[The paper outlines a general theory of grammatical categories. These fall into two main types, descriptive and taxonomic. Descriptive categories are either specific or generic. There are three kinds of specific categories: overt (phenotype), covert (cryptotype), and isosemantic; each of these is subdivided into selective categories and modulus categories. All terms in this classification are defined and illustrated.]

[In the Daiva Inscription of Xerxes, artā- is an instrumental coordinate with the preceding acc. Auramazdām; barazamanaiya is to be normalized brazmaniya on the evidence of the Elamite transcription, and is nom. sg. masc. ‘prayerful, reverent’ = Skt. brahmaṇyá-ḥ ‘religious’. The clause means ‘worship Ahuramazda and Arta reverent(ly)‘. Previous views are listed with a critique.]

This paper attempts to generalize the term MORPHEME SO as to apply not only to sequences of successive phonemes but also to broken sequences. In so doing, it offers a method of expressing one of the possible relations between morphemes as previously understood. The relation in question is that which obtains between two or more morphemes that always occur together (in a given environment). The essence of the method is that any two or more continuous morphemes which always occur together shall be considered to constitute together a single new morpheme. Since this relation between continuous morphemes is a type of grammatical agreement, the method here proposed obviates the necessity of separately treating this type of agreement.

In the fifth book of the Iliad is a tangle of textual and linguistic problems. The poet, after telling how Diomedes stabbed Aphrodite's arm near the hand, finishes (if we disregard for a moment lines 340–2, which will turn out to be an interpolation) his description of the encounter (339):

The intensive-durative prefix ∗ande- which appears in Ir. andfocul, annocul ‘habitual saying’ (cf. G Anderoudus ‘sehr rot’, Pedersen, VKG 2.10) can also be identified in the following Irish words: an[d]-ainmne ‘perseverance’, LB 261b26 (= ACL 3.320.17, where 23 N 10 reads ainmnet against the metre: the text has been assigned to the 9th century by Strachan, Trans. Phil. Soc. 1894.516); and-glondas ‘habitual cruelty’ (A1 2 168.5 = O'Dav. 135); and-locht ‘habitual fault’; and-lonn ‘condiment’; Andobor ‘the river Anner’; andord ‘musical strain’; and-ró ‘trouble’.

The name Vištāspa-, which the Greeks represented by Hystaspes, belongs to a number of personages in Iranian history: the East Iranian ruler who was Zoroaster's first royal convert, the father of Darius the Great, and several minor princely descendents of Darius. That Zoroaster's patron was identical with Darius's father is most unlikely, though it has been and is still maintained by some scholars; but that problem lies outside the field of the present discussion. Only so far does it at the moment interest me, that the name Vištāspa appears to have gained prestige from the fact that it was the name of Zoroaster's patron, and therefore enjoyed a vogue in the line of Achaemenian princes.

[Turfan Median kēč ‘somebody’ is probably *kahyačit, and Persian čīz ‘something’, *čahyačit; Skt. kím ‘what’ is a Middle Indic form; the IE interrogative was probably *ku̯os ‘who’ (stem *ku̯o-) and *ku̯id ‘what’ (stem *ku̯i-).]

The parts of the body in Semitic were dealt with in two important studies: F. Dietrich, Ueber die Gliedernamen im Semitischen, in his Abhandlungen für semitische Sprachforschung 101–258 (Leipzig, 1844); and H. Holma, Die Namen der Körperteile im Assyrisch-Babylonischen (Leipzig, 1911). Dietrich has a large collection of the names of the parts of the body in the Semitic languages known at the time; but his etymological explanations can be disregarded in view of the tendency, characteristic for that epoch, to explain the meaning of the words by two radicals of the root. Holma's book is a valuable collection, with sound etymological explanation. It would be desirable for Akkadian scholars to complete this collection systematically from the Akkadian texts discovered since 1911.

[Just as homonyms may conflict, so historically independent words through a series of gradual shifts in meaning tend to develop irreconcilable senses; obsolescence is thus closely connected with semantic change. Illustrations from the New English Dictionary and the Linguistic Atlas of New England.]

[The paper delimits the group of pure-relational suffixes by removing some morphemes that have hitherto generally been included in this group. An important morphological criterion of these suffixes, the oblique stem, is examined, and as a result of this examination pure-relational suffixes are divided into two classes: case suffixes and suffixed postpositions, the latter forming the transition from the case suffixes to the free postpositions.]

[This is intended as the first of a series of detailed discussions of the so-called Tocharian palatalization. All that is attempted here is an examination of the actual appearance of the sound change, to determine what are the original and what are the secondary consonants.]

[On the basis of a complete record of the word and of its derivatives in ancient and modern dialects, an attempt is made to link OSp. (a)quexar ‘to press, to squeeze, to crush’ with OPtg. queixo, OSp. quex-ada, Cat. queix ‘jaw’ < capsu; the further semantic development shows no anomalies. In the course of the discussion of the suggested etymology, reference is made to questions of method in Romance word-studies; to contrasts between the Old Spanish and the Old Portuguese vocabularies and their subsequent interpenetration; to synonyms of (a)quexar and of quexarse and their relation to the neologism; to various suffixes involved (-ūra, -ūmen, -ōsus, -itāre, -icāre); to individual words, including Mozar. quexdar < quaesitare, Sp. quejigo, Sp. requejo, Sp. caja, and its cognates.

[A system of four pitch phonemes enables us to differentiate all the phonemically (i.e. perceptually and meaningfully) different pitch contours of English that have come to the writer's notice. If new contours should come to light, they could probably be treated without the introduction of any additional pitch phonemes. The system applies to the problem of English pitches the methods already applied to segmental phonemes as well as to stresses and junctures.]

The IE origin of Chin. mi ‘honey’, ancient and archaic Chin. miět, first pointed out by E. Polivanov, is by now generally accepted. The question from which IE language the word was borrowed, however, is still unanswered. During the latter half of the first millennium B.C., the Chinese had contact not only with the Tokharians, but also with other IE peoples. In a letter to the present writer, dated August 29, 1938, E. Sapir wrote, ‘I suspect strongly that at least two distinct IE languages appeared in Chinese Turkestan and West China at a quite early date and that these two languages belonged to distinct branches of IE.‘ If the two words discussed in this paper are IE loan words in Chinese, as I believe they are, they might be of some help in clarifying the problem of the IE languages spoken in the western Chinese borderlands.

[A reconsideration of the available orthographic and orthoepistic evidence and a new explanation of how k and g disappeared from the initial groups kn and gn.]

1. In the better known South Dravidian languages, the verb system is such that two principal parts, or two stems, suffice as a basis on which the whole system may be built. The second stem is in general derivable from the first by the addition of a suffix *-i- or a suffix containing a dental stop (sometimes disguised by morphophonemic alternations), and is the basis for the past tense alone, or, in some of the languages, for present-future and past tenses, or for other combinations of forms. On the first stem are built all the remaining positive forms and the negative forms. Vocalic alternation within the stems, either of quality or of quantity, is very rare, but in the extremely few verbs where it occurs tends to be persistent throughout the history of all the languages; e.g. the verb ‘see’ has stems *ka·ṇ-, *kaṇṭ- in Tamil, Malayalam, Kannada, Coorg, and Toda, while Kota shows variation between ka#x0387;ṇ- and kaṇ- in the first stem, with leveling begun but not yet completed, or perhaps completed but with the disturbing long vowel reappearing by loan from other languages of the group. Likewise, any deviation from the suffixes given above as forming the second stem is rare.