The intransitive verb ki- (‘to say’) can be used as the main verb of the framing part of the quotative construction.6 The sentence in (4) shows that the addresser ‘I’ is identified by the (unmarked) absolutive internal argument of this verb. I also call the reader’s attention to the morphosyntactic expression of the addressee ‘you all’ by the postpositional phrase headed by the perspectival marker heke.

The element heke can mark an ‘about’ meaning, a perspective, and the external cause of a transitive eventuality. Heke is cognate to genetically related forms in other Carib languages. The proto-form can be reconstructed as *pôkô, an abstract notion approximately translatable as ‘about’ (Meira & Franchetto 2005).

In Franchetto (2010: 134–140), I proposed a continuum of its different contexts of occurrence and an extension of a specific notion of ‘perspective’ from the field of spatial relations to the quantification or individuation and actualization of a potential member within a set, and finally, to the external cause of a transitive verb. I gloss heke as PRSP (perspective/perspectival) when it is not the head of a DP external argument of a transitive verb (i.e., its subject), reserving the gloss ERG for this latter function. I illustrate the contrast between these two uses of heke in (5) and (6). In (5), the postpositional phrase tüngisão ingilü heke identifies the specific condition of the intransitive verb anügü, while in (6) iheke is the pronominal external cause (the agent) of the transitive verb ingilü.

Perspectival heke is also used with the intransitive verb ki- ‘to say’ in contexts where it is not a reported speech framer, as shown in (7) and (8).

In Kuikuro, as well as in many Amerindian languages, direct quoted speech is by far more frequent than indirect quoted speech, not just in narratives, but also in other kinds of genres, and even in ordinary speech. In narrative texts, direct quoted speech accounts for between 20 and 40 percent of the whole text.7 Direct quotes contain verbs inflected by performative moods, interjections, ideophones and an abundance of epistemic markers. These modulate and vivify the atitudes and communicative intentions of the interacting characters, as well as expressing their inner thoughts.

Like other framers, ki- follows the quoted speech, as exemplified by (9) and (10):8

Observe in (10) that the addressee ngikogo ‘Indian’ follows the verb ki, inflected with the punctual aspect, and is expressed by the postpositional phrase headed by the ‘perspectival’ heke as described above.

Reported speech contructions framed with the verb ki are not limited to the narrative genre, but are also found in another genre of Kuikuro verbal-musical art. Among the Kuikuro, tolo is a feast or ritual where dances and songs are executed exclusively by women, and that form a ritual and musical complex. It contrasts with the kagutu flutes, a complementary masculine domain that is prohibited to women. Tolo means ‘bird’ as tolo songs are made to fly. The word also means ‘pet’, as the possessed form of tolo refers also to one’s lover.9

Many tolo songs show quotatives recursive structures. This is characteristic of a large part of Amerindian poetical echolalia, where speeches of others (humans, dead persons, dead enemies) are made present by the voice of a singer or a shaman. In many tolo songs, the expression uhisü kilü uheke ‘my younger brother said to me’ is the quotative framer of embedded quotes. The example in (11) is the transcription and translation of the song auga imitoho ‘for the tuvira fish to wake up at dawn’. The terms hisü ‘younger brother’ and tühüninhü ‘the one who is missing’ are paradigmatically used in framing expressions as poetic parallelistic play.

The lovers will go toward each other, since the place of the encounter was previously agreed. These images are frozen in quotative recursive structures, embedded, in turn, in parallelisms with minor variations (substitution, inversion). There is, in each tolo, a repeated core of meanings that often ends up with the mark of a reported speech: uhisü kilü uheke ‘my sweet love told to me’; tühüninhü kilü uheke ‘the one who is missing told to me’; tühüninhü heke ukilü ‘I told to the one who is missing’; ukilü ‘I said’. The woman singing performs a speech/song that was made to fly (tolotelü) in the past by another woman or a man addressing her or his lover.

In narratives, particularly interesting is the frequent use of the forms ta(gü) and nü(gü) as quote framers (12), either immediately after the reported speech, even when it is an inner thought (13), or inside it after a constituent, in most cases, a vocative or an interjection (14).10 Ta(gü) and nü(gü) are the durative and the punctual aspectual inflectional morphemes of Class 2, the major morphological class of Kuikuro verbs. The addresse is always a 3rd person pronominal form prefixed to the perspectival postposition heke. My analysis is that we are facing a transitive verb ‘to say’ that is not phonologically realized, although their inflectional class markers do at least partially appear. The short form of these aspectual morphemes -ta and -nü followed by the 3rd person addressee marked by the perspectival heke are frequently reduced to a single phonological unit, ta^jheke, nü^jheke, spoken in a voice so low as to be almost inaudible.

Note that if both the addresser and addressee are explicit, both are marked by heke, as in (15), an example of direct quoted speech in everyday colloquial interaction: the ‘perspectival’ heke marks the addressee, while the ergative heke marks the addresser.

Example (16) shows co-occurence of frames in a parallelistic repetition which allows the identification of the addresser.

The co-occurence of frames is also found in ordinary speech, as in (17), where the direct quotation is enclosed between a pre-framer and a reduced post-framer.

In example (18) we see that the quote framers ta(gü) and nü(gü) can be completely elided, leaving only the addressee (marked by heke) explicit.

Looking at the Kuikuro facts exemplified in this section, the possibility of a non phonologically realized ‘say’ verb is corroborated by a phonologically realized aspectual inflection (-nügü, -tagü). Kuikuro is a case of the phenomenon that Spronck & Nikitina (2019: 126–129) call “defenestration”: in many languages the realization of M, the framing clause of a directly reported speech construction, is often reduced or even absent, as an optional element. Cross-linguistically, meanings associated with M do not always receive structural expression, and ‘framing’ clauses that are M-less clauses are ‘defenestrated’. Examples (12) to (18) show increasing degrees of defenestration, that is, of M-less clauses. The ‘defenestration’ phenomenon in Kuikuro reaches its maximum manifestation when direct reported speech occurs without any frame, as we will see in Section 3.3.

As mentioned, Kuikuro exhibits the phenomenon that Spronck & Nikitina (2019: 126–129) call defenestration: the framing clause or less-than-a-clause of a directly reported speech construction can be completely absent. According to Spronck & Nikitina, the absence of an explicit framer does not present a semantic lacuna, since it can always be recovered thanks to the use of interjections, ideophones, epistemics, kinship terms, vocatives, among other clues. During the execution of a narrative, the brief questions posed by the itüinhi—the story-teller’s formal interlocutor or ‘what-sayer’—can provide this function of recoverability, when it is beyond the storyteller’s itüinhi immediate understanding of who is talking to whom.

The absence of any quoted speech frame (M) is by far the most frequent case in Kuikuro narratives. Expressive prosody, the context, and, above all, the prior and shared knowledge of the narrative, knowledge from which the non-native researcher is excluded, provide the clues for the recovery of addresser and addressee referents. The mastery of unframed quoted speech is a salient characteristic of the performance of an experienced story-teller, introducing dramatic movements and passages that distinguish between scenes and characters.

Example (19) presents one of the blocks in parallel sequence from the narrative itaõ kuẽgü etĩkipügü ‘the transformation of hyper-women’. It illustrates the way inner thoughts are realized as direct quotatives. In this case, the narrator is hidden in the forest during a fishing trip, and is watching the transformation of men into hyper-peccaris.11

From the same narrative, the same scene depicted in the above example is repeated, but now as the unframed direct quoted speech of the same character who reports what he just saw in the forest to his mother in the village.

The pronominal strategy is the main clue for the recovering of conjoint or disjoint reference between the subjects of a main clause and a dependent clause. However, this is true only when a 3rd person is involved. If a non-3rd person is the pronominal absolutive argument, its prefixed phonological exponent is always obligatorily present in the dependent verb, as shown in (21), where the dependent clause is an adverbial headed by the morpheme -tomi.

The situation is different with 3rd person clauses. Having no subordinative conjunction, Kuikuro mobilizes what could be called a pronominal strategy in indirect quotatives, as well as in other complex constructions, to encode coincident (SS, same subject) or distinct (DS, different subject) cross-reference relations between the arguments of the matrix clause and a 3rd person in the dependent clause. Compare the paired direct (a) vs. indirect (b) quotes in each of the following examples.

Direct quote:

Indirect quote SS:

Direct quote:

Indirect quote DS:

In these examples the SS (same subject) pronominal form in the dependent verb indicates whether we are hearing a direct or an indirect quotation. The sentence (22b) is a clear example of SS cross-reference, where the reflexive 3rd person morpheme t-/tü- indicates that the subject of the dependent verb (-telü) is coreferent with the subject of the matrix clause. However, when we face a DS (different subject or disjoint reference) construction, as in (23b), there is no difference between direct and indirect quotative constructions, leading to an ambiguous interpretation. This problem doesn’t arise when the disjoint reference is between a non-3rd person subject of the main clause and the subject of the embedded clause, as in (24b).

Direct quote:

Indirect quote:

In (25a) and (25b), the pronominal strategy is at work not only in the contrast between the 1st person and the 3rd person ergative arguments of the verb in the quoted sentence, but also through another deictic: the person markers of a possessed nominal in the direct and indirect quotatives. Here, the distinction between the dual inclusive (kuk-) and the 1st plural exclusive (tis-) is at stake. Moreover, different emphatic morphemes, atsange and akatsange, are used. Finally, there is a difference between the simple ego-centered proximal locative adverb ãde in the indirect quotative (25b) and the double ego-centered proximal locative adverbs ãde and ĩde in the direct quotative (25a).

Direct quote:

Indirect quote:

We saw in Section 2.1, the use of the perspectival or ‘about’ hekeP for the identification of the addressee in the framing part of a direct quote construction. This postpositional phrase is once more in use when the indirect quoted speech contain a transitive verb. Compare the direct (a) and indirect speech (b) forms of the sentences below: in the indirect quote constructions, kanga engelü heke iheke in (26b) and eingilü heke iheke in (27b) are the ‘about’ postpositional phrases that specify the content of the quotation.12

Direct quote:

Indirect quote:

Direct quote:

Indirect quote:

Sometimes indirect quoted speech is introduced by the verb -ki (‘to say’). However the data currently available show that another verbum dicendi—iha- (‘to point at, to show’)—is preferred when the reported speech act is indirect. Compare the direct quotation in (28a) with the indirect quotation in (28b):

The above sentences are bi-clausal focus constructions. They again show the pronominal strategy at work for distinguishing SS from DS, with the reflexive prefix on the dependent verb marking SS, as also exemplified in (29a).

The verb iha- is the only possible option in complex constructions like (30), where a full sentence is the internal argument of iha-, and in recursive constructions like (31).

The verb iha- is also the only possible option in interrogative constructions, as in the examples below, from (32) to (34).

(32), (33), and (34) are examples of interrogatives where the questioned argument is the object of a transitive verb like iha (‘to point at/say’): the object marker ng- prefixed to the verb is coindexed with the question particle tü, always in sentence-initial position, and the inflectional suffix -nümi is, in fact, the exponent of a fusion of the Punctual aspect short form (-nü-) and non-verbal copula -i.