say so of Belmar and not without expressing their own strong doubt. Later came Mechling, who did the best he could with manuscript vocabularies placed at his disposal by Peñafel, and broke up the family into several groups. More recently Lehman proposed other groupings still. But all this was avowedly not much more than mere guessing. As Mechling says, the material was not only much too scanty, but it was obtained generally by persons without any linguistic training. Anyone who has seen some of the grammars published by the early friars knows that they are pretty nearly worthless from every point of view. Copies are very rare now. The late Mr. Gracida, who had devoted a lifetime of labor to the antiquities of his native state of Oaxaca, possessed a splendid library which he placed at my disposal while I was in Mexico. I derived not a little amusement from the fantastic grammars done on a Greco-Latin model by the good priests. I will admit that a good many Zapotecan sounds are rather puzzling to the Spanish ear and present quite a problem in phonetic transcription. But it had never occurred to me that the easiest way out of the difficulty was to write something entirely different. Mexican scholars of a later date like Belmar, Leon, Pimentel, Peñafel, and others have done somewhat better work, and they command our respect for their zeal and interest at a time when philology was limited to the Indo-Germanic languages, but for the purposes of modern comparative inguistics their data are too inaccurate to repay the labor of critical examination. I tried often with my native informants to elucidate some of the sentences, examples or even wordlists given in their books, but usually we could not even guess what it was they had meant to say.

I was therefore fully prepared to find the amount of similarity between these languages rather scanty. I will set down a comparative list of a few words only. I have omitted the tone marks and reduced the phonetic transcription to its simplest expression. The omission of the tones accounts for the identity of certain words within the same tongue, as in Chocho for I and thou, two and five, etc.

A mere glance at the above table will show that the resemblances are not striking, at least not for those among us who demand something pretty clear to base semantic relationship upon. And I want to add that this is a very fair sample. I have not picked the words in this list with a view to showing either the differences or the similarities. I have simply taken a score of those words which are most apt to have a fundamental identity in all language. I say that the similarities are not striking; I do not say that they are entirely absent. There are clearly resemblances between Zapotec and Chatino, between Mixtec and Cuicatec, between Mazatec and Chocho. And in almost every case the Chinantec word may be compared to one or another of the other six. Finally by stretching the point one could base a relationship on the following argument: if the above three pairs are related, and there are sufficient cases where one or the other member of a pair is related to a member of another pair, this is fair presumptive evidence that they are all related. But this is getting pretty thin. The semasiology alone would never have led me to group these languages together. On the contrary, at this stage of my investigation I believed that they had nothing in common.

It came to me therefore as a surprise when I found one language after another conforming to the same type of morphology that I had already found in Zapotec. There was an undeniable air de famille among them all. The same importance of the pitch tone element. The same division into animal, man, and thing. The same semi-suffixing of pronouns involving tone changes. The same sparsity of structure. The same method of expressing temporal ideas by means of prefixes-prefixes almost iden

tical in form, although having passed over from one aspect to another. I will give here a table of these temporal prefixes:

I think this is striking enough by itself, but it was all the more striking after the entirely different systems of the other two languages of the same region, Chontal and Mixe.

Another point of morphological similarity is the differentiation of the first person plural into plural exclusive and plural inclusive. This is found in all the group except Chatino, Chocho and Chinanteco.

I began more and more to have the feeling that all these languages belonged together, all except perhaps Chinanteco, which seemed in some ways to stand apart. This tongue fascinated me from the first. All the early accounts of the friars spoke of it as a frightful language, made up of grunts through the nose, where all the words were the same, and the devil himself could not learn it. All the good Oaxaqueños who heard of my intention to study it laughed at me. Their imitations of Chinanteco speech sounded indeed like pig with a cold in the head. As a matter of fact I found Chinanteco to be an extremely simple language, once you understand its monosyllabic and tonal nature. It is a typical monosyllabic language of the isolating type, with the characteristic "empty words" of such languages, with an exuberance of tone, and with a strong tendency to idiomatic expression. Just one sentence will give a good idea of the language. It is taken from a tale where a horse admonished by his fellow traveler donkey, not to waste his strength foolishly, retorts with: "When you grow up to horse size, then you will run like me."

BRARY

In this sentence the word li is a good example of "empty-wotd." Its primary meaning of "to do, to act" is here entirely lost. In its combination by means of word-order with ni (another empty-word used to i̟dicate the future), and with nu (a semi-suffixed pronoun), it comes to have the sense of "to grow." In the second part of the sentence the same li could as well be translated by "then." This may perhaps appear somewhat vague and ill delimited, but I am sure that anyone who has studied Chinese will recognize a familiar face. It is small wonder though, that the padres thought all the words were the same! This is the typical isolating language habit, very difficult to describe in the usual grammatical terms of convention, but unmistakable after one has handled a few such languages. And this is precisely what gave such an air de famille to the other languages I was studying, only I did not know then what it was; I did not suspect that in Chinanteco I had in its pure undiluted state the morphology through which all the other tongues of the group had recently passed.

But I am anticipating. I did not make that discovery until I went south on a field trip to collect all the dialectical variations of Zapotec that I could find. I was amazed as I penetrated further and further into the mountains of the southern range to find the suffixes more and more detachable. The semasiology was not one whit altered, but one after another of the fused endings appeared at first as agglutinated suffixes, then as pure morfemas of the empty-word type, and finally as semantemas used now concretely, now relationally. I was now in the heart of the Miahuatlan district, and to my increasing amazement I found that the "Miahuateco" dialect was as strictly monosyllabic and isolating as Chinanteco. I could now go back and read again all my notes on the other dialects of Zapotec and understand everything that had puzzled me in the morphology: as forms but recently evolved from an isolating prototype, everything was clear.

It was then that I formulated in my own mind the hypothesis that all these languages formed a group having in common a primary understructure of monosyllabic morphology, whatever their diversity in semantics might be.

It is very important to make it clear that this hypothesis does not necessarily imply a common genetic origin for the languages thus grouped together. They may all have descended from parents belonging to different linguistic stocks, and then have become penetrated by the same morphology. That there exists such a "regional" factor in the borrowing of morphological structure has already been demonstrated in several

cases. The point here is that the levelling would in this case have proceeded so far that all these languages would form a real group, linguistically.

Whether my hypothesis is correct or not can only be judged from a full presentation of the facts. I have a very large amount of material which I hope to be able to present soon in texts, vocabularies, and comparative morphological studies of all these languages. The material is all ready and needs only to be incorporated in the work on Oaxaca which will be published by the Department of Anthropology of Mexico under the Direction of Manuel Gamio, and of which it is a part.