Comparison of six clitics across Quechuan languages

Comparison of six clitics across Quechuan languages* Florian Matter, University of Bern

The Quechuan language family of South America is spread over a large area, with considerable differences between varieties. A feature common to all varieties is a set of elements that are used for information structure management, expression of evidentiality and negation. Using data and descriptions from different Quechuan languages, I compare the form and function of these elements across the family. Based on common criteria for word-hood, I demonstrate that they behave like clitics and that their functions are largely the same in all varieties, and show that the cliticisation of ari ‘yes’ is a recent innovation that played a role in the development of the conjectural marker =ch(a/i).

1 Introduction Quechuan languages show rather big differences in phonology, morphosyntax, and lexicon, to the point of no mutual intelligibility (Adelaar et al. 2004: 168). One feature that can be found in all varieties is the presence of what are often called independent or sentential suffixes, or clitics (Adelaar et al. 2004: 209). A subset of these elements is used for information structure management, claimed to mark topic and focus/comment, respectively. These elements include the topic/contrast marker =qa, the negation/polar question marker =chu, and the three evidential markers =mi, =s(h)i and =cha/=chi (Adelaar et al. 2004: 209ff). Cusco Quechua also has a topic marker =ri, and other varieties have elements of the same or a similar form used for e.g. emphasis. In Ancash, the two functions of =chu are performed by two different morphemes =ku and =tsu. In order to establish a basis for comparison and possibly enable further work towards a history of these elements, I aim to provide an overview of their morphosyntactic behaviour and the functions they assume in nine different varieties. These varieties were chosen to cover the main subgroups of Quechua, the list can be seen in table 1. *

This is a term paper written for the course What is a word?, Autumn Semester 2014

1

Since these elements are often called “suffixes”, I attempt to assess whether they are in fact suffixes, or actually more clitic-like. The term “clitic” generally refers to morphemes that are phonologically integrated into a host, but are syntactically more independent elements than suffixes (Haspelmath et al. 2010: 196). Dixon et al. (2003) propose a set of criteria to determine the “wordhood” or “affixhood” of an element. Due to the limited information available in the respective language descriptions and the fact that an exhaustive discussion of all criteria for every suffix would be too much for this paper, I limit myself to five (resp. four) criteria in discussing these elements. The first criterion is selectivity: What kinds of hosts does the element allow (Dixon et al. 2003: 44f)? The second criterion, intimately connected with the first one, is how the element’s position in the sentence is determined: Is it conditioned by grammatical, phonological or some other criteria (Dixon et al. 2003: 45–47)? I will list the types of hosts mentioned in the literature for each variety and each morpheme, and if no such information is given or if the morpheme takes more hosts, I will provide example sentences. A third criterion is the (syntactic) scope: Does the element have clause-, phrase- or word scope (Dixon et al. 2003: 55f)? Here, the main types of hosts include noun phrases, verb phrases, and adverbial phrases. While these first three criteria are included in detail in the discussion of every element for each variety separately, I discuss the last two criteria, phonological integration (Dixon et al. 2003: 49–50) and position in relation to what can be defined as affixes (Dixon et al. 2003: 53), in a more general manner. Section 2 focuses on the three evidential/focus markers =mi, =s(h)i and =cha/=chi. Section 3 discusses the non-factual marker =chu, and specifically the situation in Ancash Quechua. Section 4 treats the marker =qa and its function in the different varieties. In section 5, I compare several elements of the form /ri/, their distribution and functions, and the outcome of the cliticisation of ari ‘yes’ in the different varieties. Section 6 discusses the matter of phonological integration and clitic-hood in general and section 7 summarizes my findings.

2

Variety Ancash Quechua Huallaga Quechua Wanka Quechua Cajamarca Quechua Ecuadorian Quechua Inga Quechua San Martín Quechua Ayacucho Quechua Cusco Quechua

Subgroup I I I IIA IIB IIB IIC IIC IIC

Table 1: Varieties under consideration (Adelaar et al. 2004: 185ff)

2 The evidential markers There are three evidential markers, =mi (also =m, =n), =shi (also =si, =sh, =s), and =cha (also =chi, =tra, =tri…). They are commonly described as marking direct knowledge, indirect knowledge and conjecture, respectively (Faller 2002: 18ff). At the same time, they commonly mark the focus of a sentence (Faller 2002: 13). The three following subsections give an overview of their behaviour and function in the nine varieties.

2.1 The direct knowledge marker =mi The evidential marker =mi has been described as expressing “direct knowledge”, “personal knowledge”, “certitudinal”, “learned by direct evidence”, “assuming responsibility for the information”, “testimonial”, “first hand information”, “witnessed knowledge”, see also the descriptions from the individual sources below. 2.1.1 =mi in Ancash Quechua The clitic =mi functions as a focus marker if the information stems from the speaker’s personal knowledge. It can attach both to NPs and VPs, with phrase-wide scope. It can also take particles as its host, such as mana ‘neg’, see Example (1). After short vowels, it has the allomorph =m (Parker 1976: 145ff). (1) a. Aw=mi. yes=dir.evid ‘Yes.’ (Parker 1976: 150)

3

b. Mana=m neg=dir.evid

warmiki=tsu. your.wife=neg

‘She is not your wife.’ (Parker 1976: 148)

2.1.2 =mi in Huallaga Quechua The clitic =mi tends to appear on new information, marking the comment of a sentence, and simultaneously framing it as direct knowledge of the speaker. It is not obligatory. It has phrase scope (Weber 1989: 397) and can be found on NPs, VPs and the negator mana, see Example (2) below. In a simple sentence, if there are sentence-initial constituents marked with =qa, =mi is pushed further to the right (Weber 1989: 419ff). (2) a. Mana=mi neg=dir.evid

kanan today

ka-yka-sha-yki-naw=qa be-impf-sub-2-sim=top

ka-nki-paq=chu. be-2.fut-fut=neg

‘You will not be like you are now.’ (Weber 1989: 338) b. Noqa=mi 1sg=mi

chaya-a-man arrive-1-cond

aywa-r=qa. go-adv=top

‘I would arrive, if I were to go.’ (Weber 1989: 421) c. Wañu-nqa-paq=mi. die-3fut-fut=dir.evid ‘It will die.’ (Weber 1989: 421)

2.1.3 =mi in Wanka Quechua The clitic =mi marks the comment when the speaker is assuming responsibility for the provided information and is presenting it as personal knowledge. After short vowels and before some other clitics (e.g. =aa, ‘certainly’), it has the form =m. It can attach to NPs, VPs and several particles. According to Cerrón-Palomino (1976: 221ff, 237ff), evidential markers are obligatory in declarative sentences, but only one of them can occur per sentence. 2.1.4 =mi in Cajamarca Quechua The marker =mi frames its host as first hand information. It appears on the first word or phrase of a sentence and can be found on NPs, VPs and mana, see Example (3). After vowels, it has the allomorph =m (Coombs et al. 2003: 21). In the example dialogues, it can frequently be found in answers. 4

(3) a. Pika-ta=m pica-acc=dir.evid

yanuykani. I.cook

‘I’m cooking pica.’ (Coombs et al. 2003: 56) b. Randinillipa=m we.buy=dir.evid

kach’i-ta, salt-acc

uchu-ta, chili-acc

arrus-ta, rice-acc

tandita-ta. bread-acc

‘(What do you buy?) We buy salt, chili, rice and bread.’ (Coombs et al. 2003: 62) c. Mana=m neg=dir.evid

kaypi=chu. here=neg

‘He is not here.’ (Coombs et al. 2003: 35)

2.1.5 =mi in Ecuadorian Quechua The independent suffix =mi signals that the speaker was witness to an event. It can attach to NPs and VPs (Carpenter 1982: 301ff, 314ff). In some central dialects, =mi has been reinterpreted as a verb root in copular constructions: (4) Allilladi good

mi-shka. mi-npk.pst

‘It’s fine.’ (Carpenter 1982: 316f) It is also noteworthy that there appear to be different past markers for personally witnessed and not personally witnessed knowledge (Carpenter 1982: 269ff), also seen in Example (4) above. It is unclear how these two markers interact with =mi and =shi. In fact, the non-personal knowledge marker -shka would contradict the meaning usually conveyed by =mi in the example above. 2.1.6 =mi in Inga Quechua The clitic =mi implies that the speaker was a witness to the described action. It attaches to the first word of the VP or alternatively to a preverbal (presumably nominal) phrase, in which case it focuses that phrase (Levinsohn et al. 1978: 25). It can thus be assumed to have phrasal scope. 2.1.7 =mi in San Martín Quechua The clitic =mi marks testimonial, certain knowledge. It usually occurs on the first constituent (implying phrase scope) of a clause, except when =ka is in that position: One 5

or several =ka push =mi further to the right. It is optional and often the evidentiality is established once at the beginning of a text (Coombs et al. 1976: 149ff). In the examples, it was found on NPs, VPs and mana: (5) a. Mana=mi neg=dir.evid

tiyu. uncle

‘No, uncle.’ (Coombs et al. 1976: 151) b. Ari, yes

ri-rka-n=mi go-pst-3sg=dir.evid

feria-man. show-dir

‘Yes, he went to the show.’ (Coombs et al. 1976: 153) c. Ari, yes

feria-man=mi show-dir=dir.evid

ri-rka-n. go-pst-3sg

‘Yes, he went to the show.’ (FM: Presumably Yes, it was to the show he went.) (Coombs et al. 1976: 152) 2.1.8 =mi in Ayacucho Quechua The “ambivalent” (i.e. not restricted to one word class as its host) suffix =mi is an assertative marker, expressing that the information comes from the speaker’s own experience or belief. It can occur on NPs, VPs and mana, see Example (6) below, and only once per clause. After vowels, it appears as =m (Hartmann et al. 1985: 103ff). (6) Mana=m neg=dir.evid

ka-n=chu. be-3=neg

‘There’s none.’ (Hartmann et al. 1985: 105)

2.1.9 =mi in Cusco Quechua The clitic =mi marks direct knowledge, or better, “the best possible source of information required for the type of event described” (Faller 2002: 18ff). Its host is focalized, which is often accompanied by fronting of the host. Its scope is the phrase and it can only occur once per clause; =qa can occur before and after it. It attaches to NPs, VPs and mana. After vowels, it takes the form =n (Itier 1997: 96ff; Sánchez 2010: 29ff).

6

2.2 The indirect knowledge marker =shi The evidential marker =s(h)i has been described as marking indirect knowledge, knowledge aquired by hearsay, non-personal knowledge, second-hand information, or reportative knowledge. Like =mi, it is claimed to also function as a focus marker. 2.2.1 =shi in Ancash Quechua The clitic =shi functions as a focus marker if the information does not stem from the speaker’s personal knowledge. It can attach both to NPs and VPs, with phrase-wide scope. It is unclear whether it can appear on mana. After short vowels, it has the allomorph =sh (Parker 1976: 145ff). 2.2.2 =shi in Huallaga Quechua The clitic =shi marks indirect, hearsay knowledge. It tends to appear on new information, marking the comment of a sentence and is not obligatory. It has phrase scope (Weber 1989: 397) and can be found on NPs, VPs and the negator mana, see Example (7). In a simple sentence, if there are sentence-initial constituents marked with =qa, =shi is pushed further to the right (Weber 1989: 419ff). (7) a. Llapan=shi all=ind.evid

chay dem

kasta family

ka-q=qa be-sub=top

wanu-n. die-3

‘All of that family dies if they do so.’ (Weber 1989: 423) b. Mana=shi neg=ind.evid

ima-si what-indef

ka-ra-n=chu. be-pst-3=neg

‘There wasn’t anything.’ (Weber 1989: 337) c. Chayshi so

utku-pa cotton-gen

muru-lla-n-ta seed-just-3-acc

tari-rku-mu-r=qa find-asp-afar-adv=top

anka-q=shi. fry-narpst=ind.evid

‘So, finding just some cotton seeds, she fried them.’ (Weber 1989: 430)

2.2.3 =shi in Wanka Quechua The clitic =shi marks the information provided by the speaker as not stemming from first-hand experience; the speaker does not assume responsibility. It is used to introduce people in narrations (e.g. “It is said there was a man”). It can attach to NPs, VPs and several particles and can occur once per clause. After vowels or before other clitics, it appears as =sh (Cerrón-Palomino 1976: 221ff, 237f). 7

2.2.4 =shi in Cajamarca Quechua The marker =shi frames the comment as second-hand information. There was little material available, but it seems to at least occur on NPs and mana, see Example (8). After vowels, it has the form =sh (Coombs et al. 2003: 21). (8) a. Mana=shi=r neg=ind.evid=emph

allin good

nanay=chu. disease=neg

‘It is not a good disease (they say).’ (Coombs et al. 2003: 113) b. Taqshu bad.air

nanay=shi=ri. disease=ind.evid=emph

‘It is a disease of bad air (they say).’ (Coombs et al. 2003: 113)

2.2.5 =shi in Ecuadorian Quechua The independent suffix =shi expresses second hand knowledge. It can occur on NPs and VPs but is optional (Carpenter 1982: 217). Interestingly, it does not occur at all in the available texts from Napo Quechua (Carpenter 1982: 433), but almost every sentence ends with nin “s/he says”. Cole (1982: 165) suggests a semantic shift of =shi from “hearsay” to “conjecture” in Ecuadorian Quechua. =shi primarily fulfills the role of marking conjectural statements, i.e. where other varieties have =ch(a/i). This explains the absence of =shi from the texts but casts some doubt on the accuracy of Carpenter’s description. 2.2.6 =shi in Inga Quechua In Inga Quechua, the clitic has the form =si, =s after vowels. It occurs on the first word of the VP, or on some preverbal element, in which case it marks that constituent as the focus of the sentence, implying phrase scope. It expresses reportative knowledge, learned from another person (Levinsohn et al. 1978: 25). 2.2.7 =shi in San Martín Quechua The clitic =shi marks the whole utterance as second hand information. Like =mi, =shi is also optional and only appears once in a text. It also interacts with =ka the same way as =mi (see section 2.1.7 and Coombs et al. 1976: 149ff). In the examples, it can be found on NPs, VPs and mana:

8

(9) a. Pero but

mana=shi neg=ind.evid

paypish machutarachu a.big.one.either

apik. I.took

‘But I didn’t take a big one either.’ (Coombs et al. 1976: 151) b. Mashu=shi bat=ind.evid

kasarakunayarkan wanted.to.marry

suk one

warmiman. woman

‘A bat wanted to marry a woman.’ (Coombs et al. 1976: 150) c. Killa=ka moon=top

wañuchinayan=shi want.to.kill=ind.evid

yaku-wan water-instr

inti=ka. sun=top

‘The moon wants to kill the sun with water.’ (Coombs et al. 1976: 150)

2.2.8 =shi in Ayacucho Quechua The ambivalent suffix has the form =si (=s after vowels) in Ayacucho Quechua. It expresses hearsay knowledge. It can occur on NPs and VPs and can only be used once per clause (Hartmann et al. 1985: 104ff). 2.2.9 =shi in Cusco Quechua As in Ayacucho and Inga Quechua, the clitic has the form =si, =s after vowels, in Cusco Quechua. It marks the new information (i.e. the comment) in a sentence as hearsay knowledge. Its host is often fronted to a sentence-initial position (Itier 1997: 96ff; Faller 2002: 18ff). (10) Alqu=s dog=ind.evid

aycha-ta meat-acc

mikhu-n. eat-3

‘It is the dog that ate the meat (they say).’ (Itier 1997: 97)

2.3 The conjecture marker =ch(a/i) This marker varies more in its form across the different varieties than the other two evidential markers, showing both differences in the onset and nucleus. It marks information as derived from conjecture or speculation. 2.3.1 =ch(a/i) in Ancash Quechua It appears as clitical =chi in Ancash Quechua, in the province of Huailas with an allomorph =ch after short vowels. It has phrase scope, attaches to NPs and VPs and can 9

only occur once per sentence. It marks the comment of a sentence as conjecture on the speaker’s part (Parker 1976: 145ff). 2.3.2 =ch(a/i) in Huallaga Quechua The clitic has the form =chi in Huallaga Quechua. It tends to appear on new information, marking the comment of a sentence as conjecture and is not obligatory. It has phrase scope (Weber 1989: 397) and can be found on NPs and VPs, see Example (11). In a simple sentence, if there are sentence-initial constituents marked with =qa, =chi is pushed further to the right, similar to =mi and =shi (Weber 1989: 419ff). (11) a. Wañu-nqa-paq=chi. die-3.fut-fut=conj.evid ‘It might die.’ (Weber 1989: 421) b. Noqa=chi 1sg=conj.evid

chaya-a-man arrive-1-cond

aywa-r=qa. go-adv=top

‘I might arrive, if I were to go.’ (Weber 1989: 421)

2.3.3 =ch(a/i) in Wanka Quechua In Wanka, the conjecture marker appears as clitical =tra or =tri ( = /t͡ʂ/), with an allomorph =tr after short vowels or before other clitics. It occurs on NPs, VPs and several particles, having phrase scope. It is obligatory in declarative sentences and can occur only once. It can be followed by other clitics, e.g. certitudinal =aa (Cerrón-Palomino 1976: 237ff). 2.3.4 =ch(a/i) in Cajamarca Quechua The clitic variably appears as =ch’i, =ch’ or =sh’ ( marks retroflex consonants). It can be assumed that the vowel-less allomorphs appear after vowels, as for the other evidential clitics in Cajamarca. The clitic marks uncertainty and often appears in clauses with future tense (Coombs et al. 2003: 82). (12) Yanapa-shaq=ch’i. help-1.fut=conj.evid ‘Maybe I’ll help them.’ (Coombs et al. 2003: 82)

10

2.3.5 =ch(a/i) in Ecuadorian Quechua Carpenter describes =cha(ri) as a dubitative marker, appearing on VPs and NPs (Carpenter 1982: 306f). Unlike =mi and =shi, it is not considered a “terminating suffix”, although that might just be an analytical decision to group it with the other clitics ending in /ri/: When /ri/ is lost, =cha keeps the (previously penultimate) stress, indicating that it is in fact word-final. Cole (1982: 164) lists all three suffixes plus the negator =chu as “validators”. (13) a. Shamu-ju-n-pacha=chari? come-prog-3p-nonfut=dub ‘Might they truly be coming?’ (Carpenter 1982: 307) b. Ima=chá? what-dub ‘What might it be?’ (Carpenter 1982: 307)

2.3.6 =ch(a/i) in Inga Quechua In Inga, the clitic has the form =char and expresses doubt about the information provided by the speaker. As the other clitics, it has phrase scope and occurs on the first word of the VP or on a preverbal phrase, acting as a focus marker (Levinsohn et al. 1978: 25). 2.3.7 =ch(a/i) in San Martín Quechua The clitic =chá expresses conjecture (Coombs et al. 1976: 154ff). It can be found on NPs as well as VPs: (14) a. Wañu-nka=chá die-3sg.fut=conj.evid

pues. then

‘Then maybe he’ll die.’ (Coombs et al. 1976: 154) b. Chay=chá dem=conj.evid

riku-ri-n. appear-inch-3sg

‘Maybe this appears.’ (Coombs et al. 1976: 154)

2.3.8 =ch(a/i) in Ayacucho Quechua The ambivalent suffix =cha has the allomorph =ch after vowels and also appears in a stressed, emphatic form =chá. It expresses doubt or uncertainty and can occur once per 11

clause (Hartmann et al. 1985: 104ff). In the examples it can be found on NPs, VPs and AdvPs: (15) a. Paqarin=cha tomorrow=conj.evid

hamu-nqa. come-3sg.fut

‘He’ll probably come tomorrow.’ (Hartmann et al. 1985: 105) b. Hamu-nqa=cha. come-3.fut=ind.evid ‘He’ll probably come.’ (Hartmann et al. 1985: 110) c. Atoq=cha fox=ind.evid

miku-ru-n. eat-pst-3

‘Maybe the fox ate it.’ (Hartmann et al. 1985: 105)

2.3.9 =ch(a/i) in Cusco Quechua The clitic =chá always appears in its stressed form. It marks conjectural statements. Marking the focus (which is then usually fronted) of a sentence, it can occur on NPs and VPs and only once per sentence, along with the other evidentials (Itier 1997: 96ff; Faller 2002: 18ff).

2.4 Comparing the evidential markers As can easily be seen from the descriptions provided above, the evidential markers look and behave rather similarly across the different varieties. Their evidential contribution is — at least superficially — essentially the same in all investigated varieties. The only possible exeption is =shi in Ecuadorian Quechua which appears to have taken on the function of =cha, with a quotative ni-n taking on its former role. The clitics do however show some differences in form and pragmatic function. 2.4.1 Form Apart from the resyllabified allomorphs of =mi and =shi after (short) vowels in some varieties and the not uncommon alternation /ʃ/ ~ /s/ in =s(h)i (compare e.g. runa simi ~ runa shimi ‘Quechua’), the form of the conjectural marker clearly shows the most heterogeneity. It features variation with regards to the nature of the onset, being either /t͡ʂ/ or /t͡ʃ/, the syllable nucleus (either /i/ or /a/), the presence or absence of a coda, and stress assignment. Concerning the place of articulation of the onset: Although there 12

are varieties which have /t͡ʃ/ simply because they merged the two affricates everywhere, there are others which still have a phoneme /t͡ʂ/ but use /t͡ʃ/ in this clitic, e.g. Ancash (Parker 1976: 41). As for the vowel, the three Quechua I varieties Ancash, Huallaga and Wanka have /i/ (Wanka also shows /a/); so do the Quechua IIA varieties Cajamarca and Lambayeque1 . The other varieties all have /a/. Finally, there are some varieties where there is a coda or even an additional syllable present. In Ecuadorian Quechua, an element /ri/ can appear, while in Cusco Quechua, =chari is still found south of Cusco, the disappearance of /ri/ only leaving a trace in the stress pattern of =chá in Cusco itself. This is due to the fact that Quechua normally has penultimate stress, which did not shift when the apocope happened. Cole (1982: 95) indicates that in Imbabura (Ecuadorian) Quechua, this element is the emphatic marker =ri, which comes from the affirmative ari. I suspect that this is also the origin of the Inga form =char. There are also varieties where combinations of the other two evidential markers plus =ri appear, for further discussion of =ri and stress assignment see section 5 below. 2.4.2 Pragmatic function As for the pragmatic dimension of the clitics, there are some differences between the sources. Some authors describe them as marking the “comment” (Parker 1976: 149, Cerrón-Palomino 1976: 237), “focus” (Levinsohn et al. 1978: 25, Cole 1982: 165), or “new information” Itier 1997, while others simply describe their interaction with =qa (Coombs et al. 1976: 150) or make no comment about their pragmatic function (Carpenter 1982). There are two not necessarily contradictory approaches to the description of how the evidential markers pattern in a sentence. For Huallaga Quechua,Weber (1989: 427) points out that an adequate explanation cannot be as simple as identifying their function as “focus”. However, he describes a pattern where =qa, if present, always precedes the evidential markers, and where this distribution expresses “a sentence’s progression from thematic to rhematic material” (Weber 1989: 429). Sánchez (2010: 7) and Levinsohn et al. (1978: 25) describe a pattern where the focused elements appear before the main VP, Sanchez explicitely describing them as “fronted”. Compare also the following examples from Cusco: (16) a. Allqu=m dog=dir.evid

kawallu-ta horse-acc

kani-n. bite-3sg

‘It is the dog that bites the horse.’ (Sánchez 2010: 47) 1

Source: Vilcabana Sánchez et al. 2007: 128, this variety was otherwise not included due to poor data.

13

b. Kawallu-ta=m horse-acc=dir.evid

allqu=qa dog=top

kani-n. bite-3sg

‘It is the horse that the dog bites.’ (Sánchez 2010: 47) c. Kani-n=mi bite-3sg=dir.evid

kawallu-ta horse-acc

allqu=qa. dog=top

‘The dog bites the horse.’ (Sánchez 2010: 47) Most of the encountered example sentences from the varieties seem to fit these patterns, i.e. =qa appearing before the evidentials, or the evidentials occuring sentence-initially, focalizing their host. It thus seems safe to characterize the evidential markers as basically marking new information. How this manifests itself in each variety specifically may vary and is beyond the scope of this paper, but there seem to be no major deviations from the basic function described above.

3 The nonfactual marker =chu In most modern Quechua varieties, there is a marker =chu that appears both in negated statements and in polar questions (Pineda-Bernuy 2014: 86). When occurring on an NP it functions as a focalizer. This happens for both functions, in questions and in negated statements, consider the following examples from Cusco (a), Ecuadorian (b), Huallaga (c), and Wanka (d — f) Quechua: (17) a. Pilar=qa P.=top

mana=(n/s/cha) neg=evid

t’anta-ta=chu bread-acc=neg

mikhu-rqa-n. eat-pst1-3

‘It wasn’t bread that Pilar ate.’ (Faller 2002: 27) b. Na neg

pay=chu 3=neg

awa-naju-n. weave-distr-3

‘It’s not them that’s weaving.’ (Carpenter 1982: 321) c. Maria M.

Hwan-ta=chu J.-acc=pol

maqa-sha? hit-perf

‘Did Mary hit John (Or someone else)?’ (Weber 1989: 19) d. Tantak-ta=chu=n bread-acc=pol.q=evid

Kunsipsiyun-piqta C.-abl

apamu-nki? carry-2sg.fut

‘Will you carry bread from Concepción?’ (Cerrón-Palomino 1976: 232)

14

e. Kunsipsiyun-piqta=chu=n C.-abl=pol=evid

tantak-ta bread-acc

apamu-nki? carry-2sg.fut

‘Is it from Concepción that you will carry bread?’ (Cerrón-Palomino 1976: 232) f. Apamu-nki=chu=n carry-2sg.fut=pol=evid

tantak-ta bread-acc

Kunsipsiyun-piqta? C.-abl

‘Will you carry bread from Concepción?’ (Cerrón-Palomino 1976: 232) Some varieties allow for combinations of =chu with evidential clitics, such as Cusco Quechua having sequences of =chu and =si/=chá, which mark dubitative questions (Itier 1997: 198ff). Similarly, Ayacucho Quechua has a wide range of combinations with evidential suffixes (Hartmann et al. 1985: 104ff). Wanka Quechua has =chun (from earlier =chu=mi) as a default question marker and also has combinations =chu=sh and =chu=tr. However, the basic function of =chu as a question marker is the same in all varieties. Its function as a negation marker is also fairly uniform across the language family, although not as uniform as the question marking function. It usually occurs in combination with the particle mana by which it is preceded, see e.g. Examples (6), (7b), or (8a) above. In many varieties, it cannot occur in subordinate clauses, the negation being expressed differently there, see e.g. Weber 1989: 341. There are some differences as to whether mana or =chu is the main negation marker, and which one is possibly optional, but these details are not relevant for the present paper. Pineda-Bernuy (2014) provides an excellent overview of the variation in the different negation strategies across the language family and concludes that, diachronally, the question marker =chu was added to negated sentences, probably due to its nonfactual nature. In some varieties it then became mandatory, and even replaced mana as the primary means of negation (PinedaBernuy 2014: 121). The only variety deviating from this pattern is Ancash Quechua, which has two different markers for the two functions: =tsu for negation and =ku for questions (Pineda-Bernuy 2014: 86; Parker 1976: 148f): (18) a. Juan J.

urya-rqa-n=ku? work-pst-3=pol.q

’Did John work?’ (Pineda-Bernuy 2014: 87) b. Juan J.

urya-rqa-n=tsu. work-pst-3=neg

’John didn’t work.’ (Pineda-Bernuy 2014: 87) Form: =tsu is cognate with the =chu of other varieties, the change /t͡ʃ/ → /t͡s/ being common in Ancash, compare e.g. Ayacucho /t͡ʃaki/ with Ancash /t͡saki/ ‘dry’ (Gobierno Re15

gional Cusco 1995: 288). The source of =ku is unclear, though. Pineda-Bernuy (2014: 122) provides two possible origins: A phonological change /ku/ -> /t͡ʃu/ or vice versa, or alternatively =ku as a loan from neighbouring languages. Potential candidates for the latter origin are listed in table 2. However, these explanation attempts are both hypothetical and no conclusive evidence for either one of them has been found so far. Form -k kaa -ki/-ka -k(i) -ki -y -ki -ka -ki kom

Language Aymara Iquito Awajun Shuar Puquina Puquina Shipibo Ashaninka Mapudungun Tehuelche

Meaning anticompletive (‘still, yet’) negative polar interrogative neglected action interrogative suffix interrogative interrogative interrogative prohibitive negative verb

Notes Appears with negative sentences only.

in wh-words in wh-words and yes-no questions in wh-words in wh-questions -ki-l (ki-nu-l)/(-la) in indicative

Table 2: Suggested sources for Ancash =ku, adapted from Pineda-Bernuy 2014: 122f

Function: Despite the rather distinct forms, Ancash =tsu and =ku fulfill the same functions as =chu in other varieties, as seen in Example (18) above. At least =tsu has the same focalizing function when applied to NPs; no example of focalizing =ku was found but it can be assumed that it works similar to =chu ‘pol.q’ in other varieties. (19) Mana=m neg=dir.evid

wamra-yki-ta-tsu child-2.poss-acc-neg

José José

rika-rqa-n. look.after-pst-3

‘It wasn’t your child that José looked after.’ (Pineda-Bernuy 2014: 105) Despite the apparently clear division of labor between =ku and =tsu, a link between =ku and the function of negation is found in its use in emphatic negation together with =taa and =raa ‘still, yet’ in the provinces of Antonio Raimondi and Huari, see Example (20). This suggests that =ku is somehow also connected to negation and not its use is not limited to questions. However, this favors neither of the two possible explanations for the origins of =ku. (20) a. Ka-n=taa=ku! be-3=yet=pol.q ‘There’s none!’ (Parker 1976: 149)

16

b. Munaa=taa=ku! give.me.the.profits=yet=pol.q ‘Don’t give me the profits!’ (Parker 1976: 149) c. Shuyaaman=raa=ku! I.should.hope-yet=pol.q ‘I shouldn’t hope!’ (Parker 1976: 149)

4 The topic marker =qa The marker =qa (or =ka) “identifies information that is discourse accessible and is a matter of common concern by speaker and addressee” (Sánchez 2010: 43). This is commonly referred to as “topic” or “theme” of a sentence, which is generally the function assigned to the marker =qa in Quechua. In the following I give a brief overview of the marker =qa in the different varieties.

4.1

=qa in Ancash Quechua

The clitic =qa marks the topic, can occur multiple times in a sentence, but is optional (Parker 1976: 149). In the examples, it can be found on NPs and subordinated VPs: (21) a. Kondor=qa sheqikacharqunaq tsayyaq=shi wamra=qa eskapakunaq. condor=top hit.pst.3 while=ind.evid boy=top escape.pst.3 ‘The condor hit her while the boy escaped.’ (Parker 1976: 150) b. Chaarirnin=qa tapurinaq. when.he.arrived=top he.asked.it ‘When he arrived, he asked him (it).’ (Parker 1976: 150)

4.2

=qa in Huallaga Quechua

The clitic =qa occurs on nouns and on discourse connectives, sometimes also on infinitives. It cannot occur on mana, and only on main constituents, i.e. it has phrase scope. It cannot occur in subordinated clauses, but multiple =qa can occur in a clause. It is never used to establish a new topic, only marking referents that have previously been mentioned or that are assumed to be general knowledge (Weber 1989: 393ff). As for the evidentials (see also section 2.4.2 above), Weber (1989: 404ff) suggests that simply 17

identifying the function of =qa as ‘top’ might be an overly simplistic approach, and characterizes =qa as “mark[ing] those constituents of a sentence which — in the speaker’s eye — are most responsible for that sentence’s relevance to its context”.

4.3

=qa in Wanka Quechua

Cerrón-Palomino describes =qa’s function as primarily “connective”, contrasting some circumstance or referent with something that has been mentioned previously. The role of marking the topic has been taken over by the definite article -kaq, a nominal suffix, see also Cerrón-Palomino 1976: 140ff. Also, =qa can occur on subordinated clauses, but its function is not quite clear (Cerrón-Palomino 1976: 230f). (22) Wakinnin=qa achka qillayta=m mañakun. some.others=top much money=dir.evid they.demand ‘And some others ask for a lot of money.’ (Cerrón-Palomino 1976: 230)

4.4

=qa in Cajamarca Quechua

The clitic =qa marks the topic, something that has been mentioned before, or something that is understood (Coombs et al. 2003: 21). It appears to have phrase scope, attaching to NP’s, e.g. ‘hot soup’ in Example (23). (23) Ama=raq mikuy=chu chay timbuchi rupaq-ta=qa neg=yet eat.2sg=neg dem boiling soup-acc=top ‘Don’t eat that hot soup yet.’ (Coombs et al. 2003: 127)

4.5

=qa in Ecuadorian Quechua

In Ecuadorian Quechua, =ka “usually indicates the topic or focus of a conversation” (Carpenter 1982: 322). It can introduce new topics, regress to previous ones, or introduce subtopics (i.e. topics related to an old topic). It occurs on NPs and AdvPs and is optional (Carpenter 1982: 322f).

4.6

=qa in Inga Quechua

The clitic =ka or =k attaches to NPs and AdvPs, to contrast them with something mentioned earlier, or to indicate that there is a progression in the story (Levinsohn et al. 1978: 25f): 18

(24) a. Pacho-ta=mi pusa-rka-ni. Maria-ta=k(a) mana. Pacho-acc=dir.evid bring-pst-1 Maria-acc=top neg ‘I brought Pacho. As for Maria, I didn’t bring her’ (Levinsohn et al. 1978: 26) pusa-rka-ni. Nispa=k cutimu-rka-ni. b. Pacho-ta=mi Pacho-acc=dir.evid bring-pst-1sg then=top leave-to.here-pst-1sg ‘I brought Pacho. Then I came here.’ (Levinsohn et al. 1978: 26)

4.7

=qa in San Martín Quechua

The marker =ka mostly occurs on NPs, although it can also be found on VPs; it is rare in subordinated clauses. Although Coombs et al. (1976) describe three disctinct functions of =ka on NPs (“new grammatical role”, “new subject”, “topic”), these can all be subsumed under the main function of “topic”. In all the listed examples, =ka marks the subject as having been mentioned previously. When occuring on VPs, =ka marks the action referred to by the VP as having been mentioned before. In subordinated clauses and particles like chaymanta “then”, its function is not quite clear but seems to express temporal or logical progression (Coombs et al. 1976: 146ff).

4.8

=qa in Ayacucho Quechua

The topic of a sentence is marked with =qa; it is possible to have multiple =qa per sentence if stronger emphasis is desired. It can also be used to change the topic and contrast it with the previous one (Hartmann et al. 1985: 105ff). In the example sentences, it can be found on NPs and AdvPs: (25) a. Taytan=qa qo-n=mi wasita churi-n-man. father=top give-3-dir.evid house son-3.poss-dir ‘The father gives the house to his son.’ (Hartmann et al. 1985: 105) b. Mana=m karu=chu=qa. neg=dir.evid far=ind.evid=top ‘It’s not far.’ (Hartmann et al. 1985: 105)

4.9

=qa in Cusco Quechua

A sentence’s topic, something that the hearer is already familiar with, is marked with =qa. It can occur on NPs and AdvPs but not in subordinated clauses (Sánchez 2010: 19

29ff; Itier 1997: 100). There is also another topic marker, =ri. Itier (1997: 100f) describes it as an “associative topic”, resulting in a kind of contrast or comparison between two referents. Also, it is always =ri that marks the topic in interrogative sentences: (26) a. Kinsa killa-lla-ña Juan-pa ripu-sqa-n, Martin-pa=ri three month-restr-accompl Juan-gen leave-prf-3sg Martin-gen=assoc.top killa-lla-raq. month-restr-inaccompl ‘Juan left three months ago, and Martin one month ago.’ (Itier 1997: 101) b. Pi-taq chay wiraqucha=ri? who-contr dem mister=assoc.top ‘And who’s that mister?’ (Itier 1997: 101)

4.10 Comparing =qa’s The form of =qa across the different varieties only varies depending on the regular sound change /q/ → /k/ and the possible deletion of /a/ after (short) vowels. It seems to be quite free in terms of host selection in all varieties, but primarily attaches to nominal hosts. Its basic function, that of marking the topic of a sentence, is also present in most varieties. The only exceptions are Wanka and Inga Quechua, where its primary functions appear to be contrasting or comparing new topics with old ones, and establishing continuity within a narration.

5 Morphemes of the form ri Cusco Quechua has a topic marker =ri (see below). There are several other, possibly related morphemes with the same phonological make-up. This section contains an overview of these morphemes, including a discussion of instances of final stress.

5.1 The “ri”-suffixes In several Quechua languages, there are verbal suffixes of the form -ri, which have the functions listed below. Even though they might be related to clitic =ri’s, they will not be discussed in detail here, due to their affigal nature. • punctual aspect in Huallaga Quechua (Weber 1989: 147)

20

• inchoative/punctual aspect (Levinsohn 1976a: 40) and reflexive/reciprocal (Levinsohn 1976b: 61; Levinsohn 1976a: 75) in Inga • repeated actions in Cajamarca Quechua (Coombs et al. 2003: 127) • punctual aspect in Cusco Quechua (Itier 1997: 92) and Ancash (Parker 1976: 125) • reflexive/reciprocal in Ecuadorian Quechua (Cole 1982: 90ff) • inchoative aspect, politeness, and emphasis in Ayacucho (Hartmann et al. 1985: 196f) • punctual/inchoative aspect in San Martín Quechua (Coombs et al. 1976: 126) In most varieties, these suffixes express some aspectual meaning or reflexivity. These functions appear to have no obvious connection to the clitical elements discussed below. However, there is a possible link in Ayacucho Quechua. There, in addition to the aspectual meaning, the suffix can also express emphasis and politeness, which are among the function of the clitics in other varieties. Also, no clitical elements of this form exist in Ayacucho.

5.2 Emphatic clitics =ri There are clitical elements of the form =ari, =ri, or =r in several varieties. Their basic function seems to be emphasis. 5.2.1 =ri in Ancash Quechua In Ancash, there is an emphatic clitic =r which always occurs in combination with the evidential clitics (Parker 1976: 151). Cerrón-Palomino (1976: 244) suggests that it is a grammaticalized ari ‘yes’. (27) Pununki=mi=r! sleep=dir.evid=emph ‘You’ll sleep!’ (Parker 1976: 151)

5.2.2 =ri in Huallaga Quechua The particle ari, which can be used as a standalone word for emphasis, see Example (28), can also cliticize (Weber 1989: 86) and also occurs in combinations with =mi and =chaq ‘surely’, as =mari and =chari, respectively (Weber 1989: 461). 21

(28) Millu-y wind-imp

ari! emph

‘Well, go ahead and wind it!’ (Parker 1976: 151)

5.2.3 =ri in Wanka Quechua In Wanka, =ari marks emphasis. It evidently comes from the affirmative particle ari, which can also occur as a separate phonological word before or after the emphasized constituent, see examples below. Both the particle and its cliticized version appear to be borrowings in Wanka (Cerrón-Palomino 1976: 243f). (29) a. Mana-laq=tr=ari neg-yet=conj=emph

alli=chu. good=neg

‘He’s probably not well yet!’ (Cerrón-Palomino 1976: 244) b. Aa=shi yes=ind.evid

ari! emph

‘They confirm it!’ (Cerrón-Palomino 1976: 244) c. Ari emph

aa=shi. yes=ind.evid

‘They certainly confirm it!’ (Cerrón-Palomino 1976: 244)

5.2.4 =ri in Cajamarca Quechua For Cajamarca, the function of the clitic =r(i) is described as marking emphasis or politeness (Coombs et al. 2003: 52f): (30) Ch’ayamuy=ri! get.here=emph ‘Get here already!’ or ‘Get here, please!’ (Parker 1976: 151)

5.2.5 =ri in Ecuadorian Quechua In Ecuadorian Quechua, there are two emphatic markers =má(ri) and =chá(ri) (Cole 1982: 164), which are combinations of the respective evidential markers and ari, ‘yes’ (Cole 1982: 95). As indicated by the parentheses, /ri/ can be omitted, resulting in a final stressed syllable which deviates from the default stress pattern in Quechua. 22

5.2.6 =ri in Inga Quechua There is a clitic =kar in Inga Quechua which is probably a combination of the topic marker =ka and cliticized ari. It is used to introduce a new, unexpected referent and puts emphasis on it (Levinsohn 1976a: 84). I suspect that the coda of the dubitative/conjecture marker =char also comes from cliticized ari. 5.2.7 =ri in San Martín Quechua There is a morpheme =rá or =rí which expresses doubt in questions. It can occur in combination with =chu (Coombs et al. 1976: 69): (31) a. Pi-wan=rá who-instr=dub

shamu-n? come-3

‘With whom might he come?’ (Coombs et al. 1976: 69) b. Ima-ta=rí what-acc=dub

rura-nka do-fut.3

wawki-ki=ka brother-2.poss=top

kaya=ka? tomorrow=top

‘What might your brother do tomorrow?’ (Coombs et al. 1976: 70)

5.2.8 =ri in Cusco Quechua As mentioned in section 4.9, there is a topic marker of the form =ri in Cusco Quechua. However, Itier (1997: 97) suggests that the normally stressed form of the conjecture marker =chá is due to apocope of /ri/ (since penultimate stress is the normal stress pattern in Quechua); an earlier form of the clitic, =chari, can still be found to the South of Cusco. Other clitics that might share a similar origin, although Itier does not say anything about this, are =má, which is used for contradiction, and =yá, which expresses obviousness. This would mean that there are — or were — two different clitical elements =ri in Cusco Quechua. It is unclear if (and how) these are related. 5.2.9 Comparing =ri’s What seems fairly clear is that in Ancash, Huallaga, Wanka, Cajamarca, Ecuadorian, and Inga Quechua, the affirmative particle ari was grammaticalized into an emphatic clitic =ari/=ri/=r, sometimes combining with evidentials or other clitics, sometimes developing secondary meanings like politeness. The situation in Cusco, San Martín, and Ayacucho Quechua is less clear. Cusco apparently also has =chári in some regions, suggesting a cliticization of ari as well. However, no other variety shares an element comparable to the Cusco topic marker =ri. San Martín has a dubitative marker =rí/=rá, which might be 23

related to ari, although it would have to be explained where the stress on =rí or the /a/ comes from. In Ayacucho, no emphatic marker of this form can be found — except as a suffix, see above. 5.2.10 Possible interplay of =chi and =ari In Ayacucho, there are stressed versions of the evidential markers (=má, =sá, and =chá), which convey emphasis. This is reminiscent of the developments in the marker =chá in Ecuadorian and Cusco Quechua, which receives its final stress from apocope of /ri/ in earlier =chari (see section 2.3 above). Based on these circumstances, one could hypothesize that the following things all come from a previously present =(a)ri after the conjectural marker =chi: • The stressed versions of the evidentials in Ayacucho Quechua. • The vowel in the clitic in Ecuadorian, Inga, San Martín, Cusco, and Ayacucho Quechua, which is /a/. This stands in opposition to /i/ in other varieties, none of which have a stressed clitic. • The final stress on =cha in San Martín Quechua. However, at least two facts speak against this explanation or at least complicate things: In Wanka, =tra can co-occur with =ari, see Example (32) below, although this might actually be variation between older =trari (from earlier *=tri=ari) and newer =tra, which just preserves the vowel, rather than two unrelated suffixes =tra and =ri. Ayacucho Quechua uses final stress to express emphasis/politeness everywhere, not just on the evidentials, and so does Huallaga Quechua (Weber 1989: 457). This suggests that =cha’s deviations in stress assignment might not necessarily always come from earlier =cha(ri). However, it can be concluded that the conjectural marker is somehow connected to =ari, mainly because of its stress pattern and alternation with =chari in Ecuadorian and Cusco Quechua, its final /r/ in Inga, and possibly its vowel in the varieties listed above. (32) Alman=tr=ari soul=conj.evid=emph

puli-yka-n! come-dur-3

‘So his soul is probably coming!’ (Cerrón-Palomino 1976: 240)

24

6 Phonological integration and clitic-hood As has become clear from the discussion of the elements in question above, the selected morphosyntactic criteria point towards a clitic-like behaviour. What has not been discussed so far is their phonological behaviour (except for stressed =chá). Most available sources do not discuss the matter of phonological integration in detail, but a good indicator is stress assignment: The default stress in Quechua is penultimate (Cerrón-Palomino 1976: 60). In some varieties,2 combinations of host and clitic can deviate from this pattern, suggesting a lesser integration of the clitics into the phonological word than affixes. In Inga, =ka does not shift the main stress when attached to a pronominal base, but =mi and presumably also other evidentials shift it when they are attached to verbs (Levinsohn 1976b: 7), as demonstrated in the following examples: (33)

wasí-ma núka máy-ma rikú-ni (Levinsohn 1976b: 7)

wasí-ma=mi núka=mi máy-ma=mi riku-ní=mi

‘to the house’ ‘I’ ‘where’ ‘I’m coming’

In San Martín Quechua, clitics behave like affixes in terms of stress assignment, compare the examples in Coombs et al. (1976: 42). In Huallaga Quechua, at least =mi shifts stress, compare Weber (1989: 459). In Ecuadorian Quechua, =chu and =qa optionally do not affect stress. Based on this, Cole (1982: 208f) suggests a diachronic origin as independent words. The last criterion for clitic-hood, that of position relative to clearly affigal elements, is rather straightforward to assess, since the relevant markers all always appear after the verbal or nominal suffixes, respectively. This, as their host selection and scope, points towards clitic-hood.

7 Discussion It can be concluded that all elements investigated here are clitical to some degree in all included Quechua varieties, i.e. none of them are affigal. The emphatic marker =ari/=ri/=r shows variation with its phonologically independent source ari ‘yes’ in at least Huallaga, Wanka, and Ecuadorian Quechua. Considering the other varieties as well, it is in different stages of phonological erosion, from =ari (Wanka) to having completely disappeared, leaving only final stress (Cusco). If the emphatic -ri in Ayacucho Quechua shared this origin, it would have been grammaticalized much further, to being an affix. 2

Varieties where the sources provided no clues about deviation or conformance to stress patterns are not discussed here.

25

The associative topic marker =ri is unique to Cusco Quechua, and nothing indicates a connection with the emphatic marker, nor another source. =ku and =tsu are unique to Ancash Quechua, but show no fundamental difference in function compared to =chu in other varieties. They only seem to differ in form, but no convincing source for this distribution has been found.

26

Variety / Marker

evidential markers

chu

qa

ri

Ancash

clitical dir/ind/conj.evid, =ch(i)

neg =tsu and pol.q =ku

=qa top

emph =r always with evidentials

Huallaga

clitical dir/ind/conj.evid, =chi)

=chu

=qa top

emph ari/=ri/=m=ari

Wanka

clitical dir/ind/conj.evid, =tr(a/i)

=chu

=qa “connective”, top is =kaq

emph ari/=ri, combines with evidentials

Cajamarca

clitical dir/ind/conj.evid, =ch’(i)

=chu

=qa top

emph/polite=r(i)

Ecuadorian

=shi conj.evid, ind.evid is nin, =cha(ri) is dub

=chu

=ka top

emph.evid =má(ri) and =chá(ri)

Inga

clitical dir/ind/conj.evid, =char

=chu

=k(a) top

“unexpected topic” =kar, dub =char

San Martín

clitical dir/ind/conj.evid, =chá

=chu

=ka top

dub.q =rá/=rí, combines with =chu

Ayacucho

clitical dir/ind/conj.evid, =ch(a)

=chu

=qa top

only affigal emph -ri

Cusco

clitical dir/ind/conj.evid, =chá

=chu

=qa top

contrastive topic =ri, conj.evid =chá

Table 3

=

27

As for the evidential markers, =mi and =s(h)i are very similar across varieties, both in form and function. =ch(a/i) is more diverse, especially in its form. This can diachronically partly be explained by its combination with =(a)ri. Concerning their function, the evidential markers tend to appear on new information, although it might be an oversimplification to call them “focus” markers, at least in some varieties. =qa appears on known information, mainly NPs and AdvPs, but also including actions (i.e. VPs) in some varieties. It also serves as a connector, establishing a progression in the discourse. The uniformity in form and function, the morphosyntactic behaviour, and the different stages of phonological integration discussed in Section 6 suggest a diachronic origin of these elements as cliticized, formerly independent words. With the exception of the emphatic marker =ri, which is arguably early in its cliticization, there is no immediately evident source for these clitics.

28

References Adelaar, Willem F.H. & Pieter C. Muysken. 2004. The languages of the Andes. Cambridge: Cambridge University Press. Carpenter, Lawrence Kidd. 1982. “Ecuadorian Quichua. Descriptive Sketch and Variation”. Ph.D. dissertation. University of Florida. Cerrón-Palomino, Rodolfo. 1976. Gramática quechua. Junín-Huanca. Lima: Ministerio de Educación. Cole, Peter. 1982. Imbabura Quechua. Amsterdam: North-Holland Publishing Company. Coombs, David, Heidi Coombs, & Robert Weber. 1976. Gramática quechua: San Martín. Lima: Ministerio de Educación. Coombs, David M., Heidi Coombs, & Blanca O. Chamán. 2003. Rimashun kichwapi. Hablemos en quechua: Una introducción al quechua cajamarquino. url: http://www.sil. org/resources/archives/29613 (visited on 04/09/2015). Dixon, Robert M. W. & Alexandra Y. Aikhenvald. 2003. Word. A cross-linguistic typology. Cambridge: Cambridge University Press. Faller, Martina T. 2002. “Semantics and pragmatics of evidentials in Cuzco Quechua”. Ph.D. dissertation. Stanford University. Gobierno Regional Cusco. 1995. Diccionario Quechua-Español-Quechua/Qheswa-EspañolQheswa Simi Taqe. url: http://www.illa-a.org/cd/diccionarios/DicAMLQuechua. pdf (visited on 04/09/2015). Hartmann, Roswith et al. 1985. “Rimaykullayki”. Unterrichtsmaterialien zum Quechua Ayacuchano. Berlin: Dietrich Reimer. Haspelmath, Martin & Andrea Sims. 2010. Understanding Morphology. London: Hodder Education. Itier, César. 1997. Parlons quechua. La langue du Cuzco. Paris: L’Harmattan. Levinsohn, Stephen H. 1976a. Una gramática pedagógica del inga. segunda parte. url: http://www- 01.sil.org/americas/colombia/pubs/21532.pdf (visited on 04/09/2015). — 1976b. Una gramática pedagógica del inga. primera parte. url: http://www-01.sil. org/americas/colombia/pubs/11421.pdf (visited on 04/09/2015). Levinsohn, Stephen H., Alonso M. Bilbao, & Domingo T. Chasioy. 1978. Apuntes sobre la gramática inga. url: http : / / www . sil . org / americas / colombia / pubs / abstract.asp?id=928474518991 (visited on 04/09/2015). Parker, Gary John. 1976. Gramática quechua: Ancash-Huailas. Lima: Ministerio de Educación. Pineda-Bernuy, Edith. 2014. “The development of standard negation in Quechua. A reconstruction”. In: The Diachrony of Negation. Ed. by Maj-Britt Mosegaard Hansen & 29

Jacqueline Visconti. Amsterdam / Philadelphia: John Benjamins Publishing: pp. 83– 129. Sánchez, Liliana. 2010. The morphology and syntax of topic and focus: Minimalist inquiries in the Quechua periphery. Amsterdam / Philadelphia: John Benjamins Publishing. Vilcabana Sánchez, Victor et al. 2007. Guía de lengua quechua para castellano-hablantes. url: http://www-01.sil.org/acpub/repository/Guia_50733.pdf (visited on 04/09/2015). Weber, David. 1989. A grammar of Huallaga (Huánuco) Quechua. Berkeley / Los Angeles: University of California Press.

30