This paper presents a series of grammatical phenomena connected with quote constructions in Panará, a Northern Jê language spoken in central Brazil. Quote constructions encompass a wide variety of syntactic strategies that a language can use for quoting both direct and indirect speech. In the case of Panará, discussing quotation means also discussing clause type and the structure of the polysynthetic verbal complex. Unless otherwise indicated, the data presented were collected by the author over the course of several fieldwork stays among the Panará since 2014.

Panará is the language spoken by the members of the Indigenous nation also called Panará, from the autonym /panãra/ ‘those that are’. Panará today has over 600 speakers who live in the officially demarcated Panará Indigenous Land on the Iriri and Ipiranga rivers. Today the Panará land is a territory of 500,000 hectares across the state border between Mato Grosso and Pará in central Brazil. However, until the 1970s their territory was much larger, occupying an area between the Cachimbo mountain range and the modern city of Sinop in several village clusters.

The Panará were eventually reduced to less than 10% of their population as a result of their first contact with neocolonial Brazilian society, when they contracted highly contagious diseases from the Brazilian government’s contact expedition and the workers that were building the BR-163 highway, which would connect Cuiabá and Santarém, across Panará land. In 1974, the 67 survivors were forcibly removed from their land and relocated to the Xingu Indigenous Park. Twenty years later, in the 1990s, they were successful in their lengthy fight to demarcate a piece of their land that still contained Amazonian rainforest and were able to move back. Since then, they have seen a demographic and social expansion and now occupy seven different villages.

Panará was classified as a Jê language following an initial hypothesis presented by Heelas (1979) that established the correspondences between their language and a series of word lists collected from a supposedly extinct Jê group that lived further south in Brazil in the 18th and 19th centuries, called the Southern Cayapó by colonists (Giraldin 1997). Linguistic work by Rodrigues and Dourado (1993) later proved that Panará is indeed a Jê language (Davis 1966), in the Macro-Jê macro-family. Since then, Panará has been classified as a Northern Jê language alongside Apinajé, Kajkwakratxi, Kĩsêdjê, Mẽbêngôkre, and the Timbira languages (Rodrigues 1999).1

The Panará polysynthetic verb complex presents morphological cross-reference with all kinds of arguments. For the sake of ease in examining the examples provided throughout the paper, the paradigms of argument cross-reference are included in Table 1. As the paradigms in Table 1 show, the irrealis portion of the reality status dimension of Panará presents a tripartite alignment in its argument cross-reference; rather than the ergative-absolutive pattern that argument clitics display in realis, in irrealis there is a different cross-reference for the ergative, for the transitive absolutive, and for the intransitive absolutive.

At the same time, as Table 1 shows, there is an impoverished morpheme inventory for the exponence of person features in the irrealis, with multiple exponence for intransitive absolutives, as illustrated in examples (1) to (3). To reflect the distinction between irrealis 1st person /Ø/ against 2nd + 3rd person /ti/, on the one hand, and 2nd person /a/ against 1st + 3rd person /Ø/, on the other, I gloss the irrealis ergative and intransitive absolutive clitics as spk ‘speaker’ and nspk ‘non-speaker’ on the left slot, and adre ‘addressee’ and nadre ‘non-addressee’ on the right slot, closest to the verb.2

Panará presents several syntactic differences from the other Jê languages, not only in the Northern branch but also elsewhere in the family, all of which are arguably innovations (Bardagil 2018; Dourado 2001). Notably, Panará lacks the templatic Jê verb-final clause structure with fixed case positions seen in (4) for Mẽbêngôkre. Instead, in Panará the Jê verb-final clause was reanalyzed as a polysynthetic verbal complex with a series of proclitics, as can be seen in (5). Panará participant clitics present discontinuous exponence, some are omittable, and some are homophonous with free pronouns. They also present omnivorous number, with a single number morpheme controlling more than one argument, in this case the dual clitic (Bardagil 2020). Person-case restrictions are also manifested in Panará, as a ban on the co-occurrence of 1st and 2nd absolutive and dative clitics (Bardagil 2019).

In Panará, rather than the verb remaining in a low VP-internal position as elsewhere in the family, the verb raises to Infl, which is also the cliticizing category. This results in the language’s non-verb-final constituent order (6), with the syntactic derivation illustrated in (7).

Unlike in the other nine Jê languages, Panará arguments always receive ergative-absolutive case marking. This has a direct connection with clause typology in the languages of the Jê family. In main clauses, the Jê verb typically appears in a short form and the case marking is nominative-accusative—with a further split intransitive distinction between unergative and unaccusative verbs. In order to build subordinate clauses, however, Jê languages require a nominalized verb, in which the case marking switches to ergative-absolutive. This is exemplified in (8) for Kĩsêdjê.

Just like in the other Jê languages, Panará subordinate clauses are head-internal. In contrast to the classic Jê pattern, however, they show all the properties of finite main clauses with respect to their argument marking, modal morphology, case marking alignment, and available left periphery (9).

In sister languages to Panará in the Northern Jê branch, embedded clauses are characterized for the lack of finiteness and an unavailability of clausal positions higher than Infl (Bardagil 2018; Bardagil & Groothuis 2023; Salanova 2007). This is not the case for Panará, which can have its anchoring TAME category, i.e., reality status or mood, manifestly present in embedded contexts, such as relative and dependent clauses, denoted by square brackets through examples, including (10–11).3

Relativization is available to all argument types (12–13). Relativization, and clause embedding more generally, is not sensitive to constraints on classes of arguments, such as syntactic ergativity, and all types of arguments can head an embedded clause.

The same properties seen for relative clauses hold for complement clauses. The case marking is ergative-absolutive, the constituent order is free, and both the pre- and postverbal positions inside the clause are available (15).

Having established the relevant morphosyntactic characteristics of Panará, the rest of this article is devoted to Panará quotation strategies.

Panará employs a wide range of syntactic strategies to indicate quotation. This section presents Panará direct and indirect quotes, and also a particular type of quotes that appear in narration contexts, which I call here narrative quotes. In what follows I describe each type of quote and the grammatical cues that signal the different types.

Since in Panará all subordinate clauses are internally headed, the syntactic shape of both direct and indirect quotations is identical to main clauses. Clausal position, case marking and argument cross-reference morphology are not particular to quotations, direct or indirect, as can be seen across the examples in (43–45).

The very free constituent order of the Panará clause means that the position of the quotation relative to the quote construction is not fixed. Quotations can both precede or follow the quotative marker and can appear on either edge of the clause (46–47).

Prosodically, direct quotations are signaled by a raise in pitch, regardless of whether the person or character uttering the frame utterance would have been using a high- or low-pitched voice.

In (48) we can see the context for what is happening in the story from which example (49) is drawn. When compared to the framing narration in (49a), the direct quotation in (49b) is both realized with a falsetto voice and with a higher pitch, which can be seen in Figure 1.

Similarly, the onomatopoeic call of the marmoset in (50) is said with a higher pitch than the rest of the sentence, and in a falsetto voice, as shown in Figure 2.

There is also a comma pause indicating the quote boundary (marked with # in the following two examples). Unlike high pitch, which is exclusive to direct quotations, the prosodic break can be present in both direct and indirect quotations.

As for the constituency of quotes, a single quote, direct or indirect, can host more than one clause (53–54).

The Panará people make extensive use of quotation in their daily lives, and their language provides them with several grammatical strategies for doing so. These are summarized in Table 2.

As we have seen, quotes are similar to phrases or full clauses in Panará in terms of their morphosyntax. In particular, they present no properties that would not be expected from complement clauses: a full verbal complex, ergative-absolutive case marking, and pre- and postverbal clausal positions. Panará has several quotatives to introduce quotations without the need of a dedicated framer, all of which are verbs—and the verb itself can sometimes undergo elision. The narrative quote has a degree of lexicalization and coexists with a closely related reportative. If a hearer is introduced, it does so as a dative participant.

Direct and indirect quotations are not entirely identical. However, as it stands, there is no clear syntactic or morphological diagnostic to tell them apart. One of the key differences is indexical shift, observed only in indirect quotations. The second diagnostic is prosodic, namely the presence of a high pitch assigned to a direct quotation.

The information brought up in this paper points towards avenues of future research on the topic of quotation in Panará and, more generally, the distinction between different types of embedded clauses. One phenomenon that did not become clear during research on this topic was the matter of extraction from quotations. It is quite clear that indirect quotes allow for extraction of an argument to the matrix clause, but the picture is not clear for direct quotes. The status of tijãri as a narrative-dedicated quotation marker and as a reportative is also an issue that needs further research, as it is one of the few grammatical elements in Panará with evidential semantics. The potential use of both tijãri and rêjãri as discourse elements that target elements of the speech act, and to what extent they contribute discourse-related information, is also a matter for further research. As an exceptional Jê language, this study of quotation in Panará also raises a diachronic question regarding the status of Panará within the Jê family. Future research will be necessary to provide information about the degree of retention and innovation of Panará concerning the grammatical properties of Jê quotation strategies.