I wish to express gratitude to Professor Bernard Bloch for suggesting the topic of this dissertation and for guiding my analysis; I am further indebted to him for much of my understanding of linguistic theory. I am grateful also for the kind encouragement and guidance throughout my graduate work of other faculty members at the University of California 1947-9, the Linguistic Institute at the University of Michigan 1948, and Yale University 1949-50: Professors Peter A. Boodberg, Denzel Carr, Y. R. Chao, .J.M. Cowan, Franklin Edgerton, Murray B. Emeneau, Mary R. Haas, Harry Hoijer, and Ralph L. Ward. I have profited from discussions with Professors Isidore Dyen, Elizabeth F. Gardner, Johannes Rahder, and Rulon S. Wells of Yale University. Other influences will be found in the Bibliography. To Dr. Eleanor Harz Jorden I am grateful for permission to use part of her data on numeral-counter compounds; to Dr. Masako Yokoyama of the University of Hawaii, and to Mrs. Miyo Okada, Mr. Toshio Kono, and Mr. Senzo Usui of Yale University, I express my thanks for their kindness in checking certain facts. The Graduate School of Yale University made the completion of this dissertation possible through the grant of a Junior Sterling Fellowship. Part of the dissertation was presented before the Linguistic Institute at Ann Arbor in 1950, which I attended on a Postdoctoral Fellowship from the American Council of Learned Societies, and the present version benefits from suggestions received at that time.

In a study of the neuter plural in Vergil, Dr. John F. Gummere has shown that the use of what is often called the ‘poetic plural‘ is in the Vergilian hexameter largely determined by metrical convenience. The same conclusion was reached with regard to Ovid by Dr. Margaret W. Herr, as the result of an investigation of the poetic or illogical plurals in the Metamorphoses.

A recent article, The importance of language universals, raises a problem that is fundamental to all linguistic science, and is of particular relevance at the present time in the light of current developments in linguistic method. The authors refer to the importance of studying those features of culture that are found to be universal in human society, one of which is language itself. They then suggest (and this is a topic that requires some elaboration) that linguistics and the study of languages will be benefited by an investigation of ‘universals of language’, features of human language exhibited everywhere, in all speech communities. They proceed to suggest and enumerate such universals in the fields of phonology, grammar, and semantics.

The purpose of the present study is to examine the persisting Latinisms in the Poema de mio Cid and other selected Old Spanish literary works, such as Roncesvolles and Elena y María, and to show how tenaciously Latin held its grip upon the emerging Spanish vernacular.

Writing in 1875, William Dwight Whitney refers to the Indo-European family of languages as more clearly drawn than most; but, he continues, some additions may yet be made to it, as perhaps Albanian and Etruscan. His list did not include Armenian, Phrygian, or Thracian, and it included Illyrian only as the possible predecessor of Albanian.

This paper presents a method for the analysis of connected speech (or writing). The method is formal, depending only on the occurrence of morphemes as distinguishable elements; it does not depend upon the analyst's knowledge of the particular meaning of each morpheme. By the same token, the method does not give us any new information about the individual morphemic meanings that are being communicated in the discourse under investigation. But the fact that such new information is not obtained does not mean that we can discover nothing about the discourse but how the grammar of the language is exemplified within it. For even though we use formal procedures akin to those of descriptive linguistics, we can obtain new information about the particular text we are studying, information that goes beyond descriptive linguistics.

Discussion during the last twenty-five years has made the relation of Latin to Osco-Umbrian a problem, where none had previously existed. Until about that time, the assumption of a unity later than West Indo-European—Uritalisch or Proto-Italic—from which both Latin and Osco-Umbrian were derived, was universally accepted; it is still the doctrine to be found in the standard handbooks of Indo-European. During the last quarter-century, however, this hypothesis has been repeatedly called in question by Italian and German linguists, and there are now many who regard Latin and Osco-Umbrian as essentially two separate branches of the Indo-European family. Since this problem, so far as I know, has not been the subject of a full-length treatment in any American journal, a review and an evaluation of the recent arguments may be of interest.

In spite of the delay in the publication of a linguistic atlas of the Iberian Peninsula, the material which is gradually becoming available on individual dialects is so plentiful that lexical areas can be delimited with a fair degree of accuracy, and studies in the reconstruction of sequences of events may be undertaken with small risk. One such problem involves the spread (in the dimensions of space and time) of five entirely unrelated Latin word-families, which, at a late stage of their Hispanic development, seem to have exerted a measure of influence on one another through mutual encroachment or avoidance. This problem is here discussed as a tentative step in the indicated direction, and also as a contribution, based on fresh material, to the general problem of homonymy and near-homonymy as factors in language history.

A number of IE roots beginning with stops or sonants show alternating forms with an added initial s-. Occasionally doublets survive in the same daughter language, with or without differentiation of meaning; examples are Engl. melt and smelt, and Gk. τέγoς and στέγoς, both meaning ‘roof’.

When in 1925 I set about the task of writing an Indo-European comparative dictionary, I was hampered by lack of material on Albanian. I turned aside from the main theme to collect what I could both inside and outside Albania, and my IE materials were enriched thereby. Walde-Pokorny appeared in 1929–32, at a time when my own work was already well advanced. My results, while happily coinciding with theirs at many points, are nevertheless more fully documented, and are drawn up according to a different plan. In this article I offer some of these results as far as they concern Albanian.

In the Greek vocabulary there are a number of words with a prefix ἀ- which have caused considerable difficulty to students of Greek morphology. The problem is usually dismissed by adopting the merely descriptive term ‘prothetic vowel’, even in comparative and historical statements; but this way of avoiding the issue is far from satisfactory.

In comparative Germanic grammar it is conventionally assumed that IE e eu became respectively PGc. i iu when a high front vowel or glide (i ī j) occurred in the following syllable. Apart from differences of opinion regarding the manner in which this development took place, the grammars show very little disagreement on the subject. Nearly all assign the change to Proto-Germanic (or at least pre-Gothic) times, although Schulze in particular has noted an apparent lack of umlaut in early runic inscriptions and foreign spellings of Germanic names. It is also a conventional assumption that IE i u became respectively PGc. e o when a low or mid vowel () occurred originally in the next syllable, provided that neither i ī j nor a cluster of nasal plus consonant intervened. In this instance, however, some scholars have doubted that the change took place in Proto-Germanic times.

In dealing with the origin of Hittite ḫ, Danish ‘laryngealists’ seem reluctant to give up the idea of deriving it from an IE k-sound. This theory originated with Bedřich Hrozny, who, in several IE etymologies for Hittite words, implied that their initial ḫ- reflected an IE guttural. The tentative conclusions of the Czech linguist were soon worked out by Carl J. S. Marstrander, on the assumption that Hitt. ḫ- continues IE *ĝh- (gh). The evidence adduced being most questionable, this view failed to carry conviction; yet other prominent linguists, later on, considered Hitt. ḫ a reflex of IE gutturals. Johannes Friedrich derived Hitt. ḫu-iš-nu-zi ‘cause to live’ from IE *gwī- ‘live’, whereas Holger Pedersen compared Hitt. ḫa-an-te-iz-zi-iš ‘first’ with OIr. cét-, cétne, W cyntaf ‘first’, Gaul. Cintugnatus.

Linguistic geography may simply present the facts about a linguistic detail on which there has been speculation; it may enable a speaker to judge the social acceptability of a linguistic form on the basis of usage rather than prejudice; or it may throw light on the history of the language, and suggest the forces that have led to the prestige of linguistic forms. Conclusions of all these types can be drawn from an examination of the occurrence or loss of initial /h-/ before semivowels in the states of the Atlantic seaboard, as recorded in the field records for the Linguistic Atlas of the United States and Canada.