1) they may only occur in roots, never in suffixes;
In the other Quechua language ch functions rather as part of the sibilant and/or fricative systems, not as part of the occlusive system (Torero1975), and the occlusive system is limited to p t k (q)
The Aymara system although with the same inventory, is nevertheless very different. None of the noted restrictions apply. Aspiration and glottalization occur in suffixes and roots alike and may occur on any occlusive in the word -- there in no theo-retical limit on the number, selection or order in any given word. Note the following Aymara forms from Suqa, Peru:
Thus, the data would lead to the conclusion that Jaqi is primarily responsible for glottalization and aspiration in Cuzco Quechua. Once such a feature is borrowed into a language, of course, these new phonological elements may extend beyond borrowed words -- in this case primarily into the realm of onomatopoeia.
An additional criticism which needs to be remarked on relative to all the various lists used to prove the common origin hypotheses, is that the items selected are not what are usually considered used the concept. Basic terms split quite clearly on family grounds. For example, the numbers 1 and 2 are not cognate, although higher numbers sometimes appear to be; also 'black' and 'white' show no similarity, although some of the other colors sometimes do -- both circumstances reflect quite clearly market interchange. The numbers especially tend to be pan-Andean and colors are much influenced by dyes, among other exchange patterns.
The specific form that a borrowing takes may permit us to establish the period in which it was borrowed: for example, the Aymara word iwisa 'sheep' was borrowed from Spanish while the Spanish language still had sh where today it has j. At times one hears uwija, which shows a reborrowing from a more modern period. Another example of the same type is from jaqaru, shupuna 'jacket' from when the Spanish still used jubones, cognate with modern French jupe 'skirt", and also before the consonant shift, that is, during approximately the first century after the conquest. the last example makes us appreciate another possibility -- that a borrowing can remain alive in the language that does the borrowing and die out in the language that does the lending. Another example of that is parlar which is no longer a Spanish word but which was borrowed into Aymara and continues in regular use in that language today as parlaña 'to talk'.
In the above argumentation, I have shown that the evidence produced to sup port the hypothesis of common origin for Quechua and Aymara is faulty and that adequate proof for the hypothesis is not available. I have also indicated that a far more likely explanation for the strong similarities, which in some respects do exist between the two languages, is the long and persistent linguistic and cultural contact.
I would like to propose a possible linguistic history of the Andes, as a speculation that would incorporate the linguistic evidence as it currently stands without in any way contradicting the archaeological, indeed incorporating it many respects. The following outline owes a great deal to the work of Alfredo Torero (1964, 1968, 1972, 1975).
The original language of the Tiwanaku builders was most likely Puquina, but this was not the language used for commercial expansion. The lingua franca of the Tiwanaku-Wari expansion period was that of proto-jaqi. Toponymics, among other evidence, support this hypothesis. When the founders of Tiwanaku crossed Lake Titicaca and settled in Cuzco. it may be presumed that they brought their home language with them -- certainly Cuzco itself was trilingual at the time of the conquest. When these people, by then known as the Incas, began their own expansion, they reserved the Puquina language for the royal family (i. e. the original conquering group) and used the readily available lingua franca, Jaqi, for expansionist purposes, until they met up with the powerful Pachacamac expansion. One might recall the great honors paid the Lord of Pachacamac even at Cajamarca. The people of Pachamac were Quechua speaking, of the Chinchay variety, and dominated the coastal area through domination of the sea. They had apparently already expanded as far as Ecuador as much as a half millennium earlier (Carpenter 1982). Thus, apparently, it seemed to be to the advanta-ge of the Incas to switch the lingua franca. (Also, Wayna Capac was in love with a woman from Pachacamac, be it noted.) Politically, such a switch was possible to legislate -- the court had no great personal loyalty to any conquest language per se. It might be remembered that the Spanish repeated, or carried on, the same pattern in expanding the use of Quechua themselves, to which they certainly felt no loyalty. Thus, only some 100 years before the European conquest, the Cuzco administrator adopted a new language. This reconstruction of events would clearly explain the enormous similarities between Aymara and Cuzco Quechua -- the entire Inca court would for a while be trilingual and all administrators drawn from conquered peoples minimally bilingual -- a situation designed for ready language interference, as well as for the convergence that is common when languages even of different families like long in interactive situations, as has also happened, e.g. in India (Emeneau 1964).
The double expansion of the Inca and the Chinchay Quechua left the Jaqi languages isolated and fragmented, particularity those closest to the Chinchay, in what is today the Departmento of Lima. But the remnants of the intensive and extensive contacts are still evident in the multiple borrowings.
From Jaqi to Quechua I propose two waves of massive borrowing: 1) during the commercial and cultural dominance of Wari, from proto-Jaqi specifically, during a period approximately 1500 to 1000 years ago; 2) during the first years of Inca expansion, while Jaqi was the official language; these borrowings would then have been from the language that is today Aymara, already separated from the sister languages closer to the coast. These borrowings would have been considerably later, for example, 400 to 700 years ago.
The major wave of borrowing from Quechua into Jaqi would have come from the last years of the empire, from the extensive use of Quechua during the virreinato, and even continuing through to today, sometimes via Spanish.
Reconstruction of the proto-Jaqi languages has led us postulate as occlusive system no less complex than that of Jaqaru today, although doubtless differing in phonetic details.  The largest difference between the system of Jaqaru/Kawki and that of Aymara is that the latter lacks the occlusive series tx tz cx. The modern reflexes of these are t ch t respectively. The result is that the modern t of Aymara comes from three sources, *tx, *cx, or *t. For example,
Therefore words with t or ch only in the modern languages do not directly give us evidence of whether they originated in the Jaqi or the Quechua family. However, when glottalization or aspiration is present, it is clearly strong evidence that these terms originated in the Jaqi languages, and that they arrived into Quechua from Jaqi. The actual shape of the words in Quechua today can indicate the approximate period of the borrowing. So, for example, the commercial expansion of Wari is reflected in the borrowings of numerals. 
Quechua has ch where proto-Jaqi had *cx, which is just what would be expected of those which is just what would be expected of those which had no cx as part of their occlusive system and thus did not hear the distinction. Even today, cx in modern Jaqaru words is perceived by Quechua speakers as ch. Other words of this early period show the same adaptation, for example:
In the later period -- and to a lesser extend than the first wave, during the direct impingement of Aymara on the early Incas -- there are direct borrowings from Aymara, when Aymara had already experienced the change from *cx to t. Thus, upon borrowing a form, Quechua borrowed if with t, and Quechua has t in those items today, for example:
To compete the picture during the modern period we have the third wave, with Southern Quechua returning to aymara words from the early period, with the *ch (re)borrowing resulting in ch, without , of course, any consequence being felt by the northern languages, for example:
Thus from the history of one sound we can glimpse the history of a region. The process can be seen also with *tx, where Quechua took ch in the first wave, but Aymara continued its own way in the development of t ; later borrowings from Aymara went into Quechua with t. An example of the first is:
Some words, of course, have their own individually unique history. let us look at:
Given the evidence of Kawki, we would be led to postulate a Jaqi origin for this word which is so identified with the inca empire. The t in Jaqaru does not got regu-lar sound change; let us look at the socio-cultural situations. Kawki is the most consonantly conservative of the Jaqi languages. In contrast, of two, Jaqaru is the most innovative. Furthermore, Tupe, where Jaqaru is spoken, was for a long period, lasting up to about 25 years ago, the cultural center of the area. Early in this century Tupe had a most distinguished educational center and peoples from all around, including Quechua and Spanish speakers as well as Jaqi speakers, went to a school in Tupe on a boarding basis. Inti is a common word in Peruvian school-books. It is, actually, used more in the school-room than in common conversation, which is more concerned with sunlight and warmth (nup'i in Jaqaru) than the star itself. Thus, inti in Jaqaru appears to be a reborrowing from Quechua through Spanish back into a Jaqi language. the same did not occur in Kawki because there was no schooling while the language was still the dominant language for the children.
These examples show the impact of socio-economic and historical factors on languages as well as the dangers of proposing genetic relationships between somewhat similar languages without giving importance to either the grammatical structure or the cultural situation. As is now clear, after two millennia of intimate contact, massive borrowing was inevitable. Making lists of correspondences in these cases is not a viable method of establishing common language origin without a concomitant recourse to the grammar.
Both language families are suffixing languages however, since suffixation is the most common morphological process in language, the above constitutes a typological statement, not a comparative one. There is a major difference in the way suffixation is handled in the two languages. Quechua suffixes are loosely tacked on, easily peeled off, in the manner that has been called 'agglutinating'. Morphophonemic rules are almost non-existent -- for most of the Quechua languages one rule only suffices. In the other words, morphological structure is transparent.
The jaqi languages, on the other hand, are of the type that has been called 'inflective'; the suffixes are complex, there is a great deal of complex morphophonemic modi-fication and suffixes are certainly not easily separable from each other. As one example the form mamshqa 'with your mother' from Jaqaru mama 'mother' plus -sa 'fourth person possessive' plus -wshqa 'subject coordinate' requires a number of complex morphophonemic rules involving both morphological and phonological conditioning to account for the surface form. Quechua has nothing like this type of morphophonemic anywhere.
The Jaqi languages are based on a system of four person, without number mark, where the distinctive features are presence or absence of first/second person. The proto-forms as currently reconstructed are:
This system is reflected throughout the grammar, noun and verb systems alike (Hardman 1975a, b, c).
Quechua, on the other hand, works with a system of three persons with (usually) number mark. Cuzco Quechua (but not all Quechua languages) has two plural suffixes which allow the language to mark the distinction between inclusive and exclusive.
In Ecuador and some other places this contrast is nonexistent. this is evidence of convergence -- when a distinction so important to one's a neighbors that is cannot be ignored is finally incorporated into one's own system (Hardman 1942;1978a,b; 1983a,b). Thus Quechua speakers came to be able to translate what they understand to be the distinction between Jaqi first and fourth persons, and, in Cuzco Quechua at least, it now forms part of that system.
The verbal person systems in each language family are congruent with the nominal person systems. In the case of Jaqi the basic four person system is develops into a verbal person system such that each suffix which marks person in interactive -- that is, subject and object are merged into the single inflective suffix, giving a paradigm of ten inflected persons in Jaqaru and Kawki and nine in Aymara. In the case of Quechua, in contrast -- in Cuzco at least -- the system in one of three persons and two numbers, allowing for the two-way contrast in first person plural for a total of seven persons, no objects included. Other varieties of Quechua have fewer persons, in some cases with little or no number (Carpenter 1982). In Quechua there are a few suffixes which may refer to objects and which may be included within the verb, but they are primarily extensions of directional suffixes as occupy a rather different position within the overall structure.
At the syntactic level, an important point of distinction in that within the Jaqi languages the sentences is defined through the use of a particular set of suffixes which we call sentence suffixes. There are some signs of this type of system within Cuzco Quechua, but without the obligatoriness and pervasiveness characteristic of the Jaqi system, nor does this use constitute the definition of the sentence within Quechua.
In summary, taking careful consideration of a body of data, including grammatical and phonological structure as well as correspondence lists together with socio-cultural history and circumstances, the only viable conclusion is that there are, indeed, at least two great language families extant in the Andes: Jaqi and Quechua. the apparent similarities between the languages of these two families are best explained by mutual borrowings and influences, which are easily accounted for by the type and extent of contract that has occurred through the centuries. It is therefor impossible to maintain any vestige of the notion of a common origin with differentiation in the Andes, and even more absurd to hold that Cuzco Quechua be considered a relatively conservative language. Rather, Aymara and Quechua can be for us an example of the innovations and adaptations that can and do occur when languages and cultures meet.
NOTES.- Also see Mannheim's "contact and Quechua-External Genetic Relationships" in this volume.
.- The Aymara language is a member of the Jaqi family of languages: Aymara, spoken today by upwards of three million people including one third of the population of boli-via, a large portion of the population of southern Peru, and in northern Chile and Argen-tina; Jaqaru, spoken by approximately five thousand in Tupe, Yauyos, and in immigrant communities in the cities of Huancayo, Chincha, Cañete, Lima, Peru; and Kawki, a dying language with a few speakers left in Cachuy, Peru.
.- Curiously enough, a recent dissertation (Davidson 1977) manages to disprove the Longacre/Orr hypothesis using, by claim of the author, only Cuzco Quechua and Aymara. Davisdon's conclusion corroborates mine, but is not genuinely independent, although the author believes it to be so, and his work was unknown to me the time he was doing it. The grammars on which he based his work were: 1) the Aymara grammar done at the University of Florida (Hardman, Yapita, & Vasquez 1975); 2) my Jaqaru grammar (Hardman 1966a [1983a]); 3) the Cuzco Quechua grammar on which I was working when I first came to the conclusion that Jaqaru and Cuzco Quechua were not related (Sole and Cusihuaman 1967).
.- All examples from the Andean languages are presented in the practical (phonemic) alphabets. The Aymara alphabet was developed by a linguist native speaker of Aymara (Yapita 1981). For further details on this alphabet see the two articles by Briggs in this volume.
ch, alveopalatal fricative, functions structurally as a stop.
The k series is velar; the q series is post-velar. Aspiration
is represented by "; glottalization by
'. j represents
a pharyngeal fricative; x a post-velar fricative.
.- For Jaqaru and Kawki the Aymara alphabet is compatible, except that x does not function as a separate letter and there are no long vowels (Hardman 1983b). The following characters are added for consonants not present in Aymara:
.- If we take into account also the Jaqaru/Kawki data, the number of items with glottalization/aspiration with no similar form in Jaqi is reduced - apparently Aymara some of the forms after "loaning" them to Quechua.
.- Part of the reconstruction work has been reported in Hardman (1975b). Work continues on the reconstruction of proto-Jaqi; with primacy given to the reconstruction of grammatical paradigms. Lists of general vocabulary correspondence have not yet been published.
.- The most common morphophonemic variation in the Jaqi languages is vowel dropping; in any syntactic position and any complex number, '100' would occur without the final vowel. Quechua permits final consonants, and therefore would be expected to borrow the word without the final vowel given that they would most certainly rarely if ever have heard the word uttered with the final vowel.