1.1. The Huarpean peoples and their languages

In general terms, the territory inhabited by Huarpean peoples has been called the central-western Argentinian sub-area in Andean archaeology. The geographical borders of this region are the Jáchal Zanjón river in the north, the Diamante River in the south, and the Andes mountains in the west (^{Bárcena, 2001, p. 561}). Thus, it is possible to consider that this region is at the southern end of the Andean Area. The geography of this region is characterized by its lakes and the presence of the Andes mountains in the west, where there are different crossing points that allowed contact between the different human populations that inhabited the area. (^{Bárcena, 2001, p. 563}).

There are two languages in the Huarpean family: Allentiac and Millcayac. Allentiac speakers inhabited the marshy region of highlands on the margins of the Guanacache and Del Rosario lagoons in southern San Juan and northwestern San Luis. Millcayac speakers inhabited the region south of the forementioned lagoons, in the north of the Santa Rosa, La Paz, Rivadavia, and General Alvear districts (^{Rusconi, 1962, p. 79}). Nowadays, both languages are sleeping languages and the only available bibliographical sources about them are, as far as we know, the grammars and texts documented by Luis de Valdivia.

Regarding their social life, ^{Schobinger (2009)} suggests that the Huarpes were sedentary people with social distinction, textile production, engraving, erect tabular deformations, and the use of tembetá and semi-underground rooms in their social practices. In historical terms, according to ^{Schobinger (2009, p. 17)}, the adaptation to the lacustrine environment of northern Mendoza dates to pre-ceramic times. However, cultural advances such as agriculture and ceramics date from approximately 800 to 300 BCE, that could have been influences from the central west, in what is now Chile, the Argentine northwest, and the central Andes. Thus, it is possible to consider that the relationships between the peoples that inhabited central-western Argentina and neighboring regions existed from the earliest archeological periods up until Spanish colonization began to restrict contact between different groups. The European colonization was accompanied by the gradual reduction of the indigenous population as a product of European settlement. Thus, for example, ^{Canals-Frau (1942)} points out that it is probable that some Huarpes spoke the Mapuche language and an unidentified Quechuan language that reached central-western Argentina. In this sense, ^{Rusconi (1962)} raises the idea of “indigenous cosmopolitanism” as a reflection of the diversity of materials found in the archaeological sites, that Bárcena suggests could indicate a complementarity of habitats (^{Bárcena, 2001, p. 581}). For example, the exchange of mollusks, whether from the Atlantic or the Pacific coasts, with the peoples of Cuyo, including the Huarpes, was not unusual (^{Bárcena, 2001, p. 478}).

Figure 1. Approximate distribution of the Huarpean languages in the 16^th century (^{Adelaar & Muysken, 2004, p. 503}) 

2.1. Negation from a typological perspective

In general terms, negation is an operator that changes the truth value of a proposition from affirmative to negative. Negation and affirmation are the two poles of the grammatical category of polarity. Negation has been studied from a wide range of perspectives which has resulted in considerable variation in the terms used for its description. This work studies negation from a typological perspective, thus the terminology used here is that found in works such as the World Atlas of Language Structures (^{Dryer, 2005}, ²⁰¹³; ^{Miestamo, 2005b}), Miestamo (^2005a), and ^{Van der Auwera and Krasnoukhova (2020)}, among others. For a generative framework see ^{Haegeman (1995)}, for an Optimality Theory framework see ^{De Swart (2010)}, for a Functional Discourse Grammar perspective see ^{Hengeveld and Mackenzie (2018)}, and for a more general perspective on negation see ^{Horn (2001)}.

The features considered in this work are: (1) type of negative marker used in standard negation, (2) order of negative marker and verb, (3) structure of the negative construction, and (4) type of prohibitive constructions. These features were selected according to the following criteria:

2.2. Negative constructions

3.1. Standard negation: marking strategies, order, and constructional structure

In Millcayac, standard negation is expressed by means of affixation, attaching the suffix «-na» to the main verb ([V-NEG]). This suffix appears before the person and number suffixes. As stated by Valdivia:

The negative verb is made in this language by attaching this particle («na») to all the affirmative verbs and has to be put immediately before the particles that form tenses and persons, that we put at the end of the fifth chapter. The first thing is that in the indicative, for present and past tense, there is no more than one negative, which is the one that negates the imperfect past. (^{Valdivia, 2011 [1607], Arte, p. 16r})¹

Our interpretation of Valdivia’s statement is that the imperfect past suffix «-eye», and its allomorphs, are obligatorily required in standard negation. Because of that, it loses its imperfective meaning in these constructions as in the following example:

The same author states that in the future tenses in Millcayac, instead of «-eye», the suffix «-e» is required, as seen in the following example:

However, the suffix «-e» does not only occur in combination with the future tenses, as seen in this present tense example:

Previous studies have analyzed the «-eye» and «-e» occurrence with some differences. Regarding its general form, ^{Ridruejo (2009, p. 146)} suggests that the imperfect past suffix «-eye» is so like the interrogative «-ye» that they are difficult to distinguish (as seen in example 27. ^{Diaz-Fernández (2014)} considers the past tense suffix to only have the form «-e», without mentioning the form «-eye». On the other hand, ^{Tornello et al. (2011)} considers that the first and third persons use the form «-eye» and the second person the form «-e». Even though we do not agree with this distribution, since, as can be seen in examples 24 and 25, the form «-e» can occur with the forms assigned to «-eye». The fact is that «-eye» and «-e» seem to behave as allomorphs. While their specific distribution and their specific meaning is beyond the scope of this work, what is relevant for this investigation is to present the asymmetry of the negation². Furthermore, the same meaning is expressed by means of the suffix «-i» when occurring in negative constructions along with the third person «-nap», as seen in the following example.

It is also noticeable that the forms «-eye/-e/-i» are not obligatory in all negative contexts. These suffixes are not required in nominalized forms, as seen in the following examples, that clarifies doubts about the independence of both suffixes and establishes the limits of the asymmetry of the construction.

In sum, apart from the negative «-na», negative constructions require the occurrence of the suffix «-eye/-e/-i», described by ^{Valdivia (2011 [1607])} as imperfect past (preterito imperfecto), so the structure of the negative constructions is asymmetric, as is seen in the contrasting examples 28 and 29. Furthermore, the type of asymmetry is A/Cat since the changes in the structure are related to the marking of a grammatical category, in this case tense-aspect.

On the other hand, in Allentiac, according to Valdivia’s grammar, standard negation is expressed by means of the particle «naha» or its reduced form «na» in preverbal position (NEG V). However, this particle appears in its reduced form «na» in most of the analyzed texts.

A possible way to evaluate whether the negative marker in Allentiac is a prefix or a preverbal word is the use of negative words in negative polar answers. In Millcayac, polar questions are answered negatively with «naha peche» or the word «pechewe», according to the grammar by Valdivia, even though it is «pechewe» that is used in all the available texts. Due to its similarity to the word «peche», which means ‘negate’, it can be argued that it is derived from this word. In contrast, polar questions in Allentiac are answered negatively by means of the negative particle «naha», the same used in standard negation. In the case of negative polar answers, the reduced form «na» is not found in the texts. It is therefore possible to suggest that while «naha» is a preverbal particle, the reduced form «na» seems to be on the path to become a prefix. Since more evidence is necessary to confirm the syntactic independence of «na», in the present investigation we consider that Allentiac presents a preverbal particle in the domain of negation.

The structure of the negative constructions in Allentiac is symmetric since apart from the addition of «na(ha)» no other differences with the affirmative can be found, as seen when comparing examples 34 and 35.

In sum, in Millcayac standard negation is expressed by means of affixation with the suffix «-na» and the negative constructions have an asymmetric structure since apart from «-na» it also requires the imperfective suffix «-eye» and its allomorphs. In Allentiac, SN is expressed by means of the preverbal particle «na(ha)» and the constructions have a symmetric structure. These values are summarized in Table 6.

3.2. Prohibitive constructions

In Millcayac, the paradigm of the imperative suffixes that also mark person and number is the following.

To negate the second person imperative, ^{Valdivia (2011 [1607])} states that there are two possibilities: using the negative suffix «-na», along with the imperfective allomorph «-e», as is also required in the negative declaratives, combined with the corresponding imperative ending (example 36), or using a special prohibitive suffix «-te» (example 37). That is, this language has two different prohibitive constructions: the one in example 36 corresponds to type 1 and the one in 37 corresponds to type 4, according to ^{Van der Auwera and Lejeune’s (2005)} typology. For further detail, see the contrast with example 38 that shows an example of the imperative.

In Allentiac, the imperative mood paradigm is detailed in Table 8:

Like in Millcayac, Allentiac has two different possible constructions to express the prohibitive meaning: one that uses the corresponding form of the second person imperative paradigm and the negative marker used in the declaratives «na(ha)» (example 39), and another that uses a special prohibitive suffix «-che», «-eche», «-tche» (example 40 and 41). This suffix has different allomorphic variants, possibly depending on the conjugation of the verb. For a clearer picture compare them to example 42 where an example of the imperative can be seen.

In sum, both Millcayac and Allentiac have two different prohibitive constructions, type 1 and type 4.

4.1. Synthesis

A summary of the comparison between the considered features in Millcayac and Allentiac is presented in Table 9.

Among the similarities between both languages, we first find the phonological similarity between the Millcayac suffix «-na» and Allentiac’s particle «na(ha)», probably due to their common origin. Furthermore, regarding prohibitive constructions, both languages have two constructions to express this meaning: one of type 1, based on the paradigm of the imperative and declarative negation, and one of type 4 with a special prohibitive marker unrelated to the imperative or declarative negation markers.

On the other hand, both languages show a series of remarkable differences. While Millcayac marks negation by post-verbal suffixation, Allentiac marks negation by means of a pre-verbal particle. Moreover, Millcayac presents an asymmetric structure of standard negation in which the presence of the imperfective suffix «-eye/-e» is required along with the suffix «-na», while in Allentiac standard negation shows a symmetric structure in which only the negative marker «na(ha)» is necessary.

4.2. Proposal for the reconstruction of standard negation in proto-Huarpean

The description of and comparison between Millcayac and Allentiac make it clear that negative constructions in both languages are strikingly different. If we consider the fact that both languages belong to the same family, their differences are even more remarkable. This leads us to look for diachronic developments in these languages that could explain this divergence. Since the only data available from these languages is synchronic, reconstructing the diachronic development is a difficult task. However, based on the available data some diachronic change scenarios can be proposed.

One possibility is that the original negative marker in Proto-Huarpean was «naha» since it is a free form that is still present in negative polar answers in both languages, though it only survived in Millcayac alongside «pechewe». In Allentiac, it kept its order in relation to the verb but has reduced to «na», a change that, based on the existence of both «na» and «naha» in Valdivia, could have been in course at the time the language was documented.

Given that the short form «na» does not occur in negative polar answers, it is possible that, along with the reduction of the free particle «naha» to «na», «na» could have been in a process of prefixation.

In Millcayac, the form «na» became suffixed. In this language, additionally, the imperfective form «-eye/-e» is required. We propose that the addition of the imperfective form in standard negation is a later development not only because of its absence in Allentiac, which seems to be more conservative in this domain, but also because the imperfective is not required in nominalized negative constructions in Millcayac, and these constructions are usually more conservative than the indicatives (see for example ^{Givón, 1999}).

Along with the internal explanations, it is also possible to consider contact influence from neighboring languages, especially given some of the similarities between the Huarpean languages and other Andean languages. An interesting similarity can be observed in the asymmetric structures of Millcayac and Aymara. In Aymara, standard negation is expressed by means of the preverbal particle «jani» and requires the co-occurrence of the interrogative suffix «-ti» attached to the verb (^{Hardman, Vasquez and Yapita, 2001, p. 185}; ^{Coler, 2014, p. 648}). Additionally, along with «jani» and «-ti», negative constructions frequently include the incompletive suffix «-k(a)» on the verb (^{Hardman, Vasquez and Yapita, 2001, p. 301}); however, since the incompletive is not completely obligatory it is not considered to be true asymmetry. If this incompletive suffix were obligatory instead of frequent, it would be classified as an A/cat type of asymmetry, like Millcayac. Despite this, the similarity must be considered.

However, despite the similarity, it must be considered that only Millcayac, the language geographically further south, and not Allentiac, more geographically close to Aymara, has this type of construction. Because of this, it is difficult to imagine a contact scenario between Millcayac and Aymaran where Allentiac would not have been at least similarly influenced. Another possibility is that contact between Millcayac and Mapudungun could have produced, or at least influenced, the development of suffixation as the means of expressing negation in Millcayac.

All the same, interdisciplinary extensions of this research could further refine and elaborate the possible contact scenarios in which these diachronic changes could have taken place.

4.3. The prohibitive and its changes

In both Allentiac and Millcayac it is possible to observe two different prohibitive constructions: one with the same negative marker found in the indicative and the other with a special prohibitive marker. In this context it is possible to suggest, following Manczak (1958, cited in ^{Campbell [1999, p. 265]}), that the special form of the prohibitive could be the older one given that the forms of the indicative show a major trend to expand towards other moods. Meanwhile, regarding the special prohibitive form, it is remarkable that while in Millcayac this form is «-te», in Allentiac it is «-che». Considering the allomorphs found in Allentiac, it is possible that the original proto-Huarpe form was a disyllabic form *«-teche » that derived to «-te» in Millcayac and to «-che», and its variants «-eche» and «-teche», in Allentiac.

To explore a possible explanation for this difference, it is important to note the similarity in form and function between «-che» and the prohibitive suffix «-cha» (example 45) from Kunza, a southern Andean language geographically close to Allentiac. This similarity, the geographic proximity between both languages, and the divergence with Millcayac all make it possible to consider that the prohibitive «-che» may have been reinforced in Allentiac because of the influence from Kunza’s form, producing an internal shift of the family in a divergent direction.