PhiN 33/2005: 84

Roland Schmidt-Riese (München)

Colonial Grammars on Nominal Case. The Quechua series

Colonial Grammars on Nominal Case. The Quechua series
The purpose of this contribution is to investigate the categorization of nominal case structures of Quechua within grammars from the colonial period. The endeavour is at the crossroads of two different interests: on the one hand, the structures of non-Indo-European languages are the primary concern of the texts being analyzed and thus, are of primary concern to the analysis. On the other, the specific assumptions of early modern grammar are what gave shape to the categorization and thus have to be taken into account. Insofar as colonial grammars extend traditional techniques onto languages unheard of, their undertaking is comparable to that of modern language typology. An introduction both into these techniques and into the case structures of Quechua as held by modern scholars is followed by a detailed analysis of the texts. The assumptions on the case category held by early modern authors prove to be amazingly adequate – and diverse.

1 Introduction

Colonial grammars are grammars produced in colonial societies.* Colonial societies are, by definition, diglossic societies. The languages analyzed in colonial grammars are those spoken by the colonized, the metalanguage being the one of the colonizers, or some language used by them. With European colonizers, the metalanguage may be Latin, though this is not often the case. Further, the colonizers bring in resources of grammatical description, coined in terms of their metalanguage. They try to make use of these resources within the colonial setting, and this is what the article is about: how to describe the nominal case system of Quechua in the sixteenth and early seventeenth century, starting from categories worked out for Latin within the Western grammatical tradition? The investigation will shed some light on the history of the inquiry into Native American languages and, thereby, on some basic procedures of comparative linguistics and typology.

The historical setting from which the texts emerge is early Spanish rule in the central Andes. The grammars analyzed are all authored by ecclesiastics, four from the regular and one from the secular clergy. They stem from a period of about half a century, the first having been published in 1560 and the last in 1619. 1 These and other texts of their kind are frequently referred to as 'missionary grammars'. This term is questionable, though, for several reasons. The most important one, from my point of view, is that the concept of 'mission' was unknown in the period under study. Those whose aim was to communicate the Christian faith in the Americas, at least up to some point in the seventeenth century, did not consider themselves missionaries, but instructors (doctrineros). The difference seems insignificant, but it is not. The concept of 'mission' is commonly considered a Christian universal, but in fact is due to certain historical developments – developments which indeed have helped to bring forth our epistemic present from the period under study.2

PhiN 33/2005: 85

The second reason not to refer to the texts as 'missionary grammars' is this: not all the authors were equally engaged in the religious instruction of the natives. In addition to monastery libraries, where native grammar had been studied in the first decades of Spanish rule, chairs of native languages were set up and invested before 1600, both at cathedrals and universities. These scholarly institutions created a proper space for grammatical studies. Thus colonial grammar did not entirely quit the ecclesiastic framework, but it did abandon its most immediate proselytic preoccupations. Certainly, this does not hold everywhere and for all the Native American languages. But it does hold for the central Andes and Quechua.3

The basic assumption of this article is that colonial grammarians are not necessarily wrong. The problems they deal with are real problems, not imagined ones. Certainly, in some instances they fail to set their intuitions straight, and leave contradictions open to question. But this does not mean by itself that there is an easy answer and that we have it. Grammars of Quechua and, for the subject under study, their chapters on case, prove to be amazingly diverse due to their unerring intuitions and contradictions, some silenced, others explicitly assumed. My aim is to see this diversity in some detail. As for conclusions on the progress of grammatical knowledge, I prefer to leave those up to the reader. There is progress of knowledge on Quechua, of course. But there is not just progress.

Next, the authors do not insert Quechua materials into Latin categories. This idea is misleading. The Western grammatical tradition, first of all, is by no means uniform. Second, categorization doesn't work like this. Categorization is not to fill materials into slots given once and for all. Categories consist in networks which result from former processing of entities and which are constantly reshaped by processing. The attribution of entities to a category does not leave the category – the possibility of further attribution – unmodified. Whether by strengthening the network in all its branches, or by strengthening just parts of it, by connecting it to neighbouring networks, some modification takes place. Science certainly tries to make category boundaries explicit and stable. It defines concepts and thus tries to control further evaluation of instances. It does not therefore abandon the basic principle of categorization, insofar as science is an activity and does not consist in resulting statements only.4

From an epistemological point of view, one must admit that Latin grammar is the common ground for both the authors and ourselves. Renaissance grammar is an offspring of the grammar of Antiquity, just as modern linguistics is.5 A certain number of categories are shared among these different programs and the case categories are certainly part of this legacy. This is not to say that definitions remain unalterable. On the contrary, shared categories produce an illusion of shared concepts and therefore constitute a risk for historical research. Further, we know linguistics better, but Renaissance grammarians, colonial or not, knew Latin better and above all, knew the edges and faults of tradition much better than most of us do, including myself. There is a risk then to miss the points they make, by grasping only what still sounds familiar to us.

PhiN 33/2005: 86

This risk increases, given Renaissance discourse conditions. Grammar is an acknowledged topic of that period, but not only this. It is the basis for all kinds of studies and, therefore a severely controlled epistemical matter. The fact that the languages described in colonial grammars are unheard of is not a licence for maintaining nonsense. On the contrary, the more exotic the matter, the more suspicious the book. Lies are unwelcome. Further, if ever an author wishes to be understood, he must use common-sense terms. Colonial grammars are designed for language learners, at least in principle, and they have to rely on traditional concepts to be useful. Even so, there is no need for commiseration. The authors manage discourse and pragmatic conditions, as we would in their place, in addition to the challenge of grasping things unknown. The point is to see how they proceed.

2 Prerequisites

Writing a book on the historical vicissitudes of the concept of nominal case would be a misplaced effort, in this instance.6 A brief reminder, though, of the structures of Latin and the traditional inaccuracy of the concept is in order. "Case is a notoriously ambiguous notion", says the International Encyclopedia of Lingustics.7 It constantly switches from content to form and vice versa, difficult and in fact useless to detain on whatever side of the river. It may be looked at from one side. But while focusing on the opposite bank, you easily lose sight of what you stand on. Anyway, a content-to-form and a form-to-content perspective may be distinguished, in principle.

TABLE I. Latin cases8

1 NOM dom-us Nominative
2 GEN (dom-i), dom-us Genitive
3 DAT dom-u, dom-ui, dom-o Dative
4 ALL (locus ad quem)
dom-um Accusative
5 VOC dom-us Vocative
6 ABL (locus a quo)
[a, de, ex] (dom-u), dom-o

8 LOC (locus in quo) dom-i  

PhiN 33/2005: 87

Table I gives a case paradigm for the Latin domus 'house', a noun inflected in a somewhat deviant manner. The word exemplifies both the dynamics of Latin structures and the problems of categorization. The point is that these are real structures and real problems. Colonial grammar might have been an easier undertaking, had things been simpler in Latin. On the one hand, domus has formal variants for the same argument relation – in the genitive, dative and ablative – on the other hand, it has no variants to offer for different argument relations – nominative and vocative have identical form. The unwelcome but existing variants are due to diachronic developments – domus is, in fact, about to switch from the fourth to the second declension class, in Priscian's count – and the variants bear stylistic significance in synchrony. But diachronic explanation and stylistic evaluation do not cancel the actual variation or lack if it. The vocative, by the way, is identical to the nominative with most Latin nouns. So one might even ask if domus actually has a vocative and, if it does, what the use of this vocative is.10

Further, domus has a locative formally distinct from the ablative, one of the few Latin nouns to do so. In domus, there is both a proper inflection and a proper locative meaning of that inflection and still, the locative is not normally recognized as a Latin case.11 The carryover of case values to inflectional forms unmarked for that value stops wherever it stops. Table I demonstrates the very dilemma of the case category. Nominal case means formal variation of nouns (center column), that indicates distinct argument relations (left column). The grammatical concepts (right column) are supposed to grasp these dimensions simultaneously, but both inflection and argument relations do escape from the grasp in certain instances.

Anything from position 6 on downward is traditionally termed ablative case. This is quite a few different relations, some of which must be specified by prepositions. Prepositions are, in principle, just another means of specifying argument relations. They serve the very same purpose as case. As Latin is a case-inflecting language, prepositions not only specify argument relations, but govern case, in turn. Within the ablative domain, the variation of both argument relations and formal procedures is such that the Roman grammarians tried to work out further distinctions, taking into account both formal requirements and syntactic surroundings.12 These projects were canceled by Priscian and the subsequent medieval mainstream. Nebrija, in turn, claims a seventh case of Latin in 1481, which he terms the effective. This case is defined by lack of prepositions and instrumental argument relation. Nebrija immediately adds that the effective is 'always similar to the ablative' (semper ablatiuo similis), and he consistently avoids dressing an effective in his paradigms.13 Latin case paradigms always have been six-layered. Also, persistent formal coincidence with the ablative would have made the effective case look useless.

Discussion of table I draws attention to two case categories present in Sanskrit and reconstructed for Proto-Indo-European – instrumental and locative – merged in Latin with the ablative. What comes out of this merger is the enormous functional array of the ablative in Latin. This is common knowledge today, but it was not in the sixteenth century. Discussion has thus foregrounded structures and problems which were backgrounded at the time. Even so, they were not absent from it. The authors are concerned in any case with the less-than-total adequacy of mainstream teachings. Nebrija's claim on the effective is just one conscious effort to arrive at more convincing solutions. It is thus part of the uncovering of structures behind inherited concepts, structures of ancestral nature. Colonial studies on Quechua are another, related part of the same inquiry. There is no reason, then, to despise the ancestors for uncovering structures known today, even if resuming modern knowledge on the topic makes their efforts appear all the more insufficient.14

PhiN 33/2005: 88

Table II gives an overview on the case system of Quechua as sustained in contemporary literature.15 This inventory nevertheless bears some problems similar to those observed for Latin. Roughly, argument relations in the upper positions are more abstract, in contrast to those at the end of the inventory, which appear to be rather concrete. Further, relations in the upper positions are generally heterogeneous. This diversity again stretches from the more concrete to the more abstract. Syntactically, it seems to be controlled by the subcategorization of the arguments. This problem is additionally sketched in Table III.

TABLE II. Quechua cases16

1 NOM  
2 GEN -V]p, -C]pa  
3 ACC -ta, (-V]kta)17 ALL
4 ILA -man18 DAT[+hum] DIR
5 BEN -pa-q PURP -na-paq19
6 LOC -pi  
7 ABL -man-ta MAT
8 INST -wan COM[+hum]
9 LIMIT -kama DIST
10 CAUS -rayku  
11 COMP -hina  
12 INT -pura  

TABLE III. Central case items by subcategories of arguments and supposed directionality of grammaticalization

[-an] [+an] [-an]  
2 -p, -pa     GEN GEN  
3 -ta ALL   ACC ACC  
4 -man DIR   DAT    
5 -pa-q     BEN PURP -na-paq
6 -pi LOC        
7 -man-ta ABL MAT      
8 -wan   INST COM    

PhiN 33/2005: 89

Case items in the upper positions of table II are monosyllabic. This holds for positions 2–6 and 8, while position 7, the ablative item, appears to be a combination of 4 and 3. It is therefore bisyllabic. The genitive, accusative and locative items correspond to the CV profile preferred for Quechua suffixes. They show cross-dialectal stability and are presumably of great time depth.20 The genitive item is reduced to one segment only, with stems ending in vowels. Reduced to its consonantal onset, it is thus integrated into the syllable structure of the noun. In contrast, items in positions 9–12 are all bisyllabic. This is not the preferred pattern of suffixes in Quechua. On the contrary, allowing for penultimate accent, these items might easily stand on their own. In fact, the limitative and interactive items continue to be used as autonomous units at the modern stage of the language, if only in certain dialects.21

Discussion of tables II and III has focused on a diachronic process which implies segmental reduction, loss of prosodic and syntactic autonomy, and a semantic change from the more sustained and concrete to the more abstract and scanty: as for case markers, the path runs most often from locative to grammatical. Grammaticalization does not need to be explained here in any detail. The point for the historiographical endeavour is to see that grammaticalization proceeds in the way of a predictable succession of changes. Further, grammaticalization is by no means synchronized within languages. Different items may represent different stages of grammaticalization at any given moment and, strictly speaking, do not even necessarily represent any of the predefined stages in an unambiguous manner.

Two consequences follow from this for grammar writing: (i) it may be impossible to assign a given lexical item to one and only one syntactic category, (ii) it may be impossible to attribute all the functionally equivalent lexical items of a given language – say, all the terms indicating argument relations – to one and the same syntactic category. Grammaticalization implies haziness, if looked upon from the perspective of synchrony. It is not possible to focus on toads, having made different progress on a toads' migration path, so as to see all of the toads sharply at the same moment. This is a serious problem for grammar writing. Because grammar (and linguistics) aims at coherence, unlike language.

The migration path of case items is commonly described thus, regarding syntactic – and prosodic – categories:22

grammaticalization path of relational items
relational nouns > secondary adpositions > primary adpositions > agglutinative case markers > fusional case affixes > Ø

Both the path and the channel metaphor in grammaticalization theory do predict that the terms pass from one stage to the next and that they do not turn back (unlike toads).23 Therefore, a case item may be ambiguous with regard to neighbouring stages. It is nevertheless impossible that it pertain randomly to any two categories which set up the path.

PhiN 33/2005: 90

3 Reading the texts

Starting from the above considerations, the first five grammars of Quechua are analyzed in what follows. The sequence chosen is the chronological one, as later authors may refer to the statements of predecessors but not vice versa. Table IV gives a brief survey. The analysis will be oriented by two question groups:

What are the basic assumptions made by the authors on the case category and on categories in general? Do they choose a content-to-form or a form-to-content perspective? Do they switch from one perspective to the other?

What are the case inventories of Quechua the authors arrive at, on the basis of their assumptions? Are differences due to different methodological prerequisites, or are they due to the observation of different facts?

TABLE IV. Quechua grammars up to 1619

text size25
Sancto Thomas 156026 OP 240.000
AVQE 158627 SJ 100.000
Gonçalez Holguin 1607 SJ 840.000
Huerta 1616   100.000
Torres Rubio 161928 SJ 90.000

3.1 Sancto Thomas 1560

Domingo Medina Navarrete is a member of the Dominican order, having chosen 'Domingo de Sancto Thomas' as his order name. He works toward the religious instruction of the natives from 1540 onward in the coastal regions of Peru, mainly south of Lima. In 1555 Sancto Thomas returns to Spain in order, among other things, to see his grammar printed. When this is achieved, he definitely returns to the Americas in 1561. Sancto Thomas is the only Dominican among the authors and the only one to describe Quechua II A, the dialects of the Pacific lowlands, which were soon lost in favor of Spanish.29

Sancto Thomas holds that there is no declension in Quechua, just as there is none in Spanish. By declension he refers most obviously to the formal variation of nouns. From the fact that declension is linked to formal variation, it follows that, for Sancto Thomas, cases covers the domain of argument relations. Argument relations, of course, do exist. They are not known as declension, but as articles added to cases – i.e. added to nouns inserted into given argument relations. These basic notions are explained first with regard to Spanish.30

PhiN 33/2005: 91

Cerca de la sexta y vltima propriedad (que es declinacion) es de notar que no ay en esta lengua declinacion ninguna, sino que todos los nombres son indeclinables en si, como en nuestra lengua española, en la qual ningun nombre se declina: y de vna misma manera haze en todos los casos, como en este nombre, hombre, dezimos en nominatiuo, hombre, y en el gentiuo [sic] dezimos, del hombre, y en el datiuo dezimos, para el hombre: de manera que todos los casos son inuariables & indeclinables, y se profieren de vna misma manera: pero conoscemos ser la declinacion del nominatiuo, o genitiuo, o datiuo: no en la variacion del, que ninguna ay como esta claro: sino en los articulos que se les añaden a los casos, como la señal del nominatiuo en la lengua española es esta diction, el, para el masculino, como dezimos, el hombre: para el feminino, la, como dezimos, la muger: por aquella particula, el, o, la, entendemos que es nominatiuo. (Sancto Thomas 1560: fol. 5r)

Sancto Thomas adheres to a realistic semiotic model. Cases, in this view, are universally given relations – real things. The question is how these relations can be recognized. This is what Spanish articles are good for: they are indices, attached to the outer limits of nouns, that reveal the argument relations. Articles, therefore, do not fill nouns into relations. They just give a hint of what's going on, of what the case of the noun is.

Articles, nevertheless, are not the only indices to do so. In fact, with regards to Spanish, articles are, strictly speaking, just the indices that signal nominatives, according to Nebrija.31 Sancto Thomas is less conclusive on category boundaries, though. In addition to articles, he uses the terms of particles, signals and dictions when he refers to case indices. Not uncommonly – conforming to Renaissance stylistics – he employs two terms at a time. He then refers to one target item by particle or article, particle or signal, article or signal, article or diction. The distribution is not random, as first impression suggests.

Sancto Thomas employs, it seems to me, article and particle as both formal and functional categories, in the sense of 'adnominal clitics pointing out argument relations', with particle being prosodically the more autonomous term. Signal, on the other hand, is just a functional category and predicts nothing about phonological weight and prosody, while diction refers to any substantial and autonomous term – to a word – and predicts little if anything on grammaticalization. The author seems to have a clear intuition of what the problem is. Quechua case indices may not belong to just one formal category.

Sancto Thomas even spells out what the criterion of a split into different formal categories would be: the question of an index forming one term with the noun. In fact, it seems to him that the case articles of Quechua do form just one term with the noun, so that it might, and in fact, it must be said that there is formal variation of nouns in Quechua, that there is – unlike in Spanish – declension. Having arrived at this claim, the opposite of the one he started from, Sancto Thomas avoids exploring the matter any further. The point is probably that some articles do form just one term with the noun, while others do not. At least, I suspect that this is what underlies the author's apparently random use of the labels.32

Y aunque esto sea assi verdad, que el nombre en si sea indeclinable: pero porque los articulos que se le añaden, paresce que se hazen vn mismo termino con el: y estos entre si son diuersos como abaxo se vera, me paresce que se puede y debe dezir, que todos los nombres y demas partes de la oracion declinables tengan vna declinacion, no por parte dellos, sino por la razon de la diuersidad de los articulos: y assi se dara declinacion del nombre y mas partes de la oracion declinables, por lo dicho: para lo qual es de notar las reglas siguientes. (Sancto Thomas 1560: fol. 5v)

PhiN 33/2005: 92

Sancto Thomas claims seven cases for Quechua: six traditional Latin cases plus the effective, proposed for Latin by Nebrija. Sancto Thomas, though, goes one step further than the author of Introductiones latinae, in that he introduces the effective into paradigm tables, the most exposed part of grammars. This option is not only licensed by the fact that no-one has ever claimed anything different about Quechua, but is sustained by target structures. In Quechua, the effective is not just 'not always similar to the ablative'. It never is: /-guan/ is different from any other index.

Nominatiuo yayanc ____________ el señor.
Genitiuo yayancpa ____________ del señor.
Datiuo yayancpac ____________ para el señor.
Accusatiuo yayancta, o yayancman ____________ al
Vocatiuo xe, o xay yayanc ____________ o señor.
Ablatiuo yayancpi
____________ en el señor, o
del señor.
Effectiuo yayancguan ____________ con el señor.

The paradigm is supposed to have seven lines, as there are seven case categories sustained in the left column. But in fact, there are nine lines, and this is not just for layout reasons.33 Additional lines spring from the center column, which gives the target language items. Having a paradigm with three columns is not an innovation. It simply corresponds to medieval grammar usage, wherein Latin holds the center column. As seen from this perspective, Latin items just seem to have been interchanged for Quechua ones. Still, as Quechua creates additional lines, not announced by the case category terms on the left, Spanish equivalents on the right do not just translate the target materials for didactic purposes either. They interpret them, additionally, with regards to the argument relations.

This is most evident for the ablative, where target locative and ablative are given different Spanish translations. It remains unnoticed within the accusative, because target accusative and allative happen to be merged in Spanish, as they are in Latin. The author fills the alternative target items into one line for the accusative, but into different lines for the ablative. In his discussion, though, he refers both alternatives to a difference in kinesis: verbs of quiescence (verbos de quietud), he says, require /-(c)ta/ accusative, motion verbs /-man/ accusative. Analogously, verbs of quiescence select /-pi/, motion verbs /-manta/ ablative.

PhiN 33/2005: 93

Las particulas del accusatiuo son dos: la vna es (ta), y la otra es (man), empero ay differencia entre el (ta) y el (man) que el (ta) sirue al accusatiuo quando se ayunta el tal accusatiuo con verbo que significa quietud: como diziendo, yo veo al señor, diremos (yayancta ricuni), pero si se ayunta el tal accusatiuo a verbo, que significa mouimiento: assi como diziendo, voy al señor, diremos (yayancman rini). (Sancto Thomas 1560: fol. 6v)

Al ablatiuo tiene dos articulos, o dictiones por señal: el vno es (pi) y el otro (manta) con differencia que el (pi) quiere dezir, en, y sirue y ayuntase con verbo que significa quietud: como diziendo, en el señor se hizo, diremos (yayancpi rurasca), el (manta) significa mouimiento, y ayuntase siempre con verbo que significa mouimiento: y significa lo que dezimos en romance, de, con mouimiento, y en latin (ex) o (abs) como diziendo, vino del señor, diremos (yayancmanta xamurca). (Sancto Thomas 1560: fol. 7r)

The analogy seems convincing. Still, when reading the text closely, there is one difference left between accusatives and ablatives. Within the accusatives, the verbal head is said to trigger the difference in kinesis exclusively. Within the ablatives, /-manta/ is attributed motion semantics. No need to recall here that from the perspective of grammaticalization, allative and accusative are virtually the same thing. Even if in Quechua, there are two different source items, one of them having been grammaticalized much further on the track towards an accusative, i.e. /-(c)ta/. For the locative and ablative merger, on the other hand, there is no universally predicted path.

The analogy Sancto Thomas outlines is in any case dependent on his claim of the effective. It would break down, had he included /-guan/ among the ablatives. The semantics of the Quechua effective, though, is not coextensive with that of the Latin effective as outlined by Nebrija. For, in contrast to the bare ablative morphology of Latin, /-guan/ covers both instrumental and comitative. This does not pass unnoticed. Sancto Thomas refers the convergence of instrumental and comitative to the functional array of Latin cum. This cross-linguistic analogy is consistent in functional, but not in formal terms: cum is a preposition, not a case suffix.34

El articulo o diction del effectiuo es (guan) que significa lo que en romance, con, o en latin (cum), vg. venço con el señor, dezimos (yayancguan atini) verdad es que esta diction (guan) tambien significa compañia, vg. dezimos: voy contigo (canguan rini) y assi en los demas. (Sancto Thomas 1560: fol. 7r)

Sancto Thomas does not insist on prosodic differences within his paradigm. Still, there is no space left between the noun and the indices except for the vocative.35 Except for the vocative then, articles do seem to form a single term with the noun, in contrast to what occurs in Spanish. Except for the vocative, Quechua seems to have declension. But the given target materials may also correspond to Latin categories only with regard to contents. They may just serve given argument relations, in whatever fashion. There would then be no declension in Quechua, as there is none in Spanish.

The latter perspective receives additional support from the discussion of prepositions. Quechua prepositions, the author explains, are autonomous case-assigning terms. Being prepositions of this kind – not being parts of complex word structures, they are nevertheless articles of Quechua cases. This is an unsettling statement. Nebrija, in fact, had claimed Spanish a and de to be prepositions, but he had claimed them to be nothing else.36 Sancto Thomas, in turn, holds that Quechua prepositions – postpositions for their location – both assign cases to nouns and indicate the assigned cases. For the latter purpose, they are termed articles. Each of these views is supported by traditional arguments, but the circularity which results from linking them is not. It seems inevitable, nevertheless, to the author.

PhiN 33/2005: 94

Y lo primero acerca de la preposicion que se antepone en la lengua latina a los casos, rigiendo y determinando los nombres, y las demas partes declinables, a casos determinados. Y al verbo, y a las demas partes de la oracion, componiendose con ellas, como consta de la lengua latina, es de notar, que en esta lengua estas preposiciones se hallan solamente, en la primera manera (conuiene a saber, rigiendo, y determinando casos, en los nombres, pronombres y participios: pero nunca se hallan juntas, con verbos, ni con las otras partes de la oracion por composicion. Las quales son todas las particulas dichas en la declinacion del nombre: porque todas ellas de tal manera son articulos de aquellos casos: que son juntamente preposiciones dellos (conuiene a saber) el (pa) es preposicion de genitiuo (pac) de datiuo. Y assi de todas las demas arriba dichas en la declinacion del nombre, &c. Porque si solamente fueran articulos no tuuieran la significacion que vemos que tienen de preposiciones, y assi son preposiciones. Y aunque lo son en la significacion, y officio: pero en el sitio son postposiciones: que siempre se postponen a los tales casos, como arriba se dixo, assi no dezimos (pac ñoca) sino (ñocapac) que significa, para mi. Y assi en todos los demas. (Sancto Thomas 1560: fol. 50v–51r)

Sancto Thomas is aware of the risk that this approach implies: if Quechua articles are prepositions, Quechua prepositions might be articles as well. The number of Quechua cases would then exceed the number of well-known cases: it would make the canon explode. The author makes an effort to contain the danger. He discusses in detail two relational items which, he says, resemble prepositions, but are not. Both /nac/ 'without' and /cama/ 'until' are found in postposition to nouns, both establish argument relations (significacion), but they refrain from assigning case.37 The argument is hard to refute, because the case would be covered anyway. Whether or not covert categories have been assigned is an awkward question. The author handles it in a way which allows him to maintain the categories he started from – a familiar procedure.

Sancto Thomas adduces additional evidence for his claim that nac and cama are not prepositions and do not assign case: nac 'without' also builds up nominalizations ('...less') – cama 'until' accepts as arguments not only NPs, but sentences.38 Both observations are relevant, but less than convincing.

Assi mismo esta diction (nac) es preposicion en la significacion, y significa (sin), y posponese a los terminos. Dixe en la significacion, porque no rige caso alguno determinado sino que se junta con el nombre y se haze quasi vn termino con el, vg. (guarmenac) sin muger, (coçanac) sin marido, &c. y diziendo, veo a la muger sin marido, diriamos (coçanacta ricuni), donde el (nac) quasi haze parte del termino (coça) como paresce. Y assi en los demas. [...].

Item esta diction (cama), que significa lo que en latin (usque) y en romance, hasta, en la significacion es preposicion, aunque en el regir del caso se junta con el nominatiuo, vg. dezimos (caycama) hasta aqui, (camcama) hasta ti, &c. Y algunas vezes se junta con verbo, como diziendo, hasta que yo coma, dezimos (micungaypac) [micungaycama] hasta que tu comas (micungayquipac) [micungayquicama]. (Sancto Thomas 1560: fol. 51r)

PhiN 33/2005: 95

Sancto Thomas, in sum, perspicuously discusses the prosodic differences among Quechua relational items. He adheres, nevertheless, to a concept of grammar which admits covert categories. Although a seventh case is claimed for Quechua, discussion aims at maintaining the boundaries of the case – adposition distinction as known from Latin. The categorization of relational items as both case articles and postpositions remains open to question.

3.2 AVQE 1586

The AVQE is printed in Lima by Antonio Ricciardi. The title sounds Franciscan.39 This may be due to the fact that Ricciardi, being from Piemont, worked in Mexico before moving to Peru in 1580. 40 The Franciscans still dominate the grammatical field in Mexico at the time and Ricciardi, who authors both the dedication and the address to the reader, may be responsible for the AVQE's title as well. The book is unusual in that the grammar follows the dictionary, not vice versa. Ricciardi states in the dedication, written in 1584, that the dictionary is completed, but he doesn't mention the grammar at all. The grammar may have been added to the dictionary later, but before 1586.

The AVQE is related to the Third Council of Lima (1582–1583), which implements, in the Andes, the resolutions of the Council of Trent. The Lima council edits canonical versions of the catechism in both Quechua and Aymara and thus sets a standard for both languages.41 The AVQE conforms to these standards, which – for Quechua – rely on the Cuzco variant.42 This choice is no accident. It overruns both the Inca and the Dominican linguistic traditions, while it flatters the traditional elites of the ancient capital. And the Jesuits have a strong hold on Cuzco.43

The author turns the terminology of Sancto Thomas upside down.44 Quechua, it seems to him, has declension, but not cases. The underlying statement apparently remains the same: there are argument relations in Quechua marked on NPs, but they are marked by means different from fusional affixes. The terminological turning-of-tables has but one inconvenience: there is no term left to refer to an individual argument relation, as declension obviously refers to the whole system. Further, the author levels out all the prosodic differences introduced by Sancto Thomas. He invariably uses the term particles to refer to the case items of Quechua. This, of course, is not just a matter of terminology. The question of whether or not a case item forms a single term with its noun is abandoned altogether.

The AVQE follows a different semiotic model, which may be termed nominalistic. Categories, in this view, may not be claimed on the basis of zero evidence. There are no such things, for instance, as covert cases. Case particles in Quechua, the author holds, are always added to one and the same case, to the nominative of nouns. The nominative is the non-oblique, the non-case, and if there is just one case, there is none, whatever its name. The argumentation holds tight to previous terminological choices: no cases in Quechua. In a different perspective, assuming that the particles are in fact agglutinated to nouns, one might interpret nominative as an ad-hoc equivalent of 'nominal stem'. For there is no term or concept available in that period which corresponds to this notion.45

La declinacion es vna sola para todos los nombres, pronombres, participios, y finalmente para todas las dictiones que se pueden declinar como nombres, aunque se ponen quatro, todas ellas se reduzen a vna, no se haze por variacion de casos como en la Latina, sino por addicion de ciertas particulas sobre el Nominatiuo, las quales siempre se posponen al nombre en la variacion y oracion. (AVQE 1586: fol. 2r)

PhiN 33/2005: 96

The paradigm table of the AVQE has nine lines, as had that of Sancto Thomas. The lines correspond, in principle, to the same categories or items. Yet there are differences. In the AVQE, every slot is filled in by just one term or sequence, if it is filled at all. There are vacant slots left not just in the category label column, but in the translation column, too. Strictly speaking, accusative in the left and al hombre in the right column are set in between lines four and five so as to cover both target terms, runacta and runaman.46 But this does not help there being nine lines, which spring from the target column.

Nomi. runa el hombre
Geniti. runap del hombre
Dati. runapac para el hombre
Acusa. runacta al hombre
Vocati. runa o hombre
Ablat. runapi en el hombre
  runamanta del hombre
  runahuan con el hombre

The author discusses Quechua declension from a content-to-form perspective. The nominative has no particle, he says. Genitive, dative and accusative have /p/, /pac/ and /cta/. In order to establish accusative /man/, nevertheless, he has to turn to the opposite perspective, explaining what /man/ signifies, thus starting from the term encountered. The author is inconclusive on the vocative. He insists that there are different vocative items, all optional and located either before or after the noun.47 The table is consistent, nevertheless, with the claim that the Quechua vocative is similar to the nominative, i.e. identical.

None of the vocative items seems to be a particle, to the author. It is intellectually honest, then, not to introduce any into the table. But things being as they are, the vocative line remains unsustained by the target. It is sustained by the Spanish translation, but this is certainly not the point of a Quechua grammar. The vocative line is thus maintained for teaching considerations or for convenience only.

Finally, the ablative category covers three different target items. The AVQE claims no effective, although it is stated that /huan/ serves the effective. This is an unsettling statement, for the author starts from target materials and arrives at case categories he does not claim. Had he followed the same procedure for /pi/, both the locative and the instrumental would have been established for Quechua. As if to escape this possibility, the author defends the idea that there are ablative endings. He suddenly abandons all the formal categories outlined thus far. For the concept of endings suggests that there are cases. There may be cases in Quechua of course, but the ablative domain would certainly be the last one to start from.

PhiN 33/2005: 97

El Nominatiuo no tiene particula, como runa hombre, cari varon, huarmi hembra. El Genitiuo vna p sola en los acabados en vocal, pa en los acabados en dos vocales, o en consonante. El Datiuo tiene, pac. El Acusatiuo tiene cta, para los acabados en vocal, y ta para los acabados en dos vocales, o en consonante, man significa a con mouimiento corporal o espiritual. El vocatiuo es semejante al Nominatiuo, aunque algunas vezes le preceden los vocatiuos de llamar, y otras vezes se posponen, como llamando el hombre a otro hombre le dize ya, o yau runa, runaya, o runa yau, hola hombre. El hablatiuo tiene tres terminaciones, pi, manta, huan, pi es lo mesmo que, en, manta de, huan significa con, y sirue para Effectiuo. (AVQE 1586: fol. 2r)

All of the particles introduced into the paradigm table are discussed once more under the heading of prepositions. The coincidence is not commented on, the reader may note it or not. There are two exceptions, though: accusative /cta/ is not mentioned. Genitive /pa/ is, but is categorized as the genitive item only.48

The AVQE, in sum, distinguishes between formal and functional aspects with admirable clarity. It gives preference to the former, though abandoning inquiry into prosody. Structure is taken seriously. In the paradigm table, vacancies are admitted. These vacancies clearly point to inconsistencies between categories and target. Still, discussion on the ablative, the most delicate matter, violates the AVQE's most genuine perspective.

3.3 Gonçalez Holguin 1607

Gonçalez Holguin arrives in Lima in 1581.49 He mainly stays in Cuzco, Juli on the shore of lake Titicaca, Quito, Chuquisaca and Arequipa until 1607, when his grammar is printed. He then moves to Paraguay. Gonçalez Holguin claims to have done better than previous authors on Quechua, and doubts that his grammar might ever be surpassed. This claim is subscribed to by both contemporary examiners and historiography, but is accepted only partially here.

Gonçalez Holguin outlines his grammar as a dialogue. The script sets two characters, disciple and master. The disciple's task is twofold. He asks for basic grammatical notions and filters in, at convenient moments, arguments sustained in the literature. He is thus both an ignorant and an intimate connoisseur of previous grammars. This literary fiction creates a virtual rhetorical frame for the author's arguments. In the initial turn of the section, declension is defined in a vertiginous manner. Formal devices are referred to simultaneously by endings, cases and particles alike, while content relations are referred to by cases, as well. The distinction between inflection and content relations which Sancto Thomas and the AVQE tried to establish is canceled by means of rhetoric.

Further, six cases are assumed without specifying for what language. The master lectures on universal grammar, apparently. For Quechua, in any case, his explanation would be invalid. The Quechua ablative, as outlined in what follows, does not have just one particle, as cases, in the master's turn, are claimed to have.

DISCIPULO. Que cosa es declinar nombres? MAESTRO. Es mostrar de quantas maneras puede vn nombre acabarse, y quantas terminaciones tiene, que se llaman casos, los quales son seys, cada vno conocido por su propria terminacion, que son estas particulas, que no significan nada sino señalar cada vna su caso. (Gonçalez Holguin 1607: fol. 1r)

PhiN 33/2005: 98

Still, with regard to Quechua, there is a distinction introduced immediately hereafter, with genitive, dative and accusative on the one hand, and nominative, vocative and ablative, on the other. The first group of cases does have particles in Quechua, the second does not. The nominative, within the second group, has no particle at all. The vocative, on the contrary, may be indicated (puede), or may not, and there is a choice of different strategies. Gonçalez Holguin even introduces a vocative item not previously mentioned, and attributes a wider pragmatic scope to it (siempre se puede). In fact, this presumed vocative item is the possessive suffix of first person singular (Yayay - '(oh) my father').

The ablative, finally, 'has as its sign and ending' any primary adposition. The second group, then, is defined in negative terms only, and is extremely heterogeneous in itself. The grouping procedure, in fact, remains discrete. It has to be deduced from the sequence of paragraphs, structured by the disciple's questions.

DISCIP. Como no ay particula para los otros tres casos, sino para la mitad nomas? MAEST. Porque el Nominatiuo no tiene particula ninguna, que esso fuera acabar todos los nombres de vna manera lo qual no puede ser: y esso mismo basta para conocer que es Nominatiuo el no tener particula ni señal ninguna, porque el Vocatiuo puede tener vna A antes de si, quando exclamamos, o llamamos en nuestro fauor, como à Dios ó Dios, y puede tener ya antes o despues de si para quando llamamos, ya hola, mas es para amos o mayores, pero siempre se puede poner y al cabo para llamar con cortesia, Yayay Padre mio. El Ablatiuo siempre tiene por su señal y terminacion una de las preposiciones simples, porque todas ellas son de Ablatiuo, y siruen de señalarle. (Gonçalez Holguin 1607: fol. 1v)

Gonçalez Holguin thus introduces category borderlines into the paradigm. For him, only genitive, dative and accusative are cases with definite particles. Nominative, vocative and ablative are cases of a different kind. While Sancto Thomas remained vague with regard to prosodic profiles and ultimately declared particles to be adpositions at the same time, while the AVQE disregarded prosodic differences and kept silence on the parallel assignment of items to different categories, Gonçalez Holguin rejects parallel assignment and cuts the paradigm into pieces, instead. At the same time, his table resembles Latin tables more than any previous one. There are six lines, as there should be.

Nominatiuo Ccari El varon
Genitiuo Ccarip Del varon
Datiuo Ccaripac Para el varon
Acusatiuo Ccaricta Al varon
Vocatiuo A, ccari ya O varon
Ablatiuo Ccari huan Con el varon

PhiN 33/2005: 99

The table perfectly resumes the previous discussion.50 The disciple, nevertheless, accustomed to the tables of Sancto Thomas and the AVQE, is unhappy with it. He asks for both the second accusative and the second and third ablative. The master's answer is the same for both questions: there is just one accusative, because there couldn't be two (no puede auer), and just one ablative, because there is just one ablative in grammar, as is perfectly known (no ay mas). The line of reasoning, though, is different in each case. As /cta/ is the accusative ending, /man/ must necessarily be something different, i.e. an adposition. Adpositions in turn, at least the primary ones (preposiciones simples), govern ablative case in Quechua.51 Therefore, any preposition may be inserted into the ablative slot of paradigm tables and, for didactic reasons, in fact, a different one is chosen each time (cada vez diferente).

DISCIPULO. Como no se pone aqui el otro Acusatiuo de mouimiento con la preposicion man? MAES. Porque no puede auer dos accusatiuos, sino vno: y porque en esta lengua, el man, que es a, no es preposicion de Acusatiuo, sino de Ablatiuo, ni man dize siempre mouimiento, ni el mouimiento aca se pone con preposicion de Acusatiuo, sino de Ablatiuo, como prouamos en su lugar, Lib. IV, Cap. 3, que no ay en esta lengua mas preposiciones que de Genitiuo y Ablatiuo solamente.

DISCI. Como no ay aqui tres Ablatiuos con manta, pi, huan? MAES. Porque no solas essas tres son preposiciones de Ablatiuo, sino otras muchas que estan en el dicho libro IV. Y porque no ay mas de vn Ablatiuo que lo señala qualquier preposicion, y no solo las tres, pi, manta, huan, basta que pongamos vna en cada declinacion, y cada vez diferente, para que se queden todas en las memoria. (Gonçalez Holguin 1607: fol. 2r)

As far as the ablative is concerned, Gonçalez Holguin then admits covered cases. He does not insist, of course, that the Quechua ablative, as he construes it, is formally identical to the nominative. This goes without saying. Nor does he insist that the target materials introduced in paradigm tables represent incompatible categories. For unlike the genitive, dative and accusative, the ablative is recognized in syntax only, by the prepositions which govern it. Prepositions – except those which govern genitive – govern ablative case. Therefore, no effective is admitted for Quechua either.

DISCIPULO. Y porque no se pone otro caso mas que es el effectiuo? MAEST. Porque en esta lengua Ablatiuo y effectiuo todo es un caso, porque el effectivo siempre se haze, no como en Latin sin preposicion, sino con una de dos preposiciones, ò huan, ò Pi, y estas son siempre de Ablatiuo. (Gonçalez Holguin 1607: fol. 1v)

Gonçalez Holguin is consistent is his formal reasoning. If /man/ governed the accusative, accusative /cta/ would have to be added to the noun first (auia de ponerse). As this is not the case, /man/ governs the ablative even if, semantically speaking, it is the equivalent of Latin /ad/. There are Quechua prepositions, in turn, which govern the genitive. In this case, the noun is suffixed by /pa/. The genitive, though, is the only alternative to the ablative in adpositional syntax.

Y lo primero que se ofrece disputar de las preposiciones, es de que caso sean, o que caso rijan, y si auemos de seguir la regla que dan los grammaticos en todas las artes para que vna preposicion sea de algun caso, o le rija, hase de poner el caso con su particula propria, p, o pac, ò cta y luego la preposicion, como para para ser ad en latin preposicion de acusatiuo, dezimos ad Deum, no ad Deus, y assi para ser de acusatiuo la preposicion man auia de ponerse sobre la particula de acusatiuo cta, como ylgesiactaman, a la iglesia. Mas no es assi, ni tal se puede dezir sino yglesiaman, y por eso man no es preposicion de acusatiuo, porque yglesia no es acusatiuo, que lo auia de ser para que se llamasse de acusatiuo, y dezir que es otro acusatiuo de mouimiento, no ay ningun acusatiuo de mouimiento en ninguna lengua ni dos acusatiuos. (Gonçalez Holguin 1607: fol. 133v–134r)

PhiN 33/2005: 100

The author explores the details of adpositional syntax further. He states that 'genitive adpositions' (i) receive possessive marking from their arguments, (ii) serve as arguments of 'ablative adpositions', (iii) may be incorporated, and – for all three reasons, not just for the last one, as the author holds – genitive adpositions are nouns 'in addition to being adpositions'.52

De las preposiciones vnas son compuestas y son todas de genitiuo, otras simples y son de ablatiuo. Estas primeras preposiciones de genitiuo tienen algunas cosas communes entre si, que no conuienen a las de ablatiuo. Lo primero, que todas estas reciben los pronombres possessiuos, y las de ablatiuo no. Lo segundo, que estas se componen con algunas de las preposiciones de ablatiuo, y las otras no se componen con otras preposiciones. Lo tercero, que estas son nombres adjectiuos casi todas, demas de ser preposiciones. (1607: fol. 134r)

Gonçalez Holguin, in sum, adduces both content-to-form and form-to-content considerations, choosing in each instance the ones which come in most handy for keeping the categories of universal grammar straight. No attempt is made at defining formal and functional aspects of case independently. Nevertheless, the implicit splitting of Quechua relational items is perspicuous. It may just be too acute to account for grammaticalization in process. The ablative is outlined by negative evidence only. It remains formally unsustained. On formal grounds, what the author has as ablative adpositions might also be nominative or vocative adpositions. All three cases are equally unmarked, as argued.

Huerta 1616

Alonso de Huerta is born in Huanuco, Peru. He probably aquires knowledge of the regional dialect of Quechua as a child. Huanuco is located east of Lima, in the Quechua I area.53 Huerta enters the secular clergy, both teaching Quechua and preaching in this language at Lima cathedral.54 His grammar thus springs from his teaching experience, as probably do all colonial Quechua grammars. The AVQE, Huerta claims, offers too little, Gonçalez Holguin too much for beginners. He keeps silent on Sancto Thomas, as the Jesuits did.

If Huerta builds on any preceding text it is the AVQE. He even copies his definition of declension from this grammar, deviating from the original only in two significant respects: (i) he avoids referring to nominal stems as 'nominatives', he simply doesn't refer to the basis of nominal inflection at all, (ii) he gives up the uniform treatment of case items as particles, and introduces instead one new formal concept, the letter. Of course, letter is not an innovative concept in orthography (in phonology), but it is in morphology. Huerta refers by letter to certain phonological materials added to nominals. The term is conspicuous insofar as only one variant of one Quechua item, genitive /-p/, is monosegmental.55 In any case, prosody is part of the concept: both letters and particles are items added, are adnominal clitics – enclitics for their relative positions. Of the two, letters a fortiori.

PhiN 33/2005: 101

La declinacion en esta lengua es vna para todo nombre, pronombre, o participio declinable. Hazese, no por variacion de casos como en la lengua Latina, sino por addicion de letras o particulas, aduirtiendo solo vna cosa, que es la letra en que se acaua el nombre que se ha de declinar, que puede ser vna vocal, consonante, o dos vocales (y todo lo que se dixere de el nombre acauado en consonante, se ha de entender de el acauado en dos vocales). (Huerta 1616: fol. 3r–v)

His paradigm table has ten lines.56 Of these, the lower five present exclusively ablative materials. Alternative accusative and vocative items are contained in one line each, as Sancto Thomas might have preferred. Nothing announces within the table that these are alternatives of a completely different kind. For as is obvious, the alternative items code different argument relations in the accusative, while in the vocative they code the very same, if any. Alternative translations are offered within the vocative line only. The identity of allative and accusative in Spanish is an accomplice of the case categories, once more, and it is virtually misleading in functional terms, with regard to Quechua.

The ablative field contrasts sharply with this. Huerta adds one argument relation, the causative (motive, 'for ...'s sake'), hardly sustained as an ablative within Latin, and he adds a variant of the inessive, dialectal in nature.57 He maintains the established sequence of inessive, ablative and instrumental in the ablative positions 1–3 and adduces the new materials in addition to that sequence, with the functionally relevant item first. Nevertheless, conformity with tradition is less than total. Huerta might have inserted the inessive variant into the existing inessive line, or he might have left it out of the paradigm altogether. Mentioning it in the comments would have sufficed. Huerta, as becomes clear from this, makes no effort to contain the proliferation of ablatives. And he proceeds consciously in this way.

NO. Yaya el Padre
GE. Yayap de el Padre
DA. Yayapac para el Padre
Ac. Yayata l. yayaman al Padre
VO. à yaya l.yayaya o el Padre, o ola Padre
AB. Yayapi en el Padre
  Yayamanta de el Padre, o segun el Padre
  Yayahuan con el Padre
  Yayaraycu por amor del Padre
  Yayachau en el Padre. CHINCHAYSUYO

The author deals with the problem of boundaries between case endings and prepositions, as did the former authors. His solution, though, is completely different from any previous one and is, it seems to me, by far the most convincing. Huerta outlines an intermediate category, which, even if termed preposition, is anything but case-assigning and covers Quechua declension as a whole. He proceeds by subsequent steps of reasoning to the destruction of common-sense prepositions and the construction of an appropriate concept. First, Quechua prepositions never precede, but always follow the noun. Second, Quechua prepositions are invariable particles joined to lexical heads, both nouns and verbs, so as to modify their meanings (particula llegada).58 Third, all the nominal cases, as set out within declension, are prepositions in this sense of the term, as are derivative and possessive items joined to nouns and verbs.59

PhiN 33/2005: 102

La proposicion es cierta particula indeclinable, que se puede llamar mejor en esta lengua posposicion, o interposicion que proposicion, por posponerse y interponerse, y no anteponerse, la qual particula llegada al nombre o verbo le declara y varia su significacion. Son vnas proposiciones de todos los casos, que son las con que se haze la declinacion, otras de composicion de nombres, o pronombres, que son las con que se componen los nombres compuestos de segunda manera, y los pronombres possessiuos. Las otras son de composiciones de verbos, con que se forman los deriuatiuos. (Huerta 1616: fol. 33r–v)

Huerta, in sum, looks for prosody. He adds one functional item to the established list of different Quechua ablatives, an item unsustained by Latin.60 His concept of preposition, though, cancels the principle of government in Quechua adpositional syntax altogether. It refers the whole matter to morphotactics. Roughly equivalent to 'both inflectional and derivative affixes and clitics', this concept would certainly not resist further additions to the paradigm table. On the contrary, additions are encouraged. Declension is no longer a closed inventory.

Torres Rubio 1619

Diego de Torres Rubio arrives in Peru in 1578. He first stays in Juli, then, from 1586 onwards in different cities of modern Bolivia, mainly Potosí and Chuquisaca. All these Jesuit emplacements fall within Aymara territory, and it is on Aymara that Torres Rubio publishes first, in 1616. His Quechua grammar, in fact, copies the structure of the Aymara grammar in its details. The address to the reader is literally the same, as are large parts of the text. By means of both grammars then, the author claims, perfect learning of the target languages is possible, if combined with and continued by immersion. Former grammars are left unmentioned.61

Torres Rubio returns to the reasoning of Sancto Thomas. This is most obvious from the fact that he re-introduces the effective, excluded since 1560, into the paradigm. At the same time, Torres Rubio suggests one new concept, exclusively functional in nature. Returns are never complete. Cases, the author says, are known in Quechua by particles or notes. The latter, innovative term is abundantly used in subsequent discussion. It resembles Sancto Thomas's signs, but has the advantage of being disconnected from medieval terminology. Notes, at the same time, is more abstract. Contiguity between the sign and the signified is no longer assumed. Torres Rubio only appears to return to the realistic semiotic model, just as rationalism does.

Covert cases are assumed anyway, and even more easily, as the represented is emancipated from representation. The notes, Torres Rubio holds, are added to the respective cases of nouns.

En esta lengua no ay mas que vna declinacion por la qual se declinan todos los nombres, y las demas partes declinables. En ella los casos no se conocen por la terminacion como en la latina, sino por ciertas particulas, o notas que se posponen a cada caso. (Torres Rubio 1619: fol. 1r)

PhiN 33/2005: 103

Torres Rubio is the first author to outline a paradigm table consisting exclusively of target items, lacking both the accustomed example nouns and Spanish translations.62 As seven cases are assumed, so there are seven lines. Formal and functional variants are carefully distinguished, the former receiving the abbreviated l. 'or', the latter just a separation point. Torres Rubio fills in possessive /-y/ for the vocative, mentioned by Gonçalez Holguin in his comment only. Quechua possessives always have postnominal position, as do case items. One traditional irregularity of the paradigm is thus ruled out. The price of this is to assume that /-y/ is both a vocative and a possessive item.63

Geniti. p l. pa
Datiuo pac
Acusa. cta l. ta. man
Vocati. y
Ablati. pi. manta
Effecti. huan

Assumptions such as these bear no problems for Torres Rubio. The attribution of relational items to both the case and the prepositional category, stated as a paradox by Sancto Thomas, is treated by Torres Rubio as a mystery. Certainly, reference to Christian dogmatics is not so explicit as to say that the items have two different natures. The author just states that they have two different 'things' (dos cosas). Things, nevertheless, refers to being prepositions and notes (ser prepocisiones – ser notas). Beginners are told they might have the terms as notes only. This does no harm, the author argues, because covert cases are correctly assigned anyway.

Even so, beginners (principiantes) might be interested in what the covert cases are they assign to nouns without knowing. For among all the mysterious identities, it remains unclear whether /huan/ assigns covert effective, as maintained in the paradigm table, or covert ablative, as maintained in the comment. Covert cases, as it appears, are less than evident.

Nota, esta particula man, de acusatiuo de mouimiento y estas, pi, manta, huan, de ablatiuo, tienen dos cosas. La vna ser prepocisiones, y la otra ser notas de los dichos casos. Aqui siruen de notas para los principiantes: porque nunca vsamos dellas sino es en acusatiuo, y ablatiuo, y assi la declinacion es. (Torres Rubio 1619: fol. 1v)

The author supplies an inventory of Quechua prepositions, starting from Latin, with a first section on items governing the accusative, and a second on items governing the ablative. Within the first section, /man/ is duly registered as an equivalent of Latin ad, in, aduersus, aduersum, contra, and /ta/ as an equivalent of Latin per. Still, /manta/ is registered twice: as an equivalent of ab, ex, de and e in the ablative section, as it should be, and as a governing term in the accusative section.64 Semantic considerations are fundamental to this way of proceeding. Formal criteria are downgraded, at least with regard to the target.

PhiN 33/2005: 104

Torres Rubio, in sum, uses an epistemic model similar to that of Sancto Thomas. He claims covert cases and assumes the parallel assignment of items to different categories, but in a more affirmative manner. His paradigm is as streamlined as Gonçalez Holguin's was. Transparency is definitely given precedence over consistency.


This analysis has set out different basic assumptions on categories and different paradigms of Quechua case items, comparing the first five grammars on Quechua. Assumptions and inventories do show correspondences. Grammar is never purely descriptive, even if teaching is its dominant purpose. There is no way to escape theory in colonial grammar, as there is none elsewhere. The scrutiny of the theoretical prerequisites may now be reduced to two questions: (i) is the assignment of one item to different categories assumed as legitimate by the authors, and (ii) are covert categories assumed as legitimate? The answers to the first question run 'yes – yes – no – no – yes'. The answers to the second question run 'yes – no – yes – no – yes'.

The answers do not necessarily correspond to each other. No historical development is immediately to be deduced from them either. Of course, a 'yes' is not necessarily identical to a preceding 'yes', nor a 'no' to a preceding 'no'. Evolution remains to be seen in the details, but it will hardly be straightforward. A tendency to make Quechua conform to Latin is already present in Sancto Thomas, but reaches its peak with Gonçalez Holguin. The opposite tendency, to modify the Latin inventory and adapt it to what is found in Quechua, is already present in Sancto Thomas as well, but reaches its peak with Huerta. None of the conflicting tendencies imposes itself definitely and none of them reaches its peak with Torres Rubio, the last author considered.

Nevertheless, one of the authors should be singled out, and this is not Gonçalez Holguin, but Huerta. Gonçalez Holguin, for all his perspicacy and rhetorical brilliance, and even for his advances towards the recognition of primary and secondary adpositions, blurs the definitions of case and declension, sets category boundaries stricter than the language has, and defends an ablative without evidence. Huerta, in turn, sees the inadequacy of the inherited case ending – preposition dichotomy and outlines an intermediate category instead, adapted to Quechua morphotactics. He does this using traditional labels, in an admirable ruse. The corresponding track of the migration path of case items, formerly unknown, is recognized anyway.

PhiN 33/2005: 105

Of course, a lot of work remains to be done, from what Huerta suggests. But, it seems to me, a straight path leads to the Quechua case paradigm as sustained by modern linguistics – given at the outset, in accordance with Cerrón-Palomino – starting from Huerta's table alone. All of the authors, though, contribute to linguistic knowledge in different ways, whether by paying attention to contrasts in weight and prosody, by inquiring into the disjunction of content and form, or by outlining categories only apparently familiar. The history of linguistics, it seems, proceeds adventurously from one less-than-tenable statement to the next.


Historical texts

Aguilar, Juan (1690): Arte de la lengua Quichua general de los indios del Peru, manuscript. Tucumán: Universidad Nacional de Tucumán 1939. Radamés A. Altieri, ed.

AVQE (1586): Arte y vocabulario en la lengua general del Peru llamada Quichua, y en la lengua Española. El mas copioso y elegante que hasta agora se ha impresso. Los Reyes: por Antonio Ricardo 1586. Sevilla: en casa de Clemente Hidalgo 1603. Los Reyes: por Francisco del Canto 1614.65

Bertonio, Ludouico, SJ (1603): Arte y grammatica muy copiosa de la lengua Aymara. Roma: por Luis Zannetti.

Gonçalez Holguin, Diego, SJ (1607): Gramatica y Arte nueua de la lengua general de todo el Peru, llamada lengua Qqichua, o lengua del Inca. Añadida y cumplida en todo lo que le faltaua de tiempos, y de la Grammatica, y recogido en forma de Arte lo mas necessario en los dos primeros libros. Con mas otros dos libros postreros de addiciones al Arte para mas perficionarla, el vno para alcançar la copia de vocablos, y el otro para la elegancia y ornato. Ciudad de los Reyes del Peru: por Francisco del Canto impressor.

Huerta, Alonso de (1616): Arte de la lengua Quechua general de los yndios de este Reyno del Piru. Los Reyes: impresso por Francisco del Canto.

Nebrija, Antonio de (1481): Introductiones latinae. Salmanticae. Introduciones latinas, contrapuesto el romance al latin. Zamora: por Antonio de Centenera 1494. Münster: Nodus 1996. Miguel Ángel Esparza/Vicente Calvo, eds.

Nebrija, Antonio de (1492): Gramatica de la lengua castellana. Salamanca. Madrid: Centro de Estudios 'Ramón Areces' 1989. Antonio Quilis, ed.

Roxo Mexia y Ocon, Iuan (1648): Arte de la lengua general de los indios del Peru. Por Iorge Lopez de Herrera, en la calle de la Carcel de Corte.66

Sancho de Melgar, Estevan (1691): Arte de la lengua general del Ynga, llamada Qquechua. Lima: por Diego de Lyra.

PhiN 33/2005: 106

Sancto Thomas, Domingo de, OP (1560): Grammatica o Arte de la lengua general de los indios de los Reynos del Peru. Valladolid: por Francisco Fernandez de Cordoua.

Torres Rubio, Diego de, SJ (1616): Arte de la lengua Aymara, Lima: por Francisco del Canto.

Torres Rubio, Diego de, SJ (1619): Arte de la lengua Quichua. Lima: por Francisco Lasso.67

Modern studies

Agud, Ana (1980): Historia y teoría de los casos. Madrid: Gredos.

Andersen, Henning (2001a): "Actualization and the (uni)directionality of change", in: Andersen ed. 2001b: 225–248.

Andersen, Henning, ed. (2001b): Actualization. Change in progress. Amsterdam/Philadelphia: Benjamins.

Aristar, Anthony R. (1997): "Marking and hierarchy types and the grammaticalization of case-markers", in: Studies in Language 21.2: 313–368.

Booij, Gert/Lehmann, Christian/Mugdan, Joachim, eds. (2000): Morphology. vol. 1. Berlin/New York: Mouton de Gruyter (= HSK 17.1).

Bright, William, ed. (1992): International Encyclopedia of Linguistics. 4 vols. New York/Oxford: OUP.

Butt, Miriam/Geuder, Wilhelm, eds. (1998): The projection of arguments. Lexical and compositional factors. Stanford: CSLI.

Bybee, Joan L./Haiman, John/Thompson, Sandra A. eds. (1997): Essays on language function and language type. Dedicated to T. Givón. Amsterdam/Philadelphia: Benjamins.

Bybee, Joan L./Hopper, Paul J. eds. (2001): Frequency and the emergence of linguistic structure. Amsterdam/Philadelphia: Benjamins.

Calvo Pérez, Julio (1994): "Las gramáticas de Nebrija y las primeras gramáticas del quechua", in: Escavy Zamora/Hernández Terrés/Roldán Pérez eds. 1994, vol. 2: 63–80.

Calvo Pérez, Julio (2000): "Las gramáticas del siglo de oro quechua: originalidad y diversidad",in: Zwartjes ed. 2000: 125–201.

Campbell, Lyle (1997): American Indian languages. The historical linguistics of Native America. New York/Oxford: OUP.

PhiN 33/2005: 107

Cerrón-Palomino, Rodolfo (1987): Lingüística quechua. Cuzco: CERA Bartolomé de las Casas.

Cerrón-Palomino, Rodolfo (1994): Quechumara. Estructuras paralelas de las lenguas quechua y aimara. La Paz: CIPCA.

Cerrón-Palomino, Rodolfo (1997): "La primera codificación del aimara", in: Zimmermann ed. 1997: 195–257.

Chafe, Wallace (2000): "Florescence as a force in grammaticalization", in: Gildea ed. 2000: 39–64.

Colombat, Bernard (1999): La grammaire latine en France à la Renaissance et à l'Âge classique. Théorie et pédagogie. Grenoble: Université Stendhal.

Comrie, Bernard (1991): "Form and function in identifying cases", in: Plank ed. 1991: 41–55.

Corrigan, Roberta/Eckman, Fred/Noonan, Michael, eds. (1989): Linguistic categorization. Amsterdam/Philadelphia: Benjamins.

Croft, William (1998): "Event structure in argument linking", in: Butt/Geuder eds. 1998: 21–63.

De Clercq, Jan/Desmet, Piet, eds. (1994): Florilegium Historiographiae Linguisticae. Études d'historiographie de la linguistique et de grammaire comparée àla mémoire de Maurice Leroy. Louvain: Peeters.

DeLancey, Scott (1997): "Grammaticalization and the gradience of categories. Relator nouns and postpositions in Tibetan and Burmese", in: Bybee/Haiman/Thompson eds. 1997: 51–69.

Escavy Zamora, Ricardo/Hernández Terrés, José Miguel/Roldán Pérez, Antonio, eds. (1994): Actas del Congreso Internacional de Historiografía Lingüística. Nebrija V. Centenario. 3 vols. Murcia: Universidad de Murcia.

Gaudin, Thierry/Hatchuel, Armand, eds. (2002): Les nouvelles raisons du savoir. Vers une prospective de la connaissance. Paris: Editions de l'Aube.

Gil, David (2000): "Syntactic categories, cross-linguistic variation and universal grammar", in: Vogel/Comrie eds. 2000: 173–216.

Gildea, Spike L.O. ed. (2000): Reconstructing grammar. Comparative linguistics and grammaticalization. Amsterdam/Philadelphia: Benjamins.

Haiman, John (1994): "Ritualization and development of language", in: Pagliuca ed. 1994: 2–28.

PhiN 33/2005: 108

Haspelmath, Martin (1999): "Why is grammaticalization irreversible?", in: Linguistics 37.6: 1043–1068.

Himmelmann, Nikolaus P. (1998): "Regularity in irregularity. Article use in adpositional phrases", in: Linguistic Typology 2.3: 315–353.

Heine, Bernd/Claudi, Ulrike/Hünnemeyer, Friederike (1991): Grammaticalization. A conceptual framework. Chicago: Chicago UP.

Hopper, Paul J. (1998): "Emergent grammar", in: Tomasello ed. 1998: 155–175.

Hopper, Paul J./Traugott, Elizabeth C. (1993): Grammaticalization. Cambridge: CUP. 22003.

Hüllen, Werner, ed. (1990): Understanding the historiography of linguistics. Problems and projects. Münster: Nodus.

Jankowsky, Kurt R. ed. (1995): History of Linguistics 1993. Papers from ICHoLS VI, Washington DC, 9–14 August 1993. Amsterdam/Philadelphia: Benjamins.

Law, Vivien A. (1990): "The history of morphology. Expression of a change in consciousness", in: Hüllen ed. 1990: 61–74.

Law, Vivien A. (2000): "Morphology. The Middle Ages", in: Booij/Lehmann/Mugdan eds. 2000: 76–90.

Lehmann, Christian (1985a): "Latin case relations in typological perspective", in: Touratier ed. 1985: 81–104.

Lehmann, Christian (1985b): "Ergative and active traits in Latin", in: Plank ed. 1985: 244–255.

Lehmann, Christian (1995): Thoughts on grammaticalization, München: Lincom Europa. Köln: Institut für Sprachwissenschaft der Universität zu Köln 1982.

Lerner, Isaias (1997): "La colonización española y las lenguas indígenas de América", in: Colonial Latin American Review 6: 7–15.

Lisi, Francesco Leonardo (1990): El Tercer Concilio Limense y la aculturación de los indígenas sudamericanos. Estudio crítico con edición, traducción y comentario de las actas del concilio provincial celebrado en Lima entre 1582 y 1583. Salamanca: Universidad de Salamanca.

Mannheim, Bruce (1991): The language of the Inka since the European invasion. Austin: Texas UP.

Mithun, Marianne (1989): "The subtle significance of the locus of morphologization", in: IJAL 55.3: 265–282.

PhiN 33/2005: 109

Nichols, Johanna (1986): "Head-marking and dependent-marking grammar", in: Language 62.1: 56–119.

O'Neill, Charles E. ed. (2001): Diccionario histórico de la Compañía de Jesús, biográfico-temático. 4 vols. Romae: Institutum Historicum Societatis Iesu.

Pagliuca, William, ed. (1994): Perspectives on grammaticalization. Amsterdam/Philadelphia: Benjamins.

Percival, W. Keith (1995): "The genealogy of general linguistics", in: Jankowsky ed. 1995: 47–54.

Plank, Frans, ed. (1985): Relational Typology. Berlin/New York: Mouton de Gruyter.

Plank, Frans, ed. (1991): From case to adposition. Konstanz: Universität Konstanz.

Ramat, Anna Giacalone/Hopper, Paul J. (eds.) (1998): The limits of grammaticalization, Amsterdam/Philadelphia: Benjamins.

Robins, Robert H. (1990): "Constraints and intentions in the organization of the history of linguistics", in: Hüllen ed. 1990: 11–19.

Rorty, Richard (1984): "The historiography of philosophy: four genres", in: Rorty/Schneewind/Skinner eds. 1984b: 49–75.

Rorty, Richard/Schneewind, John B./Skinner, Quentin, eds. (1984): Philosophy in history. Cambridge: CUP.

Schlieben-Lange, Brigitte (2000): 'Idéologie'. Zur Rolle von Kategorisierungen im Wissenschaftsprozeß. Heidelberg: Winter.

Schmidt-Riese, Roland (2003): "Catégorisation cognitive et grammaticale au XVIIe siècle. Propriétés des choses et mots de propiété en kariri", in: Romanistisches Jahrbuch 53 (2002): 383–407.

Schmidt-Riese, Roland (2005): "Arts grammaticaux du Nouveau Monde: recherche, progrès, transparence", in: Vérin/Dubourg-Glatigny eds. 2005 appear.

Serbat, Guy (1981): Cas et fonctions. Études des principales doctrines casuelles du Moyen Âge à nos jours. Paris: PUF.

Serbat, Guy (1994): "Le septimus casus en latin: un cas-fantôme?", in: De Clercq/Desmet eds. 1994: 159–172.

Taylor, Gerald (1994): Estudios de dialectología quechua. Chachapoyas, Ferreñafe, Yauyos. Lima: Universidad Nacional de Educación 'Enrique Guzmán y Valle'.

Tomasello, Michael, ed. (1998): The new psychology of language. Cognitive and functional approaches to language structure. Mahwah, NJ/London: Lawrence Erlbaum.

PhiN 33/2005: 110

Touratier, Christian, ed. (1985): Syntaxe et latin. Aix-en-Provence: Université de Provence.

Vérin, Hélène (2002): "Généalogie de la 'réduction en art'. Aux sources de la rationalité moderne", in: Gaudin/Hatchuel eds. 2002: 29–41.

Vérin, Hélène/Dubourg-Glatigny, Pascal, eds. (2005): Réduire en art. La techno-logie du XVIe au XVIIIe siècle, Paris: Muséum National d'Histoire Naturelle/CNRS, to appear.

Vogel, Petra M./Comrie, Bernard, eds. (2000): Approaches to the typology of word classes. Berlin/New York: Mouton de Gruyter.

Wischer, Ilse/Diewald, Gabriele, eds. (2002): New reflections on grammaticalization. Amsterdam/Philadelphia: Benjamins.

Zimmermann, Klaus, ed. (1997): La descripción de las lenguas amerindias en la época colonial. Frankfurt a.M.: Vervuert.

Zwartjes, Otto, ed. (2000): Las gramáticas misioneras de tradición hispánica (siglos XVI – XVII). Amsterdam/Atlanta: Rodopi.

Zwartjes, Otto/Hovdhaugen, Even, eds. (2004): Missionary linguistics – lingüística misionera. Selected papers from the First International Conference on Missionary linguistics, Oslo 13–16 march 2003. Amsterdam/Philadelphia: Benjamins.


* This investigation was carried out within the project "Wissenstraditionen in der Christianisierung Amerikas" (B5), part of the interdisciplinary research area "Pluralisierung und Autorität in der Frühen Neuzeit" (SFB 573), run at the University of Munich by Deutsche Forschungsgemeinschaft (DFG). Thanks to Anna Larsson for revising the English version of the paper. All remaining shortcomings are mine.

1 There are three other Quechua grammars written before 1700, not included in this analysis: Roxo Mexia y Ocon 1648, Aguilar 1690 and Sancho de Melgar 1691.

2 For discourse conditions on colonial grammars see Zimmermann ed. 1997, Zwartjes ed. 2000, Zwartjes/Hovdhaugen eds. 2004 and bibliographies, also Schmidt-Riese 2005. For a wider perspective on the Renaissance project of arts see Vérin 2002.

PhiN 33/2005: 111

3 Another reason for preferring 'colonial grammar' is that this term spans a certain historical period. This limitation seems preferable to me, as I consider colonial grammar a historical, not an anthropological matter. Further, 'missionary grammar' fits all too nicely into globalization rhetoric. It is not just descriptive.

4 See Corrigan/Eckman/Noonan eds. 1989, Schlieben-Lange 2000, Schmidt-Riese 2003, for a theory-oriented approach to linguistic categories. See also Gil 2000.

5 To complicate the matter, Renaissance grammar and linguistics are not just different offshoots of one and the same tradition at different moments. For linguistics itself springs from Renaissance grammar – being posterior to it – and perhaps springs from Indian grammar. Renaissance grammar, in turn, may spring from Hebrew grammar, in addition to Latin grammar. See Law (1990: 66) regarding the latter hypothesis, Robins (1990: 13) regarding the former. See also Percival 1995.

6 Also, there are books written, see Agud 1980, Serbat 1981. For a wider perspective of rare expertise see Colombat 1999.

7 Cited from Bright ed. 1992 (vol. 1: 217).

8 Nom – nominative, Gen – genitive, Dat – dative, All – allative, Acc – accusative, Voc – vokative, Abl – ablative, Erg – ergative, Inst – instrumental, Com= comitative, Loc= locative.

9 For a discussion on ergative traits in Latin see Lehmann 1985b.

10 See Aristar 1997 for the claim that case systems might not be invariant across nominal classes within a language.

11 Comrie (1991: 49) claims in fact a locative for Latin on the basis of what he calls distributional case. This concept is essentially the same as that of traditional grammar: "The tradition of describing Latin operates essentially on the principle that if any nominal evinces a formal case distinction, then this case distinction must be carried over to all nominals."

12 See Serbat 1994 for the details of different suggestions on a seventh case of Latin made in Antiquity. A prominent syntactic pattern apparently distinct from the rest is ablatiuus absolutus.

13 See, for instance, the modern re-edition of the second edition of the bilingual version of his Introductiones latinae [Nebrija 1996: 8–16, 106].

14 For the legitimacy of calling past scientific achievements insufficient see Rorty 1984. It's an inevitable consequence of taking historical texts seriously.

15 Taken from Cerrón-Palomino (1987: 270, 1994: 89–95).

16 Ila – ilative 'to', Ben – benefactive, Purp – purposive 'good for', Mat – material 'out of', Limit – limitative 'until', Dist – distributive, Caus – causative 'for …'s sake', Comp – comparative 'just as/in the manner of', Int – interactive 'among [+ hum, Pl]'.

17 The variant /-kta/ is attached to stems ending in short vowel, but only in certain varieties. Cerrón-Palomino (1987: 133–134) claims that /-kta/ represents a certain development stage of Proto-Quechua. Varieties with /-kta/ prove to be archaic in further respects.

18 In some modern dialects, both /-kta/ and /-man/ may indicate the 'core dative' function with third arguments of verbs of giving, see Cerrón-Palomino (1994: 91). Colonial grammars associate both directionals to the accusative. In fact, /-kta/ is grammaticalized into an accusative to a considerable degree, while /-man/ seems to enter the very same channel, in some dialects.

PhiN 33/2005: 112

19 Cerrón-Palomino (1987: 135) suggests that /-q/ might be identical to a nominalizing suffix of the same shape. It is most evidently preceded by the genitive marker /-pa/. The purposive value requires, additionally, previous suffixation by /-na/. I leave open to question whether this is actually an inflectional process. For comparative and cross-dialectal evidence see Cerrón-Palomino (1994: 80, 92).

20 Cerrón-Palomino (1987: 207) argues in this way for accusative /-ta/ and instrumental /-wan/. Notably, the locative holds different items in different dialects. Quechua I has locative /-chaw/, see Cerrón-Palomino (1987: 136). Cross-linguistically, genitive and accusative are expected to be the most grammaticalized cases, see below.

21 See Cerrón-Palomino (1994: 94–95). The author further points to the fact that /rayku/ has been adopted into Aymara as /layku/. The liquid change presupposes an originally word-initial position for /r-/ in Quechua. In Aymara, /r-/ is unlicenced in that position and therefore turned into /l-/. There would have been no need for liquid changes, on the other hand, had /rayku/ been a suffix at the time of borrowing.

22 Secondary adpositions would also be used as non-adpositions, though not as heads of NPs, as relational nouns. In adpositional use, secondary adpositions would prefer genitive constructions, in addition to being morphologically and phonologically more complex than primary ones. Primary adpositions are used as adpositions only and prefer accusative constructions. For details see Lehmann (1985a: 92, 1995: 112), Himmelmann (1998: 320), see also DeLancey 1997, Aristar 1997. For basic work on grammaticalization see further Heine/Claudi/Hünnemeyer 1991, Hopper/Traugott 1993, for wider perspectives Haiman 1994, Hopper 1998, Ramat/Hopper eds. 1998, Chafe 2000, Bybee/Hopper eds. 2001, Wischer/Diewald eds. 2002, for nominal and verbal marking procedures Nichols 1986, Mithun 1989. For a basic notion of argument relations see Croft 1998.

23 For a recent discussion of the irreversibility hypothesis see Haspelmath 1999, for a somewhat different perspective Andersen 2001a.

24 OP – Ordo Praedicatorum, Dominican order. SJ – Societas Jesu, Jesuit order.

25 The text size indicated here is an estimate. It includes blanks and spaces. The actual number of signs used is certainly lower. Nevertheless, the numbers approach the actual sizes and most certainly represent size relations among the texts.

26 I give author names in the original spellings from title pages. In the case of Sancto Thomas it seems an advantage to me that this spelling is different from that of Thomas Aquinas in modern Spanish. Domingo de Sancto Thomas OP is not Aquinas.

27 AVQE is an anonymous grammar, indicated here by its title initials. Its attribution to the Jesuit order is undocumented, though convincingly argued for, see below.

28 Calvo Pérez (1994: 74, 2000: 144, 169) suggests that there might have been an earlier edition of this grammar, published in 1603 maybe in Rome, maybe in Seville, maybe elsewhere. This is not impossible, of course, but it is just a guess. Actually, the AVQE is re-edited in Seville in 1603, being overtly declared a Jesuit grammar by the IHS on the title page. The Society of Jesus would hardly both re-edit the AVQE and publish a different Quechua grammar in the same city the same year, with both grammars – the actual and the imagined one – of equal scope and size.The dedicace of the re-edition, in fact, is signed 'Diego de Torres'. This person is, in any case, the very same as the one who signs the approval of Bertonio's comprehensive Aymara grammar in December 1602 in Rome, 'Didacus de Torres'. Not necessarily Diego de Torres Rubio, though.

PhiN 33/2005: 113

29 For Quechua dialects see Cerrón-Palomino (1987: 324–349), Mannheim (1991: 11, 114), Campbell (1997: 188), for an alternative classification proposal, see Taylor 1994. Quechua is divided into several major branches with possibly no intercomprehension between them. Quechua I groups the dialects of the Andes east of Lima, the presumed area of origin of the language. Quechua II groups all the dialects of the areas of presumed later expansion. Quechua II A refers to the dialects of the central coast, probably integrated into the language domain first. Quechua II B and II C, in turn, presumably spread from the Quechua II A area, II B north and II C south of central Quechua I, both principally in the Andean highlands. Quechua II B spreads as far north as southern Colombia, II C as far south as northern Argentina, though not in a continuous manner. The densest bundle of isoglosses within the whole area is the one which separates Quechua I and II C, clearly perceived already in the colonial period. The varieties north and south of the isogloss are then termed Chinchaysuyo and Incaysuyo, respectively. Nevertheless, the language of the Inca administration is not based on Quechua II C, but on the coastal II A dialects, which connect northern and southern provinces of the Tahuantinsuyo. It is different therefore from Cuzco Quechua, the variant of the Inca capital, situated – at the time of Spanish conquest – on the language border with Aymara and definitely part of the II C area.

30 Quotes from the originals are systematically left untranslated. Translations would easily miss the point, so I prefer to gloss the quotes within my comments. Bold type is always mine, italics are when rendering abbreviations in the original. When indicating target materials, italics may be original, or may be mine. They are mine in quotes from Sancto Thomas. Nevertheless, I maintain both commas and parentheses which indicate quotes in the original.

31 See Nebrija (1492: fol. 34v) [1989: 189].

32 The indices of genitive and accusative, /-p(a)/ and /-(c)ta/, /-man/, the most grammaticalized cases, are not termed dictions by Sancto Thomas. The article or signal of nominatives, he says, is to have none – quite an adequate description of functional ø. "El articulo o señal del nominatiuo es no tener señal alguna." (1560: fol. 5v). The indices of dative, ablative and effective, on the contrary, are invariably introduced as articles or dictions. In fact, /-pac/, /-pi, -manta/ and /-guan/ have more solid profiles. Finally, the vocative index is referred to by diction only. Citation see below, note 34.

33 I reproduce the line structure from the original, Sancto Thomas (1560: fol. 7r–v). Opening parentheses in lines five and eight are obviously intended to remedy unwelcome line breaks. Further, there is a page break between lines five and six (from recto to verso).

34 Cum takes the shape of an agglutinative case marker with certain pronouns, as in tecum 'with thee'. It is then found postposed to its argument, the argument being marked as oblique only, not as ablative. With nouns, cum assigns ablative morphology.

35 This layout procedure corresponds to a former discussion: "El vocatiuo tiene por señal esta diction xe, o xay, que quiere dezir lo que en romance dezimos ola, o en latin o, y assi diremos, vocatiuo xe o xay yayanc." (1560: fol. 6v [1995: 27]). In terming the vocative items dictions, Sancto Thomas is not as explicit as Nebrija, who termed Spanish o an adverb (1492: fol. 34v) [1989: 189]. But nearly so.

36 See Nebrija (1492: fol. 34v) [1989: 189], once more. No covert cases are claimed in Grammatica castellana.

37 For /nac/, the author holds that it assigns no case at all. For /cama/, that it is added to the nominative, see citation below. This is roughly the same argument. Nominative appears as an equivalent of 'no (oblique) case', at least in the formal sense.

PhiN 33/2005: 114

38 Further, monosyllabic /nac/ is attributed a higher degree of adhesion to the noun. However, forming just one term with the noun had been used before as an argument for sustaining the attribution of target items to the case category, refuted in the present context.

39 Both the use of en and the doubling of target language and metalanguage (en la lengua general del Peru llamada Quichua, y en la lengua Española) are Franciscan or, in any case, mendicant title resources, unusual both in South America and within Jesuit grammar.

40 Ricciardi is the first printer of South America, see Lerner (1997: 11). The first book he prints in Lima is Doctrina Christiana, y Catecismo para instruccion de los Indios in 1584, followed by Confessionario para los Curas de Indios and Tercero Cathecismo y Exposicion de la Doctrina Christiana, por sermones, both in 1585. For a facsimile edition of all three volumes, see Doctrina christiana y catecismo para instruccion de indios, Madrid: CSIC 1985. The AVQE, then, is the fourth book printed in Lima.

41 The Quechua and Aymara versions of the catechism are contained, in addition to grammatical notes on both languages, in Doctrina Christiana 1584, see preceding note. For the Aymara standard proposed by the concile, see Cerrón-Palomino 1997. For Quechua dialect groupings note 28, above.

42 The proceedings of the Council address Quechua as lingua Cuzquensis, see Lisi (1990: 224).

43 The subtitle of the AVQE reads El mas copioso y elegante que hasta agora se ha impresso. This is clearly an affront to Sancto Thomas, the only Quechua grammar printed so far. The title page of the AVQE lacks any reference to the Jesuit order, as the book lacks ecclesiastic approval and printing licenses altogether. The license of the Seville re-edition of 1603, though, afforded by the Council of Castile, holds that the first edition was produced in the Jesuit office of Lima, "que se hizo y se imprimio en la Compañia de Iesus de Lima, con aprobacion del Concilio Prouincial celebrado en la dicha ciudad". The AVQE was reprinted a second time within the colonial period, in 1614, in Lima.

44 The Dominicans are present in the Andes from the very beginning of Spanish rule in 1534. The Jesuits, in turn, arrive only in 1568. They have been trained in Quechua before going, on the basis of the grammar of Sancto Thomas. Part of the Valladolid edition of 1560 remained in Spain. The AVQE author doubtless knew this book.

45 Medieval Latin grammar has at its disposal the notions of vox 'word' and terminatio 'ending', but none for what is left of the word when the ending is taken away, see Law (1990: 63). In the period under study, radix 'root' is just about to be imported from Hebrew grammar.

46 See AVQE (1586: fol. 2v–3r). I can't reproduce this here.

47 The items cited are further different from the ones adduced by Sancto Thomas: xe/xay – ya/yau.

48 "Pa, p siruen al genitiuo solamente." AVQE (1586: fol. 27bis v) [Foliation is erroneously repeated, should be 35v].

49 For Jesuit biographies see O'Neill 2001, Gonçalez Holguin pp. 1784–1785, Torres Rubio p. 3826.

50 See that /huan/, in the ablative, is separated by space. Only genitive, dative and accusative have segments added to the noun stem in the layout. The 'possessive vocative' /-y/ would cut across the borderlines, so it is left out. Gonçalez Holguin (1607: fol. 1v–2r).

51 For the difference between simple and complex adpositions as explored by the author, see below.

PhiN 33/2005: 115

52 Gonçalez Holguin supposes discrete categories, though, as for case assignment. A genitive preposition invaribly assigns genitive case, even if its argument lacks genitive and the relational head lacks possessive marking: "Ccaylla, o cispa. De tres maneras se componen todas estas preposiciones con el genitiuo y possessiuo, y de aqui queda dicho para todas. Yglesiap ccayllan, junto a la yglesia. Con genitiuo y possessiuo. Yglesia ccayllan. Sin genitiuo expresso, y con possessiuo. Yglesia ccaylla. Sin genitiuo ni possessiuo." (1607: fol. 134r). There would be no transition then from genitive to ablative, from secondary to primary adpositions.

53 For Quechua dialects see note 28, once more.

54 Huerta obtains his doctorate in theology. He later holds the Quechua chair at San Marcos university, in addition to the one at Lima cathedral.

55 In Hebrew grammar, in contrast, letter may refer to the consonantal onset of an affix of CV structure. As seen from this perspective, Huerta might refer by letter to further Quechua items, genitive /pa/, accusative /ta/ and locative /pi/. He does not go into details on this.

56 Huerta (1616: fol. 4r). Capitals in Chinchaysuyo are mine. The author gives an alternative translation for the second ablative in order to make clear the difference between genitive and ablative (proper), homonymous in Spanish, both rendered by de. In the flow text he distinguishes 'de with possession' (con possession) and 'de with removal' (con apartamiento).

57 The Latin equivalent to /rayku/ would be propter. It governs the accusative case. The insertion of /chau/ is consistent within a form-to-content perspective. Huerta remains close to the AVQE in this respect.

58 The author relies on the ambivalence of the concept as inherited from Antiquity. A clearcut boundary between prefixation and prepositional syntax is only drawn in the Middle Ages, see Law (2000: 84). Martinus de Dacia suggests around 1270 that inseparable prepositions might not be prepositions at all, but syllabic adjections. Thus, Huerta recovers for Quechua a secondary meaning from an earlier tradition, the one Sancto Thomas had rejected explicitly, in consonance with Modistae positions. See above citation, from Sancto Thomas (1560: fol. 50v–51r): "[...] pero nunca se hallan juntas, con verbos, ni con las otras partes de la oracion por composicion."

59 Compositions of the 'second manner' (segunda manera), in the author's terms, are in fact derivations: "La segunda manera de nombres compuestos se haze de vn nombre de suyo significatiuo, y vna particula que varia, o aumenta su significacion." (Huerta 1616: 6r–v).

60 In this disregard for semantics, Huerta proceeds as Gonçalez Holguin did, when he claimed that accusative /man/ governs the covert ablative.

61 Torres Rubio's grammar is reprinted in the colonial period twice, in 1701 and 1754, each time in Lima.

62 Torres Rubio (1619: fol. 1r). Italics of l. (uel) are mine. The author gives lexically filled paradigm tables later. He then chooses runa, the preferred example noun of the AVQE. Coincidence of line number and case category number is shared with Gonçalez Holguin.

63 The insertion of /-y/ in the vocative slot might have been favored by homonymy with an Aymara vocative particle /y/ which is not simultaneously possessive of first person singular, see Torres Rubio's Arte de la lengua Aymara (1616: fol. 1r). The choice remains questionable. The author admits in his comment that in Quechua /-y/ is by no means obligatory.

PhiN 33/2005: 116

64 In the accusative section, it is presented as an equivalent of Latin extra, foris, foras. See Torres Rubio 1619 [1754: 40r–42v. Foliation of 41 and 42 is interchanged].

65 The main title is slightly changed in the 1603 edition: Grammatica y vocabulario en la lengua general del Peru llamada Quichua, y en la lengua Española. El mas copioso y elegante que hasta agora se ha impresso. In 1614, the original title is entirely restored.

66 The title page lacks an indication of the publication city, Lima.

67 There is a re-edition from 1701, Lima: por Ioseph de Contreras y Alvarado, revised by Juan de Figueredo SJ, and a reprint of this re-edition, Lima: en la Imprenta de la Plazuela de San Christoval 1754. Both are often cited by the reviser's name. Titles differ from the original.