It is generally taken for granted that the present-day pronunciation of Arm. p' t' k' represents the original status: Meillet (Esquisse 23) simply states this view without discussion: ‘Il y avait trois séries d'occlusives ... chacune d'elles existant sous forme de sourde non aspirée, de sourde aspirée (c'est-à-dire où l'explosion était suivie d'un souffle) et de sonore'. In spite of the general agreement upon this matter, it seems worth while to take it up again and test the common view of p' t' k', since this view appears to be based only on the present-day situation.

In the present state of our techniques we may assume that we know how to isolate morphemes properly—that is, unequivocally and without unaccountable residue. It is not so certain that we know how to isolate words, and hence how to separate morphology from syntax. Given a viable definition of the word for some particular language, we come finally to the problem of assigning words properly to relevant syntactic units such as phrases, clauses, and sentences. To illustrate one possible technique of discovering the syntactic framework of a language, I shall analyze a short text of Ponapean, a Micronesian language of the Eastern Caroline Islands.

The village of Tetelcingo is six kilometers north of Cuautla, Morelos, on the highway to Mexico City.

My wife and children and I lived in Tetelcingo for the better part of the years from 1940 to 47 and during several short periods since. We were shown no little kindness by the inhabitants of this village and learned from them their mother tongue—Māsiewalli.

The student who turns to our many excellent handbooks for information on the stops and spirants of Proto-Germanic finds that they agree completely in some respects, but disagree just as completely in others. All of them assume for Proto-Germanic a set of voiceless spirants and a set of voiceless stops: the spirants /f þ h/ as in English fish, thin, horn, derived from PIE /p t k/ as in Latin piscis, tenuis, cornu; and the stops /p t k/ as in English lip, two, kin, derived from PIE /b d g/ as in Latin labium, duo, genus. All of them also assume that there were some further sounds derived from PIE /bh dh gh/ and, by Verner's Law, from /p t k/, and that these sounds were voiced; but there is no agreement whatever as to whether these sounds were stops, or spirants, or both.

There have been a number of successful attempts, such as Kent's, to explain variations in scansion or in the development of Vulgar Latin forms in terms of variations in syllable division. It is also rather commonly accepted that Latin had some means of marking word boundaries—that is to say, had a juncture phoneme. There has not been (to my knowledge at least) any attempt to put syllable division and juncture together in a systematic fashion, to see what each can tell us about the other and what both can tell us about the phonemic structure of the language. Such a synthesis is the task here attempted. It is the contention of this paper that when the two are put together, three situation types emerge: normal syllable division, normal syllable division accompanied by juncture, and syllable division of a type that requires juncture.

The verbal reduplicating prefix. If the verbal stem in Gothic contained an initial cluster of two consonants, only the initial consonant was transferred to the prefix, except for the indivisible clusters sk and st (no examples of sp are recorded). The few cases of these two types of transference preserved in Wulfila's Gothic are: type 1 fraisan ‘to attempt’: faífrais, gretan ‘to weep’: gaígrot, flokan ‘to lament’: faíflok, slepan ‘to sleep’: saíslep; type 2 skaidan ‘to divide’: skaískaiþ, ga-staldan ‘to acquire: ga-staístald. The digraph ƕ must have also represented two indivisible phonemes hw (and will here be so written for convenience), since hw, not h alone, was transferred to the prefix: hwopan ‘to boast': hwaíhwop. The indivisibility of the clusters sk and st may have been due to the fact that the IE stops *k and *t after s were not shifted, and hence could not be divorced from the preceding s. But this was not the case with the initial clusters fr, gr, fl, sl. To explain the loss of the second phonemes in the prefix we are justified in assuming that in pre-Gothic (i.e. in Proto-Germanic) these clusters were transferred to the prefix, but that the second phonemes r, l were later lost through dissimilation: *fraífrais > faífrais, etc. This loss brought the form of the prefix into the pattern of the more frequent type of reduplication with a single initial consonant; thus, faífrais like faífalþ (from falþan ‘to fold'), saíslep like saíso (from saian ‘to sow'), etc. One may object to this explanation on the ground that analogy alone could have produced the same result, since the great majority of reduplicating verbs in Gothic had a single initial consonant. Perhaps the safest assumption is that the leveling was brought about by a combination of phonetic change and analogy.

One of the fundamental changes from Proto-Indo-European to the Germanic dialects is that from a verbal system based on aspect to a system based on tense. By the time of our Germanic records the change was accomplished; aspect to be sure was still conveyed in the Germanic languages, but through morphological processes that were not consistently applied to all verbs, such as the ga- prefix. Tense, however, was a category marked in every verb. Moreover, unlike the PIE situation, where the imperfective (present) was sometimes marked to distinguish it from other aspects—a situation maintained in Gk. tithēmi ‘I place’ beside the unmodified stem form of the aorist éthēka T placed'—in Germanic only the preterit was marked, with rare exceptions like Go. standan beside stōþ, bidjan beside baþ. Accordingly the Gmc. verbal system differed from the IE system in the grammatical meaning which it characteristically expressed, and in its method of marking that meaning.

In the Jaina Māhārāṣṭrī tales excerpted by the late Hermann Jacobi from the monk Devendra's commentary on the Jain canonical work Uttarādhyayana Sūtra, and published by him in his Ausgewählte Erzählungen in Mâhârâshṭrî (Leipzig, 1886), there is a puzzling stanza in the story of King Bambhadatta (Sanskrit Brahmadatta), for which I wish to offer an interpretation. This story is from the commentary on Chapter 13 of the Uttarādhyayana Sūtra, which is entitled Citta and Saṃbhüya (Sanskrit Citra and Sambhūta), the names of two Jain monks. One of these, Saṃbhūya, has been reborn as the universal emperor Bambhadatta, and the other is again a Jain monk; the chapter consists of a conversation between the two. The context, however, is the story of the two in a previous existence when they were the twin brothers Citta and Sambhūya.

Ever since it was discovered that Hittite, the language in which the bulk of the cuneiform tablets from Boğazköy is composed, is an Indo-European language or a language related to Indo-European, historians and linguists alike have been interested in the question as to the time when these people entered Asia Minor. The tablets were mostly written between 1400 and 1200 b.c., but the history of the Hittite state can be traced back to about 1900 b.c.; at that time, at the latest, Hittites must have been present in Anatolia. The problem of their origin has been complicated by the fact that the Hittite language is not isolated in the peninsula, but forms part of an ‘Anatolian’ group of languages to which Luwian and Palaic—likewise attested in the Boğazköy material—and furthermore the language of the inscriptions written in Hittite hieroglyphs also belong. The latter can at present be traced back to the middle of the 2nd millennium but for the most part cover the early centuries of the 1st millennium.

This is a phonemic analysis of the Sanskritized literary Tamil spoken in Madras by those whose vernacular is Tamil and whose fluent knowledge of Sanskrit is due to their heritage and upbringing. This variety of the language is distinct, on the one hand, from the several geographically determined dialects and, on the other, from Taṉittamiḻ, ‘pure Tamil’, which, free (at least in intention) of all Indo-Aryan influences, has records possibly two thousand years old and is in use today by certain people in the Tamil area. Sanskritized Tamil is a cultivated, literary language subject to strict canons of grammar and style, but it is a living speech and not an artificial medium used only by scholars.

It has been generally held that Gothic possessed a formal aspectual system which was very much like the system in OCS. In each language a simple imperfective is thought to have become perfective by the addition of a preverb—in Gothic especially by the addition of ga-, which in most instances did not change the meaning but merely the aspect of the basic verb (PIAG 103). And in each language every perfective verb, simple or derived, had future meaning in the present tense; though in Gothic, because there was no equivalent to the OCS iterative, a perfective verb also had present meaning. This tense ambivalence of the Gothic perfective present has been referred to as a flaw in an otherwise clear-cut aspectual system.

[Recent experiments with synthetic speech have shown that the perception of voiceless stops depends in large measure on the context, not on the characteristics of the consonants alone. The same phenomenon was investigated in actual speech by recording voiceless stops on tape before certain vowels, cutting them away, and splicing them back before other vowels. The change in context caused certain instances of [k] to be perceived as [p] or [t]. The experiment shows that in actual speech the context of an initial voiceless stop is an important factor in its perception; the results agree strikingly with those obtained for synthetic speech. It is further suggested that aspiration ought properly to be regarded as part of the following vowel rather than of the preceding consonant.]

The present article is intended to display a procedure of morphemic analysis that makes maximum use of morph-segmentation, and classifies morphs into morphemes in terms of recurrent partials within a fixed frame of reference. The frame of reference consists of a text in which substitutions are made at various points and the resulting other changes are observed. Thus, in English we find one boy, one girl, one man, but two boys, two girls, two men. From this one separates out the partials one, boy, girl, man, two, -s, and replacement of a by e (this example is deliberately put in orthographic terms); -s and the replacement of a by e are then grouped together in a class of morphs, a morpheme. We can now give this morpheme a label, if we wish, such as plural, but we could just as well call it something else; the referential meaning (meta-linguistic or ‘real’ meaning) does not concern us in the analysis, but there is no need to ignore it in choosing our labels.