The major portion of this article is devoted to a criticism of Twaddell's monograph ‘On Defining the Phoneme.‘ The criticism is adverse throughout. Still, if the result of negating a negative is positive, it may be justifiable to presume that if this is a contribution at all, it is of a constructive character. Some negative assertions are made in said monograph concerning the results thus far achieved in the physical analysis of speech sounds. If we do not misconstrue those assertions, they are at variance with the literature on the subject. We wish to present the evidence.

The discussion which follows might perhaps better be left to shift for itself, but a brief statement may be risked in defense, or in explanation, of a feature which will possibly by some readers be regarded as a fault—namely the length of the discussion. This feature can be charged principally, I believe, to two things: the purpose of the investigation, and the method

[The logicians of the Vienna Circle have independently reached the conclusion of physicalism: any scientifically meaningful statement reports a movement in space and time. This confirms the conclusion of A. P. Weiss and other American workers: the universe of science is a physical universe. This conclusion implies that statements about ‘ideas’ are to be translated into statements about speech-forms.]

Spenser's grammar, in spite of its obvious archaisms and occasional eccentricities of inflection and syntax, conforms in general to the standard and usage of Elizabethan writers. Spenser differs grammatically frorn his contemporaries only in his emphasis upon certain elements of style and in the direction in which his poetic temperament and literary principles took him; in all its essential aspects, however, his grammar is as representative of the 16th century as Shakespeare's is. Differences between the grammar of Spenser and that of Shakespeare are for the most part unimportant. Compared with Shakespeare, Spenser is careful of his grammar; he rarely makes errors of case and agreement. His grammar is more formal, more learned, more self-conscious than Shakespeare's. Shakespeare's free and unregulated grammar more nearly approaches that of ordinary spoken English: Shakespeare was not constrained or influenced by any theories of diction; he had no linguistic program to carry out, no mission to perform, no mannerisms to indulge in. Spenser's form and syntax, however, are conditioned by his theories of diction, by his eagerness to enrich the language of poetry, by the limitations of his subject matter, and by the requirements of his stanza and rime scheme.

The studies of Gilliéron and his followers in linguistic geography have made us familiar with the doctrine that two words of different origin which become homonyms by regular sound-changes may interfere with one another to such an extent that one is ultimately excluded from the vocabulary of a given dialect. The lack of a linguistic atlas of England has prevented the application of this principle, and likewise of others expounded by Gilliéron, to English dialects. For it was the comparison of maps which enabled Gilliéron and Roques to show, for example, that only in those French dialects where mulgere would have become homonymous with molere, is it replaced by other words for ‘milk’, such as tirer, traire, ajuster, aria, blechi, etc. Although neat demonstration of this sort is seldom possible in English, conclusions about the conflict of homonyms may sometimes be reached, in the absence of an atlas, by more roundabout methods. Enough information for reasonable deductions may be obtained from the combined materials of the English Dialect Dictionary, which gives the distribution of a large part of the vocabulary in modern dialects, and the New English Dictionary, which gives a detailed semantic history and the approximate time of a word's disappearance from the language.

One of the questions concerning the reconstruction of IE forms which needs thorough reconsideration because of the new evidence from Hittite and Tocharian is the extent of the development of middle forms in the parent language; more particularly, whether IE possessed both of those sets of endings which are conventionally called ‘primary’ and ‘secondary’, and even special endings for the perfect, or whether there was only one, as in Latin and Gothic. As long as Aryan and Greek were the only language groups which had well developed middle paradigms, at least without the r-element, it was natural to base on them all conclusions concerning the IE status, and to assume that the parent language also had sets of both primary and secondary endings, but that Italic and Keltic were the innovators in showing no probable trace of the former. However, now that both Hittite and Tocharian are seen to have had complete middle paradigms with and without r, and these paradigms agree not with Aryan and Greek, but with the other IE languages in showing only one set of endings, the weight of evidence has shifted to the other side. It is therefore necessary to re-examine the various middle paradigms to determine which features can be traced to a common source, and whether those which are probably not IE can be plausibly explained as originating in the individual languages.

[The work of Renaissance scholars on language, especially in Italy, deserves more attention than it has usually received. Certain sixteenth-century writers (Tolomei, Castelvetro, Scaliger, and others) anticipated nineteenth-century developments: firstly, in regarding language as a social phenomenon; secondly, in recognizing change as an essential element of linguistic history; and thirdly, in a more scientific approach, particularly in regard to phonetic law.]

No satisfactory etymology has been given of Greek ἀτύζομαι ‘to be distraught (from fear), amazed, bewildered’, also 'to be distraught (with grief)' and, with accusative, 'to be amazed at (a thing)'. This verb is generally used in the passive voice; its rare active, ἀτύζω 'to strike with terror', may be considered an informal causative of the basic intransitive ‘to be distraught’, though the latter is naturally, from the strictly formal standpoint, merely a passive of the active. A derivative adjective is ἀτυζηλός ‘terrifying’. The future of the active, ἀτύξω, and the passive aorist participle, ἀτυχθϵίς, show clearly that ἀτύζομαι is a -yo- derivative from a basic *ἀτυγ-. There is no such Greek base quotable from forms lying outside the verb paradigm, apart from the secondary form ἀτυζηλόs. ἀτύζομαι may also conceivably be a simple thematic denominative from an archaic i-stem *atugi-: *atugy-o-mai, but there is no such stem available in Greek.

Linguistic science is still based on the formulation that phonetic change is regular and invariable, provided the conditions in which the sounds stand are the same. The moderate chorus of dissent in recent years has come largely from the dialect-geographers, who have found variations which to them indicate the invalidity of the formulation; but in reality their findings demonstrate rather that the differences of environmental conditions are unexpectedly complex in character: dialectal areas are unexpectedly small, social dialects are found to exist, dialect mixtures are omnipresent, especially where the movement of population is free.

The theory of phonemes has received a great deal of attention from linguists within the last few years, and the applications of the theory to descriptive linguistics have become reasonably clear, though there still remain problems in need of further study. However, there has been less interest in the theory on the part of students of the history of language, many of whom practically ignore the theory altogether; or assume, on the other hand, that the sound changes which we can prove to have taken place in the past were practically coextensive with phonemic shifts. It seems, therefore, worth while to attempt a tentative examination and classification of the relations between sound change and phonemic change in the hope that the applications of the theory of phonemes to the historic study of language may be to some extent clarified.

Friedrich and Goetze have recently shown that Hittite ša-ak-nu-(wa-)an-za means ‘ritually unclean’. The word is used of persons who engage in ritual acts or enter a temple when they are ritually impure. A distinction is sometimes drawn between ‘pure tables’ (GIŠBANŠUR. ḪI.A pár-ku-i-ya-aš) and ‘impure tables’ (GISBANŠUR.ḪI.A ša-ak-nu-wa-an-da-aš), both of which are used in the ritual.

[In Venetic inscription no. 249, Ethnographic Museum of Vienna, the r in the second line goes with tola in the first and tolar is a verb, third singular present middle, and not a noun. It belongs with OIr. canar and has an indicative and personal meaning ‘he dedicates’. canar means in the Old Irish glosses ‘is sung’, not ‘there shall be singing’. The comparison of Old Irish, Latin, Oscan (fufans), Venetic, Greek (ἔτλᾱν), Lithuanian, Old Ch. Slav., and perhaps Tocharian, indicates that this ā-formation in PIE was essentially indicative. Its ‘injunctive’ value is simply a modal use of the indicative.]

As far back as 1922, in an article which in many respects remains the most significant contribution yet made to the larger understanding of the status of Tocharian, Eduard Hermann pointed out the importance of Tibetan because of its far-reaching morphological influence on Tocharian. Little attention, however, seems to have been paid to this aspect of the Tocharian problem, yet it is safe to say that one can not take a step in interpreting the forms of Tocharian, syntax and morphology both, unless one bears in mind the moulding influence exerted on Tocharian, both A and B, by this non-Indo-European language. In brief, Tocharian is a Tibetanized Indo-European idiom. Once this is understood, the much discussed problem of whether both A and B were spoken in Chinese Turkestan or B (‘Kuchean‘) alone, the A manuscripts appearing in the eastern part of the area merely because they were brought there by speakers of B from a real or supposed western home in Tokharestan (Bactria), falls by the wayside as an utterly pointless question. If dialect A has specifically Tibetan features in its structure—and it has—how could these have developed except through actual contact between speakers of Tocharian A and speakers of Tibetan, and where could such contact be had except in Chinese Turkestan itself, at a point not far removed from linguistically Tibetan territory?

[Discussion of ta, takku, tamais, tan, tayukas, takiya, also (in connection with takiya) nakkus.]

The purpose of the present paper is not to advance yet another etymology for Greek Φοῖνιξ ‘Phoenician’, but rather to reduce the number of acceptable suggestions with the aid of new evidence from an independent quarter. Although the material to be considered is predominantly linguistic, archaeological sources will also be brought into discussion.