2.1Salience, the Chipaya SMEs and further discourse-structuring devices of Chipaya

The Chipaya SMEs, =l, =m and =ʐ, attach to constituents and elements that a speaker wishes to make salient in discourse (see Tables 1 and 2). According to Chiarcos (2011Chiarcos, Christian 2011 “The Mental Salience Framework: Context-Adequate Generation of Referring Expressions.” In Salience: Multidisciplinary Perspectives on Its Function in Discourse, ed. by Michael Grabski, Claus Berry, and Christian Chiarcos, 105–139. Berlin: De Gruyter Mouton. , 107–108), two basic types of salience can be distinguished: one relates to “hearer salience (accessibility/givenness) Hearer salient information is known and easily retrievable for the hearer” (Chiarcos 2011Chiarcos, Christian 2011 “The Mental Salience Framework: Context-Adequate Generation of Referring Expressions.” In Salience: Multidisciplinary Perspectives on Its Function in Discourse, ed. by Michael Grabski, Claus Berry, and Christian Chiarcos, 105–139. Berlin: De Gruyter Mouton. , 108; emphasis in the original), while the other one is “speaker salience (importance/newsworthiness) Speaker salient information is speaker-private and relevant, e.g. new for the hearer, not predictable or something the speaker wants to put special emphasis on” (Chiarcos 2011Chiarcos, Christian 2011 “The Mental Salience Framework: Context-Adequate Generation of Referring Expressions.” In Salience: Multidisciplinary Perspectives on Its Function in Discourse, ed. by Michael Grabski, Claus Berry, and Christian Chiarcos, 105–139. Berlin: De Gruyter Mouton. , 107; original emphasis). It is the latter I am concerned with, as I suggest that the SMEs mark information that is salient in the sense of speaker salience, i.e. new to the hearer, important and/or not predictable. In accordance with Chiarcos’ (2011Chiarcos, Christian 2011 “The Mental Salience Framework: Context-Adequate Generation of Referring Expressions.” In Salience: Multidisciplinary Perspectives on Its Function in Discourse, ed. by Michael Grabski, Claus Berry, and Christian Chiarcos, 105–139. Berlin: De Gruyter Mouton. , 109) observation that “salience is a necessary condition for shifts of attention”, the Chipaya enclitics indicate shifts in salience in a discourse. I propose that these are the main functions of the Chipaya SMEs: to make their host salient and thereby to indicate a shift in salience. Because of this, the Chipaya enclitics play a role in creating discourse coherence, as they occur at paratactic and hypotactic transitions which signal the beginning or end of a segment or mark quotations, for instance (see Section 2.2). Elements or constituents carrying one of these enclitics draw the addressee’s attention to the marked element or constituent, which is thereby singled out as being new and/or important, making the addressee aware of a shift in salience. They are thus “discourse operators” in the sense of Redeker (2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 340), where discourse operator is defined as follows:

A discourse operator is any expression that is used with the primary function of bringing to the listener’s attention a particular kind of relation between the discourse unit it introduces and the immediate discourse context.(Redeker 2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 341; original emphasis)

The Chipaya SMEs are coreferential with the subject of a clause regarding number and gender but are not based on the respective personal pronouns, with the possible exception of the second-person singular.33.In cases where an SME attaches to a third-person direct object and co-references either a third-person singular masculine, first-person plural inclusive, second-person plural or third-person plural subject, it may seem as if the SME is actually coreferential with the direct object rather than the subject (see e.g. lines 5 to 5b and 6 in Section 5.2). However, SMEs are always coreferential with the subject, as is also confirmed by our language consultant. Following Zwicky (1977Zwicky, Arnold 1977 On Clitics. Bloomington: Ohio State University [Reproduced by the Indiana University Linguistics Club]. http://​web​.stanford​.edu​/~zwicky​/on​_clitics​.pdf (27 December 2019)., 6), the Chipaya SMEs can therefore be called special clitics. Such clitics are “not necessarily derived from corresponding full forms in a transparent way and they may or must be placed at a different position than the corresponding full form if there is one” (Gerlach and Grijzenhout 2000Gerlach, Birgit, and Janet Grijzenhout 2000 “Clitics from Different Perspectives.” In Clitics in Phonology, Morphology and Syntax, ed. by Birgit Gerlach, and Janet Grijzenhout, 1–29. Amsterdam/Philadelphia: John Benjamins Publishing Company., 1; see also e.g. Siewierska 2004Siewierska, Anna 2004 Person. Cambridge: Cambridge University Press. , 26). Table 1 summarises the forms of the enclitics and the respective pronouns.

The Chipaya enclitics are not obligatory and long stretches of discourse can occur without them. Moreover, their use is genre-specific (see also Sections 1, 2.2 and 4). They co-reference only human and quasi human subjects, with the latter referring to characters in traditional folk tales, where animals behave like humans. The SMEs can occur in main and in subordinate clauses (for the latter, see e.g. line 2 in Section 5.1) and show no restrictions with respect to verbal tense marking.

The Chipaya SMEs have no canonical position in a clause, with one exception: enclitics that are coreferential with the referent of a first- or second-person subject do not attach to the pronominal subject itself. This relates to their function of marking speaker-salient referents: since first- and second-person subject referents are speech-act participants and as such hearer-salient, they do not take the enclitics that mark speaker-salient information (cf. Cerrón-Palomino 2006Cerrón-Palomino, Rodolfo 2006 El chipaya o la lengua de los hombres del agua [Chipaya or the Language of the People of the Water]. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú., 172). Thus, the Chipaya SMEs can be described as “floating clitics” (Aikhenvald 2002Aikhenvald, Alexandra Y. 2002 “Typological Parameters for the Study of Clitics, with Special Reference to Tariana.” In Word: A Cross-Linguistic Typology, ed. by Robert M. W. Dixon, and Alexandra Y. Aikhenvald, 42–78. Cambridge: Cambridge University Press., 46; see also Cerrón-Palomino 2006Cerrón-Palomino, Rodolfo 2006 El chipaya o la lengua de los hombres del agua [Chipaya or the Language of the People of the Water]. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú., 172; Section 1).44.Aikhenvald (2002Aikhenvald, Alexandra Y. 2002 “Typological Parameters for the Study of Clitics, with Special Reference to Tariana.” In Word: A Cross-Linguistic Typology, ed. by Robert M. W. Dixon, and Alexandra Y. Aikhenvald, 42–78. Cambridge: Cambridge University Press., 47) further mentions that floating clitics often express focus or emphasis. Her observation is thus in accordance with the function of the Chipaya enclitics as salience-marking devices proposed here. However, there are clearly preferred constituents and elements the enclitics attach to: with lexically expressed third-person subject referents, the enclitics preferably attach to the subject noun phrase, while with first- and second-person subject referents, the enclitics often go either onto adverbs, adjectives or locative adjuncts or onto the direct object.55.Locative adjuncts as hosts of an SME are attested only twice. In negative clauses, the enclitics frequently attach to the negative particle ana ‘no, not’ (see Cerrón-Palomino 2006Cerrón-Palomino, Rodolfo 2006 El chipaya o la lengua de los hombres del agua [Chipaya or the Language of the People of the Water]. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú., 172). All other positions are occupied only occasionally. Table 2 provides an overview of the hosts of the Chipaya SMEs, where – marks a position not attested in my data sample.66.That there are more than twice as many third-person subject enclitics than first- or second-person subject enclitics relates to the fact that a considerable part of the Chipaya corpus consists of folk tales, where subjects are usually a third person.

Of the 58 hosts, only 14 are a subject, while in the remaining cases it is a non-subject constituent. According to Siewierska (2004Siewierska, Anna 2004 Person. Cambridge: Cambridge University Press. , 175, referring to Ariel 1990Ariel, Mira 1990 Accessing Noun Phrase Antecedents. London: Croom Helm.), subjects are higher in salience than non-subjects. This means that the majority of the hosts of the SMEs show low salience with respect to their grammatical function. Excluding negations, adverbs and adjectives and with the arguable exception of two instances, the remaining hosts are third-persons and as such non-participants.77.The two exceptions refer to line 15 in Section 5.1. There, speaker and referent are identical, but the speaker uses a third-person expression – ‘father’ – to refer to himself. Non-participants are lowest in salience when compared to speakers and addressees (ibid.). Moreover, 17 of the hosts’ referents are inanimate and therefore less salient than human or animate referents (ibid.). Thus, practically all hosts of the Chipaya SMEs are rather low on the salience hierarchy and many combine several non-salient features: they are non-participants, often non-subject constituents and/or inanimate. Therefore, attaching an SME clearly contributes to making otherwise non-salient participants salient. This emphasises the function of the Chipaya enclitics as salience-markers.

The SMEs are of course not the only means Chipaya makes use of in structuring discourse. Other devices include lexical discourse markers like xaʎa ‘then’, neq^hʂtan ‘then, later’ and nuʐ and nuʐkiʂ ‘thus, so’. As yet, the scope and meaning of the Chipaya lexical discourse markers are not entirely clear, but I tentatively propose that xaʎa ‘then’ and neq^hʂtan ‘then, later’ denote temporal succession, while nuʐ and nuʐkiʂ ‘thus, so’ express the result(s) of a preceding action. The latter is based on Schiffrin’s (1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. , 201–202) analysis of English so, where I assume a semantic likeness between thus and so. These are very preliminary interpretations of the Chipaya lexical discourse markers and further research is required. Another device that contributes to creating discourse coherence in Chipaya and which I frequently refer to in my analyses is the declarative marker =tʂa (see Sections 5.1 and 5.2). It marks a predication (see Cerrón-Palomino 2006Cerrón-Palomino, Rodolfo 2006 El chipaya o la lengua de los hombres del agua [Chipaya or the Language of the People of the Water]. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú., 166) and as such attaches to the predicate of a clause, which as a default is the clause-final finite verb. Still other means that play a role in structuring discourse are, for instance, the topic marker =ki and anaphorical and zero coding of discourse referents. However, while these forms and structures play a role in creating discourse coherence in Chipaya texts and also partly interact with the SMEs, the focus of this paper lies on the SMEs and I will not elaborate further on other discourse-structuring means of Chipaya.

2.2Discourse coherence and discourse transitions

A pivotal issue of studies on coherence relations is the exploration of the hierarchical structuring of texts (see Redeker and Gruber 2014Redeker, Gisela, and Helmut Gruber 2014 “Introduction: The Pragmatics of Discourse Coherence.” In The Pragmatics of Discourse Coherence, ed. by Helmut Gruber, and Gisela Redeker, 1–22. Amsterdam/Philadelphia: John Benjamins Publishing Company. , 1–11, among others). Accordingly, coherence relations are defined by Redeker and Gruber (2014Redeker, Gisela, and Helmut Gruber 2014 “Introduction: The Pragmatics of Discourse Coherence.” In The Pragmatics of Discourse Coherence, ed. by Helmut Gruber, and Gisela Redeker, 1–22. Amsterdam/Philadelphia: John Benjamins Publishing Company. , 2) as describing “how parts of a discourse combine recursively to form larger chunks and eventually the whole structure”. Put the other way around: a text can be broken up into paragraphs and segments, each of which consists of sentences and clauses. The semantic-pragmatic relationships of coordination and subordination that exist between these units ultimately produce a coherent text (ibid.).88.Semantic-pragmatic relationships of coordination and subordination are, of course, also reflected in the formal structure of a text, as is the case with, for instance, markers of syntactic subordination.

I am concerned with the transitions between these different types of discourse units and segments, as it is there that the SMEs occur. Redeker (2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 344–345; emphasis in the original) distinguishes two basic types of discourse segment transitions: “paratactic transitions between segments that follow each other at the same level” and “hypotactic transitions involving interruption or suspension of an incomplete unit with parenthetical material”. Paratactic transitions are indicated by next-segment and/or end-of-segment markers (Redeker 2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 344). Parenthetical material which is marked by hypotactic transitions includes digressions, interruptions, specifications, paraphrases, explications, clarifications, background information, comments, repairs, quotations and returns (Redeker 2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 344, 345).99.Redeker (2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 345) further distinguishes between “‘push’-markers signalling the beginning of a parenthetical segment, and ‘pop’-markers signalling the return from a parenthetical segment”. This distinction is not taken into consideration here, as the Chipaya markers that appear at the beginning and at the end of parenthetical material do not differ formally. Figure 1 provides a schematic overview of the discourse segment transitions.

Figure 1.Discourse segment transitions (after: Redeker 2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 344–345)

Because of their pragmatic function to make referents and elements salient, thereby indicating shifts in salience, the Chipaya enclitics occur at paratactic and hypotactic transitions in Chipaya discourse (see also Table 4 in Section 4). In this way, the Chipaya SMEs contribute to structuring a discourse. However, not every discourse transition is accompanied by an SME; instead, they are used only when a transition involves a shift in salience. Thus, the function of the Chipaya SMEs is not to indicate a discourse transition per se but to highlight referents or elements that a speaker wishes to make salient. As such, however, the Chipaya enclitics frequently occur at discourse transitions, as these often, although not always, involve a shift in salience. Transitions without SMEs do not involve a shift in salience (see Section 5.1).

5.1A folk story: The fox and the armadillo (DAT 26–1)

The text was produced by a middle-aged, male Chipaya speaker, trilingual in Chipaya, Aymara and Spanish, who is an accomplished narrator. The recording was done semi-spontaneously, i.e. during a recording session the speaker volunteered to provide the text.

In line 1, the narrator gives a one-line summary of the introduction to his tale, which he had just delivered (not displayed here). The summary also marks the beginning of the actual story. Line 1 is introduced by the lexical discourse marker nuʐkiʂ ‘thus’, summarising the previous meeting of the princess with the armadillo and her pregnancy resulting from it (see Schiffrin 1987Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. , 201–202 on the English lexical discourse marker so; see also Section 2.1). Line 1 is closed with the declarative marker =tʂa, which attaches to the clause-final finite verb paq-tʃi=ki=tʂa ‘it grew’. The lexical discourse marker and the declarative marker thus enclose the summary in line 1, thereby marking it as a discourse segment of its own.

lines 2 to 8 comprise a new segment with a paratactic transition from line 1 to line 2. The new segment is introduced in line 2 by the lexical discourse marker nuʐkiʂ ‘thus’, which, again, refers to the outcome of the princess’s and the armadillo’s meeting. The SME =ʐ attaches to the third-person singular masculine subject wawa ‘child’, which is thus made salient and established as a major character of the discourse segment. This includes a shift of salience from the princess’s encounter with the armadillo and her pregnancy, which was described in the introduction of the tale, to the now born child. With the birth of the child, the story picks up speed and it serves as a catalyst for the calling out of the animals (line 7), in the context of which the two main characters of the story, the fox and the armadillo, also make their appearances (lines 13 and 17, respectively). The SME attaches to the subject of a subordinate clause, which suggests that syntactic salience and pragmatic salience do not necessarily always coincide. Presumably because of the high salience of the child’s birth the narrator repeats that the child is born in line 3.

Lines 2 to 6 form one syntactic unit, where the main clause comes in line 4. There, the utterance verb k^hi-tʃi=ki=tʂa ‘he said’ carries the declarative marker =tʂa, which marks it as the main verb, while the third-person singular masculine subject of the main clause is ni rey=ki ‘the king’. In Chipaya, the verb base k^hi- ‘say’ indicates the beginning and/or end of verbatim language and thus belongs to the “formulas of quotation” (Longacre 1996Longacre, Robert E. 1996 The Grammar of Discourse. Second edition. New York/London: Plenum Press. , 11; see also Holt 1996Holt, Elizabeth 1996 “Reporting on Talk: The Use of Direct Reported Speech in Conversation.” Research on Language and Social Interaction 29: 219–245. , 224, on English). The utterance verbs k^hi-tʃi=ki=tʂa ‘he said’ in line 4 and k^hi-kan ‘saying’ in line 6 bracket the king’s direct and indirect speech delivered in lines 5 and 6. Although there are hypotactic transitions from lines 4 to 5 and 6 to 7, no enclitic is used as there is no shift in salience that would warrant use of an SME.

Lines 7 and 8 present a new sentence, where the main clause is in line 7, marked as such by the declarative =tʂa on the main verb q^haw-ʂi=ki=tʂa ‘he called’. The subordinate clause in line 8 represents an almost verbatim repetition of the king’s indirect speech provided in line 6.

Lines 9 to 12 present a new segment with a paratactic transition from line 8 to line 9. This is indicated by the lexical discourse marker neq^hʂtan ‘then’ at the beginning of line 9, which temporally links the new segment with the preceding one. In lines 9 to 12, no major referent is introduced and those referents that are mentioned for the first time – the animals and birds in lines 10 and 11 – are not further specified and do not play a role in the remainder of the story. Instead, the discourse referents that were active in the preceding segment – the king and the child – also remain active in this new segment and there is thus no shift in salience (see also Section 2.2). Accordingly, no SME is used here. This suggests that the SME in line 2 has scope not only over the segment from lines 2 to 8, but also over this segment from lines 9 to 12 (see also Section 4). The segment is closed in line 12 by the lexical discourse nuʐkiʂ ‘thus’, which summarises the result of the animals parading in front of the child: the child does not recognise any of the assembled animals as his father. The sentence in line 12 is marked as being syntactically independent by the declarative marker =tʂa on the verb pax=ki=tʂa ‘recognise’.

The paratactic transition from line 12 to line 13, which introduces a new segment (lines 13 to 16), is once more marked by the lexical discourse marker nuʐkiʂ ‘thus’, which expresses that the appearance of the fox results from the king’s order that all animals shall parade in front of the child. The SME =ʐ (preceded by the reportative marker =ki) attaches to the third-person singular masculine subject qiti ‘fox’. Making the fox – with whom one of the two major characters of the folk tale enters the scene – salient entails a shift in salience, away from the king and the child in the preceding two segments (i.e. lines 2 to 12) to the newly introduced referent of the fox. The scope of the enclitic comprises the current segment. The motion verb t^hon- ‘come’ introduces the fox. Despite the missing declarative marker on the verb t^hon-ʧi ‘he came’, the clause in line 13 is syntactically a main clause. The verb t^hon-ʧi ‘he came’ is inflected with the tense marker -ʧi, which co-references a third-person singular masculine subject in the completive past and attaches to main verbs only. Usually, the main verb of a Chipaya clause takes the declarative marker (see lines 1, 4, 7, 9, 12, 14) and omitting it is marked. Of course, non-use of the declarative marker does not automatically create a pragmatically marked utterance but does so in this particular context where a verb marked as finite by the tense marker -ʧi does not take the expected declarative marker. The marked form of the main verb in line 13 draws further attention to the clause, thus singling it out as a particularly important clause. If this interpretation is correct, then the clause in line 13 is made salient by three different devices: by the lexical discourse marker that announces the beginning of a new segment; by the SME on the subject of the clause, which highlights the fox as a major character within the clause and the segment and signals a shift in salience; and, finally, by the missing declarative marker on the main verb.

The clauses in lines 14 to 16 form one syntactic unit, with the main clause appearing in line 14. In line 15, the direct speech of the fox is delivered and marked by the utterance verb k^hi- ‘say’, which carries the subordination marker -kan, expressing coreferentiality and simultaneousness with the subject of the main clause (here: the fox). The phrase ‘say father [call me father]’ is doubled and accordingly, the SME =ʐ, attached to tata ‘father’, also occurs twice. This is a hypotactic transition. By taking the enclitic, the word tata ‘father’ is made salient and marked as the most important constituent within the direct speech of the fox. Moreover, salience is momentarily shifted from the fox to the figure of the father (although the referent of both expressions is the fox). However, the alleged father is salient only within the direct speech of the fox, i.e. line 15, and the salience-marking =ʐ has scope only over this clause (see Section 4).1515.Quotes are, of course, transitions of speakers where a narrator indicates that the following part of the discourse is not be attributed to him- or herself but to (a) character(s) within the discourse universe (see e.g. Holt 1996Holt, Elizabeth 1996 “Reporting on Talk: The Use of Direct Reported Speech in Conversation.” Research on Language and Social Interaction 29: 219–245. , 220–221, referring to Coulmas 1986Coulmas, Florian (ed.) 1986 Direct and Indirect Speech. Berlin: Mouton de Gruyter. , 2). Two issues require a closer look: the choice of the word for ‘father’, tata, and that of the enclitic itself. The use of tata ‘father’ is remarkable because it is based on Quechua taita ‘father’; the Chipaya word for ‘father’ is ep^h (see line 6). The reason for this choice is not overly clear, but it may have to do with markedness again. Using a loan word is allegedly more marked than using the native lexical item (see Myers Scotton 1983Myers Scotton, Carol 1983 “The Negotiation of Identities in Conversation: A Theory of Markedness and Code Choice.” International Journal of the Sociology of Language 44: 115–136. ; see also Gardner-Chloros 2009Gardner-Chloros, Penelope 2009 Code-Switching. Cambridge: Cambridge University Press. , 67, 69) and this marked lexical choice is another means to draw attention to the discourse referent of the alleged father (alongside use of the SME) (Gumperz 1982Gumperz, John J. 1982 Discourse Strategies. Cambridge: Cambridge University Press. , 61, 75–76; see also Gardner-Chloros 2009Gardner-Chloros, Penelope 2009 Code-Switching. Cambridge: Cambridge University Press. , 67, 69). Should this be correct, then this pattern is reminiscent of the one found in line 13, where the lexical discourse marker, the SME and the omitted declarative marker each contribute to making the clause particularly salient. The choice of the SME =ʐ is actually an error by the narrator: the verb form used in line 15 is an imperative and the subject is thus a second-person singular (since the fox addresses the child directly by saying: ‘call me father’). The appropriate enclitic would have been =m instead of =ʐ, which co-references a third-person singular masculine subject (see Table 1). This error probably reflects the narrator who is coming through here, i.e. although the narrator lends his voice to the character of the fox, he fails to reflect this change of perspective in the choice of the enclitic.

With line 17 a new segment begins (the rest of which is not shown here) and there is another paratactic transition from line 16 to line 17. Line 17 is parallel in structure and content to line 13, in which the fox arrives. Here, the second main character of the folk tale enters the scene, the armadillo. The new segment in line 17 is opened by a lexical discourse marker neq^hʂtan ‘then’, which indicates that the following events succeed those presented in the preceding segment. The lexically expressed third-person subject kerka ‘armadillo’ takes the SME =ʐ (as well as the reportative marker =ki) and the function of the enclitic is the same as in line 13: it directs the addressee’s attention to this new character, signalling that the armadillo is pivotal in the following segment. This involves a shift in salience, from the fox in the preceding segment to the armadillo. Again, the main verb is t^hon- ‘come’, which introduces the armadillo and does not take a declarative marker, either. I propose that the reason for omitting the declarative marker is the same as discussed for line 13. Thus, the clauses in lines 13 and 17 are marked thrice: by the lexical discourse marker, the SME and the omission of the declarative marker.

5.2A descriptive text: Animales ‘animals’

The text was provided by the same speaker as in Section 5.1. It was also provided semi-spontaneously.

The text comprises only two segments, the first of which includes lines 1 to 3. In line 1, the speaker announces that he is going to talk about the mouse. This statement is repeated in line 2, where the speaker now uses the lexical discourse marker niʐta ‘thus’, thereby summarising his statement from line 1.1616.Both nuʐ and niʐta are translated as ‘so, thus’ (see Section 2.1). Usually, lexical discourse markers introduce or close a segment (see Section 5.1 and lines 4 and 8 here). It is thus remarkable that the lexical discourse marker only comes in the middle of the segment and not at its beginning and/or end. This can be explained if one consults the audio recording. After having finished the previous paragraph about the toad (not shown here), it seems that the idea of talking about the mouse next strikes the speaker rather spontaneously (line 1; see also Section 4). In line 2, the speaker starts again with atʃiku=ki ‘the mouse’, which is followed by a break of 1.46 seconds (symbolised by a dash) before the speaker starts again with the lexical discourse marker niʐta ‘thus’. It therefore seems that the speaker was not entirely certain about how to proceed after his initial statement that he is going to talk about the mouse, thus repeats the upcoming topic (i.e. the mouse) in line 2, takes a break and, after having decided upon what to say next, starts anew, this time with a lexical discourse marker which marks the actual beginning of the paragraph on the mouse. The segment is then closed in line 3 by the statement that the mouse is a thief.

There is a paratactic transition from line 3 to line 4, introducing the second segment from lines 4 to 8. Regarding contents, the paratactic transition is not as pronounced as some of those discussed in Section 5.1 and line 4 could also be understood as a specification, i.e. a hypotactic transition, of the proposition uttered in line 3. However, the speaker sets the topic for the following segment – the people’s dislike of the mouse and the reasons for it – in line 4 and this, in conjunction with the two lexical discourse markers, suggests an interpretation as a paratactic transition. Although it has been proposed that the lexical discourse marker xaʎa ‘then’ expresses a temporal succession (see Section 2.1), it appears that here it is used instead in a causal sense, not unlike the second lexical discourse marker niʐta ‘thus’.1717.The form niʐti instead of niʐta in lines 4 and 8 is caused by the following separative marker -kiʂtana (see Cerrón-Palomino 2006Cerrón-Palomino, Rodolfo 2006 El chipaya o la lengua de los hombres del agua [Chipaya or the Language of the People of the Water]. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú., 87), which also contributes to the causal interpretation of niʐti-kiʂtan(a) as ‘therefore’. However, this is subject to further research. A hypotactic transition follows line 4, where line 5 presents an explication of why people hate mice: because they eat the people’s food. Accordingly, the direct-object noun phrase ʐon-ʐ tʂ^heri-naka=ʐ ‘the people’s food’ takes the enclitic and is thus marked as being salient, since it provides the reason for why people hate mice. Salience is shifted from the people’s hatred to the food eaten by the mice. Lines 5a and 5b present further hypotactic transitions from the proposition given in line 5 and are specifications of it. In both lines, the kind of food eaten by the mice is further detailed and consequently, the constituents expressing the specification, kula-naka=ʐ ‘quinoa’ and arusa-naka=ʐ ‘rice’, are each made salient by means of an enclitic. This involves shifting salience from the more general expression ‘the people’s food’ in line 5 to the specifications in lines 5a and 5b. Note the parallel structure of these two lines. Line 5c presents a summary of the preceding two lines, where the speaker states that whatever food there is, it is eaten by the mice. Line 6 is a further explication of the speaker’s initial statement that people hate mice in line 4: not only do mice eat food, they also destroy the sacks in which the food is kept, thereby providing another reason for disliking mice. Line 6 is thus another hypotactic transition from the speaker’s introducing statement in line 4. Here, salience shifts from the foodstuff discussed previously to the sacks and woven sacks that are destroyed by the mice and in accordance with this, the direct-object noun phrase saku-naka maʐa-naka=ʐ ‘sacks [and] woven sacks’ is marked with an enclitic. As the expressions are coordinated, the enclitic =ʐ has scope over both. In line 7, the speaker repeats his statement from line 3 that the mouse is a thieving animal. Finally, the segment is closed in line 8 by a summarising repetition of line 4, which initiated the segment. Again, line 8 contains the lexical discourse markers xaʎa niʐta ‘then therefore’, which close the segment (see Section 3, Table 4). There is a structural parallelism between lines 4 and 8 and the lexical discourse markers function as next-segment and end-of-segment markers (see Redeker 2006Redeker, Gisela 2006 “Discourse Markers as Attentional Cues at Discourse Transitions.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 339–358. Oxford/Amsterdam: Elsevier., 345; see also Section 2.2). All clauses in these segments are syntactically independent and there is no syntactic subordination as is the case in the folk story discussed in Section 5.1. This may relate to the more predetermined structure of folk stories (see Section 4), where syntactically more complex constructions may be memorised as part of the story and neatly delivered upon recounting the tale, whereas in a spontaneously provided description, like the one here, speakers tend to produce syntactically less complex sentences.