Etymological Dictionary of Proto-Nostratic Online

Reconstructing Proto-Nostratic Online; Comparative Phonology, Morphology, and Vocabulary

CHAPTER ONE

Introduction, History of Research, and Methodology

1.1. Introduction

Distant (or long-range) linguistic comparison seeks to investigate the possibility that certain languages or language families, not previously thought to be genetically related, at least not “closely” related, might indeed be part of still larger groupings, which may be called “macrofamilies”.

This book will focus on Indo-European. The purpose is to show that Indo-European is not genetically isolated but, rather, that it is distantly related to certain other language families of northern and central Eurasia, the Indian subcontinent, and the ancient Near East. Where appropriate, issues concerning the other language families with which Indo-European is most likely related will also be discussed.

1.2. History of Research

From the very earliest days of Indo-European comparative linguistics, there have been speculations about the possible genetic relationship of Indo-European to other language families. Though, in the course of study, many striking similarities were noted between Indo-European and certain other language phyla, notably Uralic and Afrasian (formerly called Hamito-Semitic, Semito-Hamitic, Afroasiatic, Erythraic, and Lisramic), truly convincing evidence of distant linguistic relationship was simply not brought forth. Indeed, much of the early work was not of high quality and did more to discredit the attempt to discover possible relatives of Indo-European than to help. Gradually, the intellectual climate, especially in the United States, became hostile to long-range comparison, and Indo-European remained an orphan with no known relatives.

In the first half of the nineteenth century, no less a figure than one of the founders of Indo-European comparative grammar, Franz Bopp, investigated possible relationship of Indo-European with Kartvelian (in 1846 and 1847) on the one hand and with Malayo-Polynesian (in 1840) on the other. In the mid-1860’s, Rudolf von Raumer (in 1863) and Graziadio Ascoli (in 1864) claimed that Indo-European and Semitic were related. At about the same time (in 1869), Vilhelm Thomsen proposed relationship between Indo-European and Finno-Ugrian. This proposal was later (in 1879) explored in depth by the Estonian Nicolai Anderson and (in 1900) by the British phonetician Henry Sweet. Unfortunately, Anderson’s work contained too many errors to be of lasting value. However, insightful and solid contributions were made concerning the possible relationship of Indo-European and Uralic during the twentieth century by the Swedish Uralicist Björn Collinder. Towards the end of the nineteenth century (1873), the Semiticist Friedrich Delitzsch investigated lexical parallels between Indo-European and Semitic. Then, at the beginning of the twentieth century, the Danish linguist Hermann Möller, in the course of several publications, attempted to show that Indo-European and Semitic might be related. Möller’s work was later continued by the French linguist Albert Cuny, whose last publications date from the mid-1940’s. Möller’s and Cuny’s efforts were generally not highly regarded by the scholarly community. One exception was Möller’s student Holger Pedersen, who not only coined the term “Nostratic” but who also expanded the definition to include Indo-European, Semitic, Samoyed and Finno-Ugrian, Turkish, Mongolian, Manchu, Yukaghir, and Eskimo. Though Pedersen never published a systematic account of his views, he did make the following insightful observations (1931:335—338):

The question of the relationship among the Indo-European and foreign families of languages came up in the first period of comparative linguistics. Relationship between Semitic and Indo-European was asserted by Rudolf von Raumer, beginning in 1863, and by Ascoli from 1864 on. But convincing proof could not be expected at that time. Resemblances in the morphology of the two families are extremely few, and proof by means of vocabulary and the laws of sounds was not then understood. Schleicher denied most positively any relationship between the two, pointing to the great dissimilarity in the forms of the roots: in Semitic the roots consist of three syllables of very simple and uniform structure, as in Arabic "atala (root form and preterite of the verb ‘to kill’), while in Indo-European the roots are monosyllabic and of widely varying — partly heavily compounded — form, as in Latin ī-re ‘to go,’ stā-re ‘to stand,’ lub-et ‘it pleases,’ vert-ō ‘I turn,’ ed-ō ‘I eat,’ and so on. At that time nobody could weaken this argument. And it might have been added, although Schleicher did not do so, that the phonetic systems of the two language families are extremely different, as may be seen from a single example: in Semitic there is an abundance of gutturals, whereas in Indo-European there is not one, not even the (to us) ordinary h. With this in view, one might feel tempted to assent to Schleicher’s exclamation: “What weight have the few similarities in roots in the two language families against these sharp contrasts?” And one might well be disposed to neglect “the few similarities” which one could not help observing.

Nothing was changed in the problem by the first step in a systematic examination of the vocabulary which Friedrich Delitzsch took in his Studien über indogermanisch-semitische Wurzelverwandtschaft (1873). But the development of Indo-European linguistics changed the problem greatly. The monosyllabic form of Indo-European roots turned out to be an entirely secondary phenomenon: in historical times the roots of the words for heaven, god, or heart may appear to be *diw- or *ʔerd-, but we have good reason to believe that in the period older than that of the Indo-European parent language these roots had forms like *däyäwä-, or *ʔärädä- …, and that the phonological system in this older period had quite a different appearance from that which we attribute to the Indo-European language.

With this background, there appeared in 1906 an extraordinarily important work by the Danish scholar Hermann Möller, Semitisch und Indogermanisch. This is a splendid attempt to discover the laws controlling the relationship between Indo-European and Semitic consonants — a successful attempt, although only the main lines of development are traced. Time alone will show how far we can advance by Möller’s method. Certain it is, however, that the comparison of the two families can never be carried out so completely and in such detail as the comparison within the fields of the individual languages of one family.

But Indo-European has been brought into connection with other families besides Semitic. Vilhelm Thomsen, as early as 1869, indicated the possibility of a relationship with Finno-Ugrian, but he did not pursue the subject very far. In 1879, the Estonian Nicolai Anderson published an extensive work on the subject, the value of which is considerably impaired by its many errors. Great interest was awakened when the English scholar Henry Sweet advocated the relationship somewhat passionately in a little popular book, The History of Language (1900). However, among the individual similarities which Sweet mentions, some are incorrect, and his space was too limited to permit of actual proof. Trustworthy studies of some length by K. B. Wiklund and H. Paasonen appeared in 1906 and 1908. After these works it seemed unnecessary to doubt the relationship further.

Moreover, the inflectional systems show much greater relationships than in the case of Semitic. The original ending of the accusative case in Finno-Ugrian was -m, which in Finnish has changed to -n. The same ending is Indo-European:

Finnish                               Cheremissian         Latin                               Greek

Nominative käsi  hand         kit                           vespera  evening            hespérā

Accusative käde-n              kið-əm                     vespera-m                     hespérā-n

The similarities in the personal endings of verbs are especially striking:

Finnish                                  Cheremissian     Greek                               Sanskrit

1st person sg. kuolen  I die    kole-m                é-phero-n  I carried          a-bhara-m

1st person pl. kuole-mme  we die                      e-phéromen  we carried

2nd person pl. kuole-tte  you die                     e-phére-te  you carried

Furthermore, there is an unmistakable similarity between the two families in a series of pronouns and in the negation ‘not’:

                                Finnish                              Latin     

                minä  I (Lappish mon)                        mē  me

                sinä  thou (s from t; Lapp. don)          tē  thee

                                                                        Sanskrit

                tä-mä  this                                         ta-

                jo-ka  who, which (relative)               ya-

                ku-ka  who? (interrogative)                ka-

                            Hungarian                            Old Norse

                              ne  not                                 ne  not

It is impossible to regard all this as the result of accident. It is noteworthy, however, that the similarities hitherto pointed out in the more concrete part of the vocabulary are very few, although some of them are as striking as Finish nimi ‘name,’ and Latin nōmen. Consideration of the problem whether sound-laws still unknown to us, or morphological developments not yet understood, have obliterated the originally more numerous points of similarity, or whether the vocabulary in one of the families was largely renewed after the period in common, we must postpone until later. But to deny relationship between the families would be overbold.

If we accept relationship, we are led yet further afield, not only to Samoyed, which cannot be separated from Finno-Ugrian, but throughout all of Northern Asia and across the Bering Strait, because similar, though fainter, resemblances like those here cited are found also in Turkish, Mongolian and Manchu, in Yukaghir, and even in Eskimo. If, on the other hand, we agree in the matter of relationship with Semitic, then we must also accept relationship with the far-flung Hamitic family, and perhaps with Basque. And squarely in the midst between our supposed Northern and Southern relatives stand the Caucasian languages, which we cannot ignore, and various extinct languages in Asia Minor and thereabout. It is not impossible that some of the non-Indo-European languages of antiquity in Asia Minor were once most closely related of all to the Indo-European family.

As a comprehensive designation for the families of languages which are related to Indo-European, we may employ the expression Nostratian Languages (from Latin nostrās ‘our countryman’). The boundaries for the Nostratian world of languages cannot yet be determined, but the area is enormous, and includes such widely divergent races that one becomes almost dizzy at the thought.

In 1969, Linus Brunner published a detailed comparison of the Indo-European and Semitic vocabularies, and this was followed in 1980 by a wider comparison of languages undertaken by Kalevi E. Koskinen. We should note also that, though the investigation of problems relating to distant linguistic comparison was generally ignored by the vast majority of mainstream linguists, the field was never completely dormant — a small but persistent group of scholars (Pentti Aalto, John Bengtson, Knut Bergsland, Václav Blažek, René Bonnerjea, Karl Bouda, Bojan Čop, Heinz Fähnrich, Joseph Greenberg, Panu Hakola, Carleton T. Hodge, Georgij A. Klimov, D. H. Koppelmann, Saul Levin, Karl H. Menges, Roy Andrew Miller, Shamil Nafiqoff, Mikolas Palmaitis, Stephen A. Tyler, Ants-Michael Uesson, C. C. Uhlenbeck, to name but a few of the many scholars working on long-range comparison) has continued to work, throughout the better part of the twentieth century and on into the twenty-first century, on binary (or, in rare cases, wider) comparisons of various languages that are currently considered to belong to the Nostratic macrofamily. For comprehensive bibliographies listing publications dealing with distant linguistic comparison, cf. Hegedűs 1992a, Landsberg 1986, Bomhard—Kerns 1994:715—864, and the list of references contained in this book.

Beginning in the mid-1960’s, the intellectual climate slowly began to turn around, and a growing number of linguists, especially in the former Soviet Union, have begun to turn attention toward investigating distant linguistic relationship. The revived interest was sparked by the work of Vladislav M. Illič-Svityč [Иллич-Свитыч] and Aaron B. Dolgopolsky [Долгополский], who first started working independently and, at a later date, through the efforts of their mutual friend Vladimir Dybo [Дыбо], cooperatively. Their work, though not without its own shortcomings, was the first successful demonstration that certain language phyla of northern and central Eurasia, the Indian subcontinent, and the ancient Near East might be genetically related. Following a proposal first made in 1903 by Holger Pedersen, they employed the name “Nostratic” to designate this grouping of languages. In particular, Illič-Svityč, in the course of several publications, culminating in his posthumous comparative Nostratic dictionary, which is still in the process of publication, included Indo-European, Kartvelian, Afrasian, Uralic, Dravidian, and Altaic in his version of the Nostratic macrofamily. From his very earliest writings, Dolgopolsky also included Chukchi-Kamchatkan and Eskimo-Aleut.

Before his tragic death in an automobile accident on 21 August 1966, Illič-Svityč had planned to prepare a comparative Nostratic dictionary listing over 600 Nostratic roots and tracing their development in detail in each of the daughter languages in which they were attested. He had published a preliminary report on his work in 1965 entitled (in English translation) “Materials for a Comparative Dictionary of the Nostratic Languages (Indo-European, Altaic, Uralic, Dravidian, Kartvelian, Hamito-Semitic)”. Working diligently, literally devoting all of his energy to the project, he had managed to prepare the entries for approximately 350 roots. After his death, Illič-Svityč’s work was prepared for publication by the dedicated efforts of Rimma Bulatova, Vladimir Dybo, and Aaron Dolgopolsky, with the result that the first volume of the dictionary appeared in 1971, containing 245 entries. A second, smaller volume appeared in 1976, listing entries 246 through 353 and ending with an index — this completed all of the material prepared by Illič-Svityč himself (by the time this volume appeared, Dolgopolsky was in the process of emigrating to Israel). Finally, the first fascicle of volume three appeared in 1984, containing entries 354 through 378, none of which was prepared by Illič-Svityč — it represents the collective efforts of a team of scholars.

In the meantime, Dolgopolsky has continued to make important contributions to Nostratic studies, especially in a 1984 paper on Nostratic pronouns, and currently has material to support the reconstruction of approximately 3,000 Nostratic roots. Unfortunately, only a small amount of this material has been published to date. Other Russian scholars have also done important research into problems affecting Nostratic — mention should be made of the work of Alexandra Y. Aikhenvald, N. D. Andrejev, M. S. Andronov, Vladimir Dybo, Eugene Helimskij, Vjačeslav V. Ivanov, G. Kornilov, Oleg Mudrak, Vitaly V. Shevoroshkin, Sergej A. Starostin, V. A. Terent'jev, Vladimir N. Toporov, and V. L. Tsymburskij, among others. Though not Russian (but clearly someone who belongs to the “Moscow School”), special recognition must be given to the Czech scholar Václav Bla²ek, who has published many important papers, most of which deal with the common Nostratic lexicon. Others who should be noted include Alexis Manaster Ramer and Irén Hegedűs — each has published a number of interesting papers on Nostratic.

Beginning with an article that appeared in Orbis in 1975, I published several studies, culminating in a 1984 book entitled Toward Proto-Nostratic: A New Approach to the Comparison of Proto-Indo-European and Proto-Afroasiatic, in which I tried to show that Indo-European and Semitic (later expanded to include all of Afrasian) might be distantly related. Reviews of this book as well as discussions with colleagues prompted me to expand the scope of my research to include other language families. This resulted in the publication in April 1994 of a joint monograph by myself and John C. Kerns entitled The Nostratic Macrofamily: A Study in Distant Linguistic Relationship. It was Kerns who prepared the chapter dealing with Nostratic morphology. This book supplies a great deal of lexical evidence from the Nostratic daughter languages to support the reconstruction of 601 Proto-Nostratic roots. In an article published in Orbis in 1995, I supplied material to support an additional 29 Proto-Nostratic roots, and another 21 etymologies were proposed in my 1996 book entitled Indo-European and the Nostratic Hypothesis. I have continued to gather lexical data, with the result that an additional two hundred Nostratic etymologies are included in this book. It should be noted that my views on Nostratic differ somewhat from those of Illič-Svityč (and others who follow his system) — the differences are discussed in §1.5 below.

The late Joseph Greenberg has prepared a two-volume work entitled Indo-European and its Closest Relatives: The Eurasiatic Language Family. The first volume, which was published at the beginning of 2000, deals with grammar, and the second, which was published at the beginning of 2002, deals with lexicon. Greenberg includes Indo-European, Uralic-Yukaghir, Altaic (Mongolian, Chuvash-Turkic, and Manchu-Tungus), Japanese-Korean (Korean, Ainu, and Japanese-Ryukyuan), Gilyak, Chukchi-Kamchatkan, and Eskimo-Aleut in his Eurasiatic language family. Unlike Illič-Svityč, Dolgopolsky, and myself, he does not include Kartvelian, Afrasian, nor Elamo-Dravidian — not because he believes that they are unrelated, but because he believes that these three language phyla are more distantly related to Indo-European than are the others, which, along with Indo-European, form a natural taxonomic subgrouping. My own opinion is close to that of Greenberg. As I see the situation, Nostratic includes Afrasian, Kartvelian, and Elamo-Dravidian as well as Eurasiatic; in other words, I view Nostratic as a higher-level taxonomic entity. Afrasian stands apart as an extremely ancient, independent branch — it was the first branch of Nostratic to separate from the rest of the Nostratic speech community. Younger are Kartvelian and Elamo-Dravidian. It is clear from an analysis of their vocabulary, pronominal stems, and morphological systems that Indo-European, Uralic-Yukaghir, Altaic, Gilyak, Chukchi-Kamchatkan, and Eskimo-Aleut are more closely related as a group than any one of them is to Afrasian, Kartvelian, and Elamo-Dravidian, and this is the reason that I follow Greenberg in setting up a distinct Eurasiatic subgroup within Nostratic. Finally, mention should be made of Sumerian, which I had investigated in previous works as a possible Nostratic daughter language. I now believe that Sumerian was not a Nostratic daughter language but that it is distantly related to Nostratic. It must be noted here that I have also changed my mind about the subgrouping of Kartvelian and Elamo-Dravidian. My present thinking is that Kartvelian is closer to Eurasiatic than what I indicated in my 1994 co-authored book and that the differences are due to the fact that Kartvelian became separated from Eurasiatic at a very early date. On the other hand, I now see Elamo-Dravidian as the second group (after Afrasian) to split from the rest of the Nostratic speech community. An attempt at subgrouping is shown in Chart 1 at the end of this chapter.

Interest in issues dealing with Nostratic has resulted in several conferences, the first of which was held in Moscow in 1972 to coincide with the publication of the first volume of Illič-Svityč’s comparative Nostratic dictionary. This was followed by a series of gatherings in Russia. Another major conference was held in Ann Arbor, Michigan, at the end of 1988. Organized by Vitaly Shevoroshkin and Benjamin Stolz, this symposium brought together scholars from East and West. A series of volumes under the editorship of Shevoroshkin has appeared as a result of this conference (published by Brockmeyer in Bochum, Germany). Shevoroshkin has also organized several smaller-scale, follow-up conferences. At the end of 1993, a workshop with the theme “The Second Workshop on Comparative Linguistics. The Status of Nostratic: Evidence and Evaluation” was organized at Eastern Michigan University, Ypsilanti, Michigan. Papers from this workshop were subsequently published in a volume co-edited by Brian Joseph and Joe Salmons (1998). Several important papers on Nostratic also appear in the recent festschrift for Vitalij Shevoroshkin (1997). In December 1997, a workshop on distant linguistic relationship was held at the Santa Fe Institute in Santa Fe, New Mexico — participants included scholars from around the world.

In early 1998, Dolgopolsky’s book entitled The Nostratic Hypothesis and Linguistic Paleontology was published. In this book, Dolgopolsky is mainly concerned with linguistic paleontology, and the focus of his attention, therefore, is on putative etyma pertaining to habitat, social organization, and material culture. Dolgopolsky’s conclusions are supported by a sample of 125 proposed cognate sets. The book ends with a reconstruction of the Proto-Nostratic phonological system and the reflexes of the consonants (but not the vowels) in the major branches of Nostratic. This book was the focus of a two-day symposium held in July 1998 under the auspices of the McDonald Institute for Archaeological Research, Cambridge University, England. The proceedings of this symposium were published in mid-1999.

A major conference on “Problems in the Study of Long-Range Linguistic Comparison at the Turn of the Third Millennium” was held at the Russian State University for the Humanities in Moscow from 29 May through 2 June 2000. The conference was organized by Sergej Starostin and covered a number of topics. The first day involved papers on Indo-European. The second day was devoted to Nostratic and included papers on lexical, morphological, and phonological comparisons, as well as more theoretical considerations. There was a session on Altaic, and Starostin gave an introduction to the Altaic etymological dictionary he was then preparing in collaboration with Anna Dybo and Oleg Mudrak (this dictionary has since been published [in 2003]). Another new etymological dictionary presented at the conference was the Semitic dictionary being prepared by Yuri Militarëv and L. E. Kogan. Afrasian linguistics was also discussed in several papers at a session on comparative linguistics and ancient Near Eastern history held in memory of the late Igor M. Diakonoff. There was also a session on Sino-Tibetan and Caucasian linguistics.

In August 2003, a Nostratic Centennial Conference, marking one hundred years since the appearance of Pedersen’s bold hypothesis, was held at the University of Pécs, Hungary. The conference proceedings were published in 2004 in a volume co-edited by Irén Hegedűs and Paul Sidwell.

The Institute of Slavistics and the Department of History and Philology of the Russian Academy of Sciences sponsored a conference in Moscow honoring the 70th birthday of V. M. Illič-Svityč on 20—22 October 2004. The conference covered problems of the comparative-historical grammar of both Indo-European and Nostratic languages, of the remote relationship of languages, and of the history of Slavic and Baltic languages and their dialects.

1.3. Methodology

Even though I have repeated the following points verbatim many times in previous works, I still read irresponsible statements being made in the literature to the effect that Nostraticists do not use “traditional methods” or that they use a “weakened form” of the Comparative Method. Nothing could be farther from the truth. Therefore, I will once again state the methodological principles used in distant linguistic comparison (this is a slightly revised and expanded version of the discussion of methodology found in my co-authored book [Bomhard—Kerns 1994:7—11]).

The founders of Indo-European comparative linguistics placed great importance on the comparison of grammatical forms, and this bias continues to the present day in Indo-European studies and has even been carried over into the study of other language families. However, this overemphasis on the comparison of grammatical forms is far too restrictive and was the reason that the Celtic languages, which have developed many unique features, were not immediately recognized as Indo-European. As noted nearly eighty years ago by Pedersen (1931:245):

That agreement in the inflectional system is an especially clear and striking proof of kinship, no one denies. But it is only an anachronism in theory, which has no significance in actual practice, when such an agreement is still designated as the only valid proof. No one doubted, after the first communication about Tocharian..., that the language was Indo-European, though at that time virtually no similarities in inflection had been pointed out. Such similarities have since been shown, but even where they are almost obliterated, proof of kinship could be adduced from the vocabulary and from sound-laws. Hardly any one will assert that it would be impossible to recognize the relationship between, say, English and Italian, even without the help of other related languages or older forms of these two languages themselves, although agreements between the inflectional systems are practically nonexistent.

From the modern point of view it must be said that proof of relationship between languages is adduced by a systematic comparison of languages in their entirety, vocabulary as well as grammar. The reason why earlier scholars felt they should disregard the vocabulary was that they knew of no method of systematic comparison in the field.

The approach to language comparison that I have followed in attempting to establish genetic relationship among the various Nostratic languages is derived, in part, from that advocated by Joseph H. Greenberg in the chapter entitled “Genetic Relationship among Languages” in his 1957 book Essays in Linguistics and, in part, from traditional methods of comparison and internal reconstruction. In my opinion, the combination of Greenberg’s methodology and more traditional methods of comparison can inform and further one another. The principles established by Greenberg are as follows:

Greenberg notes that the only way to establish hypotheses about genetic relationship is by comparing languages. However, the problem is in knowing which languages to compare and in knowing what to compare since not all aspects of language are equally relevant to comparison. To be meaningful, comparison must strive to eliminate chance resemblances and to separate borrowings from native elements. This is often easier said than done; however, Greenberg lays out two main techniques for detecting borrowed lexical items. First, he notes that borrowing is most commonly confined to certain semantic spheres (for example, cultural items) and certain grammatical categories (nouns far more often than verbs). Second, borrowed words can be distinguished from native vocabulary by expanding the range of comparison to include additional languages.

The simplest way to establish genetic relationship is by identifying a large number of similar morphs (or allomorphs), especially irregularities, in similar environments in the languages being considered. Another significant indicator of probable genetic relationship is the presence of similar rules of combinability. Unfortunately, historical processes over the passage of time tend to bring about the gradual transformation and eventual elimination of such similarities. The longer the period of separation, the lesser the chances will be that similarities of morphological forms and rules of combinability will be found.

Fortunately, there remain other factors that can be helpful in determining possible genetic relationship. One significant factor is the semantic resemblance of lexical forms. Here, it is important to be able to establish recurrent sound-meaning correspondences for a reasonably large sample of lexical material. Lexical forms with identical or similar meanings have the greatest value. Next in value come forms that, though divergent in meaning, can convincingly be derived, through widely-attested semantic shifts, from earlier forms of identical or similar meaning. The chances that lexical resemblances indicate genetic relationship increase dramatically when additional languages are brought into the comparison and when these new languages also exhibit a very large number of recurrent sound-meaning correspondences. Greenberg has termed this method “mass comparison” (more recently, he has used the term “multilateral comparison”). He considers the comparison of basic vocabulary from a large number of languages from a specific, wide geographic area to be the quickest and most certain method to determine possible genetic relationship. To Greenberg, lexical data are of paramount importance in attempting to establish genetic relationship among languages, especially in the initial stages of comparison.

The basic principles underlying the Comparative Method may be summarized as follows: The first step involves the arduous task of data gathering, placing special attention on gathering the oldest data available. Once a large amount of lexical material has been gathered, it must be carefully analyzed to try to separate what is ancient from what is an innovation and from what is a borrowing. After the native lexical elements have been reasonably identified in each phylum, the material can be compared across phyla to determine potential cognates. Once a sufficient body of potential cognates has been identified, one can begin to work out the sound correspondences. Not only must the regular sound correspondences (that is, those that occur consistently and systematically) be defined, exceptions must also be explained. Here, widely-attested sound changes (palatalization, metathesis, syncope, assimilation, dissimilation, etc.) provide the key to understanding the origin of most exceptions. In other cases, the analysis of the influence that morphology has exerted provides an understanding of how particular exceptions came into being. Some exceptions, though clearly related, simply defy explanation. All of these must be noted. The final step involves the reconstruction of ancestral forms and the formulation of the sound laws leading to the forms in the descendant languages, identifying the laws that have produced the regular sound correspondences as well as the exceptions. The same principles apply to the reconstruction of grammatical forms and rules of combinability and to the identification of the historical transformations leading to the systems found in the daughter languages. Invariably, it takes the dedicated efforts of several generations of scholars to work out all of the details. Here, we may cite the case of Indo-European — as even the most casual reading of Lehmann’s book (1993) on the Theoretical Bases of Indo-European Linguistics shows, after nearly two full centuries of investigation of what must surely be the most thoroughly-studied language family on the face of the earth, there still remain many uncertainties about the reconstruction of the Indo-European parent language. The following works are excellent introductions to Comparative-Historical Linguistics: Arlotto 1972; Bynon 1977; Campbell 1999a; Hock—Joseph 1996; Lehmann 1973 [1992]; Sihler 2000 — more advanced are: Anttila 1972 and 1989; Hock 1986 [1991].

At this point, we may note that the description of the Comparative Method and Internal Reconstruction given by Schwink (1994:9) is virtually identical to the procedure outlined in the preceding paragraph:

Let us now proceed to the nuts and bolts of reconstruction. Winter (1970:149) describes the comparative method in the following terms. First one carries out “inspection”. This is looking at a number of languages for “a sufficient number of apparently recurrent correspondences”. One should look at the oldest stages of languages, judge which languages have the most archaic features or residues (Lehmann 1990). Inspection is followed by “sorting” which involves a complete listing of the correspondences discovered although without interpretation (Winter 1970:149). Thereafter comes the reduction of the material to major correspondence classes. If there are irregularities in distribution, one looks for specific factors which may condition the difference. This is now an interpretive procedure. The label chosen for an entity of a major correspondence class should have “a maximum of similarity with the items labeled” (p. 152). In this selection, the question of archaicity of daughter languages will be taken into account. After assumption that the label represents some earlier stage of the languages being looked at, an attempt may be made to look at the labels of parts of systems.

The comparative method does not produce temporal distinctions... It produces a proto-language which is a potpourri of features. It will be the job of internal analysis to sort out this proto-language.

As noted in the first paragraph of this section, it was necessary to discuss these issues in order to address concerns that have been raised about the applicability of traditional methods of comparison to long-range comparison. It must be made perfectly clear that the same principles are just as applicable to long-range comparison as they are to any other type of linguistic comparison. The fact is, these are the only tools we have. Moreover, they work — their efficacy has been proven over and over again.

Furthermore, claims that these methodologies break down when one tries to apply them beyond a certain time limit, say 5,000 to 10,000 years ago, can be shown, without a shadow of doubt, to be false. One can cite, for example, the case of the aboriginal languages of Australia. Archaeological evidence indicates that Australia has been inhabited by human beings for at least 40,000 years, and possibly even longer. Though there remain many unsettled questions, such as exactly when a putative Proto-Australian might have been spoken (probably at least 30,000 years ago), or about how the different languages should be subgrouped, and so on, it has been suggested (though not proven) that all extant languages belong to the same family (cf. Ruhlen 1987:188), and comparative work on these languages is continuing apace (cf. Dixon 1980 and 2002). Another example that can be cited is the case of the Afrasian language family. Due to the extremely deep divisions among the six branches of Afrasian (Semitic, Egyptian, Berber, Omotic, Cushitic, and Chadic), which are far greater than those found, by way of comparison, among the earliest attested branches of Indo-European, the Afrasian parent language must be placed as far back as 10,000 BCE (cf. Diakonoff 1988:33, fn. 15), or perhaps even earlier, according to some scholars (Hodge [1993:99], for example, dates Proto-Afrasian [his Lisramic] at 13,000 BCE). This extremely ancient date notwithstanding, the major sound correspondences have been determined with great accuracy (see especially Diakonoff 1992), excellent progress is being made in reconstructing the common lexicon (two new Afrasian etymological dictionaries have recently been published, one by Vladimir E. Orël and Olga V. Stolbova [1995] and another by Christopher Ehret [1995]), and scholars are beginning to piece together the original morphological patterning, though progress here lags behind other areas. A good survey of the Afrasian languages is (in English translation): Languages in the Ancient and Modern World: Hamito-Semitic Languages, edited by David Cohen (Paris: Centre National de la Recherche Scientifique, 1988). A good introduction — though now somewhat out of date — to Afrasian comparative phonology and morphology is Afrasian Languages (Moscow: Nauka, 1988) by Igor M. Diakonoff. Finally, it should be noted that Edward Lipiński brings in a lot of useful data from related Afrasian languages in his Semitic Languages: Outline of a Comparative Grammar (Leuven/Louvain: Peters Publishers and Department of Oriental Studies, 1997).

One last point needs to be made: Reconstructed languages should be thought of as real languages in every sense of the term. Of course, our reconstructions are, in a sense, purely formulaic, and one can only hope to approximate, not fully recover, all of the features of the actual proto-language. Nevertheless, our reconstructions can be surprisingly accurate, as can be seen, for instance, when reconstructed Proto-Romance is contrasted with so-called “Vulgar Latin”. When we undertake the task of trying to recover the salient features of this or that proto-language, we must be very careful not to reconstruct anything that is not characteristic of language in general: our goal should be to strive for reality in our reconstructions (cf. Labov 1994:17). The prudent use of the insights gained from linguistic typology can be extremely valuable in helping to arrive at realistic reconstructions. Now, a few more conservative linguists have questioned the propriety of using typological data in Historical-Comparative Linguistics, their main argument running somewhat along the lines: “since we cannot possibly know all of the languages that currently exist or that have ever existed, we cannot say that such and such a type was impossible, unnatural, or has never existed” — that is to say, our “database” of linguistic systems will always be incomplete. Of course, there is no arguing with this line of reasoning. However, these linguists miss an important point: from all of the data that have been collected to date — from an extremely large sample of the world’s languages — there emerge consistent, regular patterns that are repeated over and over again. There are, to be sure, rare types — typological isolates, so to speak —, but these are less important (though no less interesting) from a statistical point of view. It is the regular patterning that has emerged from the analysis of the data from a great number of languages that is most important to Historical-Comparative Linguistics. These data are important in two respects: (A) they provide a control against which our reconstructions can be evaluated and (B), when part of a system has been reconstructed, they provide a means to deduce what the rest of the system might have been like, that is to say, they can be used as a discovery procedure by making use of “implicational universals”. Concerning the consistent, regular patterning that has been observed, it should be noted that the basis for some of this patterning is human physiology, and, in such cases, we can speak of true universals. Given this regular patterning, it is disturbing when our reconstructions contradict it, as in the case of one form of the traditional reconstruction of Proto-Indo-European, for instance. To say merely that “Indo-European was a unique type” or some such statement only means that the person making such a statement chooses not to confront the issues involved. We should not hesitate to use every means at our disposal to help us arrive at realistic reconstructions. To be sure, we should be fully cognizant of the work of our predecessors and adhere closely to the time-honored methodologies — the Comparative Method and Internal Reconstruction — that have served Comparative-Historical Linguistics well since the days of Bopp, Rask, and Grimm. However, we must not stop here — we must also make full use of recent advances in phonological theory that have broadened our understanding of sound change and of new insights gained from typological studies, and our proposals must be consistent with the data. For a superb overview of the relevancy of typological studies to diachronic linguistics, cf. Schwink 1994.

In attempting to determine whether or not particular lexical items from the various language families might be related, I have made extensive use of Carl Darling Buck’s A Dictionary of Selected Synonyms in the Principal Indo-European Languages as a control for the semantic development of the proposed lexical parallels. It may be noted that, in examining the lexicons of Kartvelian, Afrasian, Uralic-Yukaghir, Elamo-Dravidian, Altaic, and Eskimo-Aleut, semantic shifts similar to those described by Buck for the Indo-European languages are found over and over again in these other language families as well. I cannot emphasize strongly enough that, in order to gain a complete understanding of how I arrived at my proposals, Buck’s dictionary must be consulted.

One final note is necessary. In recent years, several scholars (most notably, Donald Ringe and Sheila Embleton) have proposed techniques based upon statistical modeling and probability analysis as a means to help us judge the validity of our proposals concerning possible genetic relationship. Properly used, these techniques can indeed provide another valuable tool, which may be used along with, but not as a replacement for, established methodologies. Moreover, these techniques have the important advantage of introducing an objective set of criteria against which our proposals can be evaluated.

1.4. The Comparative Method

In the previous section, we discussed the methodologies used in long-range linguistic comparison and showed that these are the same methodologies used in any other type of linguistic comparison. In this section, we will explore the Comparative Method in greater detail, repeating and expanding upon what was said in the previous section and using data from the Nostratic daughter languages to illustrate the principles involved.

First, let us begin with a formal definition of the Comparative Method (cf. Kimball 1992:274):

Comparative Method examines items (e.g. phonemes, morphemes, or syntactic constructions) from two or more languages to establish genetic relationship and reconstruct ancestral forms. Unlike typological comparison, which ignores genetic affiliation, the comparative method assumes that the languages compared are (or may be) cognate languages: the descendants of a common ancestor.

Moreover, Hock (1991:567) further defines the purpose of reconstruction:

The ultimate proof of genetic relationship, and to many linguists’ minds the only real proof, lies in the successful reconstruction of the ancestral forms from which the systematically corresponding cognates can be derived. (Note that just as in courts of law, the terms ‘proof’, ‘prove’ here are used in the sense of ‘establish beyond a reasonable doubt’. In fact, the general tenet of historical linguistics is that all hypotheses, whether they concern genetic relationship, ‘language-internal’ developments like sound change or analogy, or contact-induced changes, should be established beyond a reasonable doubt. It must be admitted, however, that this tenet is often ignored in practice.)

Hock’s statement is extremely important and pinpoints the crux of the problem in attempts to establish genetic relationship, especially long-range genetic relationship — it seems that no one can agree on the threshold beyond which “reasonable doubt” has been dispelled. For some, the threshold is set so low that highly unlikely proposals can slip by, while, for others, the threshold is set so high that even well-established language families have difficulty passing.

Next, Kimball (1992:275) notes that “[t]he comparative method makes three assumptions”:

a)       The relationship between sound and meaning is arbitrary; therefore, widespread similarity in form and meaning between two languages cannot be accidental.

b)       Corresponding features of cognate languages continue features inherited from an ancestral stage or proto-language.

c)       Completed sound changes are exceptionless.

As previously stated, the first step involves the arduous task of data gathering, placing special attention on gathering the oldest data available. Once a large amount of lexical material has been gathered, it must be carefully analyzed to try to separate what is ancient from what is an innovation and from what is a borrowing. This is not a simple task — the problem of borrowing is particularly acute within Altaic, for instance. Greenberg has addressed this problem by laying out two main techniques for detecting borrowed lexical items. First, he notes that borrowing is most commonly confined to certain semantic spheres (for example, cultural items) and certain grammatical categories (nouns far more often than verbs). Second, borrowed words can be distinguished from native vocabulary by expanding the range of comparison to include additional languages. Moreover, there are important clues that can assist us in identifying borrowings. First, a knowledge of the history or, in the case of reconstructed languages, the prehistory of a language can tell us which languages were in contact or might have been in contact with the language or languages under analysis at different stages in its history. Next, knowledge of the different levels of material culture achieved by population groups speaking these languages at particular times in their history will give us a clue about the probable direction of borrowings. Archeology can be of value here by providing us with a description of the artifacts of the material cultures in question, by giving us a glimpse of the salient characteristics of the societies using those artifacts, and by identifying probable trade routes and population movements.

Let us turn once again to Kimball (1992:275) to see what she has to say on this matter:

However, languages can resemble each other for other reasons. Onomatopoetic words, ‘baby-talk’, and words showing sound symbolism are excluded from consideration; in these, the relationship between sound and meaning is not entirely arbitrary. Similarity can result from borrowing and other effects of language contact, or even from sheer chance — factors which must be eliminated in a list of potential cognates.

Sometimes knowledge of the external history of a language allows us to exclude borrowing as a cause of similarity. For example, we know that many English words resemble French words because English has borrowed extensively from French since the 11th century. Where language contact is less well documented or prehistoric, similarity resulting from borrowing can be excluded with reasonable certainty by selecting items unlikely to have been borrowed. For instance, words referring to technology or material culture, which are often borrowed along with cultural or technological innovations, may make poor candidates for comparison. By contrast, basic vocabulary — kinship terms, numerals, pronouns, pre- and postpositions, and common verbs, adverbs, adjectives, and nouns — are less likely under most circumstances to be borrowed, and are usually more helpful to the comparativist.

After the native lexical elements have been reasonably identified in each phylum, the material can be compared across phyla to determine potential cognates. Once a sufficient body of potential cognates have been identified, one can begin to work out the sound correspondences. Let us illustrate this by looking at a few cognates from the Nostratic languages (only the reconstructed forms will be given for each language group) — I have also included data from Sumerian:

1.       Proto-Indo-European *b^hor-/*b^h3- ‘to bore, to pierce’; Proto-Afrasian *bur- ‘to bore, to pierce’; Proto-Uralic *pura ‘borer, auger’; Proto-Dravidian *pur- ‘(vb.) to bore, to perforate; (n.) borer, gimlet’; Proto-Altaic *burV- ‘to bore through, to pierce’. Cf. Sumerian bùr ‘to bore through, to pierce’.

2.       Proto-Indo-European *b^her-, *b^hru- ‘brown’; Proto-Afrasian *bor- ‘dark-colored’; Proto-Altaic *bor^yV- ‘gray, brown’ (< ‘dark-colored’).

3.       Proto-Kartvelian *bur- ‘to cover, to enclose’; Proto-Afrasian *bur- ‘to cover, to wrap up’; Proto-Dravidian *pōr- ‘(vb.) to wrap around (the body), to cover, to enclose; (n.) a cover, covering, wrapping’; Proto-Altaic *būri- (~ -Ôū-, -e) ‘to cover, to enclose’.

4.       Proto-Indo-European *b^hek’-/*b^hok’- ‘to cut or split apart, to break apart’; Proto-Afrasian *bak’- ‘to cleave, to split, to break open’; Proto-Dravidian *pak- ‘to split, to rend; to be split’; Uralic: Proto-Finno-Ugrian *pakka- ‘to burst, to rend, to split’; Eskimo-Aleut: Proto-Inuit *pakak- ‘to knock into’.

The correspondence, in initial position, of Proto-Indo-European *b^h-, Proto-Kartvelian *b-, Proto-Afrasian *b-, Proto-Uralic *p-, Proto-Dravidian *p-, Proto-Altaic *b-, and Proto-Eskimo *p- allows us to reconstruct Proto-Nostratic *b-.

1.       Proto-Indo-European *p^her-/*p^hor-/*p^h3- ‘to fly, to flee’; Proto-Kartvelian *par-, *pr-en- ‘to fly’; Proto-Dravidian *par- ‘to fly, to flee; to hasten, to hurry’.

2.       Proto-Indo-European *p^her-/*p^h3- ‘to bear, to bring forth’; Proto-Afrasian *pir- ‘to bring forth, to bear fruit’; Proto-Dravidian *per- ‘to get, to beget, to bear’; Proto-Altaic *p^hŭri ‘seed, offspring’.

3.       Proto-Indo-European *p^het^h-/*p^hot^h- ‘to fly, to rush, to pursue; to fall, to fall down’; Proto-Kartvelian *petk- ‘to quiver, to tremble, to vibrate, to explode’; Proto-Afrasian *pat- ‘to flutter, to quiver, to tremble; to fall down’; Proto-Dravidian *pat- ‘to hurry; to flutter, to quiver, to shake; to be flurried, impatient, overhasty’; Proto-Eskimo *patta^w- ‘to clap or slap’.

4.       Proto-Indo-European *p^hes-/*p^hos- ‘penis’; Uralic: Proto-Finno-Ugrian *pas^yз ‘penis’; Proto-Dravidian *p`(y)-/*pac- ‘descendant, offspring’; Proto-Altaic *p^h[Ôa]s- (?) ‘male genitals’. Cf. Sumerian peš ‘sperm, semen’, peš ‘descendant, offspring, son’.

In these examples, the correspondence, in initial position, of Proto-Indo-European *p^h-, Proto-Kartvelian *p-, Proto-Afrasian *p-, Proto-Uralic *p-, Proto-Dravidian *p-, Proto-Altaic *p^h-, and Proto-Eskimo *p- allows us to reconstruct Proto-Nostratic *p^h-.

1.       Proto-Indo-European *me-/*mo- 1st person personal pronoun stem (oblique cases); Proto-Kartvelian *me-, *men- 1st person personal pronoun stem; Proto-Afrasian *m[i]- 1st person personal pronoun stem (only in Chadic, with relics in Cushitic); Proto-Uralic *me 1st person singular personal pronoun stem: ‘I, me’, *me 1st plural personal pronoun stem; Proto-Altaic (nom. sg.) (*mi >) *bi ‘I’, (oblique stem) *min-; Proto-Chukchi-Kamchatkan *(kə-)m ‘I’ (*kə- is a marker of independent pronouns); Eskimo-Aleut: West Greenlandic 1st sg. relative possessive suffix -ma. Note here also Etruscan mi ‘I’, mini ‘me’ and Sumerian (Emesal) ma(-e), me-a, me-e ‘I’, (1st pl. possessive suffix) -me ‘our’.

2.       Proto-Indo-European *mo- demonstrative stem (preserved vestigially in Celtic); Proto-Kartvelian *ma- demonstrative stem: ‘this, he’; Proto-Finno-Ugrian *mu ‘other, another’; Altaic: Common-Turkic (nom. sg.) (*mū/*mō >) *bū/*bō ‘this’, (oblique stem) *mu-n-; Mongolian mön deictic word serving as a demonstrative pronoun, adjective, adverb, and copula.

3.       Proto-Indo-European *me-/*mo- interrogative and relative pronoun stem (preserved in Hittite and Tocharian, with vestiges in Celtic); Proto-Kartvelian *mi-n- interrogative pronoun, *ma- ‘what’; Proto-Afrasian *ma- ~ *mi- relative and interrogative pronoun stem; Proto-Uralic *mi interrogative and relative pronoun stem; Proto-Altaic *mV interrogative stem; Proto-Eskimo enclitic particle *mi ‘what about?’. Cf. Sumerian me-na-àm ‘when?’, me-a ‘where?’, me-šè ‘where to?’.

4.       Proto-Indo-European *mer-/*mor- ‘to twist, to turn’; Proto-Afrasian *mar-/ *mər- ‘to twist, to turn’; Proto-Dravidian *mur- ‘to bend, to be bent, to turn round, to twist; (n.) rope, cord; bend, curve’, *mur- ‘to twist, to twine, to tighten’; Proto-Altaic *mura- ‘(vb.) to turn, to return; (adj.) round’.

Here, the correspondence, in initial position, of Proto-Indo-European *m-, Proto-Kartvelian *m-, Proto-Afrasian *m-, Proto-Uralic *m-, Proto-Dravidian *m-, Proto-Altaic *m-, and Proto-Eskimo *m- allows us to reconstruct Proto-Nostratic *m-.

These correspondences can be summarized as follows:

Abbreviations: PN = Proto-Nostratic; PIE = Proto-Indo-European; PK = Proto-Kartvelian; PAA = Proto-Afrasian; PU = Proto-Uralic; PD = Proto-Dravidian; PA = Proto-Altaic; PE = Proto-Eskimo.

Not only must the regular sound correspondences (that is, those that occur consistently and systematically) be defined (a full set of Nostratic sound correspondences can be found at the end of Chapter 12), exceptions must also be explained. Here, widely-attested sound changes (palatalization, metathesis, syncope, assimilation, dissimilation, etc.) provide the key to understanding the origin of most exceptions. In other cases, the analysis of the influence that morphology has exerted provides an understanding of how particular exceptions came into being. Some exceptions, though clearly related, simply defy explanation. All of these must be noted. The final step involves the reconstruction of ancestral forms and the formulation of the sound laws leading to the forms in the descendant languages, identifying the laws that have produced the regular sound correspondences as well as the exceptions. The same principles apply to the reconstruction of grammatical forms and rules of combinability and to the identification of the historical transformations leading to the systems found in the daughter languages.

Let us now look at some exceptions to the regular sound correspondences that have been established and provide explanations for these exceptions:

1.       Pre-Proto-Indo-European *k^hab- > (with progressive voicing assimilation) Proto-Indo-European *k^hap^h-ro- ‘he-goat, male sheep, buck, ram’ ~ Proto-Afrasian *kab- ‘he-goat, male sheep, buck, ram’.

In this example, the correspondence of Proto-Indo-European *-p^h- ~ Proto-Afrasian *-b- is irregular — instead, we would expect Proto-Indo-European *-b^h- as the regular correspondence of Proto-Afrasian *-b-. Now, it is well-known that Indo-European had a root-structure constraint against the appearance of both a voiced (aspirated) stop and a voiceless (aspirated) stop in a root, that is to say, that they had to agree in voicing (cf. Benveniste 1935:170; Lehmann 1952:17) — thus, *t^heb^h- and *b^het^h- were not allowed. However, comparison with the other Nostratic languages indicates that the forbidden root types must have once existed. Therefore, a rule of progressive voicing assimilation may be set up to account for the elimination of the forbidden root types. This means that *t^heb^h- would have become *t^hep^h-, and *b^het^h- would have become *b^hed^h-. This is confirmed by other examples, such as:

2.       Pre-Proto-Indo-European *d^yək^wh-/*d^yak^wh- > (with progressive voicing assimilation and depalatalization of initial *d^y) Proto-Indo-European *d^heg^wh-/*d^hog^wh- ‘to blaze, to burn’ ~ Proto-Afrasian *d^yak^w- ‘to blaze, to be bright’.

Another exception is found in the following examples:

1.       Proto-Indo-European *(s)t^hek’-/*(s)t^hok’- ‘to cover’ ~ Proto-Kartvelian *t’q’aw- ‘skin, hide’; Proto-Afrasian *t’ak’- ‘to cover, to obscure’.

2.       Proto-Indo-European *t^hek’-/*t^hok’- ‘to knock, to beat, to strike’ ~ Proto-Kartvelian *t’k’aD- ‘to hit, to strike’; Proto-Afrasian *t’uk’-, *t’ok’- ‘to knock, to beat, to strike, to pound’; Proto-Finno-Ugrian *tukз- (*tu^wз-) ‘to break, to crush’; Proto-Dravidian *tuk- ‘to tread down, to trample on, to step on; to beat, to strike, to pound, to mash’, *tukk- ‘to push, to shove’. Cf. Sumerian dugú-ga ‘to strike, to beat, to hit, to smite, to kill’.

In these examples, the correspondence of Proto-Indo-European *t^h- ~ Proto-Kartvelian *t’- and Afrasian *t’- is irregular — instead, we would expect Proto-Indo-European *t’- as the regular correspondence of Proto-Kartvelian *t’- and Proto-Afrasian *t’-. In traditional terms, Proto-Indo-European had a constraint against the appearance of two plain voiced stops within a root (cf. Benveniste 1935:170; Lehmann 1952:17), that is to say that a root could not both begin and end with a plain voiced stop. In terms of the Glottalic Theory (see Chapter 3, §3.4, for a discussion of the Glottalic Theory), this constraint is reinterpreted as a restriction against the co-occurrence of two glottalics in a root. This means that roots of the type *t’ek’- (*deg- in traditional terms) are not allowed. It may be noted that a similar constraint is found in a number of other languages having glottalics. However, comparison with the other Nostratic languages indicates that the forbidden root types must have once existed. Therefore, a rule of regressive deglottalization may be set up to account for the elimination of the forbidden root types. This means, for example, that *t’ek’- would have become *t^hek’-. This rule finds a close parallel in Geers’ Law in Akkadian (for details on Geers’ Law, cf. Ungnad—Matouš 1969:27).

Now, up until this point, we have been using mostly reconstructed forms to illustrate the principles involved in the Comparative Method. However, reconstructed forms contain a sufficiently high enough margin of error by their very nature to render such comparisons suspect. This means that, ultimately, we must base our conclusions about possible genetic relationship on an examination and analysis of the actual attested forms found in each daughter language. It is my contention that a comparison based on the actual attested forms alone, without recourse to the reconstructed forms, is sufficient to demonstrate the genetic relationship of the various Nostratic daughter languages. Let us illustrate this by looking at the data which support the reconstructions given in several of the examples above — we will look at one from each set.

First, let us look again at the words for ‘to bore, to pierce’:

1.       (a) Proto-Indo-European *b^hor-/*b^h3- ‘to bore, to pierce’; (b) Proto-Afrasian *bur- ‘to bore, to pierce’; (c) Proto-Uralic *pura ‘borer, auger’; (d) Proto-Dravidian *pur- ‘(vb.) to bore, to perforate; (n.) borer, gimlet’; (e) Proto-Altaic *burV- ‘to bore through, to pierce’. Cf. Sumerian bùr ‘to bore through, to pierce’.

Here are some of the attested data from within each language family to support this example:

a)       Indo-European: Old English borian ‘to bore, to pierce’; Old High German boro ‘auger’; Latin forō ‘to bore, to pierce’ (Latin f- < *b^h-); Greek öáñüù, öáñÜù ‘to plow’.

b)       Afrasian: Aramaic bəraz ‘to bore, to pierce’; Tigre (reduplicated) bärabära ‘to pierce’; Geez / Ethiopic barra, barara ‘to pierce, to penetrate, to go through’; Somali burur ‘broken piece’; Saho burūr ‘broken piece’.

c)       Uralic: Finnish pura ‘borer, auger, (big) awl’; Vogul / Mansi pore, porä ‘awl’; Ostyak / Xanty põr ‘borer, auger’; Hungarian fúr- ‘to bore, to drill’; Yurak Samoyed / Nenets parõ ‘borer, auger’; Selkup Samoyed pur ‘borer, auger’.

d)       Dravidian: Tamil purai ‘tubular hollow, tube, pipe, windpipe’; TuJu perevuni ‘to be bored, to be perforated’, perepini ‘to bore to perforate’, burma, burmu ‘a gimlet’, berpuri ‘borer, auger’.

e)       Mongolian bur^wui- ‘a piece of wire used to clean a smoking pipe’; Turkish bur- ‘to bore a hole’; Tatar borau ‘borer, auger’.

f)        Sumerian bùr ‘to bore through, to pierce’.

The second example which we will explore in depth is the words for ‘to flee, to fly’:

2.       (a) Proto-Indo-European *p^her-/*p^hor-/*p^h3- ‘to fly, to flee’; (b) Proto-Kartvelian *par-, *pr-en- ‘to fly’; (c) Proto-Dravidian *par- ‘to fly, to flee; to hasten, to hurry’.

Here are some of the attested data from within each language family to support this example:

a)       Indo-European: Sanskrit parzá-­ ‘wing, feather’; Hittite pár-aš-zi ‘to flee’; Russian Church Slavic perC, pÞrati ‘to fly’, pero ‘feather’; Czech perchnouti ‘to flee’; Polish pierzchnaB ‘to flee’; Serbo-Croatian prhati ‘to fly up’; Russian porxát' [порхать] ‘to flit, flutter, to fly about’.

b)       Kartvelian: Georgian pr-ena ‘to fly’, (m)prinveli ‘bird’; Mingrelian purin- ‘to fly’; Laz purtin- ‘to fly’.

c)       Dravidian: Tamil para ‘to fly, to hover, to flutter, to move swiftly, to hasten, to be in a hurry; to be greatly agitated; to be scattered, dispersed; to disappear’, (reduplicated) parapara ‘to hasten, to hurry’, paravai ‘bird, wing, feather, bee’; Malayalam parakka ‘to fly, to flee’; KannaTa pari, paru ‘flying, running swiftly’; TuJu pāruni ‘to run, to fly, to escape’; Telugu paracu ‘to run away, to flee, to flow; to cause to escape’, pāru ‘to run, to flow’.

The final example is the words for ‘I, me’:

3.       (a) Proto-Indo-European *me-/*mo- 1st person personal pronoun stem (oblique cases); (b) Proto-Kartvelian *me-, *men- 1st person personal pronoun stem; (c) Proto-Afrasian *m[i]- 1st person personal pronoun stem (only in Chadic, with relics in Cushitic); (d) Proto-Uralic *me 1st person singular personal pronoun stem: ‘I, me’, *me 1st plural personal pronoun stem; (e) Proto-Altaic (nom. sg.) (*mi >) *bi ‘I’, (oblique stem) *min-; (f) Proto-Chukchi-Kamchatkan *(kə-)m ‘I’ (*kə- is a marker of independent pronouns). Note here also Etruscan mi ‘I’, mini ‘me’ and Sumerian (Emesal) ma(-e), me-a, me-e ‘I’, (1st pl. possessive suffix) -me ‘our’.

Here are some of the attested data from within each language family to support this example:

a)       Indo-European: Sanskrit (acc. sg.) mā, mZm ‘me’; Greek (acc. sg.) ìå, dìÝ ‘me’; Latin (acc.-abl. sg.) mē ‘me’; Gothic (acc. sg.) mik ‘me’; Lithuanian (acc. sg.) manę ‘me’; Old Church Slavic (acc. sg.) mę, mene ‘me’.

b)       Kartvelian: Old Georgian me ‘I’; Mingrelian ma- ‘I’; Laz ma, man ‘I’; Svan mi ‘I’.

c)       Afrasian: Chadic: Hausa (pl.) maa ‘we’, (indirect object pl.) manà ‘us, to us, for us’, (pl.) muu ‘we, us, our’, (past tense subj. pl.) mun ‘we’, (continuous tense subj. pl.) munàa ‘we’; (indirect object sg.) minì ‘me, to me, for me’; Kotoko mi ‘we, us’; Mandara ma ‘we, us’; Musgu mi ‘we, us’, mu ‘I, me’; Bole mu ‘we, us’.

d)       Uralic: Finnish minä/minu- ‘I, me’; Lapp / Saami mon/mú- ‘I, me’; Mordvin mon ‘I, me’; Zyrian / Komi me ‘I’, (acc.) menõ ‘me’; Selkup Samoyed man, mat ‘I, me’; Kamassian man ‘I, me’; Yukaghir met ‘I, me’.

e)       Altaic: Mongolian (nom. sg.) bi ‘I’, (gen. sg.) minu ‘my, of me’, (gen. pl. exclusive) manu ‘our, of us’; Manchu bi ‘I, me’, (gen. sg.) mini ‘my’; Old Turkish (nom. sg.) män (rarely bän) ‘I’, (acc. sg.) mäni ‘me’.

f)        Chukchi ^wə-m ‘I’ (in predication: -i^wəm ~ -e^wəm).

g)       Etruscan mi ‘I’, mini ‘me’.

h)       Sumerian (Emesal) ma(-e), me-a, me-e ‘I’, (1st pl. possessive suffix) -me ‘our’.

It is thus perfectly clear that we are able to establish phonological correspondences on the basis of an analysis of the actual attested data from the individual Nostratic daughter languages alone, without recourse to reconstructions. Moreover, not only are we able to establish the regular sound correspondences by such an analysis, we are also able to identify and explain exceptions. And, it is on this basis as well that we are able to reconstruct the Proto-Nostratic forms. This is identical to what was done in Indo-European and which continues to be done in Comparative-Historical Linguistics — the Indo-European parent language was reconstructed on the basis of a direct comparison of the actual attested data from the individual Indo-European daughter languages without recourse to reconstructed Proto-Indo-Iranian, Proto-Italic, Proto-Greek, Proto-Germanic, etc. That is to say that it was not necessary to reconstruct every intermediary level before one could tackle the problems of reconstructing the Indo-European parent language. Of course, reconstruction is still both important and necessary. Reconstruction, including the reconstruction of intermediary levels, allows us to make powerful statements about the (pre)historical development of each daughter language, especially about how and why particular features came into being or became extinct. Finally, the understanding of what has taken place historically in one daughter language often provides an explanation of what has taken place in another daughter language.

In any attempt to establish genetic relationship, one is going to come across chance resemblances. By “chance resemblances”, one means unexpected, and sometimes rather striking, instances of identical or nearly identical vocabulary items or, in rare cases, even grammatical forms in two or more totally unrelated languages or in languages that, if they are related, are distant enough apart to make it otherwise unlikely that they would share such items. The example that Kimball (1992:275) gives is the word for ‘man’, wiro, in the extinct Timucua language, formerly spoken in northern Florida and southeastern Georgia, which resembles Latin vir ‘man’. Chance resemblances of this type do occur and, it goes without saying, do not indicate genetic relationship. Chance resemblances can range from a mere handful of examples up to several dozen depending upon how much latitude one is willing to allow in both forms and meanings. As noted above, one of the main assumptions of the Comparative Method is that “the relationship between sound and meaning is arbitrary; therefore, widespread similarity in form and meaning between two languages cannot be accidental”. Thus, when the languages under analysis exhibit a large number of recurrent sound-meaning correspondences, we are not dealing with chance resemblances.

1.5. Critique of Moscovite Views on Nostratic

Let me begin by stating unequivocally that I have the highest admiration for what Moscovite scholarship (especially the work of V. M. Illič-Svityč and A. B. Dolgopolsky — some of the work done by other Russian scholars is not on the same level) on Nostratic has achieved. Their research has opened up new and exciting possibilities and given Nostratic studies new respectability. However, this does not mean that I agree with everything they say. I regard their work as a pioneering effort and, as such, subject to modification in light of advances in linguistic theory, in light of new data from the Nostratic daughter languages, and in light of findings from typological studies that give us a better understanding of the kind of patterning that is found in natural languages as well as a better understanding of what is characteristic of language in general, including language change.

Let us begin by looking at phonology: In 1972 and 1973, the Georgian scholar Thomas V. Gamkrelidze and the Russian scholar Vjačeslav V. Ivanov jointly proposed a radical reinterpretation of the Proto-Indo-European stop system. According to their reinterpretation, the Proto-Indo-European stop system was characterized by the three-way contrast glottalized ~ voiceless (aspirated) ~ voiced (aspirated). In this revised interpretation, aspiration is viewed as a redundant feature, and the phonemes in question could also be realized as allophonic variants without aspiration. Paul J. Hopper made a similar proposal at about the same time (Hopper 1973). I should point out here that, even though I support the revisions proposed by Gamkrelidze, Hopper, and Ivanov, my views are not dependent upon any particular reconstruction of the Proto-Indo-European stop system — the sound correspondences I have proposed can be maintained using the traditional reconstruction as well. What the new views of Proto-Indo-European consonantism did was bring into light the implausibility of certain Nostratic sound correspondences established by Illič-Svityč and Dolgopolsky (see below for details). Moreover, this new interpretation opened new possibilities for comparing Proto-Indo-European with the other Nostratic daughter languages, especially Proto-Kartvelian and Proto-Afrasian, each of which had a similar three-way contrast. The simplest and most straightforward assumption would be that the glottalized stops posited by Gamkrelidze, Hopper, and Ivanov for Proto-Indo-European would correspond to glottalized stops in Proto-Kartvelian and Proto-Afrasian, while the voiceless stops would correspond to voiceless stops and voiced stops to voiced stops. This, however, is quite different from the correspondences proposed by Illič-Svityč and Dolgopolsky. They see the glottalized stops of Proto-Kartvelian and Proto-Afrasian as corresponding to the traditional plain voiceless stops of Proto-Indo-European, while the voiceless stops in the former two branches are seen as corresponding to the traditional plain voiced stops of Proto-Indo-European, and, finally, the voiced stops to the traditional voiced aspirates of Proto-Indo-European. Illič-Svityč and Dolgopolsky then reconstruct the Proto-Nostratic phonological system on the model of Kartvelian and Afrasian, with the three-way contrast glottalized ~ voiceless ~ voiced in the series of stops and affricates.

The mistake that Illič-Svityč and Dolgopolsky made was in trying to equate the glottalized stops of Proto-Kartvelian and Proto-Afrasian with the traditional plain voiceless stops of Proto-Indo-European. Their reconstruction would make the glottalized stops the least marked members in the Proto-Nostratic labial series and the most marked in the velar series. Such a reconstruction is thus in contradiction to typological evidence, according to which glottalized stops uniformly have the opposite frequency distribution (most marked in the labial series and least marked in the velar series [for details, cf. Gamkrelidze 1978]). The reason that Illič-Svityč’s and Dolgopolsky’s reconstruction contradicts the typological evidence is as follows: Illič-Svityč and Dolgopolsky posit glottalics for Proto-Nostratic on the basis of a small number of seemingly solid examples in which glottalics in Proto-Afrasian and/or Proto-Kartvelian appear to correspond to traditional plain voiceless stops in Proto-Indo-European. On the basis of these examples, they assume that, whenever there is a voiceless stop in the Proto-Indo-European examples they cite, a glottalic is to be reconstructed for Proto-Nostratic, even when there are no glottalics in the corresponding Kartvelian and Afrasian forms! This means that the Proto-Nostratic glottalics have the same frequency distribution as the Proto-Indo-European plain voiceless stops. Clearly, this cannot be correct. The main consequence of the mistaken comparison of the glottalized stops of Proto-Kartvelian and Proto-Afrasian with the traditional plain voiceless stops of Proto-Indo-European is that Illič-Svityč and Dolgopolsky are led to posit forms for Proto-Nostratic on the basis of theoretical considerations but for which there is absolutely no evidence in any of the daughter languages. The following examples illustrate the ad hoc nature of these reconstructions:

1.       Dolgopolsky (1998:17) reconstructs a second singular personal pronoun *sü > *si ‘thou’, with an initial glottalized dental, on the basis of data from Indo-European, Afrasian, Uralic, and Mongolian. When one looks at the attested forms in the daughter languages, one cannot find a single form anywhere that begins with a glottalized consonant. Indeed, in natural languages having glottalized consonants, these sounds tend to be underrepresented in pronoun stems and inflectional affixes. What, then, is the basis for the reconstruction *sü? — nothing more than an ad hoc rule set up by Illič-Svityč.

2.       Dolgopolsky (1998:17) also reconstructs an interrogative stem *"o- ‘who?’ (see also Illič-Svityč 1971—  .I:355—356, no. 232, *Áo ‘who’). As in the preceding example, there is no evidence in any of the Nostratic daughter languages to support the reconstruction of an initial glottalized velar in this stem.

Do these criticisms completely invalidate the cognate sets proposed by Illič-Svityč and Dolgopolsky in which glottalics in Kartvelian and Afrasian appear to correspond to plain voiceless stops in Indo-European? Well, no, not exactly — it is not quite that simple. In some cases, the etymologies are correct, but the Proto-Nostratic reconstructions are wrong. This applies to the examples cited above — for the second person personal pronoun, I would reconstruct Proto-Nostratic *t^hi, and, in place of *"o- ‘who?’, I would reconstruct Proto-Nostratic *k^wha-. Other examples adduced by Illič-Svityč and Dolgopolsky admit alternative explanations, while still others are questionable from a semantic point of view and should be abandoned. Once the questionable examples are removed, there is an extremely small number (no more than a handful) left over that appear to support their position. However, compared to the massive counter-evidence in which glottalized stops in Kartvelian and Afrasian correspond to similar sounds (the traditional plain voiced stops) in Proto-Indo-European, even these residual examples become suspect (they may be borrowings or simply false cognates). Finally, there are even some examples where Dolgopolsky’s and Illič-Svityč’s comparison of glottalized stops in Proto-Kartvelian and Proto-Afrasian with plain voiceless stops in Proto-Indo-European is correct. This occurs in the cases where two glottalics originally appeared in a Proto-Nostratic root: *C’VC’-. Such roots are preserved without change in Proto-Kartvelian and Proto-Afrasian, while in Proto-Indo-European, they have been subject to a rule of regressive deglottalization: *C’VC’- > *CVC’-.

Another major shortcoming is in Illič-Svityč’s reconstruction of the Proto-Nostratic vowel system, which, according to him, is essentially that of modern Finnish. It simply stretches credibility beyond reasonable bounds to assume that the Proto-Nostratic vowel system could have been preserved unchanged in Finnish, especially considering the many millennia that must have passed between the dissolution of the Nostratic parent language and the emergence of Finnish (Serebrennikov 1986:75 makes the same point). No doubt, this erroneous reconstruction came about as a result of Illič-Svityč’s failure to deal with the question of subgrouping. The Uralic-Yukaghir phylum, of which Finnish is a member, belongs to the Eurasiatic branch of Nostratic. Now, Eurasiatic is several millennia younger than Afrasian, which appears to be the oldest branch of the Nostratic macrofamily. Therefore, Afrasian must play a key role in the reconstruction of the Proto-Nostratic vowel system, and the Uralic-Yukaghir vowel system must be considered a later development that cannot possibly represent the original state of affairs.

In closing, we may note that Alexis Manaster Ramer (1997:94—96) arrived at the same conclusions reached here regarding the need to reexamine the Nostratic sound correspondences proposed by Illič-Svityč (and, by implication, Dolgopolsky as well) in light of typological considerations. Specifically, he writes:

6.1. Finally, quite recently, I decided to see what would happen if one counted up the occurrences of the different stops (voiceless vs. voiced vs. glottalized as well as labial vs. coronal vs. velar) reconstructed for Nostratic by Illich-Svitych. I only performed the experiment on root-initial stops, with the following results: (they are given as approximations because there is a problem arriving at exact figures given that there [are] some cases where it is difficult to tell whether one is dealing with a single Nostratic form or two, or whether a particular form should begin with this or that stop):

        *b  50+                                  *d  20+                                  *g  40+

        *p  15+                                  *t  15+                                   *k  50+

        *p’  40+                                 *t’  30+                                  *k’  60+

        The first observation (see Manaster Ramer in press a) was that … the relative frequencies of the three phonation types (voiced, voiceless, glottalized) posited for Proto-Nostratic stops, as reflected in the sets of cognates compiled by Illich-Svitych, seem to be inconsistent with typological predictions. Specifically, at least in initial position, the series of stops reconstructed as glottalized is much more frequent at all points of articulation than the series reconstructed as (plain) voiceless.

Since one expects glottalized stops to be more marked and hence less frequent than plain voiceless, in particular, something was amiss. However, just as in the case of the clusters and affricates discussed above, the solution turned out to be quite simple. Given the markedness considerations, I would suggest that the “glottalized” series was actually plain voiceless in Proto-Nostratic, while the “voiceless” series represented some more marked phonation type, glottalized or perhaps aspirated. This is consistent with the fact that the Nostratic series Illich-Svitych wrote as “glottalized” is in fact realized as glottalized only in parts of Afro-Asiatic and in Kartvelian, and in the latter it is easy to imagine that this could be a contact-induced development.

This reinterpretation of Nostratic … naturally calls to mind the glottalic theory of Indo-European. As it happens, the stop series reconstructed by Illich-Svitych as plain voiceless and by me as glottalized (or aspirated) comes out in Proto-Indo-European as that series of stops which is traditionally reconstructed as voiced (media) but which many scholars have recently interpreted as glottalized.

      Nostratic                Nostratic                Indo-European         Indo-European

  (Illich-Svitych)   (Manaster Ramer)          (Traditional)               (Glottalic)

        *t                     *t’ (or *t^h)                        *d                                *t’

        *t’                           *t                              *t                                 *t

        *d                           *d                             *dh                              *d

Totally unexpectedly, typological considerations provide us with arguments for reinterpreting the Nostratic stop series in a way that fits quite well with the glottalic theory of Indo-European. Of course, there is no reason in general to expect the phonetics of related languages and proto-languages to agree in this way, and such a convergence cannot be regarded as a criterion or an argument for relatedness among languages, since that would entail the “misuse of similarity” which Hamp (1992) cautions against. But it is not an unwelcome development when it occurs.

1.6. Evidence for Nostratic

The following evidence provides the basis for setting up a Nostratic macrofamily:

1.       First and foremost, the descendant languages can be shown to share a large common vocabulary. In an article published in 1965, Illič-Svityč listed 607 possible common Nostratic roots, but only 378 have been published to date in his posthumous comparative Nostratic dictionary. It should be noted that there are differences between the etymologies proposed in 1965 and the items included in the later dictionary: first, some of the items listed in 1965 do not appear in the dictionary; next, minor changes have been made to several of the earlier etymologies. Dolgopolsky currently claims to have approximately 3,000 common Nostratic roots, but only a small sampling of this material has been published to date. In the joint monograph (1994) by myself and John C. Kerns, entitled The Nostratic Macrofamily: A Study in Distant Linguistic Relationship, I supply a great deal of lexical material from the Nostratic daughter languages to support 601 common Nostratic roots. It should be mentioned here as well that, in Volume 2 (2002) of his book Indo-European and Its Closest Relatives: The Eurasiatic Language Family, Greenberg also presents a substantial body of lexical material, though Greenberg’s Eurasiatic is not the same as Nostratic.

2.       As is to be expected, the various branches of Nostratic investigated to date exhibit regular sound correspondences (see the table of Nostratic sound correspondences at the end of Chapter 12 for details), though, it should be mentioned, there are differences in interpretation between Illič-Svityč and Dolgopolsky on the one hand and myself on the other.

3.       Finally, a substantial number of common grammatical formants have now been recovered — many of these are listed in Illič-Svityč’s comparative Nostratic dictionary; see also the chapter on Nostratic morphology by John C. Kerns in Bomhard—Kerns (1994:141—190), Volume 1 of Greenberg’s Indo-European and Its Closest Relatives: The Eurasiatic Language Family (Greenberg 2000), Dybo (2004), and Bomhard (2002a and 2004c). The grammatical formants that have been recovered to date are discussed in detail in Chapter 16 of this book, while a systematic reconstruction of Proto-Nostratic morphology is attempted in Chapter 17.

Notable among the lexical items uncovered by Illič-Svityč, Dolgopolsky, Greenberg, and myself is a solid core of common pronominal stems (these are listed below in Table 1, though only the stems represented in Indo-European are given — the Proto-Nostratic reconstructions are given according to my system; for information on other pronoun stems, cf. Dolgopolsky 1984). These pronominal stems have particular importance, since, as forcefully demonstrated by John C. Kerns (1985:9—50), pronouns, being among the most stable elements of a language, are a particularly strong indicator of genetic relationship (Ruhlen 1994a:92—93 makes the same point). Kerns (1985:48) concludes (the emphasis is his):

The results are overwhelming. We are forced to conclude that the pronominal agreements between Indo-European and Uralic, between Uralic and Altaic, and between Indo-European and Altaic, did not develop independently, but instead were CAUSED by some UNIQUE historical circumstance. In short, it is extremely unlikely that the three pronominal systems could have evolved independently.

Likewise, Collinder (1966:200):

It has been said that identical pronouns do not even give an indication of affinity, because you will find such identities anywhere, even if you compare two manifestly unrelated languages. The random checks I have made seem to indicate that this does not hold good. Outside the nostratic group, there are identities, but only a few, from one to four. Within the nostratic group the number of identities varies from, let us say, seven to ten. As the probability of mere chance decreases in geometric, not in arithmetic, proportion to the increasing number of identities, seven to ten identities means quite another level of probability than one to four.

The conclusion seems inescapable that the consistent, regular phonological correspondences that can be shown to exist among the Nostratic daughter languages as well as the agreements in vocabulary and grammatical formants that have been uncovered to date cannot be explained as due to linguistic borrowing or mere chance but can only be accounted for in terms of common origin, that is, genetic relationship. To assume any other possibility would be tantamount to denying the efficacy of the Comparative Method. This does not mean that all problems have been solved. On the contrary, there remain many issues to be investigated and many details to be worked out, but the future looks extremely exciting and extremely promising.

At this stage of research, we can confidently say that the following languages/ language families are to be included in the Nostratic macrofamily: Afrasian, Elamo-Dravidian, Kartvelian, and Eurasiatic. Eurasiatic, in turn, includes the following: Etruscan, Indo-European, Uralic-Yukaghir, Altaic, Chukchi-Kamchatkan, Gilyak, and Eskimo-Aleut. Each of these languages/language families will be discussed in more detail in Chapter 2. The Nostratic family tree may be represented as follows (note here, for comparison, the computer-generated family tree given by Starostin [1999c:66]):

Chart 1: The Nostratic Macrofamily

Table 1: The Distribution of Nostratic Pronoun Stems

A. Personal Pronoun Stems

NOTES:

1.       Indo-European: The 1st sg. stem *mi-/*me- is used in the oblique cases (except in the Celtic branch, where it has spread into the nominative as well); the 1st pl. inclusive stem *ma-/*mə- is preserved in 1st person plural verb endings; the 1st pl. stem *wa-/*wə- is preserved as an independent 1st person plural pronoun stem and in 1st person dual and/or plural verb endings; the 2nd sg. reconstructions *t^h³, *t^he- represent later, Post-Anatolian forms.

2.       Kartvelian: The 1st pl. stem *na-/*nə- is found in Svan näj ‘we’.

3.       Afrasian: The 1st sg. stem *mi-/*me- and 1st pl. inclusive stem *ma-/*mə- are found only in Chadic as independent pronouns; the 1st sg. stem *mi-/*me- serves as the basis of the 1st sg. verbal suffix in Highland East Cushitic; the 1st pl. stem *wa-/*wə- is found in Egyptian and Chadic (in Egyptian, wy means ‘I, me’).

4.       Elamo-Dravidian: The 2nd sg. stem *t^hi-/*t^he- is found in Elamite in the 2nd sg. and pl. personal class marker -t(i/a) (cf. Khačikjan 1998:34) and in Dravidian in, for example, the Parji appositional marker -t of the 2nd sg. in pronominalized nouns and as a verb suffix of the 2nd sg.

5.       Altaic: The 1st sg. stem *mi- has become bi ‘I’ in the Altaic daughter languages, while the 1st pl. stem *ma- has become ba in Mongolian (= 1st pl. exclusive); the initial *m- is preserved in the oblique cases, however; the 2nd sg. stem *t^hi- has become Di ‘you’ in Mongolian.

6.       Sumerian: ma(-e), me-a, me-e ‘I’ are Emesal forms; -me is a 1st pl. possessive suffix, ‘our’; -zu is a 2nd sg. possessive suffix, ‘your’.

7.       Etruscan: The 1st sg. stem *mi-/*me- is preserved in (nominative) mi ‘I’, (accusative) mini ‘me’; the 2nd sg. stem may be preserved in the pronoun stem †i, but this is uncertain since the meaning of the Etruscan form is unknown — however, the 2nd sg. stem *t^hi-/*t^he- is clearly reflected in the Etruscan verbal imperative endings -ti, -θ, -θi.

8.       Chukchi-Kamchatkan: The pronouns of the 1st and 2nd persons sg. and pl. are as follows in Chukchi:

                                                Singular                                 Plural

                                1              γə-m                                       mu-ri

                                2              γə-t                                         tu-ri

9.       Gilyak: The 1st pl. inclusive stem *ma-/*mə- is preserved in the 1st pl. inclusive pronoun me-r, mi-r ‘we’ (note also 1st dual me-ge, me-gi); the 1st plural stem *na-/*nə- is found in the 1st pl. exclusive pronoun ńyŋ ‘we’; the 2nd sg. stem *t^hi-/*t^he- is preserved in the 2nd sg. pronoun či ‘you’. (The forms cited are from the Amur dialect [cf. Gruzdeva 1998:25—26].)

10.    Eskimo-Aleut: The 1st sg. stem *mi-/*me- is preserved in the West Greenlandic 1st sg. relative possessive suffix -ma, while the 2nd sg. stem *t^hi-/ *t^he- is preserved in the 2nd sg. absolutive possessive suffix -(i)t. The plural forms are -ma and -tit respectively.

B. Demonstrative Pronoun Stems

NOTES:

1.       Indo-European: The stem *d^yi-/*d^ye- is only preserved as a suffixed particle   *-d^he; the stem *ne-/*no- has a derivative *ʔe-no-/*ʔo-no-.

2.       Altaic: The stem *t^ha-/*t^hə- is used as the distant demonstrative in Altaic: Mongolian (nom. sg.) tere (< *te-r-e) ‘that’, (nom. pl.) tede ‘those’; Tungus (Solon) tari ‘that’; Manchu tere ‘that’.

3.       Sumerian: The demonstrative stem *ʔi-/*ʔe- is found in e ‘hither, here’.

4.       Etruscan: The proximate stem *t^ha-/*t^hə- is preserved in ita, ta ‘this’; the stem *k^ha-/*k^hə- is preserved in eca (archaic ika), ca ‘this’.

5.       Gilyak: The proximate stem *t^ha-/*t^hə- is preserved in (proximate) tyd' ‘this (the nearest to the speaker, visible and available in the present situation)’; the stem *k^ha-/*k^hə- is preserved in kud' ‘that (absent in the present situation, formerly referred to in the previous discourse)’. (The forms cited are from the Amur dialect.)

6.       Eskimo-Aleut: The stem *t^ha-/*t^hə- is preserved in the Inuit (also called Inupiaq) prefix ta-, which may be added to any demonstrative form whose coreferent has already been focused.

C. Relative and Interrogative Stems

NOTES:

1.       Kartvelian: The relative/interrogative stem *ʔya- is found in Svan (inter-rogative) jär ‘who?’, (relative) jerw\j ‘who’, (indefinite) jer ‘somebody, something’.

2.       Altaic: The interrogative stem *mi-/*me- is found in the Turkish interrogative particles mi, mı, mu, mü and in the Middle Mongolian suffixed interrogative particle -mu, -mi.

3.       Sumerian: The interrogative stem *mi-/*me- occurs in me-na-àm ‘when?’,   me-a ‘where?’, me-šè ‘where to?’.

4.       Chukchi-Kamchatkan: The interrogative stem *mi-/*me- is preserved in me-in ‘who?’.

5.       Eskimo-Aleut: The interrogative stem *k^wha-/*k^whə- is preserved in the Proto-Eskimo interrogative pronoun *ki(na) ‘who?’ and in *qa–a ‘when?’, *qavcit ‘how many?’, *qaku ‘when (in future)?’. The interrogative stem *mi-/*me- is preserved in the Proto-Eskimo enclitic particle *mi ‘what about?’.

Reconstructing Proto-Nostratic Online; Comparative Phonology, Morphology, and Vocabulary

CHAPTER TWO

A Survey of the Nostratic Languages

2.1. Indo-European

The Indo-European (in German, Indogermanisch — occasionally translated as “Indo-Germanic” in older works) language family includes the following branches: Anatolian (Hittite-Luwian), Italic, Celtic, Germanic, Tocharian, Greek, Baltic, Slavic, Albanian, Armenian, and Indo-Iranian. There are also a number of poorly-attested Indo-European daughter languages such as Thracian, Phrygian, Venetic, Illyrian, Ligurian, and several others. Phrygian may be the ancestor of Armenian, but this is not absolutely certain. Indo-European languages cover all of Europe except for Basque (found in northern Spain and the southwestern corner of France), Turkish (found in the Balkans), and Uralic (Finnish, Estonian, Hungarian, and several others with extremely small numbers of speakers), modern Iran, parts of Central Asia north of Iran, Afghanistan, and northern and central India. European colonization has also spread Indo-European languages to the New World, where they have mostly supplanted Native American languages, to Australia and New Zealand, and to large parts of Africa and Asia, where they are used as languages of administration and/or learning. The extinct Hittite and Luwian (along with Palaic, Hieroglyphic Luwian, Lycian, Lydian, Carian, and several other poorly-attested dialects and/or languages) were spoken in what is now Turkey, while the Tocharian dialects, which are also extinct, were spoken in what is now the Xīnjiāng (Sinkiang; formerly called Chinese Turkestan) Uighur Autonomous Region (Xīnjiāng Wéiwú’ěr Zìzhìqū) of the People’s Republic of China (Zhōnghuá Rénmín Gònghéguó).

The Indo-European language family has been subjected to thorough study for the past two centuries, and there is broad agreement among scholars on essentials, which is not to say that all problems have been resolved or that there are still not controversial issues. Several languages have extremely old records and/or literatures, such as Hittite, whose earliest records go back to around 1800 BCE, though the majority of documents date from 1500 to 1200 BCE; Mycenaean Greek, whose earliest inscriptions date from 1300 BCE; Sanskrit, with the oldest part of the Rig-Veda (composed in an archaic dialect of Old Indic) probably going back as far as 1200 BCE; Avestan, the liturgical language of Zoroastrianism, whose most ancient scriptures date from about 600 BCE; Old Persian, which begins with the Achaemenid Records from about 500 to 400 BCE; and Italic, with the oldest Latin inscription dating from the sixth century BCE, and with the earliest Oscan-Umbrian records dating from about the fifth century BCE. Records do not begin to appear for the other Indo-European daughter languages until the middle to later half of the first millennium CE.

Two large dialect groups are conventionally recognized: (A) the so-called “centum” languages and (B) the so-called “satəm” languages. This dialectal division is based upon the different treatment of the gutturals in each group. In the satəm languages, sibilants (s and z), palato-alveolar fricatives (š and ²), and affricates correspond to velars in the centum languages, while velars and affricates in the former group correspond to reflexes of earlier labiovelars in the latter group. There are other correspondences as well, found in a small number of examples, in which velars in the centum languages correspond to velars in the satəm languages. Though much attention has been devoted in the literature to this division, its significance is greatly overrated.

Morphologically, Proto-Indo-European was a highly inflected language — except for particles, conjunctions, and certain quasi-adverbial forms, all words were inflected. The basic structure of inflected words was as follows: root + suffix (one or more) + inflectional ending. A notable morphophonemic characteristic was the extensive use of a system of vocalic alternations (“Ablaut” in German) as a means to mark morphological distinctions. Verbs were strongly differentiated from nouns. For nouns and adjectives, three genders, three numbers, and as many as eight cases have been reconstructed (mainly on the basis of what is found in Classical Sanskrit), though it is doubtful that all of these features were ancient — it is indeed possible to discern several chronological layers of development. The traditional reconstruction of the Proto-Indo-European verbal system sets up two voices, four moods, and as many as six tenses. Syntactically, Proto-Indo-European seems to have had many of the characteristics of an SOV language, though there must, no doubt, have been a great deal of flexibility in basic word order patterning. Proto-Indo-European morphology is discussed at length in Chapter 18 of this book, while earlier developments are discussed in Chapter 19.

It is generally agreed that the homeland of the Indo-Europeans is to be located to the north of and between the Black and Caspian Seas (cf. Anthony 2007). Alternative proposals are far less convincing. See Chapter 13 for more information about homelands.

The subgrouping of the Indo-European daughter languages has long been controversial. Though Sturtevant (following a suggestion by Emil Forrer) attempted to show that the Anatolian languages were the first to split off from the remainder of the Indo-European speech community, up until recently, most Indo-Europeanists did not follow him on this (a notable exception being Warren Cowgill). Sturtevant renamed the parent language “Indo-Hittite” to reflect this early split. The question about whether Baltic and Slavic are two independent branches or whether they are descended from a common Balto-Slavic is still contentious, as is the question of Italo-Celtic unity. Recently, Donald Ringe and a group of linguists from the University of Pennsylvania have returned to the problem of subgrouping. By using a computational cladistic model, they have arrived at the following conclusions (Ringe—Warnow—Taylor—Michailov—Levison 1998:406—407):

The important features of this tree can be summarized as follows. The Indo-Hittite hypothesis, according to which Anatolian is one first-order subgroup of the IE family and all other branches together are the other first-order subgroup is supported — but by only one character, the presence of a thematic aorist in the verb system... The satem core emerges as an extremely robust subgoup, always with the traditional internal structure (which is not surprising). More interestingly, there is always a subgroup including Greek and Armenian, as has been suspected in the past… Most interesting of all, Italo-Celtic emerges as a robust subgroup, as suggested by Jasanoff 1994.

They further note that Tocharian also split off from the rest of the speech community at a very early date — it was the next branch to break away after Anatolian. Finally, they conclude that Germanic was originally part of the dialect continuum that included Balto-Slavic but that it later was in contact with and shared several common developments with Pre-Proto-Celtic and Pre-Proto-Italic.

The conclusions reached by Ringe and his colleagues are both sober and persuasive. Consequently, it is their views on the subgrouping of the Indo-European daughter languages that are followed in this book.

2.2. Kartvelian

Kartvelian (also referred to as South Caucasian), which is one of the three indigenous language families of the Caucasus Mountains, includes the following languages: Georgian, Mingrelian, Laz, and Svan. These languages fall into two main groupings, namely, Svan, on the one hand, and Georgian, Laz, and Mingrelian, on the other. Laz and Mingrelian, in turn, form the Zan subbranch. Svan preserves many archaic characteristics. Except for Laz, which is spoken in Turkey, and the Ingilouri dialect of Georgian, which is spoken in Azerbaijan, the Kartvelian languages are spoken in the westernmost parts of the Caucasus Mountains within the borders of the Republic of Georgia.

The Kartvelian family tree may be represented as follows (cf. Tuite 1997:4; Schmidt 1962:13; Hewitt 1995:2; Gamkrelidze—Mačavariani 1982:20; Fähnrich—Sardshweladse 1995:5):

Georgian, which has its own distinctive alphabet, has a literary tradition going back 1500 years, the earliest text being a translation of the Bible dating from the 5th century CE, only fragments of which still exist. The early literature was exclusively religious, and it was only with the so-called “Golden Age” (12th century CE) that secular literature began to appear. There are a number of distinct Georgian dialects, which differ not only in vocabulary and phonology but also in morphology and syntax.

A notable feature of Kartvelian phonology is the existence of complex consonant clusters — Georgian, for example, tolerates 740 initial clusters, which can have upwards of six members (Fähnrich 1993:20 lists eight), and 244 final clusters. In Svan, on the other hand, initial consonant clusters are far less complex than in Georgian, while final clusters can be far more complex. Old Georgian had both voiceless and glottalized uvular stops, but only the glottalized member is retained in Modern Georgian. Both are still found in Svan. Unlike Georgian, Svan does not distinguish /v/ and /w/ as distinct phonemes — it only has /w/.

Morphologically, the Kartvelian languages are all highly inflected; Georgian, for example, has six basic grammatical cases as well as eleven secondary cases. A notable characteristic of noun declension is the distinction of ergative and absolutive cases; the ergative case is used to mark the subject of transitive verbs, while the absolutive case is used to mark direct objects and the subject of intransitive verbs. It is the dative case, however, that is used to mark the subject of so-called “inverted verbs”. There are several other departures from canonical ergative-type constructions, so much so in Mingrelian, for instance, that this language no longer possesses any true ergative features. Adjectives normally precede the nouns they modify. Postpositions are the rule. Verb morphology is particularly complicated — for example, Deeters lists eleven distinctive functional elements that may be arrayed around a given verb root, though they may not all appear simultaneously; the overall scheme is as follows:

                        1. Preverb(s)

                        2. Personal prefix(es) (subjective or objective)

                        3. Character or version vowel

                                        ROOT

                        4. Passive suffix

                        5. Causative suffix(es)

                        6. Plural suffix (for nominative-absolutive noun)

                        7. Present stem formant

                        8. Imperfect suffix

                        9. Mood vowel

                        10. Personal ending

                        11. Subjective plural suffix

Syntactically, the predominant word order is SOV, though SVO is not uncommon.

2.3. Afrasian

Afrasian (also called Afroasiatic, Hamito-Semitic, Semito-Hamitic, Erythraic, and Lisramic) includes the following branches: Semitic, Egyptian, (Libyco-)Berber, Cushitic, Omotic, Chadic, and Ongota (for an attempt at subgrouping, see Chapter 7 of this book). Except for Semitic, all of the Afrasian languages are found in northern and eastern Africa. In ancient times, Semitic was primarily located in the Near East, but Muslim conquests beginning in the 7th century CE have spread a single Semitic language, namely, Arabic, across the greater part of northern Africa, where it has totally replaced Egyptian (Coptic) as a spoken language and has greatly restricted, but has not totally supplanted Berber, which now exists only in isolated pockets. Though no longer spoken, Coptic is still used as the liturgical language of the Christian Coptic Church in Egypt. It is estimated that there are at least 250 languages in the family.

The following chronology may be established for the branching off of the various branches of Afrasian (cf. Ehret 1995:483—490): Omotic, which appears to contain many distinctive features, must have been the first branch to split from the rest of the Afrasian speech community. The next split was between Cushitic on the one hand and Chadic, Egyptian, Berber, and Semitic on the other. Finally, Chadic split off, followed by Egyptian and Berber. Within Semitic, Akkadian is the most archaic language as a whole, though Arabic preserves the original phonological structure better than any of the other Semitic languages. Tuareg is usually viewed as the most conservative Berber language, as are Beja (also called BeTawye) and Saho-Afar within Cushitic.

The study of Afrasian as a whole is still not far advanced. Several branches, such as Semitic and Egyptian, for example, have written records going back many millennia and have been scientifically investigated rather thoroughly, while other Afrasian languages are scarcely even known. Egyptian, whose earliest inscriptions date from about 3400 BCE, and Akkadian, whose earliest inscription dates from the reign of King Lugalzagesi of Uruk (roughly 2352 to 2327 BCE), were the languages of great civilizations of antiquity, while Hebrew and Arabic are the liturgical languages of Judaism and Islam respectively. The Semitic languages exhibit great internal consistency as a group, with fairly straightforward correspondences in morphology, with close resemblance in their phonological systems, and with a large common vocabulary. In contrast, the internal divisions in the other branches, except for Egyptian, of course, which is a single language, are far more pronounced.

Proto-Afrasian was most likely highly inflected. It is simply not possible, however, given the present level of knowledge, to reconstruct the morphological structure of the parent language in detail, though some common features (such as the distinction of grammatical gender, the existence of two verbal conjugation systems, at least one of which, namely, the prefix conjugation, probably goes back to Proto-Afrasian, and a common set of pronominal stems) have been noted. Syntactically, the classical Semitic languages, Egyptian, and the Berber languages are VSO, the majority of the Cushitic languages are SOV, and most Chadic languages are SVO.

2.4. Uralic-Yukaghir

As the name implies, Uralic-Yukaghir has two divisions, namely, Uralic and Yukaghir. Yukaghir consists of a single branch, while Uralic is divided into Finno-Ugrian and Samoyed. There are about 30 Uralic languages. The internal subgrouping of the Uralic languages is still not fully settled. Finno-Ugrian is thought to have become separated from Samoyed some time between 4,000 to 2,000 BCE. Yukaghir is located in northeastern Siberia, while Uralic languages are spread across northern Eurasia, from Scandinavia and central Europe in the west to north-central Siberia east of the Ural Mountains in the east.

Hungarian is the first Uralic language for which there are written records. Though the first printed text did not appear until 1527, Hungarian words are cited as early as the 9th and 10th centuries CE in Arabic and Byzantine documents. Finnish literature did not begin until 1548, with a translation of the Bible. An Estonian translation of the Bible first appeared in 1632. Yukaghir has no written literature.

Morphologically, the Uralic languages are predominantly agglutinating, though many of the modern languages, especially Estonian, which has innovated considerably, have deviated from the original type. Proto-Uralic nominal inflection had at least three numbers (singular, dual, and plural), two grammatical cases (accusative and genitive), and three local cases (dative, locative, and ablative). Verb morphology distinguished two conjugational types, namely, subjective and objective. A large number of suffixes existed, each with its own distinctive morphological function. The original syntactic structure seems to have been SOV, and this is fairly well preserved in the modern Samoyed and Ob-Ugric languages (Ostyak [Xanty] and Vogul [Mansi]) and Cheremis (Mari). The basic word order in the other languages is SVO, though, as a general rule, word order in all of the Uralic languages is rather flexible. Hungarian stands apart, word order being determined here more by topic-comment considerations than in the other Uralic languages, so that neither SOV nor SVO can be said to be dominant. Yukaghir is also basically agglutinating, though a certain amount of fusion has taken place in the verb. There are few prefixes but numerous suffixes. Postpositions are the rule. Syntactically, the basic word order is SOV.

2.5. Elamo-Dravidian

Dravidian has four branches: South Dravidian, South-Central Dravidian, Central Dravidian, and North Dravidian. Though the vast majority of Dravidian languages are concentrated in southern India, there are also pockets of Dravidian in northern India, in Pakistan, in Nepal, in northern and eastern Sri Lanka, and on the Maldive Islands. At least 25 Dravidian languages are spoken. There is still uncertainty over the subgrouping of several languages. Elamite, which is now extinct, was located primarily in southwestern Iran in the vicinity of the Zagros Mountains as well as the adjacent plains of Khuzistan and to the south along the coast of the Persian Gulf. There is good reason to believe that Elamite once occupied all or nearly all of the Iranian plateau. The inscriptions of the Indus Valley (Harappan) Civilization may have been written in an early Dravidian language (cf. Fairservis 1992:14—23 and Parpola 1994; but see Zide—Zvelebil [eds.] 1976 for a critical assessment of attempts to decipher the Indus Valley script).

The earliest Elamite text is the “Treaty of Narām-Sin”, which dates from before 2200 BCE. After that, only cuneiform texts composed in a slightly deviant form of Akkadian are found until around 1300 BCE, when Elamite cuneiform texts begin to appear. The literature of the Dravidian languages, especially Tamil, is enormous. In addition to Tamil, Malayalam, KannaTa, and Telugu are fully-developed literary languages, while the remaining Dravidian languages have extensive oral traditions. The oldest Tamil literature probably dates from around the 2nd or 3rd centuries CE.

Morphologically, the Dravidian languages are agglutinating. The basic root type was monosyllabic, though there is some indication that an extremely small number of bisyllabic roots may have to be reconstructed at the Proto-Dravidian level as well. This is, however, by no means certain, and it is best at present to regard Proto-Dravidian roots as exclusively monosyllabic. Inflectional categorization was achieved by means of suffixes added directly to the lexical roots or to the lexical roots extended by means of derivational suffixes. Prefixes were not used. Any vowel, long or short, could appear in a root, but only a, i, or u could appear in a suffix. Two basic parts of speech were differentiated in Proto-Dravidian: (A) nominals, which included nouns and adjectives, and (B) verbs. Nouns were inflected for case, person, number, and gender. Eight cases (nominative, accusative, sociative, dative, genitive, instrumental, locative, and ablative), two numbers (singular and plural), and two genders (animate and inanimate) are assumed to have existed in Proto-Dravidian. There were separate first person plural inclusive and exclusive pronouns. Verbs were inflected for tense and person. There were two tenses (past and non-past) and two moods (modal and indicative). Indeclinables existed as a separate stem type distinct from nouns and verbs. Syntactically, the basic word order was SOV.

Elamite was also agglutinating. Three basic parts of speech were differentiated: (A) verbs, (B) nominals, and (C) indeclinables. The basic verbal stem form was (C)VC(V). Grammatical categorization was achieved by means of suffixation. In the nominal stems, case relationships were mostly indicated by the use of postpositions. Verb morphology was extremely simple. Word order structure was SOV. Cf. Grillot-Susini 1987; Hinz—Koch 1987; Khačikjan 1998; Paper 1955; McAlpin 1981; Reiner 1969; Stolper 2004.

2.6. Altaic

Altaic has at least three branches: Mongolian, (Manchu-)Tungus, and (Chuvash-) Turkic. Mongolian languages are spoken in Mongolia proper, in northern China in the so-called “Inner Mongolian Autonomous Region”, in eastern Siberia in areas bordering on Mongolia, and (Moghol) in Afghanistan; (Manchu-)Tungus languages are spoken in eastern Siberia and (Manchu) in northeastern China in what was formerly known as Manchuria, but which is now divided between the provinces of Hēilóngjiāng, Jílín, and Liáoníng and the Inner Mongolia Autonomous Region (Něi Měngz Zìzhìqū) and is populated mostly by ethnic Chinese (Hàn); and (Chuvash-) Turkic languages are spoken in a large, discontinuous band, stretching from Turkey in the west, across Central Asia and western China in the middle, and on to northeastern Siberia in the east. Korean and Japanese-Ryukyuan are also considered by some specialists to be Altaic languages.

The oldest Turkic texts are the Orkhon inscriptions of the Kül-Tegin stele, written in a type of runic and dating from 735 CE. The earliest Mongolian inscription is only five lines long and mentions the nephew of Genghis Khan (1154-1227 CE). The longest early literary work in Mongolian is The Secret History of the Mongols, an imperial chronicle written in Uighur script and thought to date from around 1240 CE. Few documents in Mongolian have survived from the period between the composition of that chronicle and the 17th century. Beginning with the 17th century, however, a rich Buddhist and historical literature began to appear. There is an extensive literature in Manchu, but most of it is of relatively recent origin and consists mainly of translations from Chinese sources.

The phonological systems of the Altaic languages are comparatively uncomplicated. Vowel harmony is a common phonological characteristic, though, in the (Chuvash-)Turkic and Mongolian branches, it is based on a front ~ back contrast, while, in the (Manchu-)Tungus branch, it is based on a high ~ low contrast. It is difficult to reconstruct the common Altaic morphological system in detail since there are deep differences among the descendant languages (the resemblances are more observable in vocabulary and syntax), though there are indeed a few common morphological elements, and all of the Altaic languages belong to the same type. Morphologically, the Altaic languages are typically agglutinating in structure. Though all Altaic languages make extensive use of suffixes, only a few of them are common to all three branches, one notable common feature here being the use of possessive suffixes. Nouns and verbs are clearly differentiated, though not as sharply as in Indo-European. There is a common stock of pronominal stems, and all Altaic languages use postpositions. Syntactically, the original structure was SOV, and this is well preserved in the modern languages, especially the Turkic languages, which are fairly strict in this regard, while more flexibility is found in the Mongolian and (Manchu-)Tungus languages.

2.7. Chukchi-Kamchatkan

The Chukchi-Kamchatkan family includes the following languages: Chukchi, Koryak, Kerek, Alyutor, and Kamchadal (also called Itelmen or Itelmic). Koryak, Kerek, and Alyutor are extremely close as a group, and these, in turn, are close to Chukchi. Kamchadal, which is now on the verge of extinction, stands apart from the others. The Chukchi-Kamchatkan languages are found in the extreme northeast corner of Siberia in the Chukotka and Kamchatka peninsulas. Though written languages were developed for Chukchi, Koryak, and Kamchadal in the 1930’s, only Chukchi is still being used in publications and education.

Chukchi consonantism is fairly simple, there being only 14 distinct consonant phonemes, while that of Koryak is more complex than Chukchi, and that of Kamchadal is even more complex than either Chukchi or Koryak, containing both plain and glottalized stops, voiced and voiceless fricatives, and three lateral phonemes. A notable characteristic of Chukchi phonology is a system of vowel harmony based on a height contrast. In this system, vowels are classified as either “dominant” (e, a, o) or “recessive” (i, e, u) — note that the vowel e appears in both series. The presence of a dominant vowel in any morpheme in a word conditions the change of any recessive vowels in the word to their corresponding dominant counterparts. A similar system is partially preserved in Koryak.

The Chukchi-Kamchatkan languages are agglutinating. In Chukchi, however, some fusion has occurred, particularly in the verb. Chukchi nouns distinguish singular from plural. There are relatively few cases. Typical of all Chukchi-Kamchatkan languages is case marking of subjects and direct objects on the basis of an ergative-absolutive system. Chukchi and Koryak also exhibit a certain degree of incorporation, though it is not as extensively used as in Eskimo-Aleut. Verbs clearly distinguish between transitive and intransitive, with the ergative being used in conjunction with transitive verbs. Chukchi employs postpositions exclusively. Chukchi word order is rather free, with OV being slightly more predominant than VO.

2.8. Gilyak

Gilyak (also called Nivkh) is usually considered to be a single language, but the two main dialects, namely, the Amur dialect, on the one hand, and the Sakhalin (or Eastern) dialect, on the other, are not mutually intelligible. Of the two, the Sakhalin dialect is more archaic. The Gilyaks are found on the lower reaches of the Amur River and on Sakhalin Island. Though a written language was developed for the Amur dialect in the 1930’s, next to nothing has appeared in it.

Gilyak tolerates highly complex consonant clusters. Furthermore, initial consonants undergo various alternations, which are conditioned both by the final segment of the preceding word and by syntactical considerations. In contrast, the vowel system is fairly simple.

Gilyak morphology is typologically similar to that found in the Altaic languages. Noun morphology is uncomplicated. Only a few cases are distinguished, including several basic spatial cases. Singular and plural are also distinguished. A system of numeral classifiers has been developed. In the pronouns, there are separate forms for first person dual and plural, while the first person plural, in turn, has a distinction between inclusive (mer) and exclusive (~‹–). Verb morphology is also simple, though one notable feature worth mentioning is the wide range of non-finite gerunds that can occur. Gilyak possesses postpositions but no prepositions. Basic word order structure is SOV.

2.9. Eskimo-Aleut

As the name implies, Eskimo-Aleut has two branches: Eskimo and Aleut. The Aleut dialects are mutually intelligible. However, this is not the case with the Eskimo dialects. Two main Eskimo dialect groups are distinguished, namely, Yupik and Inuit (also called Inupiaq). Yupik speakers are concentrated in southwestern Alaska, beginning at Norton Sound and extending southward along the western and southern coasts and inland. An extremely small enclave of Yupik speakers is found in northeastern Siberia as well. Inuit speakers are found north of Norton Sound all the way to the northern coast of Alaska and extending eastward across all of the northernmost parts of Canada and on into Greenland. Aleut is spoken on the Aleutian Islands and the Commander Islands.

The Proto-Eskimo vowel system was relatively simple (Proto-Eskimo had only four vowels: *i, *a, *u, *‹ — phonemic length probably did not exist), while the consonant system resembled that of Proto-Uralic. The phonological systems found in the Eskimo dialects are far more complex than that of Proto-Eskimo. In contrast, Aleut phonology is less complicated. Nouns differentiate between singular, dual, and plural. The case system is reminiscent of that found in Chukchi-Kamchatkan, though it differs by using suffixes to indicate the plural. The verb makes no tense distinctions but has four moods and separate transitive and intransitive conjugations. The absolutive case is used as the subject of intransitive verbs and as the direct object of transitive verbs, while a different case is used as the subject of transitive verbs. Conjunctions and other particles are absent in most Eskimo dialects. A notable characteristic is that incorporation has been developed to such an extent that whole phrases may be expressed in a single word.

2.10. Etruscan

Etruscan was spoken in central and northern Italy. Its earliest texts date from the 7th century BCE, and it probably ceased to be a spoken language around the first half of the first century CE, being replaced by Latin. It was written in a special alphabet derived from Greek. There are about 13,000 Etruscan inscriptions currently known, most of which are found on tombs and sarcophagi or on artifacts. These inscriptions are extremely short, repetitive, and formulaic in nature. A few longer texts also exist, such as the Pyrgi quasi-bilingual (Etruscan and Phoenician) discovered in 1964. Unfortunately, no literary texts have survived. Though there still remain problems, the majority of what has survived can be read and understood.

Etruscan is now known to be related to the poorly-attested Lemnian (spoken on the island of Lemnos) and to Raetic (spoken in northeastern Italy in present-day Tyrol). Together, they form the Tyrrhenian language family.

The Etruscan phonological system was composed of plain voiceless stops, voiceless aspirates, and fricatives, as well as two nasals (m and n), two liquids (l and r), and h. There were no voiced stops. There were only four vowels (a, e, i, u).

Etruscan was an inflectional language. Though there probably was no grammatical gender, special suffixes were used to indicate females. Etruscan nouns and adjectives distinguished several cases as well as two numbers (singular and plural). Verb morphology is not as well known due to the nature of the material that has survived.

2.11. Sumerian

Sumerian, which is now extinct, was spoken in southern Iraq, extending from around Babylon in its northernmost limits to the tip of the Persian Gulf in the south. From the time of the earliest texts, several dialects can be distinguished — the two most important dialects are called eme-…ir÷ý and eme-sal (eme means ‘speech, language’) by the Sumerians themselves. Moreover, during the three thousand or so years in which Sumerian was recorded, several distinct stages of development can be discerned — Old Sumerian, Neo-Sumerian, Old Babylonian Sumerian, etc. As noted in the previous chapter, Sumerian is not a Nostratic daughter language but is distantly related to Nostratic.

The earliest Sumerian inscriptions date from around 3200 BCE, though the oldest intelligible literary texts date from about 2600 BCE, and the language was probably still spoken as late as the 3rd century BCE. The Sumerian writing system was based exclusively on the cuneiform syllabary, which exhibits several marked stages of development over the course of Sumerian literary history.

Though the Sumerian phonological system was simple, there are still many uncertainties about underlying phonemic distinctions. For example, the traditional transcription shows a voiced ~ voiceless contrast in the stops, but this may well have been a voiceless unaspirated ~ voiceless aspirated contrast instead. There is still not, even after more than a century of intensive study, widespread agreement among experts in the field on many fundamental questions of Sumerian grammar. Nevertheless, the overall structure is clear. Morphologically, Sumerian was an agglutinating language. Three word classes were distinguished: (A) nouns, (B) verbs, and (C) adjectives. Though grammatical gender in the strictest sense did not exist, nouns fell into two classes, namely, animate and inanimate, which were only differentiated in 3rd person actor verbal and possessive pronoun affixes and in the relative pronoun. Ten cases and two numbers (singular and plural) were distinguished. The plural was indicated either by means of the suffix -ene, which was used only with animate nouns, or by reduplication. In later texts, the plural could also be indicated by the form hi-a, which was used with inanimate nouns and which was originally an independent word meaning ‘mixed, various, unspecified’, or by -me-eš, which was properly the enclitic copula with plural suffix. Sumerian differentiated between ergative and absolutive in nouns. In pronouns, however, the patterning was that of a nominative-accusative system. Sumerian verbs were formed by adding various prefixes and/or affixes directly to the verbal root. Verbal constructions fell into one of two categories, namely, finite forms or non-finite forms. Finite verbal stems distinguished three conjugational types: (A) the intransitive conjugation, (B) the transitive hamsu conjugation, and (C) the transitive marû conjugation. Intransitive forms were noted by means of pronominal suffixes, while transitive forms were noted by means of either prefixes, suffixes, or both. The basic word order structure was SOV.

Reconstructing Proto-Nostratic Online; Comparative Phonology, Morphology, and Vocabulary

References