Twenty years ago I discussed the Hittite development of the original diphthongs. My views on the vowel system of Proto-Indo-Hittite and of Hittite have undergone several important changes in the past twenty years; but I still hold essentially the same opinion as then about the Hittite development of IH long diphthongs, and for about the same reasons. I find, however, that more or less serious doubts prevail about this feature of Hittite phonology; I must try again.

The original key to the interpretation of Hittite cuneiform writing was, of course, furnished by our knowledge of Akkadian writing. Since, however, Akkadian possessed no diphthongs, we cannot expect to learn from that source how the Hittites wrote whatever diphthongs their language may have possessed.

In the introduction to his grammar of literary Japanese, published in 1877, W. G. Aston said of the relation of spoken to written Japanese: ‘The spoken dialect of Japan differs so considerably in its grammar from the written idiom that it almost deserves to be regarded as a new language.’ A few years later B. H. Chamberlain said: ‘The business of life—whether in books, letters, or newspapers—is consistently carried on in a dialect ... whose grammar differs notably from that of spoken speech.’ And in his Handbook of Colloquial Japanese, he stressed the distinction thus: ‘A peculiarly intricate system of writing is not the sole barrier that divides the Colloquial from the language of books. The Japanese, like other Easterners, still remain at the stage in which we were during the Middle Ages—they do not write as they speak, but use an antiquated and partly artificial dialect whenever they put pen to paper. This is the so-called “Written Language”.’

The Latin alphabet is one of the most conspicuous items of cultural inheritance derived by the West from ancient Rome. It was not necessarily the most ideally suited to the sounds of the various European idioms, but by main force it was adapted to each of them and made to serve. Nowhere do we have a first-hand account by anyone who created such native orthographies in the Middle Ages, so we must draw our conclusions concerning the procedure by observing the results. It is clear that the scribes often did little more than carry over their habits of writing from Latin to the new tongues, without considering too closely the rationale of what they were doing. But there were also innovators and experimenters, such as the German Notker and the English Orm, who deliberately sought to better the quality of the traditional orthography. The only one of this tribe who has left an account of orthographic principles and a persuasive program of reform is the anonymous Icelandic author of the so-called First Grammatical Treatise, written in the middle of the twelfth century. This First Grammarian (here abbreviated FG) intended by his Treatise (here abbreviated FGT) to establish a system of writing that would identify to the reader each significant shade of sound. His accomplishment in this respect is unmatched in medieval Europe and is sufficiently extraordinary to merit the attention of modern students of language, to whom the relation of sound and spelling is again a crucial problem. He stands close to the very threshold of Scandinavian writing in the Latin alphabet, being probably a son of the very generation that first entrusted the Norse tongue to the new letters. In the words of the Icelandic scholar Björn M. Ólsen, the FGT ‘is for its time a notable, indeed, as far as I know, a unique work. It testifies to the author's delicate perception of sound and capacity for observation, his extensive learning and natural perspicacity, but above all to his independence of spirit.‘

The origin and development of speech habits in individuals, through their life cycles from birth to death, can appropriately be termed linguistic ontogeny. In contrast to this, the subject matter of historical linguistics—changes through decades and centuries in the speech patterns of communities—is of course linguistic phylogeny. The present paper deals with certain relations between these two: specifically, with the mechanisms whereby continuity of linguistic tradition is maintained in a community in the face of the constant turnover of population through birth and death, immigration and emigration. Jespersen has discussed this problem, and has connected it with historical linguistics by asking what relation there may be between these mechanisms and the fact that languages change in the course of time. But Jespersen, as is well known, did not accept the assumption that phonetic change is regular.

This study attempts a structural analysis of the inflected forms of 8th-centur Japanese, applying the principles and techniques of current descriptive linguistic theory. Although an interest in historical linguistics has prompted this undertaking, this is not a historical study, but a purely descriptive one. Grammars based on this same material have been written by native and foreign scholars. Although these are important works which have contributed greatly to the general knowledge of ancient Japanese, certain limitations are found in them. These should be attributed, not to the particular grammarians but to the period in which the grammars were written. Descriptive linguistics as we know it today is hardly two decades old. Its purpose and methodology differ radically from those found in earlier linguistic studies. Any criticism made of these particular studies of the Japanese language can, in general, be made of many grammars which are contemporaneous with them.

[Reconstruction by the comparative method (as distinct from internal reconstruction based on alternations between phonemes in a paradigm) is essentially a problem in phonemics, in which the place of allophones is taken by sets of sound correspondences that are partially alike (share one component) and in complementary distribution. The principle is illustrated by the IE dental and labial stops as reconstructed from Sanskrit and Germanic, by IE *s in Greek and Latin, and by the IE aspirates in Italic]

An essential part of Bloomfield's definition of linguistic forms and constructions is the attachment to them of stimulus-reaction features, which are his definition of meaning. The linguistic scholar is still bedeviled by the many problems involved in talking about meaning, and he still lacks a body of dogma to which he can refer when meaning is talked about. This paper is an attempt to explore one of the problems and to establish a postulate which may be added to Bloomfield's Set of Postulates.

Leonard Bloomfield, in his book Language, makes the statement:

Two later writers, discussing the reconstruction of earlier stages of related languages, have made relevant statements in this connection. Bruno Migliorini says

The great number of definitions proposed for the term Vulgar Latin mirrors the plurality of Vulgar Latin theories, which cannot all be correct. They not only do not all coincide, which is but natural, but some of them are radically opposed to each other. Since all of them are based on the same records, the divergence of theories must spring from the interpretation and not from the nature of the evidence.

One may completely discard, as the majority of scholars have done, the idea that spoken Vulgar Latin is the chronological successor (a corruption) of Classical Latin. Apart from this antiquated view, there are, roughly, two types of Vulgar Latin theories. First, there are theories which propose that there was, especially during the Empire and the early Middle Ages, a linguistic unity of popular speech throughout the Roman and Romanized world; according to some scholars this unity dissolved in the 5th or 6th century of our era, according to others not before the end of the 8th or 9th century. Second, there are theories which insist on early dialectalization of Latin, or indeed maintain that there never was, outside of early Latium, a single unified Latin, or anything but a number of local dialects, especially in the Romania outside of Italy.

As early as 1886, Hermann Paul pointed out that all borrowing by one language from another is predicated on some minimum of bilingual mastery of the two languages. For any large-scale borrowing a considerable group of bilinguals has to be assumed. The analysis of borrowing must therefore begin with an analysis of the behavior of bilingual speakers. A vast literature has come into being on the subject of borrowing, particularly in the historical studies of individual languages; but there is still room for discussion of the relationship between the observed behavior of bilingual speakers and the results of borrowing as detected by linguists. Any light that can be thrown on the question by a study of bilingual speakers should be welcome to all students interested in borrowing and in the general linguistic problems associated with this process. In the present article an effort will be made to define more precisely the terminology used in the linguistic analysis of borrowing, and to set up certain hypotheses concerning the process of borrowing. It should then be possible to test these by their usefulness of application to particular studies of bilingualism and borrowing.

[The thesis that the saṃprasaraṇa reduplication is a zero-grade reduplication of a root beginning with a laryngeal, is proved by the study of the representation of the roots belonging to this group in other IE languages, particularly in Greek and Tocharian. By using the results gained from the analysis of the saṃprasaraṇa reduplication, a new explanation of the Attic reduplication is given, and a formula is established which seems satisfactory for most of the IE perfect types: Full grade of the reduplication = C1C2-e-C1C2- > C1-e-C1C2- > e-C1C2.]

In comparative Germanic grammar it is a commonplace assumption that IE ē first became æ in Proto-Germanic and then developed into ā in the parent forms of both North and West Germanic. Prokosch, for example, remarks, Two changes [affecting IE long vowels] are generally ascribed to Primitive Germanic: that of ā to ō and of ē to ǣ. The latter is largely a “Phonetic Interpolation”, i.e., the assumption of a transitional stage between ē and ā. In Gothic, ǣ reverts to ē by a different process of tensing (raising and fronting); in North and West Germanic, ā results; this develops secondarily into ǣ in Old English, and ē in Frisian.'