, and Yawarana (§3), and the innovative argument structure of the Makushi and Akawaio negative constructions (§4).

Presented at CRLAO, Paris, 25 March, 2015

When a language has some ergative alignment pattern, it is usually found in only a subset of grammatical contexts, creating what has been called split ergativity. In one common pattern, tense-aspect-based split ergativity, ergative alignment occurs only in past tense or perfective aspect clauses, while some other alignment occurs elsewhere (Dixon 1994). Theoreticians have tried to explain these splits as a grammatical expression of abstract notions like viewpoint, transitivity, inherent agentivity, and ontological salience. In recent research, we have discovered that multiple languages from the Cariban and Jê families (spoken in northern and central South America) present the opposite pattern, in which ergative alignment occurs only in nonpast and imperfective clauses.  Trivially, such facts contradict the putative universal about split ergativity; less trivially, they call into question the proposed explanations for that universal. This talk proposes to explain both the universal and counter-universal patterns with reference to the concrete details of historical processes that create individual clause types with their distinct alignments (cf. Gildea 2012).
The metaphor of "split" ergativity (or any other form of “split” alignment) encourages us to misconceptualize the phenomenon as a kind of “surface” division of some “deeper”, somehow fundamentally unified phenomenon. However, the evolution of such alignments suggests a completely different metaphor, that of a patchwork quilt: different constructions (with their different alignment) are like new patches added to the existing quilt of main clause grammar. This metaphor derives from the evolution of “splits", but it also helps to frame the synchronic analysis of such alignments in a more useful way. This sort of explanation illustrates the potential of diachronic typology, both to help us understand the prevalence of typological universals (without lamenting the discovery of counter-examples) and to bring a new perspective to explanations of these universals (and their counter-examples).

Dixon, R M W.  1994.  Ergativity. Cambridge: Cambridge University Press.
Gildea, Spike.  2012. Linguistic Studies in the Cariban Family. Handbook of South American Languages, ed. by Lyle Campbell and Veronica Grondona, 441-494.  Berlin: Mouton de Gruyter.

Workshop: Voice systems in diachrony: a comparative perspective. University of Pavia, Italy, 11 September, 2014; revised and expanded for a talk at the University of Leiden 16 Feb, 2015.

Despite the relatively rich descriptive literature in the Cariban family, there is no canonical passive construction that is widely attested in the various sister languages. In descriptions available thus far, modern Cariban languages utilize four distinct constructions to code a passive function: the reflexive and three structurally more canonical passives, with patient subjects and optional oblique agent phrases. Of these four, only the reflexive-as-passive is reconstructible to Proto-Cariban with this function; the other three constructions are young enough to be readily reconstructible to their distinct sources.  This paper will summarize the reconstructions of each different passive, then discuss how the four distinct constructions interact over time to share the encoding of the passive function.

• Reflexive > Middle > Passive
Every attested Cariban language has a modern reflex of the DETRANSITIVIZING prefix, which goes on transitive verbs and derives an intransitive verb whose single argument is either simultaneously agent and patient (i.e., reflexive) or a patient (middle/passive). In this construction, it is not possible to explicitly indicate an agent via an oblique phrase, nor does an absent agent appear to control any grammatical properties at all, a significant difference from the Romance reflexives documented in Cennamo (1993). Crucially for the definition of passive in Givón (2001) and Heine (2002), the detransitive verbs sometime appear in clauses where there is an (unexpressable) semantic agent distinct from the patient (1-2). In these cases, the construction is arguably an agentless passive with a patient subject.

• Causative-Reflexive > Agentive Passive
In Ye’kwana and Bakairi, there is a construction in which the patient is the subject, the agent an optional PP, and the passive verb bears both the reflexive and causative morphemes (3-4). This development has been reported only recently, in work in progress by Cáceres (for Ye’kwana) and Meira (for Bakairi).  Although newly reported in the Cariban family, the causative-reflexive source of passive has been documented for other languages (Haspelmath 1990).

• Adjectival-Stative Participle > Passive
In Carib of Surinam and Apalaí, there is a construction in which the patient is subject of an optional copular auxiliary, agent is optionally expressed via an oblique PP, and the verb bears participial morphology (5-6).  This has been reconstructed (Gildea 1997) to the cognate construction found in most modern Cariban languages: like an adjective, the participle codes a state (often the result of an event), which can be a stative predicate (cf. ‘the window was already broken’), but with no grammatical means to express the agent; to form the passive, the predicate participle gains eventive semantics and an innovative agent phrase is added. In a further development, the cognate construction serves as an ergative main clause in Tiriyó, Wayana, and Katxúyana.

• Patient Nominalization > Passive
In Panare, there is a construction in which the patient is subject of an optional copular auxiliary, agent is optionally expressed in an oblique PP, and the verb bears the suffix -sa’ ‘PAST PARTICIPLE’ (7).  This form has been reconstructed to a patient (actually absolutive) nominalization (Gildea 1998: 120), a function that is still attested in modern Panare (Payne & Payne 2013). In a further development, the cognate construction in the Pemón Group (Akawaio, Makushi, Pemón) is an ergative perfect clause.